@ocean90 I've made a big change by separating terms by type, with more clear lists that we can keep or comment out, for testing.

Let me know your thoughts on this please.

I like the direction where this goes! But please make sure you update your branch with the latest changes from #1395 as it changes how gp_sort_glossary_entries_terms(), now gp_glossary_add_suffixes(), works.

@pedro-mendonca Do you plan to continue working on this? Do you need any help?

Hi, I do, just had no time yet. I'll get back to it.
Thanks!

@ocean90 I have a .po with a set of originals and a .csv for the glossary to test this.
I can share here if you want.
I'll do some unit tests for all these cases.

After looking at most rules, I've made an array of rules with the same structure:

array(
  'endings'  => array(
    'y',  // Lady.
  ),
  'preceded' => '[b-df-hj-np-tv-xz]',  // Preceded by any consonant.
  'change'   => 'i',                   // Change to 'i'.
  'add'      => 'es',                  // Add 'es'.
),

Then I made a single logic that runs on every set of rules, for every part_of_speech type.

We can add rules or remove them by adding or removing these arrays.

Feedback please :)

@ocean90 I see two possible places where to put the unit tests:
https://github.com/GlotPress/GlotPress/blob/develop/tests/phpunit/testcases/test_template_helper_functions.php
and
https://github.com/GlotPress/GlotPress/blob/develop/tests/phpunit/testcases/tests_things/test_thing_glossary_entry.php
Which one should we go for?

@ocean90 I've added all the Unit Tests for the suffixes, for nouns and verbs.
Feel free to review when you can.

@amieiro I've completed the full update related to your above suggestions.

NOUNS

Here is a summary:

4 ❌ Incorrect. Ending with '-an'. Suffix: '-en'. Words ending with 'an" add a "s" to make the plural. Some examples:
Plan -> Plans, Can -> Cans, Pan -> Pans, Fan -> Fans, Clan -> Clans, Span -> Spans, Human -> Humans.
In English we have some words with irregular plurals (Woman -> Women, Man -> Men, Child -> Children, Foot -> Feet, Tooth -> Teeth). I think it doesn't make sense to add a full list of irregular plurals.
We should remove the unit test test_map_glossary_entries_to_translation_originals_with_nouns_ending_with_an_in_glossary (Test 5)

Fixed, removed all the entries and tests, old and new, about the irregular noun plurals for 'man' and 'woman'.

6 ❌ Incorrect. Fallback. We have this regex: "\w(?<!z|x|sh|ch|s|y|an|fe)". We need to remove the "an" (point 4) and add the "f" (point 5). This is the proposed regex: "\w(?<!z|x|sh|ch|s|y|f|fe)". Adds "s".
Visual examples:
chef -> chefs.
Unit test: test_map_glossary_entries_to_translation_originals_with_nouns_fallback_not_ending_with_s_in_glossary (Test 8)

Fixed, removed the exception 'an'. No need to add 'f', as it's matched by the fallback default '-s'.

❓Comment. Have you considered the plural of some "-s" and "-z" endings? For some nouns that end in "-s" or "-z", you have to double the "-s" or "-z" and add "-es". For example:
fez –> fezzes
gas –> gasses
As it is not a rule for all endings, we can avoid it.

Not added, as it's not a regular case.

VERBS

1 ❌ Incorrect. Third-person singular. Ending in "-ss," "-z," "-x," "-sh," "-ch," or "-tch." -> Add "es". "-tch" considered the same than "-ch". We need to add the "-s" situation (e.g. bias->biases).
Visual examples:
Pass -> passes.
Quiz -> quizes. (The correct plural is quizzes).
Fix -> fixes.
Push -> pushes.
Watch -> watches.
Unit test: test_map_glossary_entries_to_translation_originals_with_verbs_3rdperson_ending_with_sibilant_in_glossary (Test 9)
Example to add:
focus -> focuses.

Fixed, removed the '-ss' and added '-s' to match both situations. Added case bias and focus.

❓Comment. Here we have some examples not checked:
Base forms which end in vowel + single consonant. This depends on the stressed syllable, so we can create both, as we won't be able to know the stressed syllable.
commit. -> committed, commited (incorrect), committing, commiting (incorrect)
Vowel + l. The consonant is doubled if the base form ends in a vowel + l, whether the last syllable is stressed or not. I think we can't double the vowel with the current approach.
travel -> travelling, travelled

Fixed: Now the ending values don't use null if no changes, now it has support for placeholders (%s or %1$s) to allow using the current ending, and duplicating for the case of doubling the ending consonants (committed, committing, etc.). The vowel-l case is matched in the added rules. As you've mentioned, it's not simple to detect the stressed syllable, so there are two matches for both single and double ending consonant.

TESTING

Created the function check_map_glossary() to simplify the testing for the functions map_glossary_entries_to_translation_originals......().

At last, do you think it would be usefull to add to the tests/phpunit/data a testing .pot and a glossary .csv despite not being used currently on any phpunit tests?

Thank you for your compreehensive review!

Related with the .pot and .csv files in the tests/phpunit/data, as we don't use it in the automatic test, I suggest you to remove it from the PR and add it to a GitHub Gist or to a own repo. I think a good option is to add them in a GitHub Gist, so we can update it easily and discuss. But I think the best option will be to read both files in each test and using their data instead of hard-coding them (we can do this in a different PR).

Hi @amieiro, thanks for your review!!
I've removed both glossary testing example files, .pot and .csv.
I've also added the missing types from check_map_glossary().