WIP

Author: calixteman

web/pdf_find_controller.js

@@ -60,7 +60,7 @@ function normalize(text) {
     // Compile the regular expression for text normalization once.
     const replace = Object.keys(CHARACTERS_TO_NORMALIZE).join("");
     normalizationRegex = new RegExp(
-      `([${replace}])|(-\\n)|(\\n)|(\\p{Mn}+)`,
+      `([${replace}])|(-\\n)|(\\n)|(\\p{Mn}+)|(\\p{Cf}+)`,
       "gum"
     );
   }
@@ -109,7 +109,7 @@ function normalize(text) {
 
   normalized = normalized.replace(
     normalizationRegex,
-    (match, p1, p2, p3, p4, i) => {
+    (match, p1, p2, p3, p4, p5, i) => {
       i -= shiftOrigin;
       if (p1) {
         // Fractions...
@@ -142,6 +142,13 @@ function normalize(text) {
         return " ";
       }
 
+      if (p5) {
+        // Invisible formatting indicators.
+        positions.push([i - shift, p5.length + shift]);
+        shift += p5.length;
+        return "";
+      }
+
       // Diacritics.
       hasDiacritics = true;
       let jj = match.length;
@@ -477,10 +484,10 @@ class PDFFindController {
       });
     }
 
-    // Replace spaces by \s+ to be sure to match any spaces.
+    // Replace spaces by [\s\p{Zs}]+ to be sure to match any spaces.
     // We must do it after the if (matchDiacritcs) block to avoid
     // wrong things with the "s".
-    query = query.replace(whitespacesRegExp, "\\s+");
+    query = query.replace(whitespacesRegExp, "[\\s\\p{Zs}]+");
 
     return query;
   }