From 62b51d9e6535fbc0b539f9d2852022ee80cda8f2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?St=C3=A9phane=20Gigandet?= <stephane@openfoodfacts.org>
Date: Thu, 28 Nov 2024 14:55:00 +0100
Subject: [PATCH] refactor: ingredients preparsing unit tests (#11063)

This PR is to change ingredients_parsing.t into ingredients_preparsing.t
The tests are exactly the same, but instead of having the expected
results listed in the .t file, we use JSON files like for other tests.

This is because I will change the output of preparsing for many tests,
and it will make the test diffs of the upcoming PR easier to see.
---
 .../ingredients_preparsing/1.json             |    6 +
 .../ingredients_preparsing/10.json            |    6 +
 .../ingredients_preparsing/100.json           |    6 +
 .../ingredients_preparsing/101.json           |    6 +
 .../ingredients_preparsing/102.json           |    6 +
 .../ingredients_preparsing/103.json           |    6 +
 .../ingredients_preparsing/104.json           |    6 +
 .../ingredients_preparsing/105.json           |    6 +
 .../ingredients_preparsing/106.json           |    6 +
 .../ingredients_preparsing/107.json           |    6 +
 .../ingredients_preparsing/108.json           |    6 +
 .../ingredients_preparsing/109.json           |    7 +
 .../ingredients_preparsing/11.json            |    6 +
 .../ingredients_preparsing/110.json           |    6 +
 .../ingredients_preparsing/111.json           |    6 +
 .../ingredients_preparsing/112.json           |    6 +
 .../ingredients_preparsing/113.json           |    6 +
 .../ingredients_preparsing/114.json           |    6 +
 .../ingredients_preparsing/115.json           |    6 +
 .../ingredients_preparsing/116.json           |    6 +
 .../ingredients_preparsing/117.json           |    6 +
 .../ingredients_preparsing/118.json           |    6 +
 .../ingredients_preparsing/119.json           |    6 +
 .../ingredients_preparsing/12.json            |    6 +
 .../ingredients_preparsing/120.json           |    6 +
 .../ingredients_preparsing/121.json           |    6 +
 .../ingredients_preparsing/122.json           |    6 +
 .../ingredients_preparsing/123.json           |    6 +
 .../ingredients_preparsing/124.json           |    6 +
 .../ingredients_preparsing/125.json           |    6 +
 .../ingredients_preparsing/126.json           |    6 +
 .../ingredients_preparsing/127.json           |    6 +
 .../ingredients_preparsing/128.json           |    6 +
 .../ingredients_preparsing/129.json           |    6 +
 .../ingredients_preparsing/13.json            |    6 +
 .../ingredients_preparsing/130.json           |    6 +
 .../ingredients_preparsing/131.json           |    6 +
 .../ingredients_preparsing/132.json           |    6 +
 .../ingredients_preparsing/133.json           |    6 +
 .../ingredients_preparsing/134.json           |    6 +
 .../ingredients_preparsing/135.json           |    6 +
 .../ingredients_preparsing/136.json           |    6 +
 .../ingredients_preparsing/137.json           |    7 +
 .../ingredients_preparsing/138.json           |    6 +
 .../ingredients_preparsing/139.json           |    6 +
 .../ingredients_preparsing/14.json            |    6 +
 .../ingredients_preparsing/140.json           |    6 +
 .../ingredients_preparsing/141.json           |    6 +
 .../ingredients_preparsing/142.json           |    6 +
 .../ingredients_preparsing/143.json           |    6 +
 .../ingredients_preparsing/144.json           |    6 +
 .../ingredients_preparsing/145.json           |    6 +
 .../ingredients_preparsing/146.json           |    6 +
 .../ingredients_preparsing/147.json           |    6 +
 .../ingredients_preparsing/148.json           |    6 +
 .../ingredients_preparsing/149.json           |    6 +
 .../ingredients_preparsing/15.json            |    6 +
 .../ingredients_preparsing/150.json           |    6 +
 .../ingredients_preparsing/151.json           |    6 +
 .../ingredients_preparsing/152.json           |    6 +
 .../ingredients_preparsing/153.json           |    6 +
 .../ingredients_preparsing/154.json           |    6 +
 .../ingredients_preparsing/155.json           |    6 +
 .../ingredients_preparsing/156.json           |    6 +
 .../ingredients_preparsing/157.json           |    6 +
 .../ingredients_preparsing/158.json           |    6 +
 .../ingredients_preparsing/159.json           |    6 +
 .../ingredients_preparsing/16.json            |    6 +
 .../ingredients_preparsing/160.json           |    6 +
 .../ingredients_preparsing/161.json           |    6 +
 .../ingredients_preparsing/162.json           |    6 +
 .../ingredients_preparsing/163.json           |    6 +
 .../ingredients_preparsing/164.json           |    7 +
 .../ingredients_preparsing/165.json           |    6 +
 .../ingredients_preparsing/166.json           |    6 +
 .../ingredients_preparsing/167.json           |    6 +
 .../ingredients_preparsing/168.json           |    6 +
 .../ingredients_preparsing/169.json           |    6 +
 .../ingredients_preparsing/17.json            |    6 +
 .../ingredients_preparsing/170.json           |    6 +
 .../ingredients_preparsing/171.json           |    6 +
 .../ingredients_preparsing/172.json           |    6 +
 .../ingredients_preparsing/173.json           |    6 +
 .../ingredients_preparsing/174.json           |    6 +
 .../ingredients_preparsing/175.json           |    6 +
 .../ingredients_preparsing/176.json           |    7 +
 .../ingredients_preparsing/177.json           |    6 +
 .../ingredients_preparsing/178.json           |    6 +
 .../ingredients_preparsing/179.json           |    6 +
 .../ingredients_preparsing/18.json            |    6 +
 .../ingredients_preparsing/180.json           |    6 +
 .../ingredients_preparsing/181.json           |    6 +
 .../ingredients_preparsing/182.json           |    6 +
 .../ingredients_preparsing/183.json           |    6 +
 .../ingredients_preparsing/184.json           |    6 +
 .../ingredients_preparsing/185.json           |    6 +
 .../ingredients_preparsing/186.json           |    6 +
 .../ingredients_preparsing/187.json           |    6 +
 .../ingredients_preparsing/188.json           |    6 +
 .../ingredients_preparsing/189.json           |    6 +
 .../ingredients_preparsing/19.json            |    6 +
 .../ingredients_preparsing/190.json           |    6 +
 .../ingredients_preparsing/191.json           |    6 +
 .../ingredients_preparsing/192.json           |    6 +
 .../ingredients_preparsing/193.json           |    6 +
 .../ingredients_preparsing/194.json           |    6 +
 .../ingredients_preparsing/195.json           |    6 +
 .../ingredients_preparsing/196.json           |    6 +
 .../ingredients_preparsing/197.json           |    6 +
 .../ingredients_preparsing/198.json           |    6 +
 .../ingredients_preparsing/199.json           |    6 +
 .../ingredients_preparsing/2.json             |    6 +
 .../ingredients_preparsing/20.json            |    6 +
 .../ingredients_preparsing/200.json           |    6 +
 .../ingredients_preparsing/201.json           |    6 +
 .../ingredients_preparsing/202.json           |    6 +
 .../ingredients_preparsing/203.json           |    6 +
 .../ingredients_preparsing/204.json           |    6 +
 .../ingredients_preparsing/205.json           |    6 +
 .../ingredients_preparsing/206.json           |    6 +
 .../ingredients_preparsing/207.json           |    6 +
 .../ingredients_preparsing/208.json           |    6 +
 .../ingredients_preparsing/209.json           |    6 +
 .../ingredients_preparsing/21.json            |    6 +
 .../ingredients_preparsing/210.json           |    6 +
 .../ingredients_preparsing/211.json           |    6 +
 .../ingredients_preparsing/212.json           |    6 +
 .../ingredients_preparsing/213.json           |    6 +
 .../ingredients_preparsing/214.json           |    6 +
 .../ingredients_preparsing/215.json           |    6 +
 .../ingredients_preparsing/216.json           |    6 +
 .../ingredients_preparsing/217.json           |    6 +
 .../ingredients_preparsing/218.json           |    6 +
 .../ingredients_preparsing/219.json           |    6 +
 .../ingredients_preparsing/22.json            |    6 +
 .../ingredients_preparsing/220.json           |    6 +
 .../ingredients_preparsing/221.json           |    6 +
 .../ingredients_preparsing/222.json           |    6 +
 .../ingredients_preparsing/223.json           |    6 +
 .../ingredients_preparsing/224.json           |    6 +
 .../ingredients_preparsing/23.json            |    6 +
 .../ingredients_preparsing/24.json            |    6 +
 .../ingredients_preparsing/25.json            |    6 +
 .../ingredients_preparsing/26.json            |    6 +
 .../ingredients_preparsing/27.json            |    6 +
 .../ingredients_preparsing/28.json            |    6 +
 .../ingredients_preparsing/29.json            |    6 +
 .../ingredients_preparsing/3.json             |    6 +
 .../ingredients_preparsing/30.json            |    6 +
 .../ingredients_preparsing/31.json            |    6 +
 .../ingredients_preparsing/32.json            |    6 +
 .../ingredients_preparsing/33.json            |    6 +
 .../ingredients_preparsing/34.json            |    6 +
 .../ingredients_preparsing/35.json            |    6 +
 .../ingredients_preparsing/36.json            |    6 +
 .../ingredients_preparsing/37.json            |    6 +
 .../ingredients_preparsing/38.json            |    6 +
 .../ingredients_preparsing/39.json            |    6 +
 .../ingredients_preparsing/4.json             |    6 +
 .../ingredients_preparsing/40.json            |    6 +
 .../ingredients_preparsing/41.json            |    6 +
 .../ingredients_preparsing/42.json            |    6 +
 .../ingredients_preparsing/43.json            |    6 +
 .../ingredients_preparsing/44.json            |    6 +
 .../ingredients_preparsing/45.json            |    6 +
 .../ingredients_preparsing/46.json            |    6 +
 .../ingredients_preparsing/47.json            |    6 +
 .../ingredients_preparsing/48.json            |    6 +
 .../ingredients_preparsing/49.json            |    6 +
 .../ingredients_preparsing/5.json             |    6 +
 .../ingredients_preparsing/50.json            |    6 +
 .../ingredients_preparsing/51.json            |    6 +
 .../ingredients_preparsing/52.json            |    6 +
 .../ingredients_preparsing/53.json            |    6 +
 .../ingredients_preparsing/54.json            |    6 +
 .../ingredients_preparsing/55.json            |    6 +
 .../ingredients_preparsing/56.json            |    6 +
 .../ingredients_preparsing/57.json            |    6 +
 .../ingredients_preparsing/58.json            |    6 +
 .../ingredients_preparsing/59.json            |    6 +
 .../ingredients_preparsing/6.json             |    6 +
 .../ingredients_preparsing/60.json            |    6 +
 .../ingredients_preparsing/61.json            |    6 +
 .../ingredients_preparsing/62.json            |    6 +
 .../ingredients_preparsing/63.json            |    6 +
 .../ingredients_preparsing/64.json            |    6 +
 .../ingredients_preparsing/65.json            |    6 +
 .../ingredients_preparsing/66.json            |    6 +
 .../ingredients_preparsing/67.json            |    6 +
 .../ingredients_preparsing/68.json            |    6 +
 .../ingredients_preparsing/69.json            |    6 +
 .../ingredients_preparsing/7.json             |    6 +
 .../ingredients_preparsing/70.json            |    6 +
 .../ingredients_preparsing/71.json            |    6 +
 .../ingredients_preparsing/72.json            |    6 +
 .../ingredients_preparsing/73.json            |    6 +
 .../ingredients_preparsing/74.json            |    6 +
 .../ingredients_preparsing/75.json            |    6 +
 .../ingredients_preparsing/76.json            |    6 +
 .../ingredients_preparsing/77.json            |    6 +
 .../ingredients_preparsing/78.json            |    6 +
 .../ingredients_preparsing/79.json            |    6 +
 .../ingredients_preparsing/8.json             |    6 +
 .../ingredients_preparsing/80.json            |    6 +
 .../ingredients_preparsing/81.json            |    6 +
 .../ingredients_preparsing/82.json            |    6 +
 .../ingredients_preparsing/83.json            |    6 +
 .../ingredients_preparsing/84.json            |    6 +
 .../ingredients_preparsing/85.json            |    6 +
 .../ingredients_preparsing/86.json            |    6 +
 .../ingredients_preparsing/87.json            |    6 +
 .../ingredients_preparsing/88.json            |    6 +
 .../ingredients_preparsing/89.json            |    6 +
 .../ingredients_preparsing/9.json             |    6 +
 .../ingredients_preparsing/90.json            |    6 +
 .../ingredients_preparsing/91.json            |    6 +
 .../ingredients_preparsing/92.json            |    6 +
 .../ingredients_preparsing/93.json            |    6 +
 .../ingredients_preparsing/94.json            |    6 +
 .../ingredients_preparsing/95.json            |    6 +
 .../ingredients_preparsing/96.json            |    6 +
 .../ingredients_preparsing/97.json            |    6 +
 .../ingredients_preparsing/98.json            |    6 +
 .../ingredients_preparsing/99.json            |    6 +
 tests/unit/ingredients_parsing.t              |  675 ---------
 tests/unit/ingredients_preparsing.t           | 1217 +++++++++++++++++
 226 files changed, 2565 insertions(+), 675 deletions(-)
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/1.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/10.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/100.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/101.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/102.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/103.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/104.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/105.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/106.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/107.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/108.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/109.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/11.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/110.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/111.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/112.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/113.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/114.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/115.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/116.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/117.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/118.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/119.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/12.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/120.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/121.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/122.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/123.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/124.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/125.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/126.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/127.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/128.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/129.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/13.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/130.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/131.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/132.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/133.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/134.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/135.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/136.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/137.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/138.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/139.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/14.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/140.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/141.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/142.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/143.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/144.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/145.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/146.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/147.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/148.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/149.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/15.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/150.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/151.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/152.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/153.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/154.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/155.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/156.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/157.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/158.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/159.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/16.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/160.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/161.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/162.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/163.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/164.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/165.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/166.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/167.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/168.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/169.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/17.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/170.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/171.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/172.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/173.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/174.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/175.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/176.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/177.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/178.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/179.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/18.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/180.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/181.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/182.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/183.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/184.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/185.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/186.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/187.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/188.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/189.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/19.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/190.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/191.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/192.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/193.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/194.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/195.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/196.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/197.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/198.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/199.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/2.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/20.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/200.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/201.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/202.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/203.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/204.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/205.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/206.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/207.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/208.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/209.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/21.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/210.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/211.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/212.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/213.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/214.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/215.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/216.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/217.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/218.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/219.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/22.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/220.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/221.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/222.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/223.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/224.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/23.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/24.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/25.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/26.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/27.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/28.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/29.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/3.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/30.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/31.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/32.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/33.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/34.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/35.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/36.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/37.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/38.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/39.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/4.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/40.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/41.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/42.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/43.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/44.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/45.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/46.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/47.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/48.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/49.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/5.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/50.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/51.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/52.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/53.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/54.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/55.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/56.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/57.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/58.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/59.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/6.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/60.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/61.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/62.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/63.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/64.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/65.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/66.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/67.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/68.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/69.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/7.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/70.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/71.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/72.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/73.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/74.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/75.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/76.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/77.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/78.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/79.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/8.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/80.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/81.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/82.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/83.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/84.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/85.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/86.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/87.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/88.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/89.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/9.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/90.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/91.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/92.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/93.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/94.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/95.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/96.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/97.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/98.json
 create mode 100644 tests/unit/expected_test_results/ingredients_preparsing/99.json
 delete mode 100755 tests/unit/ingredients_parsing.t
 create mode 100755 tests/unit/ingredients_preparsing.t

diff --git a/tests/unit/expected_test_results/ingredients_preparsing/1.json b/tests/unit/expected_test_results/ingredients_preparsing/1.json
new file mode 100644
index 0000000000000..1c50458a03f1e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/1.json
@@ -0,0 +1,6 @@
+{
+   "id" : "1",
+   "ingredients_text" : "Sel marin, blé, lécithine de soja",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sel marin, blé, lécithine de soja"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/10.json b/tests/unit/expected_test_results/ingredients_preparsing/10.json
new file mode 100644
index 0000000000000..2c3cd00f9f6b7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/10.json
@@ -0,0 +1,6 @@
+{
+   "id" : "10",
+   "ingredients_text" : "Huiles végétales de palme et d'olive",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales d'olive"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/100.json b/tests/unit/expected_test_results/ingredients_preparsing/100.json
new file mode 100644
index 0000000000000..cb4192301de61
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/100.json
@@ -0,0 +1,6 @@
+{
+   "id" : "100",
+   "ingredients_text" : "huile de tournesol, cacao maigre en poudre 5.2%",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile de tournesol, cacao maigre en poudre 5.2%"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/101.json b/tests/unit/expected_test_results/ingredients_preparsing/101.json
new file mode 100644
index 0000000000000..39969c138dbec
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/101.json
@@ -0,0 +1,6 @@
+{
+   "id" : "101",
+   "ingredients_text" : "regulatory kwasowości: kwas cytrynowy i cytryniany sodu.",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "regulatory kwasowości: kwas cytrynowy i cytryniany sodu."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/102.json b/tests/unit/expected_test_results/ingredients_preparsing/102.json
new file mode 100644
index 0000000000000..4067ec0c13470
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/102.json
@@ -0,0 +1,6 @@
+{
+   "id" : "102",
