hoverlooknary av kthenrique

• Detects the hovered word in normal webpage text
• Selecting a word (double-click or drag) looks that exact text up instead of the hovered token, which helps when the hovered token is split across inline elements (e.g. play + ground for playground)
• Detects the page language from the `` attribute
• Queries the matching Wiktionary MediaWiki API for the word's wikitext
• Parses the page's language section for IPA, part of speech, and the first short definitions
• Shows a tooltip with the word and its IPA on the right, and one line per entry: <word class> <first meaning> (IPA is omitted when the page has no pronunciation data)
• When a word is an inflected/derived form (e.g. a conjugated verb), the root lemma (e.g. the infinitive) is fetched from Wiktionary and shown underneath the derived word
• Results are cached in memory for the session; requests are debounced and aborted while the pointer is moving

The parser is config-driven; each language carries its own section headings, POS markers, pronunciation templates, and cleanup rules. Supported out of the box:

[cols="1,1,3", options="header"]
|===
| Code | Wiktionary | Notes

| pt
| pt.wiktionary
|

| en
| en.wiktionary
|

| de
| de.wiktionary
|

| fr
| fr.wiktionary
| Reference behavior

| es
| es.wiktionary
| IPA only when the page provides it

| nl
| nl.wiktionary
|

| it
| it.wiktionary
|
|===

Pages whose `` maps to an unsupported code fall back to English.