"isJapanese" check move (#730)

* Move isStringPartiallyJapanese out of ClipboardMonitor

* Create isStringPartiallyJapanese function

* Add textMayBeTranslatable

* Rename API function

* Rename internal function

* Add helper

* Update translatable check

* Pass language to TextScanner

* Pass language explicitly

* Use textMayBeTranslatable

* No redundant translatable check

* Update eslint

* Remove double newline

* Collapse

* Rename

5 files changed, 49 insertions, 18 deletions

diff --git a/ext/js/display/display-generator.js b/ext/js/display/display-generator.js
index 22912e9f..0b3236e9 100644
--- a/ext/js/display/display-generator.js
+++ b/ext/js/display/display-generator.js
@@ -20,7 +20,8 @@ import {ExtensionError} from '../core/extension-error.js';
 import {isObject} from '../core/utilities.js';
 import {getDisambiguations, getGroupedPronunciations, getTermFrequency, groupKanjiFrequencies, groupTermFrequencies, groupTermTags, isNonNounVerbOrAdjective} from '../dictionary/dictionary-data-util.js';
 import {HtmlTemplateCollection} from '../dom/html-template-collection.js';
-import {distributeFurigana, getKanaMorae, getPitchCategory, isCodePointKanji, isStringPartiallyJapanese} from '../language/ja/japanese.js';
+import {distributeFurigana, getKanaMorae, getPitchCategory, isCodePointKanji} from '../language/ja/japanese.js';
+import {getLanguageFromText} from '../language/text-utilities.js';
 import {createPronunciationDownstepPosition, createPronunciationGraph, createPronunciationText} from './sandbox/pronunciation-generator.js';
 import {StructuredContentGenerator} from './sandbox/structured-content-generator.js';
 
@@ -991,12 +992,7 @@ export class DisplayGenerator {
      * @param {string} [language]
      */
     _setTextContent(node, value, language) {
-        if (typeof language === 'string') {
-            node.lang = language;
-        } else if (isStringPartiallyJapanese(value)) {
-            node.lang = 'ja';
-        }
-
+        this._setElementLanguage(node, language, value);
         node.textContent = value;
     }
 
@@ -1008,11 +1004,7 @@ export class DisplayGenerator {
     _setMultilineTextContent(node, value, language) {
         // This can't just call _setTextContent because the lack of <br> elements will
         // cause the text to not copy correctly.
-        if (typeof language === 'string') {
-            node.lang = language;
-        } else if (isStringPartiallyJapanese(value)) {
-            node.lang = 'ja';
-        }
+        this._setElementLanguage(node, language, value);
 
         let start = 0;
         while (true) {
@@ -1029,6 +1021,22 @@ export class DisplayGenerator {
     }
 
     /**
+     * @param {HTMLElement} element
+     * @param {string|undefined} language
+     * @param {string} content
+     */
+    _setElementLanguage(element, language, content) {
+        if (typeof language === 'string') {
+            element.lang = language;
+        } else {
+            const language2 = getLanguageFromText(content);
+            if (language2 !== null) {
+                element.lang = language2;
+            }
+        }
+    }
+
+    /**
      * @param {string} reading
      * @param {import('dictionary').TermPronunciation[]} termPronunciations
      * @param {string[]} wordClasses
diff --git a/ext/js/display/display.js b/ext/js/display/display.js
index f6efb5ac..80f5e9ae 100644
--- a/ext/js/display/display.js
+++ b/ext/js/display/display.js
@@ -425,6 +425,7 @@ export class Display extends EventDispatcher {
             readingMode: options.parsing.readingMode,
             useInternalParser: options.parsing.enableScanningParser,
             useMecabParser: options.parsing.enableMecabParser,
+            language: options.general.language,
             scanning: {
                 inputs: scanningOptions.inputs,
                 deepContentScan: scanningOptions.deepDomScan,
@@ -1834,6 +1835,7 @@ export class Display extends EventDispatcher {
         }
 
         const {scanning: scanningOptions, sentenceParsing: sentenceParsingOptions} = options;
+        this._contentTextScanner.language = options.general.language;
         this._contentTextScanner.setOptions({
             inputs: [{
                 include: 'mouse0',
diff --git a/ext/js/display/query-parser.js b/ext/js/display/query-parser.js
index d27b9394..f6c26ce7 100644
--- a/ext/js/display/query-parser.js
+++ b/ext/js/display/query-parser.js
@@ -92,7 +92,7 @@ export class QueryParser extends EventDispatcher {
     /**
      * @param {import('display').QueryParserOptions} display
      */
-    setOptions({selectedParser, termSpacing, readingMode, useInternalParser, useMecabParser, scanning}) {
+    setOptions({selectedParser, termSpacing, readingMode, useInternalParser, useMecabParser, language, scanning}) {
         let selectedParserChanged = false;
         if (selectedParser === null || typeof selectedParser === 'string') {
             selectedParserChanged = (this._selectedParser !== selectedParser);
@@ -115,6 +115,7 @@ export class QueryParser extends EventDispatcher {
             if (typeof scanLength === 'number') {
                 this._scanLength = scanLength;
             }
+            this._textScanner.language = language;
             this._textScanner.setOptions(scanning);
         }
         this._textScanner.setEnabled(true);
diff --git a/ext/js/display/sandbox/structured-content-generator.js b/ext/js/display/sandbox/structured-content-generator.js
index 1dfde39b..90a47158 100644
--- a/ext/js/display/sandbox/structured-content-generator.js
+++ b/ext/js/display/sandbox/structured-content-generator.js
@@ -16,7 +16,7 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 
-import {isStringPartiallyJapanese} from '../../language/ja/japanese.js';
+import {getLanguageFromText} from '../../language/text-utilities.js';
 
 export class StructuredContentGenerator {
     /**
@@ -163,8 +163,11 @@ export class StructuredContentGenerator {
         if (typeof content === 'string') {
             if (content.length > 0) {
                 container.appendChild(this._createTextNode(content));
-                if (language === null && isStringPartiallyJapanese(content)) {
-                    container.lang = 'ja';
+                if (language === null) {
+                    const language2 = getLanguageFromText(content);
+                    if (language2 !== null) {
+                        container.lang = language2;
+                    }
                 }
             }
             return;
diff --git a/ext/js/display/search-display-controller.js b/ext/js/display/search-display-controller.js
index e23d5d50..00f5efc6 100644
--- a/ext/js/display/search-display-controller.js
+++ b/ext/js/display/search-display-controller.js
@@ -103,7 +103,7 @@ export class SearchDisplayController {
         this._searchBackButton.addEventListener('click', this._onSearchBackButtonClick.bind(this), false);
         this._wanakanaEnableCheckbox.addEventListener('change', this._onWanakanaEnableChange.bind(this));
         window.addEventListener('copy', this._onCopy.bind(this));
-        this._clipboardMonitor.on('change', this._onExternalSearchUpdate.bind(this));
+        this._clipboardMonitor.on('change', this._onClipboardMonitorChange.bind(this));
         this._clipboardMonitorEnableCheckbox.addEventListener('change', this._onClipboardMonitorEnableChange.bind(this));
         this._display.hotkeyHandler.on('keydownNonHotkey', this._onKeyDown.bind(this));
 
@@ -271,9 +271,26 @@ export class SearchDisplayController {
     }
 
     /** @type {import('application').ApiHandler<'searchDisplayControllerUpdateSearchQuery'>} */
-    _onExternalSearchUpdate({text, animate = true}) {
+    _onExternalSearchUpdate({text, animate}) {
+        void this._updateSearchFromClipboard(text, animate, false);
+    }
+
+    /**
+     * @param {import('clipboard-monitor').Events['change']} event
+     */
+    _onClipboardMonitorChange({text}) {
+        void this._updateSearchFromClipboard(text, true, true);
+    }
+
+    /**
+     * @param {string} text
+     * @param {boolean} animate
+     * @param {boolean} checkText
+     */
+    async _updateSearchFromClipboard(text, animate, checkText) {
         const options = this._display.getOptions();
         if (options === null) { return; }
+        if (checkText && !await this._display.application.api.isTextLookupWorthy(text, options.general.language)) { return; }
         const {clipboard: {autoSearchContent, maximumSearchLength}} = options;
         if (text.length > maximumSearchLength) {
             text = text.substring(0, maximumSearchLength);