Use DOMTextScanner (#536)

* Use DOMTextScanner instead of TextSourceRange.seek*

* Move getNodesInRange to dom.js

* Move anyNodeMatchesSelector to dom.js

* Remove unused functions

* Update tests

* Add layoutAwareScan option

* Use layoutAwareScan for source and sentence scanning

* Remove unused IGNORE_TEXT_PATTERN

6 files changed, 19 insertions, 5 deletions

diff --git a/ext/bg/data/options-schema.json b/ext/bg/data/options-schema.json
index 0379fa75..5885e036 100644
--- a/ext/bg/data/options-schema.json
+++ b/ext/bg/data/options-schema.json
@@ -321,7 +321,8 @@
                                     "enablePopupSearch",
                                     "enableOnPopupExpressions",
                                     "enableOnSearchPage",
-                                    "enableSearchTags"
+                                    "enableSearchTags",
+                                    "layoutAwareScan"
                                 ],
                                 "properties": {
                                     "middleMouse": {
@@ -383,6 +384,10 @@
                                     "enableSearchTags": {
                                         "type": "boolean",
                                         "default": false
+                                    },
+                                    "layoutAwareScan": {
+                                        "type": "boolean",
+                                        "default": false
                                     }
                                 }
                             },
diff --git a/ext/bg/js/options.js b/ext/bg/js/options.js
index 97368a0b..170e4799 100644
--- a/ext/bg/js/options.js
+++ b/ext/bg/js/options.js
@@ -203,7 +203,8 @@ function profileOptionsCreateDefaults() {
             enablePopupSearch: false,
             enableOnPopupExpressions: false,
             enableOnSearchPage: true,
-            enableSearchTags: false
+            enableSearchTags: false,
+            layoutAwareScan: false
         },
 
         translation: {
diff --git a/ext/bg/js/search-query-parser.js b/ext/bg/js/search-query-parser.js
index addfc686..97e98b40 100644
--- a/ext/bg/js/search-query-parser.js
+++ b/ext/bg/js/search-query-parser.js
@@ -75,15 +75,17 @@ class QueryParser {
     async _search(textSource, cause) {
         if (textSource === null) { return null; }
 
-        const searchText = this._textScanner.getTextSourceContent(textSource, this._options.scanning.length);
+        const {length: scanLength, layoutAwareScan} = this._options.scanning;
+        const searchText = this._textScanner.getTextSourceContent(textSource, scanLength, layoutAwareScan);
         if (searchText.length === 0) { return null; }
 
         const {definitions, length} = await api.termsFind(searchText, {}, this._getOptionsContext());
         if (definitions.length === 0) { return null; }
 
-        const sentence = docSentenceExtract(textSource, this._options.anki.sentenceExt);
+        const sentenceExtent = this._options.anki.sentenceExt;
+        const sentence = docSentenceExtract(textSource, sentenceExtent, layoutAwareScan);
 
-        textSource.setEndOffset(length);
+        textSource.setEndOffset(length, layoutAwareScan);
 
         this._setContent('terms', {definitions, context: {
             focus: false,
diff --git a/ext/bg/search.html b/ext/bg/search.html
index de08cdae..4a28dd88 100644
--- a/ext/bg/search.html
+++ b/ext/bg/search.html
@@ -79,6 +79,7 @@
         <script src="/bg/js/dictionary.js"></script>
         <script src="/bg/js/handlebars.js"></script>
         <script src="/fg/js/document.js"></script>
+        <script src="/fg/js/dom-text-scanner.js"></script>
         <script src="/fg/js/source.js"></script>
         <script src="/mixed/js/audio-system.js"></script>
         <script src="/mixed/js/display-context.js"></script>
diff --git a/ext/bg/settings-popup-preview.html b/ext/bg/settings-popup-preview.html
index fe92f24f..5eecd005 100644
--- a/ext/bg/settings-popup-preview.html
+++ b/ext/bg/settings-popup-preview.html
@@ -126,6 +126,7 @@
         <script src="/mixed/js/text-scanner.js"></script>
 
         <script src="/fg/js/document.js"></script>
+        <script src="/fg/js/dom-text-scanner.js"></script>
         <script src="/fg/js/popup.js"></script>
         <script src="/fg/js/source.js"></script>
         <script src="/fg/js/popup-factory.js"></script>
diff --git a/ext/bg/settings.html b/ext/bg/settings.html
index 118a13b9..77b61aef 100644
--- a/ext/bg/settings.html
+++ b/ext/bg/settings.html
@@ -400,6 +400,10 @@
                     <label><input type="checkbox" id="auto-hide-results" data-setting="scanning.autoHideResults"> Automatically hide results</label>
                 </div>
 
+                <div class="checkbox">
+                    <label><input type="checkbox" id="layout-aware-scan" data-setting="scanning.layoutAwareScan"> Layout-aware scan</label>
+                </div>
+
                 <div class="checkbox options-advanced">
                     <label><input type="checkbox" id="deep-dom-scan" data-setting="scanning.deepDomScan"> Deep DOM scan</label>
                 </div>