+   "ingredients_text" : "Wasser, Kohlensäure, Farbstoff Zuckerkulör E 150d, Süßungsmittel Aspartam* und Acesulfam-K, Säuerungsmittel Phosphorsäure und Citronensäure, Säureregulator Natriumcitrat, Aroma Koffein, Aroma. enthält eine Phenylalaninquelle",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Wasser, Kohlensäure, Farbstoff : Zuckerkulör e150d, Süßungsmittel : Aspartam* und Acesulfam-K, Säuerungsmittel : Phosphorsäure und Citronensäure, Säureregulator Natriumcitrat, Aroma Koffein, Aroma. enthält eine Phenylalaninquelle"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/103.json b/tests/unit/expected_test_results/ingredients_preparsing/103.json
new file mode 100644
index 0000000000000..ce0aeb9e18e7c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/103.json
@@ -0,0 +1,6 @@
+{
+   "id" : "103",
+   "ingredients_text" : "Farbstoffe Betenrot, Paprikaextrakt, Kurkumin",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Farbstoffe : Betenrot, Paprikaextrakt, Kurkumin"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/104.json b/tests/unit/expected_test_results/ingredients_preparsing/104.json
new file mode 100644
index 0000000000000..f3cb9b049a12c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/104.json
@@ -0,0 +1,6 @@
+{
+   "id" : "104",
+   "ingredients_text" : "Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs, weiß und gelb",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8.5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs weiß und gelb"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/105.json b/tests/unit/expected_test_results/ingredients_preparsing/105.json
new file mode 100644
index 0000000000000..8e2b8d0636b36
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/105.json
@@ -0,0 +1,6 @@
+{
+   "id" : "105",
+   "ingredients_text" : "Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs (weiß und gelb)",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8.5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs weiß und gelb"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/106.json b/tests/unit/expected_test_results/ingredients_preparsing/106.json
new file mode 100644
index 0000000000000..11d9c296acddb
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/106.json
@@ -0,0 +1,6 @@
+{
+   "id" : "106",
+   "ingredients_text" : "graisse végétale bio (colza)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "graisse végétale bio de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/107.json b/tests/unit/expected_test_results/ingredients_preparsing/107.json
new file mode 100644
index 0000000000000..e22665e8ebc22
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/107.json
@@ -0,0 +1,6 @@
+{
+   "id" : "107",
+   "ingredients_text" : "huiles végétales* (huile de tournesol*, huile de colza*). *Ingrédients issus de l'agriculture biologique",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huiles végétales Bio (huile de tournesol Bio, huile de colza Bio )."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/108.json b/tests/unit/expected_test_results/ingredients_preparsing/108.json
new file mode 100644
index 0000000000000..e0831dc8ea94c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/108.json
@@ -0,0 +1,6 @@
+{
+   "id" : "108",
+   "ingredients_text" : "huile biologique (tournesol, olive)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile biologique de tournesol, huile biologique d'olive"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/109.json b/tests/unit/expected_test_results/ingredients_preparsing/109.json
new file mode 100644
index 0000000000000..e4657be746d99
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/109.json
@@ -0,0 +1,7 @@
+{
+   "comment" : "xyz: test an unrecognized oil -> do not change",
+   "id" : "109",
+   "ingredients_text" : "huile biologique (tournesol, xyz)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile biologique (tournesol, xyz)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/11.json b/tests/unit/expected_test_results/ingredients_preparsing/11.json
new file mode 100644
index 0000000000000..ce8c621ff33e2
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/11.json
@@ -0,0 +1,6 @@
+{
+   "id" : "11",
+   "ingredients_text" : "Huiles végétales de palme, de colza et de tournesol",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza, Huiles végétales de tournesol"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/110.json b/tests/unit/expected_test_results/ingredients_preparsing/110.json
new file mode 100644
index 0000000000000..e4c888eeba0d1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/110.json
@@ -0,0 +1,6 @@
+{
+   "id" : "110",
+   "ingredients_text" : "huiles biologiques (tournesol, olive)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huiles biologiques de tournesol, huiles biologiques d'olive"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/111.json b/tests/unit/expected_test_results/ingredients_preparsing/111.json
new file mode 100644
index 0000000000000..cfb13880c6510
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/111.json
@@ -0,0 +1,6 @@
+{
+   "id" : "111",
+   "ingredients_text" : "huiles (tournesol*, olive). * : bio",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huiles de tournesol Bio, huiles d'olive."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/112.json b/tests/unit/expected_test_results/ingredients_preparsing/112.json
new file mode 100644
index 0000000000000..b03e9d12ae74f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/112.json
@@ -0,0 +1,6 @@
+{
+   "id" : "112",
+   "ingredients_text" : "huiles* (tournesol*, olive vierge extra), sel marin. *issus de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huiles Bio de tournesol Bio, huiles Bio d'olive vierge extra, sel marin."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/113.json b/tests/unit/expected_test_results/ingredients_preparsing/113.json
new file mode 100644
index 0000000000000..61d8addc9fa4d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/113.json
@@ -0,0 +1,6 @@
+{
+   "id" : "113",
+   "ingredients_text" : "riz de Camargue (1), sel. (1): IGP : Indication Géographique Protégée.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "riz de Camargue IGP, sel."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/114.json b/tests/unit/expected_test_results/ingredients_preparsing/114.json
new file mode 100644
index 0000000000000..8333baa3d291c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/114.json
@@ -0,0 +1,6 @@
+{
+   "id" : "114",
+   "ingredients_text" : "cacao (1), sucre (2), beurre de cacao (1). (1) : Commerce équitable. (2) Issue de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "cacao Commerce équitable, sucre Bio, beurre de cacao Commerce équitable."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/115.json b/tests/unit/expected_test_results/ingredients_preparsing/115.json
new file mode 100644
index 0000000000000..5139dbd7ccf7e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/115.json
@@ -0,0 +1,6 @@
+{
+   "id" : "115",
+   "ingredients_text" : "Céréales 63,7% (BLE complet 50,5%*, semoule de maïs*), sucre*, sirop de BLE*, cacao maigre en poudre 3,9%*, cacao en poudre 1,7%*, sel, arôme naturel. *Ingrédients issus de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Céréales 63.7% (BLE complet 50.5% Bio, semoule de maïs Bio ), sucre Bio, sirop de BLE Bio, cacao maigre en poudre 3.9% Bio, cacao en poudre 1.7% Bio, sel, arôme naturel."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/116.json b/tests/unit/expected_test_results/ingredients_preparsing/116.json
new file mode 100644
index 0000000000000..6fc9bc3c8153a
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/116.json
@@ -0,0 +1,6 @@
+{
+   "id" : "116",
+   "ingredients_text" : "émulsifiant : mono - et diglycérides d'acides gras.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "émulsifiant : mono- et diglycérides d'acides gras."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/117.json b/tests/unit/expected_test_results/ingredients_preparsing/117.json
new file mode 100644
index 0000000000000..0bc6032e5fd88
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/117.json
@@ -0,0 +1,6 @@
+{
+   "id" : "117",
+   "ingredients_text" : "Sucre. Fabriqué dans un atelier qui utilise des fruits à coques.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sucre. Traces éventuelles : fruits à coques."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/118.json b/tests/unit/expected_test_results/ingredients_preparsing/118.json
new file mode 100644
index 0000000000000..5aa19d9a57203
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/118.json
@@ -0,0 +1,6 @@
+{
+   "id" : "118",
+   "ingredients_text" : "Sucre. Fabriqué dans un atelier utilisant des fruits à coques et du sésame.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sucre. Traces éventuelles : fruits à coques, Traces éventuelles : sésame."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/119.json b/tests/unit/expected_test_results/ingredients_preparsing/119.json
new file mode 100644
index 0000000000000..d433a9f937a53
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/119.json
@@ -0,0 +1,6 @@
+{
+   "id" : "119",
+   "ingredients_text" : "Sucre. Fabriqué dans un atelier qui manipule du lait, de la moutarde et du céleri.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sucre. Traces éventuelles : lait, Traces éventuelles : moutarde, Traces éventuelles : céleri."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/12.json b/tests/unit/expected_test_results/ingredients_preparsing/12.json
new file mode 100644
index 0000000000000..04596bc56aa52
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/12.json
@@ -0,0 +1,6 @@
+{
+   "id" : "12",
+   "ingredients_text" : "Huiles végétales de palme, de colza, de tournesol",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza, Huiles végétales de tournesol"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/120.json b/tests/unit/expected_test_results/ingredients_preparsing/120.json
new file mode 100644
index 0000000000000..6188ba1762669
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/120.json
@@ -0,0 +1,6 @@
+{
+   "id" : "120",
+   "ingredients_text" : "Sucre. Peut contenir des fruits à coques et du sésame.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sucre. Traces éventuelles : fruits à coques, Traces éventuelles : sésame."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/121.json b/tests/unit/expected_test_results/ingredients_preparsing/121.json
new file mode 100644
index 0000000000000..a3cc90e4d4dcf
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/121.json
@@ -0,0 +1,6 @@
+{
+   "id" : "121",
+   "ingredients_text" : "vegetable oil (coconut & rapeseed)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "coconut vegetable oil, rapeseed vegetable oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/122.json b/tests/unit/expected_test_results/ingredients_preparsing/122.json
new file mode 100644
index 0000000000000..a2c910f6340e9
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/122.json
@@ -0,0 +1,6 @@
+{
+   "id" : "122",
+   "ingredients_text" : "Masse de cacao°, Quinoa° (1,8%). °Produits issus de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Masse de cacao Bio, Quinoa Bio (1.8%)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/123.json b/tests/unit/expected_test_results/ingredients_preparsing/123.json
new file mode 100644
index 0000000000000..1cdecd4aee8ab
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/123.json
@@ -0,0 +1,6 @@
+{
+   "id" : "123",
+   "ingredients_text" : "Emulgator (Sojalecithine, Mono - und Diglyceride von Speisefettsäuren, Sorbitantristearat)",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Emulgator (Sojalecithine, mono- und Diglyceride von Speisefettsäuren, Sorbitantristearat)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/124.json b/tests/unit/expected_test_results/ingredients_preparsing/124.json
new file mode 100644
index 0000000000000..ef2b54be88acb
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/124.json
@@ -0,0 +1,6 @@
+{
+   "id" : "124",
+   "ingredients_text" : "Tomates* (20%). *Ingrédients Bio",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Tomates Bio (20%)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/125.json b/tests/unit/expected_test_results/ingredients_preparsing/125.json
new file mode 100644
index 0000000000000..6bcf2c2ea2761
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/125.json
@@ -0,0 +1,6 @@
+{
+   "id" : "125",
+   "ingredients_text" : "Tomates* (20%). *Ingrédients biologiques",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Tomates Bio (20%)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/126.json b/tests/unit/expected_test_results/ingredients_preparsing/126.json
new file mode 100644
index 0000000000000..d73d312c21dcb
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/126.json
@@ -0,0 +1,6 @@
+{
+   "id" : "126",
+   "ingredients_text" : "Chocolat. Contient du lait et des noisettes. Peut contenir du blé, du soja et des crustacés.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Chocolat. Substances ou produits provoquant des allergies ou intolérances : lait, Substances ou produits provoquant des allergies ou intolérances : noisettes. Traces éventuelles : blé, Traces éventuelles : soja, Traces éventuelles : crustacés."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/127.json b/tests/unit/expected_test_results/ingredients_preparsing/127.json
new file mode 100644
index 0000000000000..b09b6e6dd29af
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/127.json
@@ -0,0 +1,6 @@
+{
+   "id" : "127",
+   "ingredients_text" : "Chocolate. Contains milk, hazelnuts and other nuts. May contain celery and mustard.",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "Chocolate. Substances or products causing allergies or intolerances : milk, Substances or products causing allergies or intolerances : hazelnuts, Substances or products causing allergies or intolerances : other nuts. Traces : celery, Traces : mustard."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/128.json b/tests/unit/expected_test_results/ingredients_preparsing/128.json
new file mode 100644
index 0000000000000..c961e66cbc4b1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/128.json
@@ -0,0 +1,6 @@
+{
+   "id" : "128",
+   "ingredients_text" : "phosphates d'ammonium et de calcium, Phosphate d'aluminium et de sodium, diphosphate d'aluminium et de sodium",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "phosphates d'ammonium, phosphates de calcium, phosphate d'aluminium et de sodium, diphosphate d'aluminium et de sodium"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/129.json b/tests/unit/expected_test_results/ingredients_preparsing/129.json
new file mode 100644
index 0000000000000..285b9f212ce1c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/129.json
@@ -0,0 +1,6 @@
+{
+   "id" : "129",
+   "ingredients_text" : "Ingrédient(s) : lentilles vertes* - *issu(e)(s) de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Ingrédients : lentilles vertes Bio"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/13.json b/tests/unit/expected_test_results/ingredients_preparsing/13.json
new file mode 100644
index 0000000000000..21e4b9e340d95
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/13.json
@@ -0,0 +1,6 @@
+{
+   "id" : "13",
+   "ingredients_text" : "Huiles végétales de palme, de colza et d'olive en proportion variable",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza, Huiles végétales d'olive"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/130.json b/tests/unit/expected_test_results/ingredients_preparsing/130.json
new file mode 100644
index 0000000000000..eadd441f5b9b0
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/130.json
@@ -0,0 +1,6 @@
+{
+   "id" : "130",
+   "ingredients_text" : "S. thermophilus, L casei, L.bulgaricus",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "streptococcus thermophilus, lactobacillus casei, lactobacillus bulgaricus"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/131.json b/tests/unit/expected_test_results/ingredients_preparsing/131.json
new file mode 100644
index 0000000000000..dad1f4f1d4e01
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/131.json
@@ -0,0 +1,6 @@
+{
+   "id" : "131",
+   "ingredients_text" : "jus de citron*. *Ingrédients issus de l'agriculture biologique Peut contenir : œuf, moutarde, graine de sésame, poisson,soja, lait,fruits à coque, céleri.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "jus de citron Bio. , Traces éventuelles : œuf, Traces éventuelles : moutarde, Traces éventuelles : graine de sésame, Traces éventuelles : poisson, Traces éventuelles : soja, Traces éventuelles : lait, Traces éventuelles : fruits à coque, Traces éventuelles : céleri."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/132.json b/tests/unit/expected_test_results/ingredients_preparsing/132.json
new file mode 100644
index 0000000000000..d3691277b1556
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/132.json
@@ -0,0 +1,6 @@
+{
+   "id" : "132",
+   "ingredients_text" : "Farine, levure. Peut contenir des traces de _soja_, _amandes_, _noisettes_ et _noix de cajou_.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Farine, levure. Traces éventuelles : _soja_, Traces éventuelles : _amandes_, Traces éventuelles : _noisettes_, Traces éventuelles : _noix de cajou_."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/133.json b/tests/unit/expected_test_results/ingredients_preparsing/133.json
new file mode 100644
index 0000000000000..1fb9caae4a953
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/133.json
@@ -0,0 +1,6 @@
+{
+   "id" : "133",
+   "ingredients_text" : "Agua, aceite de girasol*. * Ingredientes ecológicos.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Agua, aceite de girasol Ecológico."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/134.json b/tests/unit/expected_test_results/ingredients_preparsing/134.json
new file mode 100644
index 0000000000000..cc9da31b13869
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/134.json
@@ -0,0 +1,6 @@
+{
+   "id" : "134",
+   "ingredients_text" : "Agua, aceite de girasol*, arroz* (5 %). (*) Ingredientes ecológicos.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Agua, aceite de girasol Ecológico, arroz Ecológico (5 %)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/135.json b/tests/unit/expected_test_results/ingredients_preparsing/135.json
new file mode 100644
index 0000000000000..ce8c4180472e8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/135.json
@@ -0,0 +1,6 @@
+{
+   "id" : "135",
+   "ingredients_text" : "Tofu* 88% (agua, habas de soja*). *cumple con el reglamento de agricultura ecológica CE 2092/91",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Tofu Ecológico 88% (agua, habas de soja Ecológico )."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/136.json b/tests/unit/expected_test_results/ingredients_preparsing/136.json
new file mode 100644
index 0000000000000..4fab815182076
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/136.json
@@ -0,0 +1,6 @@
+{
+   "id" : "136",
+   "ingredients_text" : "agua, almendra* (5,5%). *= procedentes de la agricultura ecológica",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "agua, almendra Ecológico (5.5%)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/137.json b/tests/unit/expected_test_results/ingredients_preparsing/137.json
new file mode 100644
index 0000000000000..1af710d0ec277
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/137.json
@@ -0,0 +1,7 @@
+{
+   "comment" : "test for bug #3273 that introduced unwanted separators before natural flavor",
+   "id" : "137",
+   "ingredients_text" : "non-gmo natural flavor",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "non-gmo natural flavor"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/138.json b/tests/unit/expected_test_results/ingredients_preparsing/138.json
new file mode 100644
index 0000000000000..8724e2651375c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/138.json
@@ -0,0 +1,6 @@
+{
+   "id" : "138",
+   "ingredients_text" : "vit. e, vitamins b2, B3 and K, vit d, vit a & c, vit. B12",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "vitamin e, vitamins, vitamin b2, vitamin B3, vitamin K, vitamin d, vitamin a, vitamin c, vitamin B12"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/139.json b/tests/unit/expected_test_results/ingredients_preparsing/139.json
new file mode 100644
index 0000000000000..9b4730c3c7efe
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/139.json
@@ -0,0 +1,6 @@
+{
+   "id" : "139",
+   "ingredients_text" : "vit. pp, vit c, vit. a et b6",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "vitamines, vitamine pp, Vitamine c, Vitamine a, vitamine b6"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/14.json b/tests/unit/expected_test_results/ingredients_preparsing/14.json
new file mode 100644
index 0000000000000..63e7792f88793
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/14.json
@@ -0,0 +1,6 @@
+{
+   "id" : "14",
+   "ingredients_text" : "Huiles végétales de palme, de colza et d'olive",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza, Huiles végétales d'olive"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/140.json b/tests/unit/expected_test_results/ingredients_preparsing/140.json
new file mode 100644
index 0000000000000..b76f3dd52a65b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/140.json
@@ -0,0 +1,6 @@
+{
+   "id" : "140",
+   "ingredients_text" : "witaminy A i D",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "witaminy, witamina A, witamina D"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/141.json b/tests/unit/expected_test_results/ingredients_preparsing/141.json
new file mode 100644
index 0000000000000..8262fe3f77f68
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/141.json
@@ -0,0 +1,6 @@
+{
+   "id" : "141",
+   "ingredients_text" : "colorant de surface : caramel ordinaire, agent de traitement de farine (E300), acide citrique",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "colorant de surface : caramel ordinaire, agent de traitement de farine (e300), acide citrique"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/142.json b/tests/unit/expected_test_results/ingredients_preparsing/142.json
new file mode 100644
index 0000000000000..12dc0a8a6c236
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/142.json
@@ -0,0 +1,6 @@
+{
+   "id" : "142",
+   "ingredients_text" : "Agua, edulcorantes (INS420, INS 960, INS N'952, INS N°954, INS°950, INS N 955), conservantes (INS.218, INS #202, INS N 216).",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Agua, edulcorantes (e420, e960, e952, e954, e950, e955), conservantes (e218, e202, e216)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/143.json b/tests/unit/expected_test_results/ingredients_preparsing/143.json
new file mode 100644
index 0000000000000..b36b98ea41ff2
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/143.json
@@ -0,0 +1,6 @@
+{
+   "id" : "143",
+   "ingredients_text" : "Vitamina E y C",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "vitaminas, vitamina E, vitamina C"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/144.json b/tests/unit/expected_test_results/ingredients_preparsing/144.json
new file mode 100644
index 0000000000000..4a1f5e30cfc41
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/144.json
@@ -0,0 +1,6 @@
+{
+   "id" : "144",
+   "ingredients_text" : "color E 124",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "color : e124"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/145.json b/tests/unit/expected_test_results/ingredients_preparsing/145.json
new file mode 100644
index 0000000000000..13de9aa23e1bc
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/145.json
@@ -0,0 +1,6 @@
+{
+   "id" : "145",
+   "ingredients_text" : "colores E (124, 125)",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "colores e124, e125"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/146.json b/tests/unit/expected_test_results/ingredients_preparsing/146.json
new file mode 100644
index 0000000000000..e6947c54d1065
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/146.json
@@ -0,0 +1,6 @@
+{
+   "id" : "146",
+   "ingredients_text" : "vitamine A, B, E e K",
+   "lc" : "it",
+   "preparsed_ingredients_text" : "vitamine, vitamina A, vitamina B, vitamina E, vitamina K"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/147.json b/tests/unit/expected_test_results/ingredients_preparsing/147.json
new file mode 100644
index 0000000000000..ec358ad273c14
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/147.json
@@ -0,0 +1,6 @@
+{
+   "id" : "147",
+   "ingredients_text" : "E 102, E-104 color, E-101(i), E101 (ii), E160a(iv), e172-i, E-160 i",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "e102, e104 color, e101i, e101ii, e160aiv, e172i, e160i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/148.json b/tests/unit/expected_test_results/ingredients_preparsing/148.json
new file mode 100644
index 0000000000000..007c1cf9d8891
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/148.json
@@ -0,0 +1,6 @@
+{
+   "id" : "148",
+   "ingredients_text" : "E102-E1400",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e102 - e1400"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/149.json b/tests/unit/expected_test_results/ingredients_preparsing/149.json
new file mode 100644
index 0000000000000..e87fcf6596720
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/149.json
@@ -0,0 +1,6 @@
+{
+   "id" : "149",
+   "ingredients_text" : "E172i-E174ii, E102(i)-E101i",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "e172i - e174ii, e102i - e101i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/15.json b/tests/unit/expected_test_results/ingredients_preparsing/15.json
new file mode 100644
index 0000000000000..ff2177ae59887
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/15.json
@@ -0,0 +1,6 @@
+{
+   "id" : "15",
+   "ingredients_text" : "phosphate et sulfate de calcium",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "phosphate de calcium, sulfate de calcium"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/150.json b/tests/unit/expected_test_results/ingredients_preparsing/150.json
new file mode 100644
index 0000000000000..a8570586b66d6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/150.json
@@ -0,0 +1,6 @@
+{
+   "id" : "150",
+   "ingredients_text" : "correcteurs d'acidité : E322/E333 E474-E475",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "correcteurs d'acidité : e322/e333, e474 - e475"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/151.json b/tests/unit/expected_test_results/ingredients_preparsing/151.json
new file mode 100644
index 0000000000000..fb03f559dba99
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/151.json
@@ -0,0 +1,6 @@
+{
+   "id" : "151",
+   "ingredients_text" : "E-330; E-331; Estabilizantes (E-327; E-418)",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e330; e331; Estabilizantes (e327; e418)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/152.json b/tests/unit/expected_test_results/ingredients_preparsing/152.json
new file mode 100644
index 0000000000000..73032afed2035
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/152.json
@@ -0,0 +1,6 @@
+{
+   "id" : "152",
+   "ingredients_text" : "E120 color",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e120 color"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/153.json b/tests/unit/expected_test_results/ingredients_preparsing/153.json
new file mode 100644
index 0000000000000..fb2d53bfa01c3
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/153.json
@@ -0,0 +1,6 @@
+{
+   "id" : "153",
+   "ingredients_text" : "E172-i",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e172i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/154.json b/tests/unit/expected_test_results/ingredients_preparsing/154.json
new file mode 100644
index 0000000000000..962a4da40e5a7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/154.json
@@ -0,0 +1,6 @@
+{
+   "id" : "154",
+   "ingredients_text" : "E172 i",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e172i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/155.json b/tests/unit/expected_test_results/ingredients_preparsing/155.json
new file mode 100644
