Regex editor for regex extraction rows

Search, filter, edit, diagnose, and batch translate regex extraction rows inside the plugin editor’s Regex workspace.

The Regex editor works on the current translation file’s dict[file].regex rows. It shares the same plugin editor shell, file list, save button, and source preview panel as the AST editor, but it shows regex extraction results instead of AST nodes.

Open the Regex editor

1. Open the i18n Manager

   Click the globe icon in the sidebar.

2. Open a plugin editor

   In the plugin translation flow, open the plugin you want to edit.

3. Switch to Regex

   In the switcher card on the left side of the editor, change AST / Regex to Regex.

Regex is a workspace inside the plugin editor. It is not a separate manager tab and it is not the settings page for regex extraction rules.

Table structure and editing

The Regex table has exactly three columns:

Column	Actual content
Source	the row’s source
Target	the row’s target, editable inline
Actions	reset the current row, or delete the current row

Click a target cell to edit it inline. When the cell loses focus, the new text is written back into the current editor state.

• reset sets target back to source
• the reset button is disabled when target === source
• the table shows an empty state when the current file has no regex rows

Disk persistence still depends on the save flow. With auto-save enabled, the editor triggers save 500ms after editing stops; otherwise you need to click the save button on the left side of the editor.

Search and filters

The search box only matches these two fields:

• source
• target

The filter has three states:

State	Condition
All	no extra filtering
Translated	target exists and target !== source
Untranslated	!target or target === source

When you switch files, the Regex editor clears the search text, resets the filter to all, and clears the current editing row.

Sidebar capabilities

The right sidebar contains a stats card, a manual insert card, a quick actions card, a diagnose card, and an AI translation card. The stats, insert, quick actions, and AI cards can be toggled from the sidebar view menu. The diagnose card is always shown.

Stats card

The stats card calculates these values from the current file’s regex rows:

• total row count
• translated row count
• untranslated row count
• progress percentage

For this card, a row is translated only when target exists and target !== source.

Manual insert card

The 新增规则 card adds one regex row to the current file.

• the source field is validated with new RegExp(pattern)
• the add button is disabled while source is invalid
• when target is empty, the stored value becomes source
• Ctrl+Enter submits the current row

Quick actions card

The quick actions card exposes three actions:

• open the current source file
• run incremental extraction for the current file
• delete untranslated rows

Incremental extraction reads the current source file, calls RegexTranslator(i18n.settings).extractTranslationsByRegex(...), and merges the result with existing rows through mergeRegexItems(...). It does not replace the full table unconditionally.

If the current plugin is already applied, incremental extraction is disabled in the UI and blocked again in the extraction logic.

Delete untranslated keeps only rows that meet all of these conditions:

• target exists
• target !== source
• target.trim() !== ''

Diagnose card

The diagnose card supports three scan modes:

• syntax diagnose
• unused scan
• security scan

Results can jump directly to the matching row. The Regex table highlights affected rows as error, unused, or security.

The diagnose card also provides these actions:

• clear current diagnose results
• delete unused rows
• restore all error rows
• run AI fix for one row when the issue is neither unused nor security-related

AI batch translation

The Regex AI panel exposes these controls:

• target language
• style
• batch size
• concurrency limit
• timeout
• overwrite existing targets or not

These controls read from and write back to the global settings fields language, llmLanguage, llmStyle, llmBatchSize, llmConcurrencyLimit, and llmTimeout.

When overwrite = false, batch translation only includes rows whose target is:

• empty
• whitespace-only
• equal to source

Before requests are sent, the panel calls estimateTokens(targetItems, 'regex') to show the estimated token usage and cost. The running task can also be stopped from the same panel.

How does the Regex editor relate to regex extraction settings?

The Regex editor maintains rows that have already been extracted. Regex extraction itself reads the current settings fields reFlags, reLength, reDatas, reRejectRe, and reValidRe. The Regex AI prompt uses llmRegexPrompt.