Sentence termination character mode (#1682)

* Change enableTerminationCharacters to terminationCharacterMode

* Update settings

* Update sentence extraction

* Update tests

* Add tests

ext/css/settings.css

@@ -2249,6 +2249,9 @@ input[type=number].dictionary-priority {
     align-content: flex-start;
     justify-content: flex-start;
 }
+.horizontal-flex.horizontal-flex-nowrap {
+    flex-wrap: nowrap;
+}
 .horizontal-flex>* {
     margin-left: 0.375em;
 }

ext/data/schemas/options-schema.json

@@ -925,7 +925,7 @@
                                 "type": "object",
                                 "required": [
                                     "scanExtent",
-                                    "enableTerminationCharacters",
+                                    "terminationCharacterMode",
                                     "terminationCharacters"
                                 ],
                                 "properties": {
@@ -934,8 +934,9 @@
                                         "minimum": 0,
                                         "default": 200
                                     },
-                                    "enableTerminationCharacters": {
+                                    "terminationCharacterMode": {
-                                        "type": "boolean",
+                                        "type": "string",
+                                        "enum": ["custom", "custom-no-newlines", "newlines", "none"],
                                         "default": true
                                     },
                                     "terminationCharacters": {

ext/js/data/options-util.js

@@ -460,7 +460,8 @@ class OptionsUtil {
             {async: true,  update: this._updateVersion8.bind(this)},
             {async: false, update: this._updateVersion9.bind(this)},
             {async: true,  update: this._updateVersion10.bind(this)},
-            {async: true,  update: this._updateVersion11.bind(this)}
+            {async: false, update: this._updateVersion11.bind(this)},
+            {async: false, update: this._updateVersion12.bind(this)}
         ];
     }
 
@@ -811,4 +812,15 @@ class OptionsUtil {
         }
         return options;
     }
+
+    _updateVersion12(options) {
+        // Version 12 changes:
+        //  Changed sentenceParsing.enableTerminationCharacters to sentenceParsing.terminationCharacterMode.
+        for (const profile of options.profiles) {
+            const {sentenceParsing} = profile.options;
+            sentenceParsing.terminationCharacterMode = sentenceParsing.enableTerminationCharacters ? 'custom' : 'newlines';
+            delete sentenceParsing.enableTerminationCharacters;
+        }
+        return options;
+    }
 }

ext/js/dom/document-util.js

@@ -69,6 +69,7 @@ class DocumentUtil {
      * @param source The text source object, either `TextSourceRange` or `TextSourceElement`.
      * @param layoutAwareScan Whether or not layout-aware scan mode should be used.
      * @param extent The length of the sentence to extract.
+     * @param terminateAtNewlines Whether or not a sentence should be terminated at newline characters.
      * @param terminatorMap A mapping of characters that terminate a sentence.
      *   Format:
      *   ```js
@@ -87,7 +88,7 @@ class DocumentUtil {
      *   ```
      * @returns The sentence and the offset to the original source: `{sentence: string, offset: integer}`.
      */
-    extractSentence(source, layoutAwareScan, extent, terminatorMap, forwardQuoteMap, backwardQuoteMap) {
+    extractSentence(source, layoutAwareScan, extent, terminateAtNewlines, terminatorMap, forwardQuoteMap, backwardQuoteMap) {
         // Scan text
         source = source.clone();
         const startLength = source.setStartOffset(extent, layoutAwareScan);
@@ -102,7 +103,7 @@ class DocumentUtil {
         let quoteStack = [];
         for (; pos1 > 0; --pos1) {
             const c = text[pos1 - 1];
-            if (c === '\n') { break; }
+            if (c === '\n' && terminateAtNewlines) { break; }
 
             if (quoteStack.length === 0) {
                 const terminatorInfo = terminatorMap.get(c);
@@ -133,7 +134,7 @@ class DocumentUtil {
         quoteStack = [];
         for (; pos2 < textLength; ++pos2) {
             const c = text[pos2];
-            if (c === '\n') { break; }
+            if (c === '\n' && terminateAtNewlines) { break; }
 
             if (quoteStack.length === 0) {
                 const terminatorInfo = terminatorMap.get(c);