index 0000000000000..d59d14cf23829
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/155.json
@@ -0,0 +1,6 @@
+{
+   "id" : "155",
+   "ingredients_text" : "(E172i)",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "(e172i)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/156.json b/tests/unit/expected_test_results/ingredients_preparsing/156.json
new file mode 100644
index 0000000000000..76d87be87cc33
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/156.json
@@ -0,0 +1,6 @@
+{
+   "id" : "156",
+   "ingredients_text" : "E102(i)-E101i",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e102i - e101i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/157.json b/tests/unit/expected_test_results/ingredients_preparsing/157.json
new file mode 100644
index 0000000000000..da09e88c6117d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/157.json
@@ -0,0 +1,6 @@
+{
+   "id" : "157",
+   "ingredients_text" : "E102(i)",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e102i"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/158.json b/tests/unit/expected_test_results/ingredients_preparsing/158.json
new file mode 100644
index 0000000000000..2e3b0f8222606
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/158.json
@@ -0,0 +1,6 @@
+{
+   "id" : "158",
+   "ingredients_text" : "S.I.N.:160 b",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "e160b"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/159.json b/tests/unit/expected_test_results/ingredients_preparsing/159.json
new file mode 100644
index 0000000000000..c41bcfbef8f79
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/159.json
@@ -0,0 +1,6 @@
+{
+   "id" : "159",
+   "ingredients_text" : "estabilizadores (E 422, E 412)",
+   "lc" : "pt",
+   "preparsed_ingredients_text" : "estabilizadores (e422, e412)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/16.json b/tests/unit/expected_test_results/ingredients_preparsing/16.json
new file mode 100644
index 0000000000000..ce5dce7a00245
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/16.json
@@ -0,0 +1,6 @@
+{
+   "id" : "16",
+   "ingredients_text" : "sulfates de calcium et potassium",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "sulfates de calcium, sulfates de potassium"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/160.json b/tests/unit/expected_test_results/ingredients_preparsing/160.json
new file mode 100644
index 0000000000000..04a7d9c724b79
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/160.json
@@ -0,0 +1,6 @@
+{
+   "id" : "160",
+   "ingredients_text" : "contiene apio y derivados de leche",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Sustancias o productos que causan alergias o intolerancias : apio, Sustancias o productos que causan alergias o intolerancias : derivados de leche."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/161.json b/tests/unit/expected_test_results/ingredients_preparsing/161.json
new file mode 100644
index 0000000000000..0b7e48b465eae
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/161.json
@@ -0,0 +1,6 @@
+{
+   "id" : "161",
+   "ingredients_text" : "E160a(ii)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e160aii"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/162.json b/tests/unit/expected_test_results/ingredients_preparsing/162.json
new file mode 100644
index 0000000000000..d4ee2d5232d79
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/162.json
@@ -0,0 +1,6 @@
+{
+   "id" : "162",
+   "ingredients_text" : "(E160a-ii)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "(e160aii)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/163.json b/tests/unit/expected_test_results/ingredients_preparsing/163.json
new file mode 100644
index 0000000000000..db8dee27c0f30
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/163.json
@@ -0,0 +1,6 @@
+{
+   "id" : "163",
+   "ingredients_text" : "colorant (E160a(ii))",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "colorant (e160aii)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/164.json b/tests/unit/expected_test_results/ingredients_preparsing/164.json
new file mode 100644
index 0000000000000..75d546dcf98c4
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/164.json
@@ -0,0 +1,7 @@
+{
+   "comment" : "# do not separate acide acétique into acide : acétique",
+   "id" : "164",
+   "ingredients_text" : "Esters glycéroliques de l'acide acétique et d'acides gras",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Esters glycéroliques de l'acide acétique et d'acides gras"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/165.json b/tests/unit/expected_test_results/ingredients_preparsing/165.json
new file mode 100644
index 0000000000000..033990680d5b5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/165.json
@@ -0,0 +1,6 @@
+{
+   "id" : "165",
+   "ingredients_text" : "acide acétique",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "acide acétique"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/166.json b/tests/unit/expected_test_results/ingredients_preparsing/166.json
new file mode 100644
index 0000000000000..134bee88bbbab
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/166.json
@@ -0,0 +1,6 @@
+{
+   "id" : "166",
+   "ingredients_text" : "мука пшеничная х/п в/с",
+   "lc" : "ru",
+   "preparsed_ingredients_text" : "мука пшеничная хлебопекарная высшего сорта"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/167.json b/tests/unit/expected_test_results/ingredients_preparsing/167.json
new file mode 100644
index 0000000000000..2b265820f73da
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/167.json
@@ -0,0 +1,6 @@
+{
+   "id" : "167",
+   "ingredients_text" : "Organic garbanzo beans (cooked w/o salt), water",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "Organic garbanzo beans (cooked without salt), water"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/168.json b/tests/unit/expected_test_results/ingredients_preparsing/168.json
new file mode 100644
index 0000000000000..d2ecd640ecb1b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/168.json
@@ -0,0 +1,6 @@
+{
+   "id" : "168",
+   "ingredients_text" : "sugar, cocoa (processed w/alkali), egg yolk",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "sugar, cocoa (processed with alkali), egg yolk"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/169.json b/tests/unit/expected_test_results/ingredients_preparsing/169.json
new file mode 100644
index 0000000000000..0b79b79ca2afe
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/169.json
@@ -0,0 +1,6 @@
+{
+   "id" : "169",
+   "ingredients_text" : "LAIT entier pasteurisé*. *ingrédient issu de l'agriculture biologique.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "LAIT entier pasteurisé Bio."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/17.json b/tests/unit/expected_test_results/ingredients_preparsing/17.json
new file mode 100644
index 0000000000000..971f16ed24a8a
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/17.json
@@ -0,0 +1,6 @@
+{
+   "id" : "17",
+   "ingredients_text" : "chlorures (sodium et potassium)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "chlorures de sodium, chlorures de potassium"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/170.json b/tests/unit/expected_test_results/ingredients_preparsing/170.json
new file mode 100644
index 0000000000000..7fde9874bb44d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/170.json
@@ -0,0 +1,6 @@
+{
+   "id" : "170",
+   "ingredients_text" : "vitamines B1, B6, B9, PP et E",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "vitamines, vitamine B1, vitamine B6, vitamine B9, vitamine PP, vitamine E"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/171.json b/tests/unit/expected_test_results/ingredients_preparsing/171.json
new file mode 100644
index 0000000000000..e285d2107975f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/171.json
@@ -0,0 +1,6 @@
+{
+   "id" : "171",
+   "ingredients_text" : "vitamines (B1, acide folique (B9))",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "vitamines, vitamine B1, acide folique, vitamine B9"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/172.json b/tests/unit/expected_test_results/ingredients_preparsing/172.json
new file mode 100644
index 0000000000000..c7b37201ff5b8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/172.json
@@ -0,0 +1,6 @@
+{
+   "id" : "172",
+   "ingredients_text" : "Chocolate (Italy, contains milk)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "Chocolate (Italy, Substances or products causing allergies or intolerances : milk.)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/173.json b/tests/unit/expected_test_results/ingredients_preparsing/173.json
new file mode 100644
index 0000000000000..5b3d6d71bb0d5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/173.json
@@ -0,0 +1,6 @@
+{
+   "id" : "173",
+   "ingredients_text" : "Chocolate (contains milk)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "Chocolate ( Substances or products causing allergies or intolerances : milk.)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/174.json b/tests/unit/expected_test_results/ingredients_preparsing/174.json
new file mode 100644
index 0000000000000..97e7bb7cfca20
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/174.json
@@ -0,0 +1,6 @@
+{
+   "id" : "174",
+   "ingredients_text" : "Chocolate. Contains (milk)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "Chocolate. Substances or products causing allergies or intolerances : milk."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/175.json b/tests/unit/expected_test_results/ingredients_preparsing/175.json
new file mode 100644
index 0000000000000..7f692b854d008
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/175.json
@@ -0,0 +1,6 @@
+{
+   "id" : "175",
+   "ingredients_text" : "Sel, sucre², graisse de palme¹, amidons¹ (maïs¹, pomme de terre¹), oignon¹ : 8,9%, ail¹, oignon grillé¹ : 1,4%, épices¹ et aromate¹ (livèche¹ : 0,4%, curcuma¹, noix de muscade¹), carotte¹ : 0,5%. Peut contenir : céleri, céréales contenant du gluten, lait, moutarde, œuf, soja. ¹Ingrédients issus de l'Agriculture Biologique. ² Ingrédients issus du commerce équitable",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sel, sucre Commerce équitable, graisse de palme Bio, amidons Bio (maïs Bio, pomme de terre Bio ), oignon Bio : 8.9%, ail Bio, oignon grillé Bio : 1.4%, épices Bio et aromate Bio (livèche Bio : 0.4%, curcuma Bio, noix de muscade Bio ), carotte Bio : 0.5%. Traces éventuelles : céleri, Traces éventuelles : céréales contenant du gluten, Traces éventuelles : lait, Traces éventuelles : moutarde, Traces éventuelles : œuf, Traces éventuelles : soja."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/176.json b/tests/unit/expected_test_results/ingredients_preparsing/176.json
new file mode 100644
index 0000000000000..d9bf73d630044
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/176.json
@@ -0,0 +1,7 @@
+{
+   "comment" : "# Russian е character",
+   "id" : "176",
+   "ingredients_text" : "е322, Куркумины e100, е-1442, (е621)",
+   "lc" : "ru",
+   "preparsed_ingredients_text" : "e322, Куркумины e100, e1442, (e621)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/177.json b/tests/unit/expected_test_results/ingredients_preparsing/177.json
new file mode 100644
index 0000000000000..f6aa708b8e26f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/177.json
@@ -0,0 +1,6 @@
+{
+   "id" : "177",
+   "ingredients_text" : "huiles végétales (palme, olive et tournesol)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huiles végétales de palme, huiles végétales d'olive, huiles végétales de tournesol"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/178.json b/tests/unit/expected_test_results/ingredients_preparsing/178.json
new file mode 100644
index 0000000000000..bd8b1ca55fc09
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/178.json
@@ -0,0 +1,6 @@
+{
+   "id" : "178",
+   "ingredients_text" : "huile végétale : colza",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/179.json b/tests/unit/expected_test_results/ingredients_preparsing/179.json
new file mode 100644
index 0000000000000..908c9239a9151
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/179.json
@@ -0,0 +1,6 @@
+{
+   "id" : "179",
+   "ingredients_text" : "huile végétale : colza, fraises",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de colza, fraises"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/18.json b/tests/unit/expected_test_results/ingredients_preparsing/18.json
new file mode 100644
index 0000000000000..cdfa19eaf55ad
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/18.json
@@ -0,0 +1,6 @@
+{
+   "id" : "18",
+   "ingredients_text" : "chlorures (sodium, potassium)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "chlorures de sodium, chlorures de potassium"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/180.json b/tests/unit/expected_test_results/ingredients_preparsing/180.json
new file mode 100644
index 0000000000000..ccd8430a4082e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/180.json
@@ -0,0 +1,6 @@
+{
+   "id" : "180",
+   "ingredients_text" : "huile végétale : colza et tomates",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de colza et tomates"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/181.json b/tests/unit/expected_test_results/ingredients_preparsing/181.json
new file mode 100644
index 0000000000000..6b496a0e3b402
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/181.json
@@ -0,0 +1,6 @@
+{
+   "id" : "181",
+   "ingredients_text" : "vegetable oil: sunflower",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "sunflower vegetable oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/182.json b/tests/unit/expected_test_results/ingredients_preparsing/182.json
new file mode 100644
index 0000000000000..2d5f69fc2d4f8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/182.json
@@ -0,0 +1,6 @@
+{
+   "id" : "182",
+   "ingredients_text" : "vegetable oil (palm)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "palm vegetable oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/183.json b/tests/unit/expected_test_results/ingredients_preparsing/183.json
new file mode 100644
index 0000000000000..12d06b3da3976
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/183.json
@@ -0,0 +1,6 @@
+{
+   "id" : "183",
+   "ingredients_text" : "vegetable oils (palm, olive)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "palm vegetable oils, olive vegetable oils"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/184.json b/tests/unit/expected_test_results/ingredients_preparsing/184.json
new file mode 100644
index 0000000000000..8ddcd79b6f7b1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/184.json
@@ -0,0 +1,6 @@
+{
+   "id" : "184",
+   "ingredients_text" : "organic vegetable oils (sunflower, colza and rapeseed)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "sunflower organic vegetable oils, colza organic vegetable oils, rapeseed organic vegetable oils"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/185.json b/tests/unit/expected_test_results/ingredients_preparsing/185.json
new file mode 100644
index 0000000000000..a6ee2ab5dbb5e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/185.json
@@ -0,0 +1,6 @@
+{
+   "id" : "185",
+   "ingredients_text" : "vegetable oils : sunflower, colza and strawberry",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "sunflower vegetable oils, colza vegetable oils and strawberry"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/186.json b/tests/unit/expected_test_results/ingredients_preparsing/186.json
new file mode 100644
index 0000000000000..a3c6440e90e56
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/186.json
@@ -0,0 +1,6 @@
+{
+   "id" : "186",
+   "ingredients_text" : "oleje roślinne (słonecznikowy)",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "oleje roślinne słonecznikowy"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/187.json b/tests/unit/expected_test_results/ingredients_preparsing/187.json
new file mode 100644
index 0000000000000..58a2a841ae609
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/187.json
@@ -0,0 +1,6 @@
+{
+   "id" : "187",
+   "ingredients_text" : "oleje roślinne: słonecznikowy",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "oleje roślinne słonecznikowy"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/188.json b/tests/unit/expected_test_results/ingredients_preparsing/188.json
new file mode 100644
index 0000000000000..6c57a92152d2c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/188.json
@@ -0,0 +1,6 @@
+{
+   "id" : "188",
+   "ingredients_text" : "oleje roślinne (słonecznikowy, rzepakowy)",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "oleje roślinne słonecznikowy, oleje roślinne rzepakowy"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/189.json b/tests/unit/expected_test_results/ingredients_preparsing/189.json
new file mode 100644
index 0000000000000..b2d368f36a1da
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/189.json
@@ -0,0 +1,6 @@
+{
+   "id" : "189",
+   "ingredients_text" : "oleje roślinne (sojowy, słonecznikowy, kokosowy, rzepakowy) w zmiennych proporcjach",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "oleje roślinne sojowy, oleje roślinne słonecznikowy, oleje roślinne kokosowy, oleje roślinne rzepakowy"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/19.json b/tests/unit/expected_test_results/ingredients_preparsing/19.json
new file mode 100644
index 0000000000000..9894f19c81c7c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/19.json
@@ -0,0 +1,6 @@
+{
+   "id" : "19",
+   "ingredients_text" : "fraises 30%",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "fraises 30%"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/190.json b/tests/unit/expected_test_results/ingredients_preparsing/190.json
new file mode 100644
index 0000000000000..264a4e67f5b96
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/190.json
@@ -0,0 +1,6 @@
+{
+   "id" : "190",
+   "ingredients_text" : "tłuszcze roślinne (palmowy nieutwardzony, shea)",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "tłuszcze roślinne palmowy nieutwardzony, tłuszcze roślinne shea"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/191.json b/tests/unit/expected_test_results/ingredients_preparsing/191.json
new file mode 100644
index 0000000000000..e43c2c5f0452c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/191.json
@@ -0,0 +1,6 @@
+{
+   "id" : "191",
+   "ingredients_text" : "tłuszcze roślinne (kokosowy i palmowy) w zmiennych proporcjach",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "tłuszcze roślinne kokosowy, tłuszcze roślinne palmowy"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/192.json b/tests/unit/expected_test_results/ingredients_preparsing/192.json
new file mode 100644
index 0000000000000..ad195858f9513
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/192.json
@@ -0,0 +1,6 @@
+{
+   "id" : "192",
+   "ingredients_text" : "mięso (wołowe, wieprzowe, cielęce)",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "mięso wołowe, mięso wieprzowe, mięso cielęce"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/193.json b/tests/unit/expected_test_results/ingredients_preparsing/193.json
new file mode 100644
index 0000000000000..83eb7af7be14a
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/193.json
@@ -0,0 +1,6 @@
+{
+   "id" : "193",
+   "ingredients_text" : "przeciery z (jabłek, bananów, marchwi)",
+   "lc" : "pl",
+   "preparsed_ingredients_text" : "przeciery z jabłek, przeciery z bananów, przeciery z marchwi"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/194.json b/tests/unit/expected_test_results/ingredients_preparsing/194.json
new file mode 100644
index 0000000000000..57a33b472d72e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/194.json
@@ -0,0 +1,6 @@
+{
+   "id" : "194",
+   "ingredients_text" : "масло (Подсолнечное)",
+   "lc" : "ru",
+   "preparsed_ingredients_text" : "масло Подсолнечное"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/195.json b/tests/unit/expected_test_results/ingredients_preparsing/195.json
new file mode 100644
index 0000000000000..dbcf4e74bcbb2
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/195.json
@@ -0,0 +1,6 @@
+{
+   "id" : "195",
+   "ingredients_text" : "Масло (подсолнечное)",
+   "lc" : "ru",
+   "preparsed_ingredients_text" : "Масло подсолнечное"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/196.json b/tests/unit/expected_test_results/ingredients_preparsing/196.json
new file mode 100644
index 0000000000000..ef493384fcdb2
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/196.json
@@ -0,0 +1,6 @@
+{
+   "id" : "196",
+   "ingredients_text" : "масло растительное (подсолнечное, соевое)",
+   "lc" : "ru",
+   "preparsed_ingredients_text" : "масло растительное подсолнечное, масло растительное соевое"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/197.json b/tests/unit/expected_test_results/ingredients_preparsing/197.json
new file mode 100644
index 0000000000000..e8ec7e4121fd3
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/197.json
@@ -0,0 +1,6 @@
+{
+   "id" : "197",
+   "ingredients_text" : "Teneur en fruits: 50gr pour 100 grammes",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Teneur en fruits: 50g pour 100 g"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/198.json b/tests/unit/expected_test_results/ingredients_preparsing/198.json
new file mode 100644
index 0000000000000..be243979188e6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/198.json
@@ -0,0 +1,6 @@
+{
+   "id" : "198",
+   "ingredients_text" : "bojilo: E 150a, tvari za rahljenje: E 500 i E 503, sol.",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "bojilo: e150a, tvari za rahljenje: e500, e503, sol."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/199.json b/tests/unit/expected_test_results/ingredients_preparsing/199.json
new file mode 100644
index 0000000000000..e7394610b7ce6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/199.json
@@ -0,0 +1,6 @@
+{
+   "id" : "199",
+   "ingredients_text" : "bojilo: E 150a, tvari za rahljenje: E 500 i, E 503, sol.",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "bojilo: e150a, tvari za rahljenje: e500 i, e503, sol."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/2.json b/tests/unit/expected_test_results/ingredients_preparsing/2.json
new file mode 100644
index 0000000000000..e6ae97208cd0f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/2.json
@@ -0,0 +1,6 @@
+{
+   "id" : "2",
+   "ingredients_text" : "Vitamine A",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Vitamine A"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/20.json b/tests/unit/expected_test_results/ingredients_preparsing/20.json
new file mode 100644
index 0000000000000..d3b67e3bd90a7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/20.json
@@ -0,0 +1,6 @@
+{
+   "id" : "20",
+   "ingredients_text" : "Marmelade d'oranges 41% (sirop de glucose-fructose, sucre, pulpe d'orange 4.5%, jus d'orange concentré 1.4% (équivalent jus d'orange 7.8%), pulpe d'orange concentrée 0.6% (équivalent pulpe d'orange 2.6%), gélifiant (pectines), acidifiant (acide citrique), correcteurs d'acidité (citrate de calcium, citrate de sodium), arôme naturel d'orange, épaississant (gomme xanthane)), chocolat 24.9% (sucre, pâte de cacao, beurre de cacao, graisses végétales (illipe, mangue, sal, karité et palme en proportions variables), arôme, émulsifiant (lécithine de soja), lactose et protéines de lait), farine de blé, sucre, oeufs, sirop de glucose-fructose, huile de colza, poudre à lever (carbonate acide d'ammonium, diphosphate disodique, carbonate acide de sodium), sel, émulsifiant (lécithine de soja).",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Marmelade d'oranges 41% (sirop de glucose-fructose, sucre, pulpe d'orange 4.5%, jus d'orange concentré 1.4% (équivalent jus d'orange 7.8%), pulpe d'orange concentrée 0.6% (équivalent pulpe d'orange 2.6%), gélifiant (pectines), acidifiant (acide citrique), correcteurs d'acidité (citrate de calcium, citrate de sodium), arôme naturel d'orange, épaississant (gomme xanthane)), chocolat 24.9% (sucre, pâte de cacao, beurre de cacao, graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de sal, graisses végétales de karité, graisses végétales de palme, arôme, émulsifiant (lécithine de soja), lactose et protéines de lait), farine de blé, sucre, oeufs, sirop de glucose-fructose, huile de colza, poudre à lever (carbonate acide d'ammonium, diphosphate disodique, carbonate acide de sodium), sel, émulsifiant (lécithine de soja)."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/200.json b/tests/unit/expected_test_results/ingredients_preparsing/200.json
new file mode 100644
index 0000000000000..b221a7661f049
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/200.json
@@ -0,0 +1,6 @@
+{
+   "id" : "200",
+   "ingredients_text" : "bojilo: E 150a, tvari za rahljenje: E 500(i), E 503, sol.",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "bojilo: e150a, tvari za rahljenje: e500i, e503, sol."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/201.json b/tests/unit/expected_test_results/ingredients_preparsing/201.json
new file mode 100644
index 0000000000000..d3ea92ffbbd4b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/201.json
@@ -0,0 +1,6 @@
+{
+   "id" : "201",
+   "ingredients_text" : "bojilo: E 150a, tvari za rahljenje: E 500i, E 503, sol.",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "bojilo: e150a, tvari za rahljenje: e500i, e503, sol."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/202.json b/tests/unit/expected_test_results/ingredients_preparsing/202.json
new file mode 100644
index 0000000000000..c6508e509e861
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/202.json
@@ -0,0 +1,6 @@
+{
+   "id" : "202",
+   "ingredients_text" : "formaggio, E 472 e, E470a.",
+   "lc" : "it",
+   "preparsed_ingredients_text" : "formaggio, e472 e, e470a."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/203.json b/tests/unit/expected_test_results/ingredients_preparsing/203.json
new file mode 100644
index 0000000000000..cad0af37030d7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/203.json
@@ -0,0 +1,6 @@
+{
+   "id" : "203",
+   "ingredients_text" : "formaggio, E 472 e E470a.",
+   "lc" : "it",
+   "preparsed_ingredients_text" : "formaggio, e472, e470a."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/204.json b/tests/unit/expected_test_results/ingredients_preparsing/204.json
new file mode 100644
index 0000000000000..2e58a9cbcc6d6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/204.json
@@ -0,0 +1,6 @@
+{
+   "id" : "204",
+   "ingredients_text" : "syr, E470 a E470a, mlieko.",
+   "lc" : "sk",
+   "preparsed_ingredients_text" : "syr, e470, e470a, mlieko."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/205.json b/tests/unit/expected_test_results/ingredients_preparsing/205.json
new file mode 100644
index 0000000000000..bf0e8b0bae6b6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/205.json
@@ -0,0 +1,6 @@
+{
+   "id" : "205",
+   "ingredients_text" : "Piments (vert, rouge, jaune)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Piments vert, Piments rouge, Piments jaune"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/206.json b/tests/unit/expected_test_results/ingredients_preparsing/206.json
new file mode 100644
index 0000000000000..879310a6c3e59
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/206.json
@@ -0,0 +1,6 @@
+{
+   "id" : "206",
+   "ingredients_text" : "pflanzliches Fett (Kokosnuss, Palmkern)",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Kokosnussfett, Palmkernfett"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/207.json b/tests/unit/expected_test_results/ingredients_preparsing/207.json
new file mode 100644
index 0000000000000..e7384ea6a8ead
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/207.json
@@ -0,0 +1,6 @@
+{
+   "id" : "207",
+   "ingredients_text" : "pflanzliche Öle und Fette (Raps, Palm, Shea, Sonnenblumen)",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Rapsöl, Palmfett, Sheafett, Sonnenblumenfett"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/208.json b/tests/unit/expected_test_results/ingredients_preparsing/208.json
new file mode 100644
index 0000000000000..89b9c81c22cd4
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/208.json
