Thread Tagging Improvements 1.0.7 Patch Level 5

Thanks to @KensonPlays for reporting thing

There was a bug where we queried the thread title with an ambiguous column name, which works fine as long as no other add-on does a JOIN with another table that has a similarly named column. (e.g. XFA Tournaments addon)

TL;DR; Fixed our add-on to explicitly query for xf_thread.title instead of title so SQL knows exactly which one we want.

• 2 new advanced options were added to allow change of:
  • Max amount of thread titles are loaded at once for processing
  • Chunk size when bulk inserting the collected (up to 50K) words

• Don't call the tag-suggest endpoint if tagging is not allowed, should fix @Faust problem
• Safer limits when collecting potential words, hopefully this fixes @vwts problem on the larger forum
  • Batch load 50k thread titles instead of 100K (doubt this is the problem)
  • Batch the bulk insert of potential words 5k at a time, before could do one bulk insert of 50K items.

Moves the Rebuilding of potential words job into postUpgrade() in case that is what makes the entity not behave correctly there...

• Potential Tag Word data is now stored in a database table.
• Sorting potential words now allow ordering by:
  • Use count
  • Word count & use count (i.e. multiple words showing first)
  • Alphabetical
• "Loner tags": Tags that should not be tagged if a similar, longer tag is already matching.
  • This can be useful if you don't want short tags to be tagged alongside similar tags that are longer (more specific) and contain the same tag word, see the example below

In this example, a thread title such as "What are common elvish names" would only be auto-tagged/suggested with the elvish names-tag...
If the elvish-tag was not a "loner"-tag, both elvish names & elvish would be added as tags to the thread.

Loner tags can be specified when editing the tag:

• The potential tags page now shows words that consist of multiple words
  • New option: Potential tags max word count, (default 3)
  • New option: Allow stop words in multi-worded potential tags (default false)
• Options interface improvements
  • Tabbed interface to better group options
  • Button to save + rebuild potential words in sections where it will be needed
• New advanced option: Regex used to split words
  • The default regex might not fit your language/needs, so it can be overridden if needed.
• Allow adding a word as a synonym directly from the Potential tags page
• New option: Improve matching for accented characters (default off)
  • If your forum has content that uses accented characters, this might be useful to turn on.
• Allow swapping tag and synonym values when editing without it raising a warning.
• Other minor tweaks to make the UI/UX better.

Small bugfix in the previous release where it could sometimes collect numbers that would later raise errors in the potential tags page

• Option to set the tag as a synonym of the target tag when merging tags
• Support for other character sets than ASCII
• Open add/edit tag pages in dialogues rather than on a new page.
• When adding/editing/merging/deleting tags, return to the same page you started on
  • The default XF logic is to go back to the first/start page of tags management, which is quite annoying if you are editing tags on e.g. the 3rd page
  • To enable this feature fully, the addon redirects filtering/ordering POST requests to the GET-equivalent.. This might not be compatible with other tagging addons, and if you experience problems on the tag page you can disable this feature under the addon options
• Bugfix for breadcrumbs rendering badly on Cyrillic chars
• Bugfix to flush synonym-map on add/delete of tags also.

Shoutout to @findozor for reporting bugs and helping with testing of fixes

• New feature that automatically populates the tags field from the title
  • Can be disabled in options (It is enabled by default to promote tag usage)
  • Option to limit amount of tags that are suggested (defaults to 5)
• Option to always show the tags field (enabled by default to promote tag usage)
  • It is otherwise hidden on "quick thread"-mode unless "minimum tags required" is defined..
• Tags now can have synonyms, that are used in following scenarios:
  • When suggesting tags during thread creation.
  • When searching for tags it suggests the "real tag" in the autocomplete
    • If someone by force tries to use/add the synonym, it will show a message to use the correct name
  • When searching for potential threads to be tagged it searches tag name and all synonyms for matches
  • If selecting "Automatically tag threads" when adding/editing a tag it will also auto-tag using synonym words
  • When searching for tags in admin interface it also searches over synonyms for matches.
  • If a tag is not found when going to example.com/tags/something it will try to look if synonym matches and redirect to the "real tag" page.

Adding/Editing tags:

Tag overview:

Auto-suggest:

Autocomplete: