make sentence parser works for spaced languages

yomidevs · Nov 27, 2024 · 62d9461 · 62d9461
1 parent 3c3f858
commit 62d9461
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 5 deletions.
diff --git a/ext/css/visibility-modifiers.css b/ext/css/visibility-modifiers.css
@@ -10,6 +10,9 @@
 :root:not([data-language=ja]):not([data-language=zh]):not([data-language=yue]) .jpzhyue-only {
     display: none;
 }
+:root:not([data-language=ja]):not([data-language=zh]):not([data-language=yue]):not([data-language=ko]) .jpzhyueko-only {
+    display: none;
+}
 :root:is([data-language=ja], [data-language=zh], [data-language=yue], [data-language=ko]) .not-jpzhyueko {
     display: none;
 }
diff --git a/ext/js/display/query-parser.js b/ext/js/display/query-parser.js
@@ -20,7 +20,7 @@ import {EventDispatcher} from '../core/event-dispatcher.js';
 import {log} from '../core/log.js';
 import {querySelectorNotNull} from '../dom/query-selector.js';
 import {convertHiraganaToKatakana, convertKatakanaToHiragana, isStringEntirelyKana} from '../language/ja/japanese.js';
-import {TextScanner} from '../language/text-scanner.js';
+import {NO_SPACES_LANGUAGES, TextScanner} from '../language/text-scanner.js';
 
 /**
  * @augments EventDispatcher<import('query-parser').Events>
@@ -43,6 +43,8 @@ export class QueryParser extends EventDispatcher {
         this._setTextToken = null;
         /** @type {?string} */
         this._selectedParser = null;
+        /** @type {boolean} */
+        this._shouldInsertSpacesBetweenTerms = false;
         /** @type {import('settings').ParsingReadingMode} */
         this._readingMode = 'none';
         /** @type {number} */
@@ -100,6 +102,9 @@ export class QueryParser extends EventDispatcher {
         }
         if (typeof termSpacing === 'boolean') {
             this._queryParser.dataset.termSpacing = `${termSpacing}`;
+            if (!NO_SPACES_LANGUAGES.has(language)) {
+                this._shouldInsertSpacesBetweenTerms = true;
+            }
         }
         if (typeof readingMode === 'string') {
             this._setReadingMode(readingMode);
@@ -305,15 +310,22 @@ export class QueryParser extends EventDispatcher {
             termNode.dataset.offset = `${offset}`;
             for (const {text, reading} of term) {
                 const trimmedText = text.trim();
+                if (trimmedText.length === 0) { continue; }
                 if (reading.length === 0) {
                     termNode.appendChild(document.createTextNode(trimmedText));
                 } else {
                     const reading2 = this._convertReading(trimmedText, reading);
                     termNode.appendChild(this._createSegment(trimmedText, reading2, offset));
                 }
                 offset += trimmedText.length;
+                if (this._shouldInsertSpacesBetweenTerms) {
+                    termNode.appendChild(document.createTextNode(' '));
+                    offset += 1;
+                }
+            }
+            if (termNode.childNodes.length > 0) {
+                fragment.appendChild(termNode);
             }
-            fragment.appendChild(termNode);
         }
         return fragment;
     }

diff --git a/ext/js/language/text-scanner.js b/ext/js/language/text-scanner.js
@@ -24,7 +24,7 @@ import {clone} from '../core/utilities.js';
 import {anyNodeMatchesSelector, everyNodeMatchesSelector, getActiveModifiers, getActiveModifiersAndButtons, isPointInSelection} from '../dom/document-util.js';
 import {TextSourceElement} from '../dom/text-source-element.js';
 
-const SCAN_RESOLUTION_EXCLUDED_LANGUAGES = new Set(['ja', 'zh', 'yue', 'ko']);
+export const NO_SPACES_LANGUAGES = new Set(['ja', 'zh', 'yue', 'ko']);
 
 /**
  * @augments EventDispatcher<import('text-scanner').Events>
@@ -478,7 +478,7 @@ export class TextScanner extends EventDispatcher {
             );
 
             if (this._scanResolution === 'word' && !disallowExpandStartOffset &&
-            (this._language === null || !SCAN_RESOLUTION_EXCLUDED_LANGUAGES.has(this._language))) {
+            (this._language === null || !NO_SPACES_LANGUAGES.has(this._language))) {
                 // Move the start offset to the beginning of the word
                 textSource.setStartOffset(this._scanLength, this._layoutAwareScan, true);
             }

diff --git a/ext/settings.html b/ext/settings.html
@@ -1555,7 +1555,7 @@ <h1>Yomitan Settings</h1>
                 </p>
             </div>
         </div>
-        <div class="settings-item"><div class="settings-item-inner">
+        <div class="settings-item jpzhyueko-only"><div class="settings-item-inner">
             <div class="settings-item-left">
                 <div class="settings-item-label">Show space between parsed words</div>
             </div>