@@ -0,0 +1,6 @@
+{
+   "id" : "208",
+   "ingredients_text" : "Huiles végétales de palme, de colza et de tournesol",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza, Huiles végétales de tournesol"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/209.json b/tests/unit/expected_test_results/ingredients_preparsing/209.json
new file mode 100644
index 0000000000000..8125909b004a8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/209.json
@@ -0,0 +1,6 @@
+{
+   "id" : "209",
+   "ingredients_text" : "arôme naturel de pomme avec d'autres âromes",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "arôme naturel de pomme et âromes"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/21.json b/tests/unit/expected_test_results/ingredients_preparsing/21.json
new file mode 100644
index 0000000000000..2a09a1f993984
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/21.json
@@ -0,0 +1,6 @@
+{
+   "id" : "21",
+   "ingredients_text" : "graisses végétales (illipe, mangue, sal, karité et palme en proportions variables)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de sal, graisses végétales de karité, graisses végétales de palme"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/210.json b/tests/unit/expected_test_results/ingredients_preparsing/210.json
new file mode 100644
index 0000000000000..aefd5687271f9
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/210.json
@@ -0,0 +1,6 @@
+{
+   "id" : "210",
+   "ingredients_text" : "Carbonate de magnésium, fer élémentaire",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Carbonate de magnésium, fer élémentaire"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/211.json b/tests/unit/expected_test_results/ingredients_preparsing/211.json
new file mode 100644
index 0000000000000..6633263b9cd2f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/211.json
@@ -0,0 +1,6 @@
+{
+   "id" : "211",
+   "ingredients_text" : "huile végétale (colza)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/212.json b/tests/unit/expected_test_results/ingredients_preparsing/212.json
new file mode 100644
index 0000000000000..99a9cf3a6e0d4
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/212.json
@@ -0,0 +1,6 @@
+{
+   "id" : "212",
+   "ingredients_text" : "huile végétale : colza",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/213.json b/tests/unit/expected_test_results/ingredients_preparsing/213.json
new file mode 100644
index 0000000000000..ad000e8a4fc8e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/213.json
@@ -0,0 +1,6 @@
+{
+   "id" : "213",
+   "ingredients_text" : "ječmeni i pšenični slad",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "ječmeni slad, pšenični slad"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/214.json b/tests/unit/expected_test_results/ingredients_preparsing/214.json
new file mode 100644
index 0000000000000..0c3bda643c598
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/214.json
@@ -0,0 +1,6 @@
+{
+   "id" : "214",
+   "ingredients_text" : "ječmeni, ječmeni i pšenični slad",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "ječmeni slad, ječmeni slad, pšenični slad"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/215.json b/tests/unit/expected_test_results/ingredients_preparsing/215.json
new file mode 100644
index 0000000000000..6346738056162
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/215.json
@@ -0,0 +1,6 @@
+{
+   "id" : "215",
+   "ingredients_text" : "Pasterizirano mlijeko (s 1.0% mliječne masti)",
+   "lc" : "hr",
+   "preparsed_ingredients_text" : "Pasterizirano mlijeko s 1.0% mliječne masti"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/216.json b/tests/unit/expected_test_results/ingredients_preparsing/216.json
new file mode 100644
index 0000000000000..3034bd0b1c901
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/216.json
@@ -0,0 +1,6 @@
+{
+   "id" : "216",
+   "ingredients_text" : "Vegetal oil (sunflower, olive and palm)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "sunflower Vegetal oil, olive Vegetal oil, palm Vegetal oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/217.json b/tests/unit/expected_test_results/ingredients_preparsing/217.json
new file mode 100644
index 0000000000000..6ef630346929b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/217.json
@@ -0,0 +1,6 @@
+{
+   "id" : "217",
+   "ingredients_text" : "vegetable oil (palm)",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "palm vegetable oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/218.json b/tests/unit/expected_test_results/ingredients_preparsing/218.json
new file mode 100644
index 0000000000000..9978164f46913
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/218.json
@@ -0,0 +1,6 @@
+{
+   "id" : "218",
+   "ingredients_text" : "vegetable oil: palm",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "palm vegetable oil"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/219.json b/tests/unit/expected_test_results/ingredients_preparsing/219.json
new file mode 100644
index 0000000000000..869c4754f8ee5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/219.json
@@ -0,0 +1,6 @@
+{
+   "id" : "219",
+   "ingredients_text" : "protéines végétales (soja, blé)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "protéine de soja, protéine de blé"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/22.json b/tests/unit/expected_test_results/ingredients_preparsing/22.json
new file mode 100644
index 0000000000000..c29b69e676eca
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/22.json
@@ -0,0 +1,6 @@
+{
+   "id" : "22",
+   "ingredients_text" : "graisses végétales (illipe, mangue, palme)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de palme"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/220.json b/tests/unit/expected_test_results/ingredients_preparsing/220.json
new file mode 100644
index 0000000000000..ae44a714abfd5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/220.json
@@ -0,0 +1,6 @@
+{
+   "id" : "220",
+   "ingredients_text" : "pflanzliche Proteine (Erbsen, Sonnenblumen)",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Erbsenprotein, Sonnenblumenprotein"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/221.json b/tests/unit/expected_test_results/ingredients_preparsing/221.json
new file mode 100644
index 0000000000000..902aa0eacae2c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/221.json
@@ -0,0 +1,6 @@
+{
+   "id" : "221",
+   "ingredients_text" : "Piments (vert, rouge, jaune, sel)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Piments (vert, rouge, jaune, sel)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/222.json b/tests/unit/expected_test_results/ingredients_preparsing/222.json
new file mode 100644
index 0000000000000..bc489ca92a861
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/222.json
@@ -0,0 +1,6 @@
+{
+   "id" : "222",
+   "ingredients_text" : "Huile de palme, noisettes et tournesol",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme, Huile de noisettes, Huile de tournesol"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/223.json b/tests/unit/expected_test_results/ingredients_preparsing/223.json
new file mode 100644
index 0000000000000..c5b7a3d80e320
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/223.json
@@ -0,0 +1,6 @@
+{
+   "id" : "223",
+   "ingredients_text" : "Huile de palme, noisettes",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme, noisettes"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/224.json b/tests/unit/expected_test_results/ingredients_preparsing/224.json
new file mode 100644
index 0000000000000..d6f476a00a1e9
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/224.json
@@ -0,0 +1,6 @@
+{
+   "id" : "224",
+   "ingredients_text" : "arôme naturel de citron, citron vert et d'autres agrumes",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "arôme naturel de citron, arôme naturel de citron vert, arôme naturel d'agrumes"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/23.json b/tests/unit/expected_test_results/ingredients_preparsing/23.json
new file mode 100644
index 0000000000000..d7f6284a349f7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/23.json
@@ -0,0 +1,6 @@
+{
+   "id" : "23",
+   "ingredients_text" : "graisses végétales (illipe)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "graisses végétales d'illipe"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/24.json b/tests/unit/expected_test_results/ingredients_preparsing/24.json
new file mode 100644
index 0000000000000..6c61262239cb4
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/24.json
@@ -0,0 +1,6 @@
+{
+   "id" : "24",
+   "ingredients_text" : "graisses végétales (illipe et sal)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "graisses végétales d'illipe, graisses végétales de sal"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/25.json b/tests/unit/expected_test_results/ingredients_preparsing/25.json
new file mode 100644
index 0000000000000..9648232f68f46
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/25.json
@@ -0,0 +1,6 @@
+{
+   "id" : "25",
+   "ingredients_text" : "gélifiant pectine",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "gélifiant : pectine"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/26.json b/tests/unit/expected_test_results/ingredients_preparsing/26.json
new file mode 100644
index 0000000000000..b3c06ebc2b64c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/26.json
@@ -0,0 +1,6 @@
+{
+   "id" : "26",
+   "ingredients_text" : "gélifiant (pectine)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "gélifiant (pectine)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/27.json b/tests/unit/expected_test_results/ingredients_preparsing/27.json
new file mode 100644
index 0000000000000..f73a84ab6362e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/27.json
@@ -0,0 +1,6 @@
+{
+   "id" : "27",
+   "ingredients_text" : "agent de traitement de la farine (acide ascorbique)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "agent de traitement de la farine (acide ascorbique)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/28.json b/tests/unit/expected_test_results/ingredients_preparsing/28.json
new file mode 100644
index 0000000000000..fa69f7c9ea0c0
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/28.json
@@ -0,0 +1,6 @@
+{
+   "id" : "28",
+   "ingredients_text" : "lait demi-écrémé",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "lait demi-écrémé"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/29.json b/tests/unit/expected_test_results/ingredients_preparsing/29.json
new file mode 100644
index 0000000000000..dad4b29b1a107
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/29.json
@@ -0,0 +1,6 @@
+{
+   "id" : "29",
+   "ingredients_text" : "Saveur vanille : lait demi-écrémé 77%, sucre",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Saveur vanille : lait demi-écrémé 77%, sucre"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/3.json b/tests/unit/expected_test_results/ingredients_preparsing/3.json
new file mode 100644
index 0000000000000..a0c2c4a378541
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/3.json
@@ -0,0 +1,6 @@
+{
+   "id" : "3",
+   "ingredients_text" : "Vitamines A, B et C",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "vitamines, vitamine A, vitamine B, vitamine C"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/30.json b/tests/unit/expected_test_results/ingredients_preparsing/30.json
new file mode 100644
index 0000000000000..e75ab63acce1d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/30.json
@@ -0,0 +1,6 @@
+{
+   "id" : "30",
+   "ingredients_text" : "colorants alimentaires E (124,122,133,104,110)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "colorants alimentaires : e124, e122, e133, e104, e110"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/31.json b/tests/unit/expected_test_results/ingredients_preparsing/31.json
new file mode 100644
index 0000000000000..310f101998b00
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/31.json
@@ -0,0 +1,6 @@
+{
+   "id" : "31",
+   "ingredients_text" : "INS 240,241,242b",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e240, e241, e242b"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/32.json b/tests/unit/expected_test_results/ingredients_preparsing/32.json
new file mode 100644
index 0000000000000..bf6a7be210a74
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/32.json
@@ -0,0 +1,6 @@
+{
+   "id" : "32",
+   "ingredients_text" : "colorants E (124, 125, 120 et 122",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "colorants : e124, e125, e120, e122"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/33.json b/tests/unit/expected_test_results/ingredients_preparsing/33.json
new file mode 100644
index 0000000000000..c08765f1bef04
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/33.json
@@ -0,0 +1,6 @@
+{
+   "id" : "33",
+   "ingredients_text" : "E250-E251",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e250 - e251"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/34.json b/tests/unit/expected_test_results/ingredients_preparsing/34.json
new file mode 100644
index 0000000000000..3e1d1061c25a8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/34.json
@@ -0,0 +1,6 @@
+{
+   "id" : "34",
+   "ingredients_text" : "E250-E251-E260",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e250 - e251 - e260"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/35.json b/tests/unit/expected_test_results/ingredients_preparsing/35.json
new file mode 100644
index 0000000000000..93656ec71b075
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/35.json
@@ -0,0 +1,6 @@
+{
+   "id" : "35",
+   "ingredients_text" : "E 250b-E251-e.260(ii)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e250b - e251 - e260ii"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/36.json b/tests/unit/expected_test_results/ingredients_preparsing/36.json
new file mode 100644
index 0000000000000..f34295366282d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/36.json
@@ -0,0 +1,6 @@
+{
+   "id" : "36",
+   "ingredients_text" : "émulsifiants : E463, E432 et E472 - correcteurs d'acidité : E322/E333 E474-E475",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "émulsifiants : e463, e432, e472 - correcteurs d'acidité : e322/e333, e474 - e475"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/37.json b/tests/unit/expected_test_results/ingredients_preparsing/37.json
new file mode 100644
index 0000000000000..c55bca92f254e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/37.json
@@ -0,0 +1,6 @@
+{
+   "id" : "37",
+   "ingredients_text" : "E100 E122",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e100, e122"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/38.json b/tests/unit/expected_test_results/ingredients_preparsing/38.json
new file mode 100644
index 0000000000000..b0ccafc305188
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/38.json
@@ -0,0 +1,6 @@
+{
+   "id" : "38",
+   "ingredients_text" : "E103 et E140",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e103, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/39.json b/tests/unit/expected_test_results/ingredients_preparsing/39.json
new file mode 100644
index 0000000000000..7038d2d08ea66
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/39.json
@@ -0,0 +1,6 @@
+{
+   "id" : "39",
+   "ingredients_text" : "E103 ET E140",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e103, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/4.json b/tests/unit/expected_test_results/ingredients_preparsing/4.json
new file mode 100644
index 0000000000000..2cf124aa06e14
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/4.json
@@ -0,0 +1,6 @@
+{
+   "id" : "4",
+   "ingredients_text" : "Vitamines (B1, B2, B6, PP)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "vitamines, vitamine B1, vitamine B2, vitamine B6, vitamine PP"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/40.json b/tests/unit/expected_test_results/ingredients_preparsing/40.json
new file mode 100644
index 0000000000000..0b0075e21322f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/40.json
@@ -0,0 +1,6 @@
+{
+   "id" : "40",
+   "ingredients_text" : "curcumine et E140",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "curcumine, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/41.json b/tests/unit/expected_test_results/ingredients_preparsing/41.json
new file mode 100644
index 0000000000000..690cd9d6e11ac
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/41.json
@@ -0,0 +1,6 @@
+{
+   "id" : "41",
+   "ingredients_text" : "E140 et tartrazine",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "e140, tartrazine"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/42.json b/tests/unit/expected_test_results/ingredients_preparsing/42.json
new file mode 100644
index 0000000000000..d239b2e29002c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/42.json
@@ -0,0 +1,6 @@
+{
+   "id" : "42",
+   "ingredients_text" : "Acide citrique, colorant : e120, vitamine C, E-500",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Acide citrique, colorant : e120, vitamine C, e500"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/43.json b/tests/unit/expected_test_results/ingredients_preparsing/43.json
new file mode 100644
index 0000000000000..7157a69a0dfad
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/43.json
@@ -0,0 +1,6 @@
+{
+   "id" : "43",
+   "ingredients_text" : "poudres à lever (carbonates acides d’ammonium et de sodium, acide citrique)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "poudres à lever (carbonates acides d'ammonium, carbonates acides de sodium, acide citrique)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/44.json b/tests/unit/expected_test_results/ingredients_preparsing/44.json
new file mode 100644
index 0000000000000..e64d6d2ddd594
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/44.json
@@ -0,0 +1,6 @@
+{
+   "id" : "44",
+   "ingredients_text" : "REAL SUGARCANE, SALT, ANTIOXIDANT (INS 300), INS 334, INS345",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "REAL SUGARCANE, SALT, ANTIOXIDANT (e300), e334, e345"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/45.json b/tests/unit/expected_test_results/ingredients_preparsing/45.json
new file mode 100644
index 0000000000000..1a0c8dcd7d819
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/45.json
@@ -0,0 +1,6 @@
+{
+   "id" : "45",
+   "ingredients_text" : "colores E (120, 124 y 125)",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "colores e120, e124, e125"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/46.json b/tests/unit/expected_test_results/ingredients_preparsing/46.json
new file mode 100644
index 0000000000000..f15351b7b72b7
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/46.json
@@ -0,0 +1,6 @@
+{
+   "id" : "46",
+   "ingredients_text" : "Leche desnatada de vaca, enzima lactasa y vitaminas A, D, E y ácido fólico.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Leche desnatada de vaca, enzima lactasa y vitaminas, vitamina A, vitamina D, vitamina E, ácido fólico."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/47.json b/tests/unit/expected_test_results/ingredients_preparsing/47.json
new file mode 100644
index 0000000000000..bc492b66403eb
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/47.json
@@ -0,0 +1,6 @@
+{
+   "id" : "47",
+   "ingredients_text" : "Leche desnatada, leche desnatada en polvo, zumo de lima, almidón de maíz, extracto de ginseng 0,19%, aromas, fermentos lácticos con Lcasei, colorante: caramelo natural, edulcorantes: sucralosa y acesulfamo K, estabilizante: goma xantana, vitaminas: D, B6, ácido fólico y B12 Origen de la feche. España. Preparación: Agitar antes de abrir.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Leche desnatada, leche desnatada en polvo, zumo de lima, almidón de maíz, extracto de ginseng 0.19%, aromas, fermentos lácticos con Lcasei, colorante: caramelo natural, edulcorantes: sucralosa y acesulfamo K, estabilizante: goma xantana, vitaminas, vitamina D, vitamina B6, ácido fólico, vitamina B12 Origen de la feche. España. Preparación: Agitar antes de abrir."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/48.json b/tests/unit/expected_test_results/ingredients_preparsing/48.json
new file mode 100644
index 0000000000000..ea85790985922
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/48.json
@@ -0,0 +1,6 @@
+{
+   "id" : "48",
+   "ingredients_text" : "edulcorantes (acesulfamo K y sucralosa) y vitaminas (riboflavina (vitamina B2) y cianocobalamina vitamina B12))",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "edulcorantes (acesulfamo K y sucralosa), vitaminas (riboflavina (vitamina B2), cianocobalamina vitamina B12))"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/49.json b/tests/unit/expected_test_results/ingredients_preparsing/49.json
new file mode 100644
index 0000000000000..54afd72d40aa3
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/49.json
@@ -0,0 +1,6 @@
+{
+   "id" : "49",
+   "ingredients_text" : "aceites vegetales [aceite de girasol (70%) y aceite de oliva virgen (30%)] y sal",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "aceites vegetales [aceite de girasol (70%), aceite de oliva virgen (30%)], sal"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/5.json b/tests/unit/expected_test_results/ingredients_preparsing/5.json
new file mode 100644
index 0000000000000..9ff98203572a0
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/5.json
@@ -0,0 +1,6 @@
+{
+   "id" : "5",
+   "ingredients_text" : "Huile de palme",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/50.json b/tests/unit/expected_test_results/ingredients_preparsing/50.json
new file mode 100644
index 0000000000000..4e9c31474344d
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/50.json
@@ -0,0 +1,6 @@
+{
+   "id" : "50",
+   "ingredients_text" : "Trazas de cacahuete, huevo y frutos de cáscara.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "Trazas : cacahuete, Trazas : huevo, Trazas : frutos de cáscara."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/51.json b/tests/unit/expected_test_results/ingredients_preparsing/51.json
new file mode 100644
index 0000000000000..8fb4433c74c57
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/51.json
@@ -0,0 +1,6 @@
+{
+   "id" : "51",
+   "ingredients_text" : "sal y acidulante (ácido cítrico). Puede contener trazas de cacahuete, huevo y frutos de cáscara.",
+   "lc" : "es",
+   "preparsed_ingredients_text" : "sal y acidulante (ácido cítrico). Trazas : cacahuete, Trazas : huevo, Trazas : frutos de cáscara."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/52.json b/tests/unit/expected_test_results/ingredients_preparsing/52.json
new file mode 100644
index 0000000000000..fbe184764419b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/52.json
@@ -0,0 +1,6 @@
+{
+   "id" : "52",
+   "ingredients_text" : "bl. a. inkl. mod. past. emulgator E322 E103, E140, E250 og E100",
+   "lc" : "da",
+   "preparsed_ingredients_text" : "blandt andet inklusive modificeret pasteuriserede emulgator e322, e103, e140, e250, e100"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/53.json b/tests/unit/expected_test_results/ingredients_preparsing/53.json
new file mode 100644
index 0000000000000..d6c44499bf79c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/53.json
@@ -0,0 +1,6 @@
+{
+   "id" : "53",
+   "ingredients_text" : "bl. a. inkl. E322 E103, E140, E250 og E100",
+   "lc" : "nb",
+   "preparsed_ingredients_text" : "blant annet inklusive e322, e103, e140, e250, e100"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/54.json b/tests/unit/expected_test_results/ingredients_preparsing/54.json
new file mode 100644
index 0000000000000..69c3a5d23431c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/54.json
@@ -0,0 +1,6 @@
+{
+   "id" : "54",
+   "ingredients_text" : "bl. a. förtjockn.medel inkl. emulgeringsmedel E322 E103, E140, E250 och E100",
+   "lc" : "sv",
+   "preparsed_ingredients_text" : "bland annat förtjockningsmedel inklusive emulgeringsmedel e322, e103, e140, e250, e100"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/55.json b/tests/unit/expected_test_results/ingredients_preparsing/55.json
new file mode 100644
index 0000000000000..ed3b75db1be72
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/55.json
@@ -0,0 +1,6 @@
+{
+   "id" : "55",
+   "ingredients_text" : "Vitaminer A, B og C. Vitaminer (B2, E, D), Hvede**. Indeholder mælk. Kan indeholde spor af soja, mælk, mandler og sesam. ** = Økologisk",
+   "lc" : "da",
+   "preparsed_ingredients_text" : "vitaminer, vitamin A, vitamin B, vitamin C. vitaminer, vitamin B2, vitamin E, vitamin D, Hvede Økologisk. Stoffer, eller produkter, som forårsager allergi eller overfølsomhed : mælk. Spor : soja, Spor : mælk, Spor : mandler, Spor : sesam."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/56.json b/tests/unit/expected_test_results/ingredients_preparsing/56.json
new file mode 100644
index 0000000000000..c5f76882c0908
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/56.json
@@ -0,0 +1,6 @@
+{
+   "id" : "56",
+   "ingredients_text" : "Vítamín (B2, E og D). Getur innihaldið hnetur, soja og mjólk í snefilmagni.",
+   "lc" : "is",
+   "preparsed_ingredients_text" : "vítamín, B2-vítamín, E-vítamín, D-vítamín. Leifar : hnetur, Leifar : soja, Leifar : mjólk."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/57.json b/tests/unit/expected_test_results/ingredients_preparsing/57.json
new file mode 100644
index 0000000000000..59ff28c936cac
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/57.json
@@ -0,0 +1,6 @@
+{
+   "id" : "57",
+   "ingredients_text" : "Vitaminer A, B og C. Vitaminer (B2, E, D). Kan inneholde spor av andre nøtter, soya og melk.",
+   "lc" : "nb",
+   "preparsed_ingredients_text" : "vitaminer, vitamin A, vitamin B, vitamin C. vitaminer, vitamin B2, vitamin E, vitamin D. Spor : andre nøtter, Spor : soya, Spor : melk."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/58.json b/tests/unit/expected_test_results/ingredients_preparsing/58.json
new file mode 100644
index 0000000000000..bd635526c3104
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/58.json
@@ -0,0 +1,6 @@
+{
+   "id" : "58",
+   "ingredients_text" : "Vitaminer (B2, E och D), Vete*. Innehåller hasselnötter. Kan innehålla spår av råg, jordnötter, mandel, hasselnötter, cashewnötter och valnötter. *Ekologisk",
+   "lc" : "sv",
+   "preparsed_ingredients_text" : "vitaminer, vitamin B2, vitamin E, vitamin D, Vete Ekologisk. ämnen eller produkter som orsakar allergi eller intolerans : hasselnötter. Spår : råg, Spår : jordnötter, Spår : mandel, Spår : hasselnötter, Spår : cashewnötter, Spår : valnötter."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/59.json b/tests/unit/expected_test_results/ingredients_preparsing/59.json
new file mode 100644
index 0000000000000..f3eca65f502e6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/59.json
@@ -0,0 +1,6 @@
+{
+   "id" : "59",
+   "ingredients_text" : "Vitamiinit A, B ja C",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "vitamiinit, A-vitamiini, B-vitamiini, C-vitamiini"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/6.json b/tests/unit/expected_test_results/ingredients_preparsing/6.json
new file mode 100644
index 0000000000000..ed1fb9c7eff84
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/6.json
@@ -0,0 +1,6 @@
+{
+   "id" : "6",
+   "ingredients_text" : "Huile (palme)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/60.json b/tests/unit/expected_test_results/ingredients_preparsing/60.json
new file mode 100644
index 0000000000000..8eb8d6ab4ffd3
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/60.json
@@ -0,0 +1,6 @@
+{
+   "id" : "60",
+   "ingredients_text" : "Vitamiinit (B1, B2, B6)",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "vitamiinit, B1-vitamiini, B2-vitamiini, B6-vitamiini"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/61.json b/tests/unit/expected_test_results/ingredients_preparsing/61.json
new file mode 100644
index 0000000000000..9c56362d0bf72
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/61.json
@@ -0,0 +1,6 @@
+{
+   "id" : "61",
+   "ingredients_text" : "mansikat 30%",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "mansikat 30%"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/62.json b/tests/unit/expected_test_results/ingredients_preparsing/62.json