ext/js/language/text-scanner.js

@@ -63,6 +63,7 @@ class TextScanner extends EventDispatcher {
         this._layoutAwareScan = false;
         this._preventMiddleMouse = false;
         this._sentenceScanExtent = 0;
+        this._sentenceTerminateAtNewlines = true;
         this._sentenceTerminatorMap = new Map();
         this._sentenceForwardQuoteMap = new Map();
         this._sentenceBackwardQuoteMap = new Map();
@@ -209,19 +210,23 @@ class TextScanner extends EventDispatcher {
             this._preventMiddleMouse = preventMiddleMouse;
         }
         if (typeof sentenceParsingOptions === 'object' && sentenceParsingOptions !== null) {
-            const {scanExtent, enableTerminationCharacters, terminationCharacters} = sentenceParsingOptions;
+            const {scanExtent, terminationCharacterMode, terminationCharacters} = sentenceParsingOptions;
-            const hasTerminationCharacters = (typeof terminationCharacters === 'object' && Array.isArray(terminationCharacters));
             if (typeof scanExtent === 'number') {
                 this._sentenceScanExtent = sentenceParsingOptions.scanExtent;
             }
-            if (typeof enableTerminationCharacters === 'boolean' || hasTerminationCharacters) {
+            if (typeof terminationCharacterMode === 'string') {
+                this._sentenceTerminateAtNewlines = (terminationCharacterMode === 'custom' || terminationCharacterMode === 'newlines');
                 const sentenceTerminatorMap = this._sentenceTerminatorMap;
                 const sentenceForwardQuoteMap = this._sentenceForwardQuoteMap;
                 const sentenceBackwardQuoteMap = this._sentenceBackwardQuoteMap;
                 sentenceTerminatorMap.clear();
                 sentenceForwardQuoteMap.clear();
                 sentenceBackwardQuoteMap.clear();
-                if (enableTerminationCharacters !== false && hasTerminationCharacters) {
+                if (
+                    typeof terminationCharacters === 'object' &&
+                    Array.isArray(terminationCharacters) &&
+                    (terminationCharacterMode === 'custom' || terminationCharacterMode === 'custom-no-newlines')
+                ) {
                     for (const {enabled, character1, character2, includeCharacterAtStart, includeCharacterAtEnd} of terminationCharacters) {
                         if (!enabled) { continue; }
                         if (character2 === null) {
@@ -841,6 +846,7 @@ class TextScanner extends EventDispatcher {
     async _findTermDictionaryEntries(textSource, optionsContext) {
         const scanLength = this._scanLength;
         const sentenceScanExtent = this._sentenceScanExtent;
+        const sentenceTerminateAtNewlines = this._sentenceTerminateAtNewlines;
         const sentenceTerminatorMap = this._sentenceTerminatorMap;
         const sentenceForwardQuoteMap = this._sentenceForwardQuoteMap;
         const sentenceBackwardQuoteMap = this._sentenceBackwardQuoteMap;
@@ -856,6 +862,7 @@ class TextScanner extends EventDispatcher {
             textSource,
             layoutAwareScan,
             sentenceScanExtent,
+            sentenceTerminateAtNewlines,
             sentenceTerminatorMap,
             sentenceForwardQuoteMap,
             sentenceBackwardQuoteMap
@@ -866,6 +873,7 @@ class TextScanner extends EventDispatcher {
 
     async _findKanjiDictionaryEntries(textSource, optionsContext) {
         const sentenceScanExtent = this._sentenceScanExtent;
+        const sentenceTerminateAtNewlines = this._sentenceTerminateAtNewlines;
         const sentenceTerminatorMap = this._sentenceTerminatorMap;
         const sentenceForwardQuoteMap = this._sentenceForwardQuoteMap;
         const sentenceBackwardQuoteMap = this._sentenceBackwardQuoteMap;
@@ -881,6 +889,7 @@ class TextScanner extends EventDispatcher {
             textSource,
             layoutAwareScan,
             sentenceScanExtent,
+            sentenceTerminateAtNewlines,
             sentenceTerminatorMap,
             sentenceForwardQuoteMap,
             sentenceBackwardQuoteMap

ext/settings.html

@@ -1257,18 +1257,30 @@
         </div></div>
         <div class="settings-item"><div class="settings-item-inner settings-item-inner-wrappable">
             <div class="settings-item-left">
-                <div class="settings-item-label">Enable sentence termination characters</div>
+                <div class="settings-item-label">Sentence termination characters</div>
             </div>
             <div class="settings-item-right">
-                <label class="toggle"><input type="checkbox" data-setting="sentenceParsing.enableTerminationCharacters"><span class="toggle-body"><span class="toggle-track"></span><span class="toggle-knob"></span></span></label>
+                <div class="horizontal-flex horizontal-flex-nowrap">
-            </div>
+                    <button class="low-emphasis" data-modal-action="show,sentence-termination-characters" id="configure-sentence-termination-characters-button" hidden>Configure&hellip;</button>
-        </div></div>
+                    <select data-setting="sentenceParsing.terminationCharacterMode"
-        <div class="settings-item settings-item-button" data-modal-action="show,sentence-termination-characters"><div class="settings-item-inner">
+                        data-transform='{
-            <div class="settings-item-left">
+                            "type": "setVisibility",
-                <div class="settings-item-label">Configure sentence termination characters&hellip;</div>
+                            "selector": "#configure-sentence-termination-characters-button",
-            </div>
+                            "condition": {
-            <div class="settings-item-right open-panel-button-container">
+                                "op": "||",
-                <button class="icon-button"><span class="icon-button-inner"><span class="icon" data-icon="material-right-arrow"></span></span></button>
+                                "value": [
+                                    {"op": "===", "value": "custom"},
+                                    {"op": "===", "value": "custom-no-newlines"}
+                                ]
+                            }
+                        }'
+                    >
+                        <option value="custom">Custom</option>
+                        <option value="custom-no-newlines">Custom, no newlines</option>
+                        <option value="newlines">Newlines only</option>
+                        <option value="none">None</option>
+                    </select>
+                </div>
             </div>
         </div></div>
     </div>

