Sentence termination character mode (#1682)

* Change enableTerminationCharacters to terminationCharacterMode

* Update settings

* Update sentence extraction

* Update tests

* Add tests

3 files changed, 30 insertions, 8 deletions

diff --git a/ext/js/data/options-util.js b/ext/js/data/options-util.js
index cb7946f7..cb58206f 100644
--- a/ext/js/data/options-util.js
+++ b/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;
+    }
 }
diff --git a/ext/js/dom/document-util.js b/ext/js/dom/document-util.js
index 8284ffa5..da4d3e61 100644
--- a/ext/js/dom/document-util.js
+++ b/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);
diff --git a/ext/js/language/text-scanner.js b/ext/js/language/text-scanner.js
index a49627f8..2ebf26e2 100644
--- a/ext/js/language/text-scanner.js
+++ b/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 hasTerminationCharacters = (typeof terminationCharacters === 'object' && Array.isArray(terminationCharacters));
+            const {scanExtent, terminationCharacterMode, terminationCharacters} = sentenceParsingOptions;
             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