new file mode 100644
index 0000000000000..edca18e75ce47
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/62.json
@@ -0,0 +1,6 @@
+{
+   "id" : "62",
+   "ingredients_text" : "sakeuttamisaine pektiini",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "sakeuttamisaine : pektiini"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/63.json b/tests/unit/expected_test_results/ingredients_preparsing/63.json
new file mode 100644
index 0000000000000..6dfdf0159ca67
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/63.json
@@ -0,0 +1,6 @@
+{
+   "id" : "63",
+   "ingredients_text" : "sakeuttamisaine (pektiini)",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "sakeuttamisaine (pektiini)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/64.json b/tests/unit/expected_test_results/ingredients_preparsing/64.json
new file mode 100644
index 0000000000000..2f420107e411f
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/64.json
@@ -0,0 +1,6 @@
+{
+   "id" : "64",
+   "ingredients_text" : "jauhonparanne (askorbiinihappo)",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "jauhonparanne (askorbiinihappo)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/65.json b/tests/unit/expected_test_results/ingredients_preparsing/65.json
new file mode 100644
index 0000000000000..4734ba98fc438
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/65.json
@@ -0,0 +1,6 @@
+{
+   "id" : "65",
+   "ingredients_text" : "E250-E251",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e250 - e251"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/66.json b/tests/unit/expected_test_results/ingredients_preparsing/66.json
new file mode 100644
index 0000000000000..08732f2f1dbe6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/66.json
@@ -0,0 +1,6 @@
+{
+   "id" : "66",
+   "ingredients_text" : "E250-E251-E260",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e250 - e251 - e260"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/67.json b/tests/unit/expected_test_results/ingredients_preparsing/67.json
new file mode 100644
index 0000000000000..600a4b31c5353
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/67.json
@@ -0,0 +1,6 @@
+{
+   "id" : "67",
+   "ingredients_text" : "E 250b-E251-e.260(ii)",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e250b - e251 - e260ii"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/68.json b/tests/unit/expected_test_results/ingredients_preparsing/68.json
new file mode 100644
index 0000000000000..adeade2f580c6
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/68.json
@@ -0,0 +1,6 @@
+{
+   "id" : "68",
+   "ingredients_text" : "E100 E122",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e100, e122"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/69.json b/tests/unit/expected_test_results/ingredients_preparsing/69.json
new file mode 100644
index 0000000000000..ae1dcba75aed8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/69.json
@@ -0,0 +1,6 @@
+{
+   "id" : "69",
+   "ingredients_text" : "E103 ja E140",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e103, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/7.json b/tests/unit/expected_test_results/ingredients_preparsing/7.json
new file mode 100644
index 0000000000000..8744a6c172f4c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/7.json
@@ -0,0 +1,6 @@
+{
+   "id" : "7",
+   "ingredients_text" : "Huile (palme, colza)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme, Huile de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/70.json b/tests/unit/expected_test_results/ingredients_preparsing/70.json
new file mode 100644
index 0000000000000..392d502559773
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/70.json
@@ -0,0 +1,6 @@
+{
+   "id" : "70",
+   "ingredients_text" : "E103 JA E140",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e103, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/71.json b/tests/unit/expected_test_results/ingredients_preparsing/71.json
new file mode 100644
index 0000000000000..38269a1dcb8df
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/71.json
@@ -0,0 +1,6 @@
+{
+   "id" : "71",
+   "ingredients_text" : "kurkumiini ja E140",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "kurkumiini, e140"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/72.json b/tests/unit/expected_test_results/ingredients_preparsing/72.json
new file mode 100644
index 0000000000000..f442c7f012a15
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/72.json
@@ -0,0 +1,6 @@
+{
+   "id" : "72",
+   "ingredients_text" : "E140 ja karoteeni",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "e140, karoteeni"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/73.json b/tests/unit/expected_test_results/ingredients_preparsing/73.json
new file mode 100644
index 0000000000000..a64c7382e0107
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/73.json
@@ -0,0 +1,6 @@
+{
+   "id" : "73",
+   "ingredients_text" : "omenamehu, vesi, sokeri. jossa käsitellään myös maitoa.",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "omenamehu, vesi, sokeri. Jäämät : maitoa."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/74.json b/tests/unit/expected_test_results/ingredients_preparsing/74.json
new file mode 100644
index 0000000000000..427a631abc46e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/74.json
@@ -0,0 +1,6 @@
+{
+   "id" : "74",
+   "ingredients_text" : "omenamehu, vesi, sokeri. Saattaa sisältää pieniä määriä selleriä, sinappia ja vehnää.",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "omenamehu, vesi, sokeri. Jäämät : selleriä, Jäämät : sinappia, Jäämät : vehnää."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/75.json b/tests/unit/expected_test_results/ingredients_preparsing/75.json
new file mode 100644
index 0000000000000..420614058e3a4
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/75.json
@@ -0,0 +1,6 @@
+{
+   "id" : "75",
+   "ingredients_text" : "omenamehu, vesi, sokeri. Saattaa sisältää pienehköjä määriä selleriä, sinappia ja vehnää.",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "omenamehu, vesi, sokeri. Jäämät : selleriä, Jäämät : sinappia, Jäämät : vehnää."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/76.json b/tests/unit/expected_test_results/ingredients_preparsing/76.json
new file mode 100644
index 0000000000000..db9910bf01589
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/76.json
@@ -0,0 +1,6 @@
+{
+   "id" : "76",
+   "ingredients_text" : "luomurypsiöljy, luomu kaura, vihreä luomutee",
+   "lc" : "fi",
+   "preparsed_ingredients_text" : "luomu rypsiöljy, luomu kaura, vihreä luomu tee"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/77.json b/tests/unit/expected_test_results/ingredients_preparsing/77.json
new file mode 100644
index 0000000000000..3c0a5b48907e3
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/77.json
@@ -0,0 +1,6 @@
+{
+   "id" : "77",
+   "ingredients_text" : "arôme naturel de citron-citron vert et d'autres agrumes",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "arôme naturel de citron, arôme naturel de citron vert, arôme naturel d'agrumes"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/78.json b/tests/unit/expected_test_results/ingredients_preparsing/78.json
new file mode 100644
index 0000000000000..ac852977a62c8
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/78.json
@@ -0,0 +1,6 @@
+{
+   "id" : "78",
+   "ingredients_text" : "arômes naturels de citron et de limette",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "arômes naturels de citron, arômes naturels de limette"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/79.json b/tests/unit/expected_test_results/ingredients_preparsing/79.json
new file mode 100644
index 0000000000000..a6ae1a198d655
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/79.json
@@ -0,0 +1,6 @@
+{
+   "id" : "79",
+   "ingredients_text" : "arôme naturel de pomme avec d'autres arômes naturels",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "arôme naturel de pomme et arômes naturels"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/8.json b/tests/unit/expected_test_results/ingredients_preparsing/8.json
new file mode 100644
index 0000000000000..945d751f82cb1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/8.json
@@ -0,0 +1,6 @@
+{
+   "id" : "8",
+   "ingredients_text" : "Huile (palme et colza)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huile de palme, Huile de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/80.json b/tests/unit/expected_test_results/ingredients_preparsing/80.json
new file mode 100644
index 0000000000000..b9b4a73e7bb20
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/80.json
@@ -0,0 +1,6 @@
+{
+   "id" : "80",
+   "ingredients_text" : "jus de pomme, eau, sucre. Traces de lait.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "jus de pomme, eau, sucre. Traces éventuelles : lait."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/81.json b/tests/unit/expected_test_results/ingredients_preparsing/81.json
new file mode 100644
index 0000000000000..42cf60d2e0a0e
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/81.json
@@ -0,0 +1,6 @@
+{
+   "id" : "81",
+   "ingredients_text" : "jus de pomme, eau, sucre. Traces possibles de céleri, moutarde et gluten.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "jus de pomme, eau, sucre. Traces éventuelles : céleri, Traces éventuelles : moutarde, Traces éventuelles : gluten."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/82.json b/tests/unit/expected_test_results/ingredients_preparsing/82.json
new file mode 100644
index 0000000000000..14ebc077359be
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/82.json
@@ -0,0 +1,6 @@
+{
+   "id" : "82",
+   "ingredients_text" : "jus de pomme, eau, sucre. Traces possibles de céleri, de moutarde et gluten.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "jus de pomme, eau, sucre. Traces éventuelles : céleri, Traces éventuelles : moutarde, Traces éventuelles : gluten."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/83.json b/tests/unit/expected_test_results/ingredients_preparsing/83.json
new file mode 100644
index 0000000000000..0af8fa6f30efd
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/83.json
@@ -0,0 +1,6 @@
+{
+   "id" : "83",
+   "ingredients_text" : "Traces de moutarde",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Traces éventuelles : moutarde."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/84.json b/tests/unit/expected_test_results/ingredients_preparsing/84.json
new file mode 100644
index 0000000000000..c89caaf574c00
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/84.json
@@ -0,0 +1,6 @@
+{
+   "id" : "84",
+   "ingredients_text" : "Sucre de canne Traces éventuelles d'oeufs",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Sucre de canne, Traces éventuelles : oeufs."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/85.json b/tests/unit/expected_test_results/ingredients_preparsing/85.json
new file mode 100644
index 0000000000000..8151c108065cc
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/85.json
@@ -0,0 +1,6 @@
+{
+   "id" : "85",
+   "ingredients_text" : "huile végétale de tournesol et/ou colza",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile végétale de tournesol, huile végétale de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/86.json b/tests/unit/expected_test_results/ingredients_preparsing/86.json
new file mode 100644
index 0000000000000..fcf3103b84a34
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/86.json
@@ -0,0 +1,6 @@
+{
+   "id" : "86",
+   "ingredients_text" : "Zucker. Kann Spuren von Sellerie.",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Zucker. Spuren : Sellerie."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/87.json b/tests/unit/expected_test_results/ingredients_preparsing/87.json
new file mode 100644
index 0000000000000..9af47e34fdfd1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/87.json
@@ -0,0 +1,6 @@
+{
+   "id" : "87",
+   "ingredients_text" : "Zucker. Kann Spuren von Senf und Sellerie.",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Zucker. Spuren : Senf, Spuren : Sellerie."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/88.json b/tests/unit/expected_test_results/ingredients_preparsing/88.json
new file mode 100644
index 0000000000000..29e4a7332a642
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/88.json
@@ -0,0 +1,6 @@
+{
+   "id" : "88",
+   "ingredients_text" : "Zucker. Kann Spuren von Senf und Sellerie enthalten",
+   "lc" : "de",
+   "preparsed_ingredients_text" : "Zucker. Spuren : Senf, Spuren : Sellerie."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/89.json b/tests/unit/expected_test_results/ingredients_preparsing/89.json
new file mode 100644
index 0000000000000..c47da40a697af
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/89.json
@@ -0,0 +1,6 @@
+{
+   "id" : "89",
+   "ingredients_text" : "Puo contenere tracce di frutta a guscio, sesamo, soia e uova",
+   "lc" : "it",
+   "preparsed_ingredients_text" : "Tracce : frutta a guscio, Tracce : sesamo, Tracce : soia, Tracce : uova."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/9.json b/tests/unit/expected_test_results/ingredients_preparsing/9.json
new file mode 100644
index 0000000000000..83c8636e99105
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/9.json
@@ -0,0 +1,6 @@
+{
+   "id" : "9",
+   "ingredients_text" : "Huiles végétales de palme et de colza",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Huiles végétales de palme, Huiles végétales de colza"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/90.json b/tests/unit/expected_test_results/ingredients_preparsing/90.json
new file mode 100644
index 0000000000000..11f249aa8705b
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/90.json
@@ -0,0 +1,6 @@
+{
+   "id" : "90",
+   "ingredients_text" : "Il prodotto può contenere tracce di GRANO, LATTE, UOVA, FRUTTA A GUSCIO e SOIA.",
+   "lc" : "it",
+   "preparsed_ingredients_text" : "Tracce : GRANO, Tracce : LATTE, Tracce : UOVA, Tracce : FRUTTA A GUSCIO, Tracce : SOIA."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/91.json b/tests/unit/expected_test_results/ingredients_preparsing/91.json
new file mode 100644
index 0000000000000..5b76e9b483ca5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/91.json
@@ -0,0 +1,6 @@
+{
+   "id" : "91",
+   "ingredients_text" : "Jus de pomme*** 68%, jus de poire***32% *** Ingrédients issus de l'agriculture biologique",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Jus de pomme Bio 68%, jus de poire Bio 32%"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/92.json b/tests/unit/expected_test_results/ingredients_preparsing/92.json
new file mode 100644
index 0000000000000..560411a8c6f77
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/92.json
@@ -0,0 +1,6 @@
+{
+   "id" : "92",
+   "ingredients_text" : "Pâte de cacao°* du Pérou 65 %, sucre de canne°*, beurre de cacao°*, sel *, lait °. °Issus de l'agriculture biologique (100 %). *Issus du commerce équitable (100 % du poids total avec 93 % SPP).",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Pâte de cacao Bio Commerce équitable du Pérou 65 %, sucre de canne Bio Commerce équitable, beurre de cacao Bio Commerce équitable, sel Commerce équitable, lait Bio."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/93.json b/tests/unit/expected_test_results/ingredients_preparsing/93.json
new file mode 100644
index 0000000000000..4cdec0f0e6e3c
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/93.json
@@ -0,0 +1,6 @@
+{
+   "id" : "93",
+   "ingredients_text" : "pâte de cacao* de Madagascar 75%, sucre de canne*, beurre de cacao*. * issus du commerce équitable et de l'agriculture biologique (100% du poids total).",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "pâte de cacao Commerce équitable Bio de Madagascar 75%, sucre de canne Commerce équitable Bio, beurre de cacao Commerce équitable Bio."
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/94.json b/tests/unit/expected_test_results/ingredients_preparsing/94.json
new file mode 100644
index 0000000000000..f5158d1e867f5
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/94.json
@@ -0,0 +1,6 @@
+{
+   "id" : "94",
+   "ingredients_text" : "Céleri - rave 21% - Eau, légumes 33,6% (carottes, céleri - rave, poivrons rouges 5,8% - haricots - petits pois bio - haricots verts - courge - radis, pommes de terre - patates - fenouil - cerfeuil tubéreux - persil plat)",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Céleri-rave 21% - Eau, légumes 33.6% (carottes, céleri-rave, poivrons rouges 5.8% - haricots - petits pois bio - haricots verts - courge - radis, pommes de terre - patates - fenouil - cerfeuil tubéreux - persil plat)"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/95.json b/tests/unit/expected_test_results/ingredients_preparsing/95.json
new file mode 100644
index 0000000000000..36079cd4f52fb
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/95.json
@@ -0,0 +1,6 @@
+{
+   "id" : "95",
+   "ingredients_text" : "poudres à lever : carbonates d'ammonium - carbonates de sodium - phosphates de calcium, farine, sel",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "poudres à lever : carbonates d'ammonium - carbonates de sodium - phosphates de calcium, farine, sel"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/96.json b/tests/unit/expected_test_results/ingredients_preparsing/96.json
new file mode 100644
index 0000000000000..381bc966d1307
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/96.json
@@ -0,0 +1,6 @@
+{
+   "id" : "96",
+   "ingredients_text" : "FD&C Red #40 Lake and silicon dioxide",
+   "lc" : "en",
+   "preparsed_ingredients_text" : "FD&C Red #40 Lake and silicon dioxide"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/97.json b/tests/unit/expected_test_results/ingredients_preparsing/97.json
new file mode 100644
index 0000000000000..7a14e0d154e05
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/97.json
@@ -0,0 +1,6 @@
+{
+   "id" : "97",
+   "ingredients_text" : "Lait pasteurisé à 1,1% de Mat. Gr.",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "Lait pasteurisé à 1.1% de Matières Grasses"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/98.json b/tests/unit/expected_test_results/ingredients_preparsing/98.json
new file mode 100644
index 0000000000000..2fe0d4dfff0aa
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/98.json
@@ -0,0 +1,6 @@
+{
+   "id" : "98",
+   "ingredients_text" : "matière grasse végétale (palme) raffinée",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "matière grasse végétale de palme raffinée"
+}
diff --git a/tests/unit/expected_test_results/ingredients_preparsing/99.json b/tests/unit/expected_test_results/ingredients_preparsing/99.json
new file mode 100644
index 0000000000000..6eb51d77789b1
--- /dev/null
+++ b/tests/unit/expected_test_results/ingredients_preparsing/99.json
@@ -0,0 +1,6 @@
+{
+   "id" : "99",
+   "ingredients_text" : "huile d'olive vierge, origan",
+   "lc" : "fr",
+   "preparsed_ingredients_text" : "huile d'olive vierge, origan"
+}
diff --git a/tests/unit/ingredients_parsing.t b/tests/unit/ingredients_parsing.t
deleted file mode 100755
index 79bf6f53029fd..0000000000000
--- a/tests/unit/ingredients_parsing.t
+++ /dev/null
@@ -1,675 +0,0 @@
-#!/usr/bin/perl -w
-
-# Tests of Ingredients::preparse_ingredients_text()
-
-use Modern::Perl '2017';
-use utf8;
-
-use Test2::V0;
-use Log::Any::Adapter 'TAP';
-
-use ProductOpener::Products qw/:all/;
-use ProductOpener::Tags qw/:all/;
-use ProductOpener::Ingredients
-	qw/normalize_a_of_b normalize_enumeration preparse_ingredients_text separate_additive_class/;
-
-#use Log::Any::Adapter 'TAP', filter => "none";
-
-is(normalize_a_of_b("en", "oil", "olive", 1), "olive oil");
-is(normalize_a_of_b("es", "aceta", "oliva", 1), "aceta de oliva");
-is(normalize_a_of_b("fr", "huile végétale", "olive", 1), "huile végétale d'olive");
-
-is(normalize_enumeration("en", "phosphates", "calcium and sodium", 1), "calcium phosphates, sodium phosphates");
-is(normalize_enumeration("en", "vegetal oil", "sunflower, palm", 1), "sunflower vegetal oil, palm vegetal oil");
-is(normalize_enumeration("fr", "huile", "colza, tournesol et olive", 1),
-	"huile de colza, huile de tournesol, huile d'olive");
-
-is(separate_additive_class("fr", "colorant", " ", "", "naturel"), "colorant ");
-is(separate_additive_class("fr", "colorant", " ", "", "carmins"), "colorant : ");
-is(separate_additive_class("fr", "colorant", " ", "", "E120, sel"), "colorant : ");
-is(separate_additive_class("fr", "colorant", " ", "", "E120 et E150b"), "colorant : ");
-is(separate_additive_class("fr", "colorant", " ", "", "caramel au sulfite d'ammonium"), "colorant : ");
-is(separate_additive_class("fr", "colorant", " ", "", "caramel au sulfite d'ammonium et rocou"), "colorant : ");
-
-my @lists = (
-
-	["fr", "Sel marin, blé, lécithine de soja", "Sel marin, blé, lécithine de soja"],
-	["fr", "Vitamine A", "Vitamine A"],
-	["fr", "Vitamines A, B et C", "Vitamines, Vitamine A, Vitamine B, Vitamine C"],
-	["fr", "Vitamines (B1, B2, B6, PP)", "Vitamines, Vitamine B1, Vitamine B2, Vitamine B6, Vitamine PP"],
-	["fr", "Huile de palme", "Huile de palme"],
-	["fr", "Huile (palme)", "Huile de palme"],
-	["fr", "Huile (palme, colza)", "Huile de palme, Huile de colza"],
-	["fr", "Huile (palme et colza)", "Huile de palme, Huile de colza"],
-	["fr", "Huiles végétales de palme et de colza", "Huiles végétales de palme, Huiles végétales de colza"],
-	["fr", "Huiles végétales de palme et d'olive", "Huiles végétales de palme, Huiles végétales d'olive"],
-	[
-		"fr",
-		"Huiles végétales de palme, de colza et de tournesol",
-		"Huiles végétales de palme, Huiles végétales de colza, huiles végétales de tournesol"
-	],
-	[
-		"fr",
-		"Huiles végétales de palme, de colza, de tournesol",
-		"Huiles végétales de palme, Huiles végétales de colza, huiles végétales de tournesol"
-	],
-	[
-		"fr",
-		"Huiles végétales de palme, de colza et d'olive en proportion variable",
-		"Huiles végétales de palme, Huiles végétales de colza, huiles végétales d'olive"
-	],
-	[
-		"fr",
-		"Huiles végétales de palme, de colza et d'olive",
-		"Huiles végétales de palme, Huiles végétales de colza, huiles végétales d'olive"
-	],
-	["fr", "phosphate et sulfate de calcium", "phosphate de calcium, sulfate de calcium"],
-	["fr", "sulfates de calcium et potassium", "sulfates de calcium, sulfates de potassium"],
-	["fr", "chlorures (sodium et potassium)", "chlorures de sodium, chlorures de potassium"],
-	["fr", "chlorures (sodium, potassium)", "chlorures de sodium, chlorures de potassium"],
-	["fr", "fraises 30%", "fraises 30%"],
-	[
-		"fr",
-		"Marmelade d'oranges 41% (sirop de glucose-fructose, sucre, pulpe d'orange 4.5%, jus d'orange concentré 1.4% (équivalent jus d'orange 7.8%), pulpe d'orange concentrée 0.6% (équivalent pulpe d'orange 2.6%), gélifiant (pectines), acidifiant (acide citrique), correcteurs d'acidité (citrate de calcium, citrate de sodium), arôme naturel d'orange, épaississant (gomme xanthane)), chocolat 24.9% (sucre, pâte de cacao, beurre de cacao, graisses végétales (illipe, mangue, sal, karité et palme en proportions variables), arôme, émulsifiant (lécithine de soja), lactose et protéines de lait), farine de blé, sucre, oeufs, sirop de glucose-fructose, huile de colza, poudre à lever (carbonate acide d'ammonium, diphosphate disodique, carbonate acide de sodium), sel, émulsifiant (lécithine de soja).",
-		"Marmelade d'oranges 41% (sirop de glucose-fructose, sucre, pulpe d'orange 4.5%, jus d'orange concentré 1.4% (équivalent jus d'orange 7.8%), pulpe d'orange concentrée 0.6% (équivalent pulpe d'orange 2.6%), gélifiant (pectines), acidifiant (acide citrique), correcteurs d'acidité (citrate de calcium, citrate de sodium), arôme naturel d'orange, épaississant (gomme xanthane)), chocolat 24.9% (sucre, pâte de cacao, beurre de cacao, graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de sal, graisses végétales de karité, graisses végétales de palme, arôme, émulsifiant (lécithine de soja), lactose et protéines de lait), farine de blé, sucre, oeufs, sirop de glucose-fructose, huile de colza, poudre à lever (carbonate acide d'ammonium, diphosphate disodique, carbonate acide de sodium), sel, émulsifiant (lécithine de soja)."
-	],
-	[
-		"fr",
-		"graisses végétales (illipe, mangue, sal, karité et palme en proportions variables)",
-		"graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de sal, graisses végétales de karité, graisses végétales de palme"
-	],
-	[
-		"fr",
-		"graisses végétales (illipe, mangue, palme)",
-		"graisses végétales d'illipe, graisses végétales de mangue, graisses végétales de palme"
-	],
-	["fr", "graisses végétales (illipe)", "graisses végétales d'illipe"],
-	["fr", "graisses végétales (illipe et sal)", "graisses végétales d'illipe, graisses végétales de sal"],
-	["fr", "gélifiant pectine", "gélifiant : pectine"],
-	["fr", "gélifiant (pectine)", "gélifiant (pectine)"],
-	[
-		"fr", "agent de traitement de la farine (acide ascorbique)",
-		"agent de traitement de la farine (acide ascorbique)"
-	],
-	["fr", "lait demi-écrémé", "lait demi-écrémé"],
-	["fr", "Saveur vanille : lait demi-écrémé 77%, sucre", "Saveur vanille : lait demi-écrémé 77%, sucre"],
-	["fr", "colorants alimentaires E (124,122,133,104,110)", "colorants alimentaires : E124, E122, E133, E104, E110"],
-	["fr", "INS 240,241,242b", "E240, E241, E242b"],
-	["fr", "colorants E (124, 125, 120 et 122", "colorants : E124, E125, E120, E122"],
-	["fr", "E250-E251", "E250 - E251"],
-	["fr", "E250-E251-E260", "E250 - E251 - E260"],
-	["fr", "E 250b-E251-e.260(ii)", "E250b - E251 - E260ii"],
-	[
-		"fr",
-		"émulsifiants : E463, E432 et E472 - correcteurs d'acidité : E322/E333 E474-E475",
-		"émulsifiants : e463, e432, e472 - correcteurs d'acidité : e322/e333, e474 - e475"
-	],
-	["fr", "E100 E122", "E100, E122"],
-	["fr", "E103 et E140", "E103, E140"],
-	["fr", "E103 ET E140", "E103, E140"],
-	["fr", "curcumine et E140", "curcumine, E140"],
-	["fr", "E140 et tartrazine", "E140, tartrazine"],
-	["fr", "Acide citrique, colorant : e120, vitamine C, E-500", "Acide citrique, colorant : e120, vitamine C, E500"],
-	[
-		"fr",
-		"poudres à lever (carbonates acides d’ammonium et de sodium, acide citrique)",
-		"poudres à lever (carbonates acides d'ammonium, carbonates acides de sodium, acide citrique)"
-	],
-
-	[
-		"en",
-		"REAL SUGARCANE, SALT, ANTIOXIDANT (INS 300), INS 334, INS345",
-		"REAL SUGARCANE, SALT, ANTIOXIDANT (e300), e334, e345"
-	],
-
-	["es", "colores E (120, 124 y 125)", "colores E120, E124, E125"],
-	[
-		"es",
-		"Leche desnatada de vaca, enzima lactasa y vitaminas A, D, E y ácido fólico.",
-		"Leche desnatada de vaca, enzima lactasa y vitaminas, vitamina A, vitamina D, vitamina E, ácido fólico."
-	],
-	[
-		"es",
-		"Leche desnatada, leche desnatada en polvo, zumo de lima, almidón de maíz, extracto de ginseng 0,19%, aromas, fermentos lácticos con Lcasei, colorante: caramelo natural, edulcorantes: sucralosa y acesulfamo K, estabilizante: goma xantana, vitaminas: D, B6, ácido fólico y B12 Origen de la feche. España. Preparación: Agitar antes de abrir.",
-		"Leche desnatada, leche desnatada en polvo, zumo de lima, almidón de maíz, extracto de ginseng 0.19%, aromas, fermentos lácticos con Lcasei, colorante: caramelo natural, edulcorantes: sucralosa y acesulfamo K, estabilizante: goma xantana, vitaminas, vitamina D, vitamina B6, ácido fólico, vitamina B12 Origen de la feche. España. Preparación: Agitar antes de abrir."
-	],
-	[
-		"es",
-		"edulcorantes (acesulfamo K y sucralosa) y vitaminas (riboflavina (vitamina B2) y cianocobalamina vitamina B12))",
-		"edulcorantes (acesulfamo K y sucralosa), vitaminas (riboflavina (vitamina B2), cianocobalamina vitamina B12))"
-	],
-	[
-		"es",
-		"aceites vegetales [aceite de girasol (70%) y aceite de oliva virgen (30%)] y sal",
-		"aceites vegetales [aceite de girasol (70%), aceite de oliva virgen (30%)], sal"
-	],
-	[
-		"es",
-		"Trazas de cacahuete, huevo y frutos de cáscara.",
-		"Trazas : cacahuete, Trazas : huevo, Trazas : frutos de cáscara."
-	],
-	[
-		"es",
-		"sal y acidulante (ácido cítrico). Puede contener trazas de cacahuete, huevo y frutos de cáscara.",
-		"sal y acidulante (ácido cítrico). Trazas : cacahuete, Trazas : huevo, Trazas : frutos de cáscara."