test/data/html/test-document1.html

@@ -157,6 +157,48 @@
         <img src="data:image/gif;base64,R0lGODdhBwAHAIABAAAAAP///ywAAAAABwAHAAACDIRvEaC32FpCbEkKCgA7" alt="よみちゃん" title="よみちゃん" style="width: 70px; height: 70px; image-rendering: crisp-edges; image-rendering: pixelated; display: block;">
     </div>
 
+    <div
+        class="test"
+        data-test-type="scan"
+        data-element-from-point-selector="span:nth-of-type(3)"
+        data-caret-range-from-point-selector="span:nth-of-type(3)"
+        data-start-node-selector="span:nth-of-type(3)"
+        data-start-offset="0"
+        data-end-node-selector="span:nth-of-type(3)"
+        data-end-offset="0"
+        data-result-type="TextSourceRange"
+        data-sentence-scan-extent="22"
+        data-sentence="ありがとございます3"
+        data-terminate-at-newlines="true"
+    >
+<span>ありがとございます1</span>
+<span>ありがとございます2</span>
+<span>ありがとございます3</span>
+<span>ありがとございます4</span>
+<span>ありがとございます5</span>
+    </div>
+
+    <div
+        class="test"
+        data-test-type="scan"
+        data-element-from-point-selector="span:nth-of-type(3)"
+        data-caret-range-from-point-selector="span:nth-of-type(3)"
+        data-start-node-selector="span:nth-of-type(3)"
+        data-start-offset="0"
+        data-end-node-selector="span:nth-of-type(3)"
+        data-end-offset="0"
+        data-result-type="TextSourceRange"
+        data-sentence-scan-extent="22"
+        data-sentence="ありがとございます1&#10;ありがとございます2&#10;ありがとございます3&#10;ありがとございます4"
+        data-terminate-at-newlines="false"
+    >
+<span>ありがとございます1</span>
+<span>ありがとございます2</span>
+<span>ありがとございます3</span>
+<span>ありがとございます4</span>
+<span>ありがとございます5</span>
+    </div>
+
     <div
         class="test"
         data-test-type="text-source-range-seek"

test/test-document-util.js

@@ -129,7 +129,8 @@ async function testDocumentTextScanningFunctions(dom, {DocumentUtil, TextSourceR
             resultType,
             sentenceScanExtent,
             sentence,
-            hasImposter
+            hasImposter,
+            terminateAtNewlines
         } = testElement.dataset;
 
         const elementFromPointValue = querySelectorChildOrSelf(testElement, elementFromPointSelector);
@@ -140,6 +141,7 @@ async function testDocumentTextScanningFunctions(dom, {DocumentUtil, TextSourceR
         startOffset = parseInt(startOffset, 10);
         endOffset = parseInt(endOffset, 10);
         sentenceScanExtent = parseInt(sentenceScanExtent, 10);
+        terminateAtNewlines = (terminateAtNewlines !== 'false');
 
         assert.notStrictEqual(elementFromPointValue, null);
         assert.notStrictEqual(caretRangeFromPointValue, null);
@@ -200,6 +202,7 @@ async function testDocumentTextScanningFunctions(dom, {DocumentUtil, TextSourceR
             source,
             false,
             sentenceScanExtent,
+            terminateAtNewlines,
             terminatorMap,
             forwardQuoteMap,
             backwardQuoteMap

test/test-options-util.js

@@ -438,7 +438,7 @@ function createProfileOptionsUpdatedTestData1() {
         },
         sentenceParsing: {
             scanExtent: 200,
-            enableTerminationCharacters: true,
+            terminationCharacterMode: 'custom',
             terminationCharacters: [
                 {enabled: true, character1: '「', character2: '」', includeCharacterAtStart: false, includeCharacterAtEnd: false},
                 {enabled: true, character1: '『', character2: '』', includeCharacterAtStart: false, includeCharacterAtEnd: false},
@@ -576,7 +576,7 @@ function createOptionsUpdatedTestData1() {
             }
         ],
         profileCurrent: 0,
-        version: 11,
+        version: 12,
         global: {
             database: {
                 prefixWildcardsSupported: false