ignore zero-width non-joiner

fixes #179

4 files changed, 31 insertions, 6 deletions

diff --git a/ext/fg/js/document.js b/ext/fg/js/document.js
index f58a64fc..0697edb3 100644
--- a/ext/fg/js/document.js
+++ b/ext/fg/js/document.js
@@ -116,7 +116,7 @@ function docSentenceExtract(source, extent) {
     const sourceLocal = source.clone();
     const position = sourceLocal.setStartOffset(extent);
     sourceLocal.setEndOffset(position + extent);
-    const content = sourceLocal.text();
+    const {text: content} = sourceLocal.text();
 
     let quoteStack = [];
 
diff --git a/ext/fg/js/frontend.js b/ext/fg/js/frontend.js
index bd652f3b..afb182d4 100644
--- a/ext/fg/js/frontend.js
+++ b/ext/fg/js/frontend.js
@@ -311,10 +311,18 @@ class Frontend {
     async searchTerms(textSource) {
         textSource.setEndOffset(this.options.scanning.length);
 
-        const {definitions, length} = await apiTermsFind(textSource.text());
+        const {text, strippedIndices} = textSource.text();
+        let {definitions, length} = await apiTermsFind(text);
         if (definitions.length === 0) {
             return false;
         }
+        for (let index of strippedIndices) {
+            if (index < length) {
+                length++;
+            } else {
+                break;
+            }
+        }
 
         textSource.setEndOffset(length);
 
@@ -338,7 +346,8 @@ class Frontend {
     async searchKanji(textSource) {
         textSource.setEndOffset(1);
 
-        const definitions = await apiKanjiFind(textSource.text());
+        const {text} = textSource.text();
+        const definitions = await apiKanjiFind(text);
         if (definitions.length === 0) {
             return false;
         }
diff --git a/ext/fg/js/source.js b/ext/fg/js/source.js
index 664dbec7..cd8f63fd 100644
--- a/ext/fg/js/source.js
+++ b/ext/fg/js/source.js
@@ -16,6 +16,9 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
+// \u200c (Zero-width non-joiner) appears on Google Docs from Chrome 76 onwards
+const IGNORE_TEXT_PATTERN = /\u200c/g;
+
 
 /*
  * TextSourceRange
@@ -32,7 +35,13 @@ class TextSourceRange {
     }
 
     text() {
-        return this.content;
+        let strippedIndices = [];
+        const text = this.content.replace(IGNORE_TEXT_PATTERN, (match, offset) => {
+            strippedIndices.push(offset);
+            return '';
+        });
+
+        return {text, strippedIndices};
     }
 
     setEndOffset(length) {
@@ -195,7 +204,13 @@ class TextSourceElement {
     }
 
     text() {
-        return this.content;
+        let strippedIndices = [];
+        const text = this.content.replace(IGNORE_TEXT_PATTERN, (match, offset) => {
+            strippedIndices.push(offset);
+            return '';
+        });
+
+        return {text, strippedIndices};
     }
 
     setEndOffset(length) {
diff --git a/ext/mixed/js/display.js b/ext/mixed/js/display.js
index 3bb78fe1..6b9c295b 100644
--- a/ext/mixed/js/display.js
+++ b/ext/mixed/js/display.js
@@ -83,7 +83,8 @@ class Display {
             const textSource = docRangeFromPoint({x: e.clientX, y: e.clientY});
             textSource.setEndOffset(this.options.scanning.length);
 
-            const {definitions, length} = await apiTermsFind(textSource.text());
+            const {text} = textSource.text();
+            const {definitions, length} = await apiTermsFind(text);
             if (definitions.length === 0) {
                 return false;
             }