-	],
-
-	###########################
-	# SCANDINAVIAN LANGUAGES  #
-	###########################
-	[
-		"da",
-		"bl. a. inkl. mod. past. emulgator E322 E103, E140, E250 og E100",
-		"blandt andet inklusive modificeret pasteuriserede emulgator E322, E103, E140, E250, E100"
-	],
-	["nb", "bl. a. inkl. E322 E103, E140, E250 og E100", "blant annet inklusive E322, E103, E140, E250, E100"],
-	[
-		"sv",
-		"bl. a. förtjockn.medel inkl. emulgeringsmedel E322 E103, E140, E250 och E100",
-		"bland annat förtjockningsmedel inklusive emulgeringsmedel E322, E103, E140, E250, E100"
-	],
-	[
-		"da",
-		"Vitaminer A, B og C. Vitaminer (B2, E, D), Hvede**. Indeholder mælk. Kan indeholde spor af soja, mælk, mandler og sesam. ** = Økologisk",
-		"Vitaminer, Vitamin A, Vitamin B, Vitamin C. Vitaminer, Vitamin B2, Vitamin E, Vitamin D, Hvede Økologisk. Stoffer, eller produkter, som forårsager allergi eller overfølsomhed : mælk. Spor : soja, Spor : mælk, Spor : mandler, Spor : sesam."
-	],
-	[
-		"is",
-		"Vítamín (B2, E og D). Getur innihaldið hnetur, soja og mjólk í snefilmagni.",
-		"Vítamín, B2-Vítamín, E-Vítamín, D-Vítamín. Leifar : hnetur, Leifar : Soja, Leifar : mjólk."
-	],
-	[
-		"nb",
-		"Vitaminer A, B og C. Vitaminer (B2, E, D). Kan inneholde spor av andre nøtter, soya og melk.",
-		"Vitaminer, Vitamin A, Vitamin B, Vitamin C. Vitaminer, Vitamin B2, Vitamin E, Vitamin D. Spor : andre nøtter, Spor : soya, Spor : melk."
-	],
-	[
-		"sv",
-		"Vitaminer (B2, E och D), Vete*. Innehåller hasselnötter. Kan innehålla spår av råg, jordnötter, mandel, hasselnötter, cashewnötter och valnötter. *Ekologisk",
-		"Vitaminer, Vitamin B2, Vitamin E, Vitamin D, Vete Ekologisk. Ämnen eller produkter som orsakar allergi eller intolerans : hasselnötter. Spår : råg, Spår : jordnötter, Spår : mandel, Spår : hasselnötter, Spår : cashewnötter, Spår : valnötter."
-	],
-	###########################
-
-	["fi", "Vitamiinit A, B ja C", "Vitamiinit, A-Vitamiini, B-Vitamiini, C-Vitamiini"],
-	["fi", "Vitamiinit (B1, B2, B6)", "Vitamiinit, B1-Vitamiini, B2-Vitamiini, B6-Vitamiini"],
-	["fi", "mansikat 30%", "mansikat 30%"],
-	["fi", "sakeuttamisaine pektiini", "sakeuttamisaine : pektiini"],
-	["fi", "sakeuttamisaine (pektiini)", "sakeuttamisaine (pektiini)"],
-	["fi", "jauhonparanne (askorbiinihappo)", "jauhonparanne (askorbiinihappo)"],
-	["fi", "E250-E251", "E250 - E251"],
-	["fi", "E250-E251-E260", "E250 - E251 - E260"],
-	["fi", "E 250b-E251-e.260(ii)", "E250b - E251 - E260ii"],
-	["fi", "E100 E122", "E100, E122"],
-	["fi", "E103 ja E140", "E103, E140"],
-	["fi", "E103 JA E140", "E103, E140"],
-	["fi", "kurkumiini ja E140", "kurkumiini, E140"],
-	["fi", "E140 ja karoteeni", "E140, karoteeni"],
-	["fi", "omenamehu, vesi, sokeri. jossa käsitellään myös maitoa.", "omenamehu, vesi, sokeri. jäämät : maitoa."],
-	[
-		"fi",
-		"omenamehu, vesi, sokeri. Saattaa sisältää pieniä määriä selleriä, sinappia ja vehnää.",
-		"omenamehu, vesi, sokeri. jäämät : selleriä, jäämät : sinappia, jäämät : vehnää."
-	],
-	[
-		"fi",
-		"omenamehu, vesi, sokeri. Saattaa sisältää pienehköjä määriä selleriä, sinappia ja vehnää.",
-		"omenamehu, vesi, sokeri. jäämät : selleriä, jäämät : sinappia, jäämät : vehnää."
-	],
-	["fi", "luomurypsiöljy, luomu kaura, vihreä luomutee", "luomu rypsiöljy, luomu kaura, vihreä luomu tee"],
-
-	[
-		"fr",
-		"arôme naturel de citron-citron vert et d'autres agrumes",
-		"arôme naturel de citron, arôme naturel de citron vert, arôme naturel d'agrumes"
-	],
-	["fr", "arômes naturels de citron et de limette", "arômes naturels de citron, arômes naturels de limette"],
-	["fr", "arôme naturel de pomme avec d'autres arômes naturels", "arôme naturel de pomme et arômes naturels"],
-	["fr", "jus de pomme, eau, sucre. Traces de lait.", "jus de pomme, eau, sucre. traces éventuelles : lait."],
-	[
-		"fr",
-		"jus de pomme, eau, sucre. Traces possibles de céleri, moutarde et gluten.",
-		"jus de pomme, eau, sucre. Traces éventuelles : céleri, Traces éventuelles : moutarde, Traces éventuelles : gluten."
-	],
-	[
-		"fr",
-		"jus de pomme, eau, sucre. Traces possibles de céleri, de moutarde et gluten.",
-		"jus de pomme, eau, sucre. Traces éventuelles : céleri, Traces éventuelles : moutarde, Traces éventuelles : gluten."
-	],
-	["fr", "Traces de moutarde", "traces éventuelles : moutarde."],
-	["fr", "Sucre de canne Traces éventuelles d'oeufs", "Sucre de canne, Traces éventuelles : oeufs."],
-	["fr", "huile végétale de tournesol et/ou colza", "huile végétale de tournesol, huile végétale de colza"],
-
-	["de", "Zucker. Kann Spuren von Sellerie.", "zucker. spuren : sellerie."],
-	["de", "Zucker. Kann Spuren von Senf und Sellerie.", "zucker. spuren : senf, spuren : sellerie."],
-	["de", "Zucker. Kann Spuren von Senf und Sellerie enthalten", "zucker. spuren : senf, spuren : sellerie."],
-
-	[
-		"it",
-		"Puo contenere tracce di frutta a guscio, sesamo, soia e uova",
-		"tracce : frutta a guscio, tracce : sesamo, tracce : soia, tracce : uova."
-	],
-	[
-		"it",
-		"Il prodotto può contenere tracce di GRANO, LATTE, UOVA, FRUTTA A GUSCIO e SOIA.",
-		"tracce : grano, tracce : latte, tracce : uova, tracce : frutta a guscio, tracce : soia."
-	],
-
-	[
-		"fr",
-		"Jus de pomme*** 68%, jus de poire***32% *** Ingrédients issus de l'agriculture biologique",
-		"jus de pomme bio 68%, jus de poire bio 32%"
-	],
-	[
-		"fr",
-		"Pâte de cacao°* du Pérou 65 %, sucre de canne°*, beurre de cacao°*, sel *, lait °. °Issus de l'agriculture biologique (100 %). *Issus du commerce équitable (100 % du poids total avec 93 % SPP).",
-		"Pâte de cacao Bio Commerce équitable du Pérou 65 %, sucre de canne Bio Commerce équitable, beurre de cacao Bio Commerce équitable, sel Commerce équitable, lait Bio."
-	],
-
-	[
-		"fr",
-		"p\x{e2}te de cacao* de Madagascar 75%, sucre de canne*, beurre de cacao*. * issus du commerce \x{e9}quitable et de l'agriculture biologique (100% du poids total).",
-		"pâte de cacao Commerce équitable Bio de Madagascar 75%, sucre de canne Commerce équitable Bio, beurre de cacao Commerce équitable Bio."
-	],
-
-	[
-		"fr",
-		"Céleri - rave 21% - Eau, légumes 33,6% (carottes, céleri - rave, poivrons rouges 5,8% - haricots - petits pois bio - haricots verts - courge - radis, pommes de terre - patates - fenouil - cerfeuil tubéreux - persil plat)",
-		"Céleri-rave 21% - Eau, légumes 33.6% (carottes, céleri-rave, poivrons rouges 5.8% - haricots - petits pois bio - haricots verts - courge - radis, pommes de terre - patates - fenouil - cerfeuil tubéreux - persil plat)"
-	],
-	[
-		"fr",
-		"poudres à lever : carbonates d'ammonium - carbonates de sodium - phosphates de calcium, farine, sel",
-		"poudres à lever : carbonates d'ammonium - carbonates de sodium - phosphates de calcium, farine, sel"
-	],
-	["en", "FD&C Red #40 Lake and silicon dioxide", "FD&C Red #40 Lake and silicon dioxide"],
-	["fr", "Lait pasteurisé à 1,1% de Mat. Gr.", "Lait pasteurisé à 1.1% de Matières Grasses"],
-	["fr", "matière grasse végétale (palme) raffinée", "matière grasse végétale de palme raffinée"],
-	["fr", "huile d'olive vierge, origan", "huile d'olive vierge, origan"],
-	["fr", "huile de tournesol, cacao maigre en poudre 5.2%", "huile de tournesol, cacao maigre en poudre 5.2%"],
-
-	[
-		"pl",
-		"regulatory kwasowości: kwas cytrynowy i cytryniany sodu.",
-		"regulatory kwasowości: kwas cytrynowy i cytryniany sodu."
-	],
-
-	[
-		"de",
-		"Wasser, Kohlensäure, Farbstoff Zuckerkulör E 150d, Süßungsmittel Aspartam* und Acesulfam-K, Säuerungsmittel Phosphorsäure und Citronensäure, Säureregulator Natriumcitrat, Aroma Koffein, Aroma. enthält eine Phenylalaninquelle",
-		"Wasser, Kohlensäure, Farbstoff : Zuckerkulör e150d, Süßungsmittel : Aspartam* und Acesulfam-K, Säuerungsmittel : Phosphorsäure und Citronensäure, Säureregulator Natriumcitrat, Aroma Koffein, Aroma. enthält eine Phenylalaninquelle"
-	],
-	["de", "Farbstoffe Betenrot, Paprikaextrakt, Kurkumin", "farbstoffe : betenrot, paprikaextrakt, kurkumin"],
-	[
-		"de",
-		"Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs, weiß und gelb",
-		"Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8.5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs weiß und gelb"
-	],
-	[
-		"de",
-		"Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs (weiß und gelb)",
-		"Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8.5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs weiß und gelb"
-	],
-
-	["fr", "graisse végétale bio (colza)", "graisse végétale bio de colza"],
-	[
-		"fr",
-		"huiles végétales* (huile de tournesol*, huile de colza*). *Ingrédients issus de l'agriculture biologique",
-		"huiles végétales bio (huile de tournesol bio, huile de colza bio )."
-	],
-
-	["fr", "huile biologique (tournesol, olive)", "huile biologique de tournesol, huile biologique d'olive"],
-
-	# xyz: test an unrecognized oil -> do not change
-	["fr", "huile biologique (tournesol, xyz)", "huile biologique (tournesol, xyz)"],
-	["fr", "huiles biologiques (tournesol, olive)", "huiles biologiques de tournesol, huiles biologiques d'olive"],
-	["fr", "huiles (tournesol*, olive). * : bio", "huiles de tournesol bio, huiles d'olive."],
-	[
-		"fr",
-		"huiles* (tournesol*, olive vierge extra), sel marin. *issus de l'agriculture biologique.",
-		"huiles Bio de tournesol Bio, huiles Bio d'olive vierge extra, sel marin."
-	],
-	["fr", "riz de Camargue (1), sel. (1): IGP : Indication Géographique Protégée.", "riz de Camargue IGP, sel."],
-	[
-		"fr",
-		"cacao (1), sucre (2), beurre de cacao (1). (1) : Commerce équitable. (2) Issue de l'agriculture biologique.",
-		"cacao Commerce équitable, sucre Bio, beurre de cacao Commerce équitable."
-	],
-
-	[
-		"fr",
-		"Céréales 63,7% (BLE complet 50,5%*, semoule de maïs*), sucre*, sirop de BLE*, cacao maigre en poudre 3,9%*, cacao en poudre 1,7%*, sel, arôme naturel. *Ingrédients issus de l'agriculture biologique.",
-		"Céréales 63.7% (BLE complet 50.5% Bio, semoule de maïs Bio ), sucre Bio, sirop de BLE Bio, cacao maigre en poudre 3.9% Bio, cacao en poudre 1.7% Bio, sel, arôme naturel."
-	],
-
-	["fr", "émulsifiant : mono - et diglycérides d'acides gras.", "émulsifiant : mono- et diglycérides d'acides gras."],
-
-	[
-		"fr",
-		"Sucre. Fabriqué dans un atelier qui utilise des fruits à coques.",
-		"Sucre. Traces éventuelles : fruits à coques."
-	],
-	[
-		"fr",
-		"Sucre. Fabriqué dans un atelier utilisant des fruits à coques et du sésame.",
-		"Sucre. Traces éventuelles : fruits à coques, Traces éventuelles : sésame."
-	],
-	[
-		"fr",
-		"Sucre. Fabriqué dans un atelier qui manipule du lait, de la moutarde et du céleri.",
-		"Sucre. Traces éventuelles : lait, Traces éventuelles : moutarde, Traces éventuelles : céleri."
-	],
-	[
-		"fr",
-		"Sucre. Peut contenir des fruits à coques et du sésame.",
-		"Sucre. Traces éventuelles : fruits à coques, Traces éventuelles : sésame."
-	],
-
-	["en", "vegetable oil (coconut & rapeseed)", "coconut vegetable oil, rapeseed vegetable oil"],
-
-	[
-		"fr",
-		"Masse de cacao°, Quinoa° (1,8%). °Produits issus de l'agriculture biologique.",
-		"Masse de cacao Bio, Quinoa Bio (1.8%)."
-	],
-
-	[
-		"de",
-		"Emulgator (Sojalecithine, Mono - und Diglyceride von Speisefettsäuren, Sorbitantristearat)",
-		"Emulgator (Sojalecithine, mono- und Diglyceride von Speisefettsäuren, Sorbitantristearat)"
-	],
-
-	["fr", "Tomates* (20%). *Ingrédients Bio", "Tomates Bio (20%)."],
-	["fr", "Tomates* (20%). *Ingrédients biologiques", "Tomates Bio (20%)."],
-
-	[
-		"fr",
-		"Chocolat. Contient du lait et des noisettes. Peut contenir du blé, du soja et des crustacés.",
-		"Chocolat. Substances ou produits provoquant des allergies ou intolérances : lait, Substances ou produits provoquant des allergies ou intolérances : noisettes. Traces éventuelles : blé, Traces éventuelles : soja, Traces éventuelles : crustacés."
-	],
-
-	[
-		"en",
-		"Chocolate. Contains milk, hazelnuts and other nuts. May contain celery and mustard.",
-		"Chocolate. Substances or products causing allergies or intolerances : milk, Substances or products causing allergies or intolerances : hazelnuts, Substances or products causing allergies or intolerances : other nuts. Traces : celery, Traces : mustard."
-	],
-
-	[
-		"fr",
-		"phosphates d'ammonium et de calcium, Phosphate d'aluminium et de sodium, diphosphate d'aluminium et de sodium",
-		"phosphates d'ammonium, phosphates de calcium, phosphate d'aluminium et de sodium, diphosphate d'aluminium et de sodium"
-	],
-
-	[
-		"fr",
-		"Ingrédient(s) : lentilles vertes* - *issu(e)(s) de l'agriculture biologique.",
-		"Ingrédients : lentilles vertes Bio"
-	],
-
-	[
-		"en",
-		"S. thermophilus, L casei, L.bulgaricus",
-		"streptococcus thermophilus, lactobacillus casei, lactobacillus bulgaricus"
-	],
-
-	[
-		"fr",
-		"jus de citron*. *Ingrédients issus de l'agriculture biologique Peut contenir : œuf, moutarde, graine de sésame, poisson,soja, lait,fruits à coque, céleri.",
-		"jus de citron Bio. , Traces éventuelles : œuf, Traces éventuelles : moutarde, Traces éventuelles : graine de sésame, Traces éventuelles : poisson, Traces éventuelles : soja, Traces éventuelles : lait, Traces éventuelles : fruits à coque, Traces éventuelles : céleri."
-	],
-
-	[
-		"fr",
-		"Farine, levure. Peut contenir des traces de _soja_, _amandes_, _noisettes_ et _noix de cajou_.",
-		"Farine, levure. Traces éventuelles : _soja_, Traces éventuelles : _amandes_, Traces éventuelles : _noisettes_, Traces éventuelles : _noix de cajou_."
-	],
-
-	# Spanish organic ingredients
-	["es", "Agua, aceite de girasol*. * Ingredientes ecológicos.", "Agua, aceite de girasol Ecológico."],
-	[
-		"es",
-		"Agua, aceite de girasol*, arroz* (5 %). (*) Ingredientes ecológicos.",
-		"Agua, aceite de girasol Ecológico, arroz Ecológico (5 %)."
-	],
-	[
-		"es",
-		"Tofu* 88% (agua, habas de soja*). *cumple con el reglamento de agricultura ecológica CE 2092/91",
-		"Tofu Ecológico 88% (agua, habas de soja Ecológico )."
-	],
-	["es", "agua, almendra* (5,5%). *= procedentes de la agricultura ecológica", "agua, almendra Ecológico (5.5%)."],
-
-	# test for bug #3273 that introduced unwanted separators before natural flavor
-	["en", "non-gmo natural flavor", "non-gmo natural flavor"],
-
-	# vit. e
-	[
-		"en",
-		"vit. e, vitamins b2, B3 and K, vit d, vit a & c, vit. B12",
-		"vitamin e, vitamins, vitamin b2, vitamin B3, vitamin K, vitamin d, vitamin a, vitamin c, vitamin B12"
-	],
-	["fr", "vit. pp, vit c, vit. a et b6", "vitamines, vitamine pp, vitamine c, vitamine a, vitamine b6"],
-	["pl", "witaminy A i D", "witaminy, witamina A, witamina D"],
-
-	[
-		"fr",
-		"colorant de surface : caramel ordinaire, agent de traitement de farine (E300), acide citrique",
-		"colorant de surface : caramel ordinaire, agent de traitement de farine (E300), acide citrique"
-	],
-
-	[
-		"es",
-		"Agua, edulcorantes (INS420, INS 960, INS N'952, INS N°954, INS°950, INS N 955), conservantes (INS.218, INS #202, INS N 216).",
-		"Agua, edulcorantes (e420, e960, e952, e954, e950, e955), conservantes (e218, e202, e216)."
-	],
-
-	# Spanish Vitamin E can be mistaken for "e" meaning "and"
-	["es", "Vitamina E y C", "vitaminas, vitamina E, vitamina C"],
-	["es", "color E 124", "color : e124"],
-	["es", "colores E (124, 125)", "colores e124, e125"],
-	["it", "vitamine A, B, E e K", "vitamine, vitamina A, vitamina B, vitamina E, vitamina K"],
-
-	# Additives normalization
-	[
-		"en",
-		"E 102, E-104 color, E-101(i), E101 (ii), E160a(iv), e172-i, E-160 i",
-		"e102, e104 color, e101i, e101ii, e160aiv, e172i, e160i"
-	],
-	["fr", "E102-E1400", "e102 - e1400"],
-	["de", "E172i-E174ii, E102(i)-E101i", "e172i - e174ii, e102i - e101i"],
-	["fr", "correcteurs d'acidité : E322/E333 E474-E475", "correcteurs d'acidité : e322/e333, e474 - e475"],
-	["es", "E-330; E-331; Estabilizantes (E-327; E-418)", "e330; e331; Estabilizantes (e327; e418)"],
-	["es", "E120 color", "e120 color"],
-	["es", "E172-i", "e172i"],
-	["es", "E172 i", "e172i"],
-	["es", "(E172i)", "(e172i)"],
-	["es", "E102(i)-E101i", "e102i - e101i"],
-	["es", "E102(i)", "e102i"],
-	["es", "S.I.N.:160 b", "e160b"],
-	["pt", "estabilizadores (E 422, E 412)", "estabilizadores (e422, e412)"],
-
-	[
-		"es",
-		"contiene apio y derivados de leche",
-		"Sustancias o productos que causan alergias o intolerancias : apio, Sustancias o productos que causan alergias o intolerancias : derivados de leche."
-	],
-
-	["fr", "E160a(ii)", "e160aii"],
-	["fr", "(E160a-ii)", "(e160aii)"],
-	["fr", "colorant (E160a(ii))", "colorant (e160aii)"],
-
-	# do not separate acide acétique into acide : acétique
-	[
-		"fr",
-		"Esters glycéroliques de l'acide acétique et d'acides gras",
-		"Esters glycéroliques de l'acide acétique et d'acides gras"
-	],
-	["fr", "acide acétique", "acide acétique"],
-
-	# russian abbreviations
-	["ru", "мука пшеничная х/п в/с", "мука пшеничная хлебопекарная высшего сорта"],
-
-	# w/ with and w/o without abbreviations
-	["en", "Organic garbanzo beans (cooked w/o salt), water", "Organic garbanzo beans (cooked without salt), water"],
-	["en", "sugar, cocoa (processed w/alkali), egg yolk", "sugar, cocoa (processed with alkali), egg yolk"],
-
-	# * ingrédient issu..
-	["fr", "LAIT entier pasteurisé*. *ingrédient issu de l'agriculture biologique.", "LAIT entier pasteurisé Bio."],
-
-	# vitamines
-	[
-		"fr", "vitamines B1, B6, B9, PP et E",
-		"vitamines, vitamine B1, vitamine B6, vitamine B9, vitamine PP, vitamine E"
-	],
-	["fr", "vitamines (B1, acide folique (B9))", "vitamines, vitamine B1, acide folique, vitamine B9"],
-
-	# (origins, contains milk)
-	[
-		"en",
-		"Chocolate (Italy, contains milk)",
-		"Chocolate (Italy, Substances or products causing allergies or intolerances : milk.)"
-	],
-	[
-		"en", "Chocolate (contains milk)",
-		"Chocolate ( Substances or products causing allergies or intolerances : milk.)"
-	],
-	["en", "Chocolate. Contains (milk)", "Chocolate. Substances or products causing allergies or intolerances : milk."],
-
-	# ¹ and ² symbols
-	[
-		"fr",
-		"Sel, sucre², graisse de palme¹, amidons¹ (maïs¹, pomme de terre¹), oignon¹ : 8,9%, ail¹, oignon grillé¹ : 1,4%, épices¹ et aromate¹ (livèche¹ : 0,4%, curcuma¹, noix de muscade¹), carotte¹ : 0,5%. Peut contenir : céleri, céréales contenant du gluten, lait, moutarde, œuf, soja. ¹Ingrédients issus de l'Agriculture Biologique. ² Ingrédients issus du commerce équitable",
-		"Sel, sucre Commerce équitable, graisse de palme Bio, amidons Bio (maïs Bio, pomme de terre Bio ), oignon Bio : 8.9%, ail Bio, oignon grillé Bio : 1.4%, épices Bio et aromate Bio (livèche Bio : 0.4%, curcuma Bio, noix de muscade Bio ), carotte Bio : 0.5%. Traces éventuelles : céleri, Traces éventuelles : céréales contenant du gluten, Traces éventuelles : lait, Traces éventuelles : moutarde, Traces éventuelles : œuf, Traces éventuelles : soja."
-	],
-
-	# Russian е character
-	["ru", "е322, Куркумины e100, е-1442, (е621)", "e322, куркумины e100, e1442, (e621)"],
-
-	# New ingredients categories + types : generalized from French to other languages
-	[
-		"fr",
-		"huiles végétales (palme, olive et tournesol)",
-		"huiles végétales de palme, huiles végétales d'olive, huiles végétales de tournesol"
-	],
-	["fr", "huile végétale : colza", "huile végétale de colza"],
-	["fr", "huile végétale : colza, fraises", "huile végétale de colza, fraises"],
-	["fr", "huile végétale : colza et tomates", "huile végétale de colza et tomates"],
-	["en", "vegetable oil: sunflower", "sunflower vegetable oil"],
-	["en", "vegetable oil (palm)", "palm vegetable oil"],
-	["en", "vegetable oils (palm, olive)", "palm vegetable oils, olive vegetable oils"],
-	[
-		"en",
-		"organic vegetable oils (sunflower, colza and rapeseed)",
-		"sunflower organic vegetable oils, colza organic vegetable oils, rapeseed organic vegetable oils"
-	],
-
-	# used to have bad output: sunflower vegetable oils, colza vegetable oilsand strawberry
-	[
-		"en",
-		"vegetable oils : sunflower, colza and strawberry",
-		"sunflower vegetable oils, colza vegetable oils and strawberry"
-	],
-
-	# Polish oils
-	["pl", "oleje roślinne (słonecznikowy)", "oleje roślinne słonecznikowy"],
-	["pl", "oleje roślinne: słonecznikowy", "oleje roślinne słonecznikowy"],
-	["pl", "oleje roślinne (słonecznikowy, rzepakowy)", "oleje roślinne słonecznikowy, oleje roślinne rzepakowy"],
-	[
-		"pl",
-		"oleje roślinne (sojowy, słonecznikowy, kokosowy, rzepakowy) w zmiennych proporcjach",
-		"oleje roślinne sojowy, oleje roślinne słonecznikowy, oleje roślinne kokosowy, oleje roślinne rzepakowy"
-	],
-	[
-		"pl",
-		"tłuszcze roślinne (palmowy nieutwardzony, shea)",
-		"tłuszcze roślinne palmowy nieutwardzony, tłuszcze roślinne shea"
-	],
-	[
-		"pl",
-		"tłuszcze roślinne (kokosowy i palmowy) w zmiennych proporcjach",
-		"tłuszcze roślinne kokosowy, tłuszcze roślinne palmowy"
-	],
-
-	# Polish meats
-	["pl", "mięso (wołowe, wieprzowe, cielęce)", "mięso wołowe, mięso wieprzowe, mięso cielęce"],
-
-	# Polish juices and concentrates
-	["pl", "przeciery z (jabłek, bananów, marchwi)", "przeciery z jabłek, przeciery z bananów, przeciery z marchwi"],
-
-	# Russian oils (more tests needed)
-	["ru", "масло (Подсолнечное)", "масло Подсолнечное"],
-	["ru", "Масло (подсолнечное)", "Масло подсолнечное"],
-	["ru", "масло растительное (подсолнечное, соевое)", "масло растительное подсолнечное, масло растительное соевое"],
-
-	# grammes -> g
-	["fr", "Teneur en fruits: 50gr pour 100 grammes", "Teneur en fruits: 50g pour 100 g"],
-
-	# test conflicts between the word "and" in some languages and additives variants. With letters i or e or a.
-	[
-		"hr",
-		"bojilo: E 150a, tvari za rahljenje: E 500 i E 503, sol.",
-		"bojilo: e150a, tvari za rahljenje: e500, e503, sol."
-	],
-	[
-		"hr",
-		"bojilo: E 150a, tvari za rahljenje: E 500 i, E 503, sol.",
-		"bojilo: e150a, tvari za rahljenje: e500 i, e503, sol."
-	],
-	[
-		"hr",
-		"bojilo: E 150a, tvari za rahljenje: E 500(i), E 503, sol.",
-		"bojilo: e150a, tvari za rahljenje: e500i, e503, sol."
-	],
-	[
-		"hr",
-		"bojilo: E 150a, tvari za rahljenje: E 500i, E 503, sol.",
-		"bojilo: e150a, tvari za rahljenje: e500i, e503, sol."
-	],
-	["it", "formaggio, E 472 e, E470a.", "formaggio, e472 e, e470a."],
-	["it", "formaggio, E 472 e E470a.", "formaggio, e472, e470a."],
-	["sk", "syr, E470 a E470a, mlieko.", "syr, e470, e470a, mlieko."],
-	# normalize category and types
-	["fr", "Piments (vert, rouge, jaune)", "Piments vert, Piments rouge, Piments jaune"],
-	# New feature:
-	["de", "pflanzliches Fett (Kokosnuss, Palmkern)", "Kokosnussfett, Palmkernfett"],
-	[
-		"de", "pflanzliche Öle und Fette (Raps, Palm, Shea, Sonnenblumen)",
-		"Rapsöl, Palmfett, Sheafett, Sonnenblumenfett"
-	],
-	[
-		"fr",
-		"Huiles végétales de palme, de colza et de tournesol",
-		"Huiles végétales de palme, Huiles végétales de colza, Huiles végétales de tournesol"
-	],
-	["fr", "arôme naturel de pomme avec d'autres âromes", "arôme naturel de pomme et âromes"],
-	["fr", "Carbonate de magnésium, fer élémentaire", "Carbonate de magnésium, fer élémentaire"],
-	["fr", "huile végétale (colza)", "huile végétale de colza"],
-	["fr", "huile végétale : colza", "huile végétale de colza"],
-	["hr", "ječmeni i pšenični slad", "ječmeni slad, pšenični slad"],
-	["hr", "ječmeni, ječmeni i pšenični slad", "ječmeni slad, ječmeni slad, pšenični slad"],
-	["hr", "Pasterizirano mlijeko (s 1.0% mliječne masti)", "pasterizirano mlijeko s 1.0% mliječne masti"],
-	["en", "Vegetal oil (sunflower, olive and palm)", "sunflower vegetal oil, olive vegetal oil, palm vegetal oil"],
-	["en", "vegetable oil (palm)", "palm vegetable oil"],
-	["en", "vegetable oil: palm", "palm vegetable oil"],
-	["fr", "protéines végétales (soja, blé)", "protéine de soja, protéine de blé"],
-	["de", "pflanzliche Proteine (Erbsen, Sonnenblumen)", "erbsenprotein, sonnenblumenprotein"],
-	# Should not develop the enumeration if it contains unknown types (like "sel" here)
-	["fr", "Piments (vert, rouge, jaune, sel)", "Piments (vert, rouge, jaune, sel)"],
-	["fr", "Huile de palme, noisettes et tournesol", "huile de palme, huile de noisettes, huile de tournesol"],
-	["fr", "Huile de palme, noisettes", "huile de palme, noisettes"],
-	[
-		"fr",
-		"arôme naturel de citron, citron vert et d'autres agrumes",
-		"arôme naturel de citron, arôme naturel de citron vert, arôme naturel d'agrumes"
-	],
-);
-
-foreach my $test_ref (@lists) {
-	my $l = $test_ref->[0];    # Language
-	my $ingredients = $test_ref->[1];
-	my $preparsed = preparse_ingredients_text($l, $ingredients);
-	print STDERR "Ingredients ($l): $ingredients\n";
-	print STDERR "Preparsed: $preparsed\n";
-	my $expected = $test_ref->[2];
-	is(lc($preparsed), lc($expected)) or print STDERR "Original ingredients: $ingredients ($l)\n";
-}
-
-done_testing();
diff --git a/tests/unit/ingredients_preparsing.t b/tests/unit/ingredients_preparsing.t
new file mode 100755
index 0000000000000..9cd4627d38502
--- /dev/null
+++ b/tests/unit/ingredients_preparsing.t
@@ -0,0 +1,1217 @@
+#!/usr/bin/perl -w
+
+# Tests of Ingredients::preparse_ingredients_text()
+
+use Modern::Perl '2017';
+use utf8;
+
+use Test2::V0;
+use Log::Any::Adapter 'TAP';
+
+use ProductOpener::Products qw/:all/;
+use ProductOpener::Tags qw/:all/;
+use ProductOpener::Test qw/compare_to_expected_results init_expected_results/;
+use ProductOpener::Ingredients
+	qw/normalize_a_of_b normalize_enumeration preparse_ingredients_text separate_additive_class/;
+
+#use Log::Any::Adapter 'TAP', filter => "none";
+
+is(normalize_a_of_b("en", "oil", "olive", 1), "olive oil");
+is(normalize_a_of_b("es", "aceta", "oliva", 1), "aceta de oliva");
+is(normalize_a_of_b("fr", "huile végétale", "olive", 1), "huile végétale d'olive");
+
+is(normalize_enumeration("en", "phosphates", "calcium and sodium", 1), "calcium phosphates, sodium phosphates");
+is(normalize_enumeration("en", "vegetal oil", "sunflower, palm", 1), "sunflower vegetal oil, palm vegetal oil");
+is(normalize_enumeration("fr", "huile", "colza, tournesol et olive", 1),
+	"huile de colza, huile de tournesol, huile d'olive");
+
+is(separate_additive_class("fr", "colorant", " ", "", "naturel"), "colorant ");
+is(separate_additive_class("fr", "colorant", " ", "", "carmins"), "colorant : ");
+is(separate_additive_class("fr", "colorant", " ", "", "E120, sel"), "colorant : ");
+is(separate_additive_class("fr", "colorant", " ", "", "E120 et E150b"), "colorant : ");
+is(separate_additive_class("fr", "colorant", " ", "", "caramel au sulfite d'ammonium"), "colorant : ");
+is(separate_additive_class("fr", "colorant", " ", "", "caramel au sulfite d'ammonium et rocou"), "colorant : ");
+
+my ($test_id, $test_dir, $expected_result_dir, $update_expected_results) = (init_expected_results(__FILE__));
+
+my @tests = (
+	{
+		id => '1',
+		lc => 'fr',
+		ingredients_text => 'Sel marin, blé, lécithine de soja'
+	},
+	{
+		id => '2',
+		lc => 'fr',
+		ingredients_text => 'Vitamine A'
+	},
+	{
+		id => '3',
+		lc => 'fr',
+		ingredients_text => 'Vitamines A, B et C'
+	},
+	{
+		id => '4',
+		lc => 'fr',
+		ingredients_text => 'Vitamines (B1, B2, B6, PP)'
+	},
+	{
+		id => '5',
+		lc => 'fr',
+		ingredients_text => 'Huile de palme'
+	},
+	{
+		id => '6',
+		lc => 'fr',
+		ingredients_text => 'Huile (palme)'
+	},
+	{
+		id => '7',
+		lc => 'fr',
+		ingredients_text => 'Huile (palme, colza)'
+	},
+	{
+		id => '8',
+		lc => 'fr',
+		ingredients_text => 'Huile (palme et colza)'
+	},
+	{
+		id => '9',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme et de colza'
+	},
+	{
+		id => '10',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme et d\'olive'
+	},
+	{
+		id => '11',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme, de colza et de tournesol'
+	},
+	{
+		id => '12',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme, de colza, de tournesol'
+	},
+	{
+		id => '13',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme, de colza et d\'olive en proportion variable'
+	},
+	{
+		id => '14',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme, de colza et d\'olive'
+	},
+	{
+		id => '15',
+		lc => 'fr',
+		ingredients_text => 'phosphate et sulfate de calcium'
+	},
+	{
+		id => '16',
+		lc => 'fr',
+		ingredients_text => 'sulfates de calcium et potassium'
+	},
+	{
+		id => '17',
+		lc => 'fr',
+		ingredients_text => 'chlorures (sodium et potassium)'
+	},
+	{
+		id => '18',
+		lc => 'fr',
+		ingredients_text => 'chlorures (sodium, potassium)'
+	},
+	{
+		id => '19',
+		lc => 'fr',
+		ingredients_text => 'fraises 30%'
+	},
+	{
+		id => '20',
+		lc => 'fr',
+		ingredients_text =>
+			'Marmelade d\'oranges 41% (sirop de glucose-fructose, sucre, pulpe d\'orange 4.5%, jus d\'orange concentré 1.4% (équivalent jus d\'orange 7.8%), pulpe d\'orange concentrée 0.6% (équivalent pulpe d\'orange 2.6%), gélifiant (pectines), acidifiant (acide citrique), correcteurs d\'acidité (citrate de calcium, citrate de sodium), arôme naturel d\'orange, épaississant (gomme xanthane)), chocolat 24.9% (sucre, pâte de cacao, beurre de cacao, graisses végétales (illipe, mangue, sal, karité et palme en proportions variables), arôme, émulsifiant (lécithine de soja), lactose et protéines de lait), farine de blé, sucre, oeufs, sirop de glucose-fructose, huile de colza, poudre à lever (carbonate acide d\'ammonium, diphosphate disodique, carbonate acide de sodium), sel, émulsifiant (lécithine de soja).'
+	},
+	{
+		id => '21',
+		lc => 'fr',
+		ingredients_text => 'graisses végétales (illipe, mangue, sal, karité et palme en proportions variables)'
+	},
+	{
+		id => '22',
+		lc => 'fr',
+		ingredients_text => 'graisses végétales (illipe, mangue, palme)'
+	},
+	{
+		id => '23',
+		lc => 'fr',
+		ingredients_text => 'graisses végétales (illipe)'
+	},
+	{
+		id => '24',
+		lc => 'fr',
+		ingredients_text => 'graisses végétales (illipe et sal)'
+	},
+	{
+		id => '25',
+		lc => 'fr',
+		ingredients_text => 'gélifiant pectine'
+	},
+	{
+		id => '26',
+		lc => 'fr',
+		ingredients_text => 'gélifiant (pectine)'
+	},
+	{
+		id => '27',
+		lc => 'fr',
+		ingredients_text => 'agent de traitement de la farine (acide ascorbique)'
+	},
+	{
+		id => '28',
+		lc => 'fr',
+		ingredients_text => 'lait demi-écrémé'
+	},
+	{
+		id => '29',
+		lc => 'fr',
+		ingredients_text => 'Saveur vanille : lait demi-écrémé 77%, sucre'
+	},
+	{
+		id => '30',
+		lc => 'fr',
+		ingredients_text => 'colorants alimentaires E (124,122,133,104,110)'
+	},
+	{
+		id => '31',
+		lc => 'fr',
+		ingredients_text => 'INS 240,241,242b'
+	},
+	{
+		id => '32',
+		lc => 'fr',
+		ingredients_text => 'colorants E (124, 125, 120 et 122'
+	},
+	{
+		id => '33',
+		lc => 'fr',
+		ingredients_text => 'E250-E251'
+	},
+	{
+		id => '34',
+		lc => 'fr',
+		ingredients_text => 'E250-E251-E260'
+	},
+	{
+		id => '35',
+		lc => 'fr',
+		ingredients_text => 'E 250b-E251-e.260(ii)'
+	},
+	{
+		id => '36',
+		lc => 'fr',
+		ingredients_text => 'émulsifiants : E463, E432 et E472 - correcteurs d\'acidité : E322/E333 E474-E475'
+	},
+	{
+		id => '37',
+		lc => 'fr',
+		ingredients_text => 'E100 E122'
+	},
+	{
+		id => '38',
+		lc => 'fr',
+		ingredients_text => 'E103 et E140'
+	},
+	{
+		id => '39',
+		lc => 'fr',
+		ingredients_text => 'E103 ET E140'
+	},
+	{
+		id => '40',
+		lc => 'fr',
+		ingredients_text => 'curcumine et E140'
+	},
+	{
+		id => '41',
+		lc => 'fr',
+		ingredients_text => 'E140 et tartrazine'
+	},
+	{
+		id => '42',
+		lc => 'fr',
+		ingredients_text => 'Acide citrique, colorant : e120, vitamine C, E-500'
+	},
+	{
+		id => '43',
+		lc => 'fr',
+		ingredients_text => 'poudres à lever (carbonates acides d’ammonium et de sodium, acide citrique)'
+	},
+	{
+		id => '44',
+		lc => 'en',
+		ingredients_text => 'REAL SUGARCANE, SALT, ANTIOXIDANT (INS 300), INS 334, INS345'
+	},
+	{
+		id => '45',
+		lc => 'es',
+		ingredients_text => 'colores E (120, 124 y 125)'
+	},
+	{
+		id => '46',
+		lc => 'es',
+		ingredients_text => 'Leche desnatada de vaca, enzima lactasa y vitaminas A, D, E y ácido fólico.'
+	},
+	{
+		id => '47',
+		lc => 'es',
+		ingredients_text =>
+			'Leche desnatada, leche desnatada en polvo, zumo de lima, almidón de maíz, extracto de ginseng 0,19%, aromas, fermentos lácticos con Lcasei, colorante: caramelo natural, edulcorantes: sucralosa y acesulfamo K, estabilizante: goma xantana, vitaminas: D, B6, ácido fólico y B12 Origen de la feche. España. Preparación: Agitar antes de abrir.'
+	},
+	{
+		id => '48',
+		lc => 'es',
+		ingredients_text =>
+			'edulcorantes (acesulfamo K y sucralosa) y vitaminas (riboflavina (vitamina B2) y cianocobalamina vitamina B12))'
+	},
+	{
+		id => '49',
+		lc => 'es',
+		ingredients_text => 'aceites vegetales [aceite de girasol (70%) y aceite de oliva virgen (30%)] y sal'
+	},
+	{
+		id => '50',
+		lc => 'es',
+		ingredients_text => 'Trazas de cacahuete, huevo y frutos de cáscara.'
+	},
+	{
+		id => '51',
+		lc => 'es',
+		ingredients_text =>
+			'sal y acidulante (ácido cítrico). Puede contener trazas de cacahuete, huevo y frutos de cáscara.'
+	},
+	{
+		id => '52',
+		lc => 'da',
+		ingredients_text => 'bl. a. inkl. mod. past. emulgator E322 E103, E140, E250 og E100'
+	},
+	{
+		id => '53',
+		lc => 'nb',
+		ingredients_text => 'bl. a. inkl. E322 E103, E140, E250 og E100'
+	},
+	{
+		id => '54',
+		lc => 'sv',
+		ingredients_text => 'bl. a. förtjockn.medel inkl. emulgeringsmedel E322 E103, E140, E250 och E100'
+	},
+	{
+		id => '55',
+		lc => 'da',
+		ingredients_text =>
+			'Vitaminer A, B og C. Vitaminer (B2, E, D), Hvede**. Indeholder mælk. Kan indeholde spor af soja, mælk, mandler og sesam. ** = Økologisk'
+	},
+	{
+		id => '56',
+		lc => 'is',
+		ingredients_text => 'Vítamín (B2, E og D). Getur innihaldið hnetur, soja og mjólk í snefilmagni.'
+	},
+	{
+		id => '57',
+		lc => 'nb',
+		ingredients_text =>
+			'Vitaminer A, B og C. Vitaminer (B2, E, D). Kan inneholde spor av andre nøtter, soya og melk.'
+	},
+	{
+		id => '58',
+		lc => 'sv',
+		ingredients_text =>
+			'Vitaminer (B2, E och D), Vete*. Innehåller hasselnötter. Kan innehålla spår av råg, jordnötter, mandel, hasselnötter, cashewnötter och valnötter. *Ekologisk'
+	},
+	{
+		id => '59',
+		lc => 'fi',
+		ingredients_text => 'Vitamiinit A, B ja C'
+	},
+	{
+		id => '60',
+		lc => 'fi',
+		ingredients_text => 'Vitamiinit (B1, B2, B6)'
+	},
+	{
+		id => '61',
+		lc => 'fi',
+		ingredients_text => 'mansikat 30%'
+	},
+	{
+		id => '62',
+		lc => 'fi',
+		ingredients_text => 'sakeuttamisaine pektiini'
+	},
+	{
+		id => '63',
+		lc => 'fi',
+		ingredients_text => 'sakeuttamisaine (pektiini)'
+	},
+	{
+		id => '64',
+		lc => 'fi',
+		ingredients_text => 'jauhonparanne (askorbiinihappo)'
+	},
+	{
+		id => '65',
+		lc => 'fi',
+		ingredients_text => 'E250-E251'
+	},
+	{
+		id => '66',
+		lc => 'fi',
+		ingredients_text => 'E250-E251-E260'
+	},
+	{
+		id => '67',
+		lc => 'fi',
+		ingredients_text => 'E 250b-E251-e.260(ii)'
+	},
+	{
+		id => '68',
+		lc => 'fi',
+		ingredients_text => 'E100 E122'
+	},
+	{
+		id => '69',
+		lc => 'fi',
+		ingredients_text => 'E103 ja E140'
+	},
+	{
+		id => '70',
+		lc => 'fi',
+		ingredients_text => 'E103 JA E140'
+	},
+	{
+		id => '71',
+		lc => 'fi',
+		ingredients_text => 'kurkumiini ja E140'
+	},
+	{
+		id => '72',
+		lc => 'fi',
+		ingredients_text => 'E140 ja karoteeni'
+	},
+	{
+		id => '73',
+		lc => 'fi',
+		ingredients_text => 'omenamehu, vesi, sokeri. jossa käsitellään myös maitoa.'
+	},
+	{
+		id => '74',
+		lc => 'fi',
+		ingredients_text => 'omenamehu, vesi, sokeri. Saattaa sisältää pieniä määriä selleriä, sinappia ja vehnää.'
+	},
+	{
+		id => '75',
+		lc => 'fi',
+		ingredients_text => 'omenamehu, vesi, sokeri. Saattaa sisältää pienehköjä määriä selleriä, sinappia ja vehnää.'
+	},
+	{
+		id => '76',
+		lc => 'fi',
+		ingredients_text => 'luomurypsiöljy, luomu kaura, vihreä luomutee'
+	},
+	{
+		id => '77',
+		lc => 'fr',
+		ingredients_text => 'arôme naturel de citron-citron vert et d\'autres agrumes'
+	},
+	{
+		id => '78',
+		lc => 'fr',
+		ingredients_text => 'arômes naturels de citron et de limette'
+	},
+	{
+		id => '79',
+		lc => 'fr',
+		ingredients_text => 'arôme naturel de pomme avec d\'autres arômes naturels'
+	},
+	{
+		id => '80',
+		lc => 'fr',
+		ingredients_text => 'jus de pomme, eau, sucre. Traces de lait.'
+	},
+	{
+		id => '81',
+		lc => 'fr',
+		ingredients_text => 'jus de pomme, eau, sucre. Traces possibles de céleri, moutarde et gluten.'
+	},
+	{
+		id => '82',
+		lc => 'fr',
+		ingredients_text => 'jus de pomme, eau, sucre. Traces possibles de céleri, de moutarde et gluten.'
+	},
+	{
+		id => '83',
+		lc => 'fr',
+		ingredients_text => 'Traces de moutarde'
+	},
+	{
+		id => '84',
+		lc => 'fr',
+		ingredients_text => 'Sucre de canne Traces éventuelles d\'oeufs'
+	},
+	{
+		id => '85',
+		lc => 'fr',
+		ingredients_text => 'huile végétale de tournesol et/ou colza'
+	},
+	{
+		id => '86',
+		lc => 'de',
+		ingredients_text => 'Zucker. Kann Spuren von Sellerie.'
+	},
+	{
+		id => '87',
+		lc => 'de',
+		ingredients_text => 'Zucker. Kann Spuren von Senf und Sellerie.'
+	},
+	{
+		id => '88',
+		lc => 'de',
+		ingredients_text => 'Zucker. Kann Spuren von Senf und Sellerie enthalten'
+	},
+	{
+		id => '89',
+		lc => 'it',
+		ingredients_text => 'Puo contenere tracce di frutta a guscio, sesamo, soia e uova'
+	},
+	{
+		id => '90',
+		lc => 'it',
+		ingredients_text => 'Il prodotto può contenere tracce di GRANO, LATTE, UOVA, FRUTTA A GUSCIO e SOIA.'
+	},
+	{
+		id => '91',
+		lc => 'fr',
+		ingredients_text => 'Jus de pomme*** 68%, jus de poire***32% *** Ingrédients issus de l\'agriculture biologique'
+	},
+	{
+		id => '92',
+		lc => 'fr',
+		ingredients_text =>
+			'Pâte de cacao°* du Pérou 65 %, sucre de canne°*, beurre de cacao°*, sel *, lait °. °Issus de l\'agriculture biologique (100 %). *Issus du commerce équitable (100 % du poids total avec 93 % SPP).'
+	},
+	{
+		id => '93',
+		lc => 'fr',
+		ingredients_text =>
+			'pâte de cacao* de Madagascar 75%, sucre de canne*, beurre de cacao*. * issus du commerce équitable et de l\'agriculture biologique (100% du poids total).'
+	},
+	{
+		id => '94',
+		lc => 'fr',
+		ingredients_text =>
+			'Céleri - rave 21% - Eau, légumes 33,6% (carottes, céleri - rave, poivrons rouges 5,8% - haricots - petits pois bio - haricots verts - courge - radis, pommes de terre - patates - fenouil - cerfeuil tubéreux - persil plat)'
+	},
+	{
+		id => '95',
+		lc => 'fr',
+		ingredients_text =>
+			'poudres à lever : carbonates d\'ammonium - carbonates de sodium - phosphates de calcium, farine, sel'
+	},
+	{
+		id => '96',
+		lc => 'en',
+		ingredients_text => 'FD&C Red #40 Lake and silicon dioxide'
+	},
+	{
+		id => '97',
+		lc => 'fr',
+		ingredients_text => 'Lait pasteurisé à 1,1% de Mat. Gr.'
+	},
+	{
+		id => '98',
+		lc => 'fr',
+		ingredients_text => 'matière grasse végétale (palme) raffinée'
+	},
+	{
+		id => '99',
+		lc => 'fr',
+		ingredients_text => 'huile d\'olive vierge, origan'
+	},
+	{
+		id => '100',
+		lc => 'fr',
+		ingredients_text => 'huile de tournesol, cacao maigre en poudre 5.2%'
+	},
+	{
+		id => '101',
+		lc => 'pl',
+		ingredients_text => 'regulatory kwasowości: kwas cytrynowy i cytryniany sodu.'
+	},
+	{
+		id => '102',
+		lc => 'de',
+		ingredients_text =>
+			'Wasser, Kohlensäure, Farbstoff Zuckerkulör E 150d, Süßungsmittel Aspartam* und Acesulfam-K, Säuerungsmittel Phosphorsäure und Citronensäure, Säureregulator Natriumcitrat, Aroma Koffein, Aroma. enthält eine Phenylalaninquelle'
+	},
+	{
+		id => '103',
+		lc => 'de',
+		ingredients_text => 'Farbstoffe Betenrot, Paprikaextrakt, Kurkumin'
+	},
+	{
+		id => '104',
+		lc => 'de',
+		ingredients_text =>
+			'Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs, weiß und gelb'
+	},
+	{
+		id => '105',
+		lc => 'de',
+		ingredients_text =>
+			'Zucker, Glukosesirup, Glukose-Fruktose-Sirup, Stärke, 8,5% Süßholzsaft, brauner Zuckersirup, modifizierte Stärke, Aromen, pflanzliches Öl (Sonnenblume), Überzugsmittel: Bienenwachs (weiß und gelb)'
+	},
+	{
+		id => '106',
+		lc => 'fr',
+		ingredients_text => 'graisse végétale bio (colza)'
+	},
+	{
+		id => '107',
+		lc => 'fr',
+		ingredients_text =>
+			'huiles végétales* (huile de tournesol*, huile de colza*). *Ingrédients issus de l\'agriculture biologique'
+	},
+	{
+		id => '108',
+		lc => 'fr',
+		ingredients_text => 'huile biologique (tournesol, olive)'
+	},
+	{
+		id => '109',
+		comment => "xyz: test an unrecognized oil -> do not change",
+		lc => 'fr',
+		ingredients_text => 'huile biologique (tournesol, xyz)'
+	},
+	{
+		id => '110',
+		lc => 'fr',
+		ingredients_text => 'huiles biologiques (tournesol, olive)'
+	},
+	{
+		id => '111',
+		lc => 'fr',
+		ingredients_text => 'huiles (tournesol*, olive). * : bio'
+	},
+	{
+		id => '112',
+		lc => 'fr',
+		ingredients_text => 'huiles* (tournesol*, olive vierge extra), sel marin. *issus de l\'agriculture biologique.'
+	},
+	{
+		id => '113',
+		lc => 'fr',
+		ingredients_text => 'riz de Camargue (1), sel. (1): IGP : Indication Géographique Protégée.'
+	},
+	{
+		id => '114',
+		lc => 'fr',
+		ingredients_text =>
+			'cacao (1), sucre (2), beurre de cacao (1). (1) : Commerce équitable. (2) Issue de l\'agriculture biologique.'
+	},
+	{
+		id => '115',
+		lc => 'fr',
+		ingredients_text =>
+			'Céréales 63,7% (BLE complet 50,5%*, semoule de maïs*), sucre*, sirop de BLE*, cacao maigre en poudre 3,9%*, cacao en poudre 1,7%*, sel, arôme naturel. *Ingrédients issus de l\'agriculture biologique.'
+	},
+	{
+		id => '116',
+		lc => 'fr',
+		ingredients_text => 'émulsifiant : mono - et diglycérides d\'acides gras.'
+	},
+	{
+		id => '117',
+		lc => 'fr',
+		ingredients_text => 'Sucre. Fabriqué dans un atelier qui utilise des fruits à coques.'
+	},
+	{
+		id => '118',
+		lc => 'fr',
+		ingredients_text => 'Sucre. Fabriqué dans un atelier utilisant des fruits à coques et du sésame.'
+	},
+	{
+		id => '119',
+		lc => 'fr',
+		ingredients_text => 'Sucre. Fabriqué dans un atelier qui manipule du lait, de la moutarde et du céleri.'
+	},
+	{
+		id => '120',
+		lc => 'fr',
+		ingredients_text => 'Sucre. Peut contenir des fruits à coques et du sésame.'
+	},
+	{
+		id => '121',
+		lc => 'en',
+		ingredients_text => 'vegetable oil (coconut & rapeseed)'
+	},
+	{
+		id => '122',
+		lc => 'fr',
+		ingredients_text => 'Masse de cacao°, Quinoa° (1,8%). °Produits issus de l\'agriculture biologique.'
+	},
+	{
+		id => '123',
+		lc => 'de',
+		ingredients_text => 'Emulgator (Sojalecithine, Mono - und Diglyceride von Speisefettsäuren, Sorbitantristearat)'
+	},
+	{
+		id => '124',
+		lc => 'fr',
+		ingredients_text => 'Tomates* (20%). *Ingrédients Bio'
+	},
+	{
+		id => '125',
+		lc => 'fr',
+		ingredients_text => 'Tomates* (20%). *Ingrédients biologiques'
+	},
+	{
+		id => '126',
+		lc => 'fr',
+		ingredients_text =>
+			'Chocolat. Contient du lait et des noisettes. Peut contenir du blé, du soja et des crustacés.'
+	},
+	{
+		id => '127',
+		lc => 'en',
+		ingredients_text => 'Chocolate. Contains milk, hazelnuts and other nuts. May contain celery and mustard.'
+	},
+	{
+		id => '128',
+		lc => 'fr',
+		ingredients_text =>
+			'phosphates d\'ammonium et de calcium, Phosphate d\'aluminium et de sodium, diphosphate d\'aluminium et de sodium'
+	},
+	{
+		id => '129',
+		lc => 'fr',
+		ingredients_text => 'Ingrédient(s) : lentilles vertes* - *issu(e)(s) de l\'agriculture biologique.'
+	},
+	{
+		id => '130',
+		lc => 'en',
+		ingredients_text => 'S. thermophilus, L casei, L.bulgaricus'
+	},
+	{
+		id => '131',
+		lc => 'fr',
+		ingredients_text =>
+			'jus de citron*. *Ingrédients issus de l\'agriculture biologique Peut contenir : œuf, moutarde, graine de sésame, poisson,soja, lait,fruits à coque, céleri.'
+	},
+	{
+		id => '132',
+		lc => 'fr',
+		ingredients_text =>
+			'Farine, levure. Peut contenir des traces de _soja_, _amandes_, _noisettes_ et _noix de cajou_.'
+	},
+	{
+		id => '133',
+		lc => 'es',
+		ingredients_text => 'Agua, aceite de girasol*. * Ingredientes ecológicos.'
+	},
+	{
+		id => '134',
+		lc => 'es',
+		ingredients_text => 'Agua, aceite de girasol*, arroz* (5 %). (*) Ingredientes ecológicos.'
+	},
+	{
+		id => '135',
+		lc => 'es',
+		ingredients_text =>
+			'Tofu* 88% (agua, habas de soja*). *cumple con el reglamento de agricultura ecológica CE 2092/91'
+	},
+	{
+		id => '136',
+		lc => 'es',
+		ingredients_text => 'agua, almendra* (5,5%). *= procedentes de la agricultura ecológica'
+	},
+	{
+		id => '137',
+		comment => "test for bug #3273 that introduced unwanted separators before natural flavor",
+		lc => 'en',
+		ingredients_text => 'non-gmo natural flavor'
+	},
+	{
+		id => '138',
+		lc => 'en',
+		ingredients_text => 'vit. e, vitamins b2, B3 and K, vit d, vit a & c, vit. B12'
+	},
+	{
+		id => '139',
+		lc => 'fr',
+		ingredients_text => 'vit. pp, vit c, vit. a et b6'
+	},
+	{
+		id => '140',
+		lc => 'pl',
+		ingredients_text => 'witaminy A i D'
+	},
+	{
+		id => '141',
+		lc => 'fr',
+		ingredients_text =>
+			'colorant de surface : caramel ordinaire, agent de traitement de farine (E300), acide citrique'
+	},
+	{
+		id => '142',
+		lc => 'es',
+		ingredients_text =>
+			'Agua, edulcorantes (INS420, INS 960, INS N\'952, INS N°954, INS°950, INS N 955), conservantes (INS.218, INS #202, INS N 216).'
+	},
+	# Spanish Vitamin E can be mistaken for "e" meaning "and"
+	{
+		id => '143',
+		lc => 'es',
+		ingredients_text => 'Vitamina E y C'
+	},
+	{
+		id => '144',
+		lc => 'es',
+		ingredients_text => 'color E 124'
+	},
+	{
+		id => '145',
+		lc => 'es',
+		ingredients_text => 'colores E (124, 125)'
+	},
+	{
+		id => '146',
+		lc => 'it',
+		ingredients_text => 'vitamine A, B, E e K'
+	},
+	# Additives normalization
+	{
+		id => '147',
+		lc => 'en',
+		ingredients_text => 'E 102, E-104 color, E-101(i), E101 (ii), E160a(iv), e172-i, E-160 i'
+	},
+	{
+		id => '148',
+		lc => 'fr',
+		ingredients_text => 'E102-E1400'
+	},
+	{
+		id => '149',
+		lc => 'de',
+		ingredients_text => 'E172i-E174ii, E102(i)-E101i'
+	},
+	{
+		id => '150',
+		lc => 'fr',
+		ingredients_text => 'correcteurs d\'acidité : E322/E333 E474-E475'
+	},
+	{
+		id => '151',
+		lc => 'es',
+		ingredients_text => 'E-330; E-331; Estabilizantes (E-327; E-418)'
+	},
+	{
+		id => '152',
+		lc => 'es',
+		ingredients_text => 'E120 color'
+	},
+	{
+		id => '153',
+		lc => 'es',
+		ingredients_text => 'E172-i'
+	},
+	{
+		id => '154',
+		lc => 'es',
+		ingredients_text => 'E172 i'
+	},
+	{
+		id => '155',
+		lc => 'es',
+		ingredients_text => '(E172i)'
+	},
+	{
+		id => '156',
+		lc => 'es',
+		ingredients_text => 'E102(i)-E101i'
+	},
+	{
+		id => '157',
+		lc => 'es',
+		ingredients_text => 'E102(i)'
+	},
+	{
+		id => '158',
+		lc => 'es',
+		ingredients_text => 'S.I.N.:160 b'
+	},
+	{
+		id => '159',
+		lc => 'pt',
+		ingredients_text => 'estabilizadores (E 422, E 412)'
+	},
+	{
+		id => '160',
+		lc => 'es',
+		ingredients_text => 'contiene apio y derivados de leche'
+	},
+	{
+		id => '161',
+		lc => 'fr',
+		ingredients_text => 'E160a(ii)'
+	},
+	{
+		id => '162',
+		lc => 'fr',
+		ingredients_text => '(E160a-ii)'
+	},
+	{
+		id => '163',
+		lc => 'fr',
+		ingredients_text => 'colorant (E160a(ii))'
+	},
+	{
+		id => '164',
+		comment => "# do not separate acide acétique into acide : acétique",
+		lc => 'fr',
+		ingredients_text => 'Esters glycéroliques de l\'acide acétique et d\'acides gras'
+	},
+	{
+		id => '165',
+		lc => 'fr',
+		ingredients_text => 'acide acétique'
+	},
+	# russian abbreviations
+	{
+		id => '166',
+		lc => 'ru',
+		ingredients_text => 'мука пшеничная х/п в/с'
+	},
+	# w/ with and w/o without abbreviations
+	{
+		id => '167',
+		lc => 'en',
+		ingredients_text => 'Organic garbanzo beans (cooked w/o salt), water'
+	},
+	{
+		id => '168',
+		lc => 'en',
+		ingredients_text => 'sugar, cocoa (processed w/alkali), egg yolk'
+	},
+	# * ingrédient issu..
+	{
+		id => '169',
+		lc => 'fr',
+		ingredients_text => 'LAIT entier pasteurisé*. *ingrédient issu de l\'agriculture biologique.'
+	},
+	{
+		id => '170',
+		lc => 'fr',
+		ingredients_text => 'vitamines B1, B6, B9, PP et E'
+	},
+	{
+		id => '171',
+		lc => 'fr',
+		ingredients_text => 'vitamines (B1, acide folique (B9))'
+	},
+	# (origins, contains milk)
+	{
+		id => '172',
+		lc => 'en',
+		ingredients_text => 'Chocolate (Italy, contains milk)'
+	},
+	{
+		id => '173',
+		lc => 'en',
+		ingredients_text => 'Chocolate (contains milk)'
+	},
+	{
+		id => '174',
+		lc => 'en',
+		ingredients_text => 'Chocolate. Contains (milk)'
+	},
+	# ¹ and ² symbols
+	{
+		id => '175',
+		lc => 'fr',
+		ingredients_text =>
+			'Sel, sucre², graisse de palme¹, amidons¹ (maïs¹, pomme de terre¹), oignon¹ : 8,9%, ail¹, oignon grillé¹ : 1,4%, épices¹ et aromate¹ (livèche¹ : 0,4%, curcuma¹, noix de muscade¹), carotte¹ : 0,5%. Peut contenir : céleri, céréales contenant du gluten, lait, moutarde, œuf, soja. ¹Ingrédients issus de l\'Agriculture Biologique. ² Ingrédients issus du commerce équitable'
+	},
+	{
+		id => '176',
+		comment => "# Russian е character",
+		lc => 'ru',
+		ingredients_text => 'е322, Куркумины e100, е-1442, (е621)'
+	},
+	# New ingredients categories + types : generalized from French to other languages
+	{
+		id => '177',
+		lc => 'fr',
+		ingredients_text => 'huiles végétales (palme, olive et tournesol)'
+	},
+	{
+		id => '178',
+		lc => 'fr',
+		ingredients_text => 'huile végétale : colza'
+	},
+	{
+		id => '179',
+		lc => 'fr',
+		ingredients_text => 'huile végétale : colza, fraises'
+	},
+	{
+		id => '180',
+		lc => 'fr',
+		ingredients_text => 'huile végétale : colza et tomates'
+	},
+	{
+		id => '181',
+		lc => 'en',
+		ingredients_text => 'vegetable oil: sunflower'
+	},
+	{
+		id => '182',
+		lc => 'en',
+		ingredients_text => 'vegetable oil (palm)'
+	},
+	{
+		id => '183',
+		lc => 'en',
+		ingredients_text => 'vegetable oils (palm, olive)'
+	},
+	{
+		id => '184',
+		lc => 'en',
+		ingredients_text => 'organic vegetable oils (sunflower, colza and rapeseed)'
+	},
+	# used to have bad output: sunflower vegetable oils, colza vegetable oilsand strawberry
+	{
+		id => '185',
+		lc => 'en',
+		ingredients_text => 'vegetable oils : sunflower, colza and strawberry'
+	},
+	# Polish oils
+	{
+		id => '186',
+		lc => 'pl',
+		ingredients_text => 'oleje roślinne (słonecznikowy)'
+	},
+	{
+		id => '187',
+		lc => 'pl',
+		ingredients_text => 'oleje roślinne: słonecznikowy'
+	},
+	{
+		id => '188',
+		lc => 'pl',
+		ingredients_text => 'oleje roślinne (słonecznikowy, rzepakowy)'
+	},
+	{
+		id => '189',
+		lc => 'pl',
+		ingredients_text => 'oleje roślinne (sojowy, słonecznikowy, kokosowy, rzepakowy) w zmiennych proporcjach'
+	},
+	# Polish meats
+	{
+		id => '190',
+		lc => 'pl',
+		ingredients_text => 'tłuszcze roślinne (palmowy nieutwardzony, shea)'
+	},
+	# Polish juices and concentrates
+	{
+		id => '191',
+		lc => 'pl',
+		ingredients_text => 'tłuszcze roślinne (kokosowy i palmowy) w zmiennych proporcjach'
+	},
+	{
+		id => '192',
+		lc => 'pl',
+		ingredients_text => 'mięso (wołowe, wieprzowe, cielęce)'
+	},
+	{
+		id => '193',
+		lc => 'pl',
+		ingredients_text => 'przeciery z (jabłek, bananów, marchwi)'
+	},
+	# Russian oils (more tests needed)
+	{
+		id => '194',
+		lc => 'ru',
+		ingredients_text => 'масло (Подсолнечное)'
+	},
+	{
+		id => '195',
+		lc => 'ru',
+		ingredients_text => 'Масло (подсолнечное)'
+	},
+	{
+		id => '196',
+		lc => 'ru',
+		ingredients_text => 'масло растительное (подсолнечное, соевое)'
+	},
+	# grammes -> g
+	{
+		id => '197',
+		lc => 'fr',
+		ingredients_text => 'Teneur en fruits: 50gr pour 100 grammes'
+	},
+	# test conflicts between the word "and" in some languages and additives variants. With letters i or e or a.
+	{
+		id => '198',
+		lc => 'hr',
+		ingredients_text => 'bojilo: E 150a, tvari za rahljenje: E 500 i E 503, sol.'
+	},
+	{
+		id => '199',
+		lc => 'hr',
+		ingredients_text => 'bojilo: E 150a, tvari za rahljenje: E 500 i, E 503, sol.'
+	},
+	{
+		id => '200',
+		lc => 'hr',
+		ingredients_text => 'bojilo: E 150a, tvari za rahljenje: E 500(i), E 503, sol.'
+	},
+	{
+		id => '201',
+		lc => 'hr',
+		ingredients_text => 'bojilo: E 150a, tvari za rahljenje: E 500i, E 503, sol.'
+	},
+	{
+		id => '202',
+		lc => 'it',
+		ingredients_text => 'formaggio, E 472 e, E470a.'
+	},
+	{
+		id => '203',
+		lc => 'it',
+		ingredients_text => 'formaggio, E 472 e E470a.'
+	},
+	{
+		id => '204',
+		lc => 'sk',
+		ingredients_text => 'syr, E470 a E470a, mlieko.'
+	},
+	# normalize category and types
+	{
+		id => '205',
+		lc => 'fr',
+		ingredients_text => 'Piments (vert, rouge, jaune)'
+	},
+	{
+		id => '206',
+		lc => 'de',
+		ingredients_text => 'pflanzliches Fett (Kokosnuss, Palmkern)'
+	},
+	{
+		id => '207',
+		lc => 'de',
+		ingredients_text => 'pflanzliche Öle und Fette (Raps, Palm, Shea, Sonnenblumen)'
+	},
+	{
+		id => '208',
+		lc => 'fr',
+		ingredients_text => 'Huiles végétales de palme, de colza et de tournesol'
+	},
+	{
+		id => '209',
+		lc => 'fr',
+		ingredients_text => 'arôme naturel de pomme avec d\'autres âromes'
+	},
+	{
+		id => '210',
+		lc => 'fr',
+		ingredients_text => 'Carbonate de magnésium, fer élémentaire'
+	},
+	{
+		id => '211',
+		lc => 'fr',
+		ingredients_text => 'huile végétale (colza)'
+	},
+	{
+		id => '212',
+		lc => 'fr',
+		ingredients_text => 'huile végétale : colza'
+	},
+	{
+		id => '213',
+		lc => 'hr',
+		ingredients_text => 'ječmeni i pšenični slad'
+	},
+	{
+		id => '214',
+		lc => 'hr',
+		ingredients_text => 'ječmeni, ječmeni i pšenični slad'
+	},
+	{
+		id => '215',
+		lc => 'hr',
+		ingredients_text => 'Pasterizirano mlijeko (s 1.0% mliječne masti)'
+	},
+	{
+		id => '216',
+		lc => 'en',
+		ingredients_text => 'Vegetal oil (sunflower, olive and palm)'
+	},
+	{
+		id => '217',
+		lc => 'en',
+		ingredients_text => 'vegetable oil (palm)'
+	},
+	{
+		id => '218',
+		lc => 'en',
+		ingredients_text => 'vegetable oil: palm'
+	},
+	{
+		id => '219',
+		lc => 'fr',
+		ingredients_text => 'protéines végétales (soja, blé)'
+	},
+	{
+		id => '220',
+		lc => 'de',
+		ingredients_text => 'pflanzliche Proteine (Erbsen, Sonnenblumen)'
+	},
+	# Should not develop the enumeration if it contains unknown types (like "sel" here)
+	{
+		id => '221',
+		lc => 'fr',
+		ingredients_text => 'Piments (vert, rouge, jaune, sel)'
+	},
+	{
+		id => '222',
+		lc => 'fr',
+		ingredients_text => 'Huile de palme, noisettes et tournesol'
+	},
+	{
+		id => '223',
+		lc => 'fr',
+		ingredients_text => 'Huile de palme, noisettes'
+	},
+	{
+		id => '224',
+		lc => 'fr',
+		ingredients_text => 'arôme naturel de citron, citron vert et d\'autres agrumes'
+	}
+);
+
+foreach my $test_ref (@tests) {
+
+	my $testid = $test_ref->{id};
+	my $l = $test_ref->{lc};    # Language
+	my $ingredients_text = $test_ref->{ingredients_text};
+	my $preparsed = preparse_ingredients_text($l, $ingredients_text);
+	$test_ref->{preparsed_ingredients_text} = $preparsed;
+
+	compare_to_expected_results($test_ref, "$expected_result_dir/$testid.json", $update_expected_results);
+}
+
+done_testing();