Support search with or without diacritics

  - get original index in using a dichotomic seach instead of a linear one;
  - remove diacritics from text using NFD decomposition and unicode regex;
  - convert the query string into a RegExp;
  - replace whitespaces in the query with \s+;
  - remove pdf_find_utils.js.

Author: calixteman

l10n/en-US/viewer.properties

@@ -168,6 +168,7 @@ find_next.title=Find the next occurrence of the phrase
 find_next_label=Next
 find_highlight=Highlight all
 find_match_case_label=Match case
+find_match_diacritics_label=Match Diacritics
 find_entire_word_label=Whole words
 find_reached_top=Reached top of document, continued from bottom
 find_reached_bottom=Reached end of document, continued from top

test/pdfs/.gitignore

@@ -321,6 +321,7 @@
 !issue4650.pdf
 !issue6721_reduced.pdf
 !issue3025.pdf
+!french_diacritics.pdf
 !issue2099-1.pdf
 !issue3371.pdf
 !issue2956.pdf

test/pdfs/french_diacritics.pdf

@@ -30,7 +30,6 @@
     "node_stream_spec.js",
     "parser_spec.js",
     "pdf_find_controller_spec.js",
-    "pdf_find_utils_spec.js",
     "pdf_history_spec.js",
     "primitives_spec.js",
     "stream_spec.js",

test/unit/clitests.json

@@ -75,7 +75,6 @@ async function initializePDFJS(callback) {
       "pdfjs-test/unit/network_utils_spec.js",
       "pdfjs-test/unit/parser_spec.js",
       "pdfjs-test/unit/pdf_find_controller_spec.js",
-      "pdfjs-test/unit/pdf_find_utils_spec.js",
       "pdfjs-test/unit/pdf_history_spec.js",
       "pdfjs-test/unit/primitives_spec.js",
       "pdfjs-test/unit/scripting_spec.js",

test/unit/jasmine-boot.js

@@ -108,7 +108,6 @@ function testSearch({
           return;
         }
         eventBus.off("updatefindmatchescount", onUpdateFindMatchesCount);
-
         expect(evt.matchesCount.total).toBe(totalMatches);
         for (let i = 0; i < totalPages; i++) {
           expect(pdfFindController.pageMatches[i].length).toEqual(
@@ -271,5 +270,130 @@ describe("pdf_find_controller", function () {
       pageMatches: [[19, 48, 66]],
       pageMatchesLength: [[8, 8, 8]],
     });
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "1/2",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+      },
+      matchesPerPage: [2],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[28, 57]],
+      pageMatchesLength: [[1, 1]],
+    });
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "½",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+      },
+      matchesPerPage: [2],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[28, 57]],
+      pageMatchesLength: [[1, 1]],
+    });
+  });
+
+  it("performs a normal search, where the text with diacritics is normalized", async function () {
+    const { eventBus, pdfFindController } = await initPdfFindController(
+      "french_diacritics.pdf"
+    );
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "a",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+        matchDiacritics: false,
+      },
+      matchesPerPage: [6],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[0, 2, 4, 6, 8, 10]],
+      pageMatchesLength: [[1, 1, 1, 1, 1, 1]],
+    });
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "u",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+        matchDiacritics: false,
+      },
+      matchesPerPage: [6],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[44, 46, 48, 50, 52, 54]],
+      pageMatchesLength: [[1, 1, 1, 1, 1, 1]],
+    });
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "ë",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+        matchDiacritics: true,
+      },
+      matchesPerPage: [2],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[28, 30]],
+      pageMatchesLength: [[1, 1]],
+    });
+  });
+
+  it("performs a search where one of the results contains an hyphen", async function () {
+    const { eventBus, pdfFindController } = await initPdfFindController();
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      parameters: {
+        query: "optimiz",
+        caseSensitive: false,
+        entireWord: false,
+        phraseSearch: true,
+        findPrevious: false,
+      },
+      matchesPerPage: [1, 4, 2, 3, 3, 0, 2, 9, 1, 0, 0, 6, 3, 4],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+    });
   });
 });

test/unit/pdf_find_controller_spec.js

@@ -2608,6 +2608,7 @@ function webViewerFind(evt) {
     entireWord: evt.entireWord,
     highlightAll: evt.highlightAll,
     findPrevious: evt.findPrevious,
+    matchDiacritics: evt.matchDiacritics,
   });
 }
 
@@ -2619,6 +2620,7 @@ function webViewerFindFromUrlHash(evt) {
     entireWord: false,
     highlightAll: true,
     findPrevious: false,
+    matchDiacritics: true,
   });
 }
 
@@ -2825,6 +2827,7 @@ function webViewerKeyDown(evt) {
               entireWord: findState.entireWord,
               highlightAll: findState.highlightAll,
               findPrevious: cmd === 5 || cmd === 12,
+              matchDiacritics: findState.matchDiacritics,
             });
           }
           handled = true;

test/unit/pdf_find_utils_spec.js

@@ -218,6 +218,7 @@ class MozL10n {
     "findcasesensitivitychange",
     "findentirewordchange",
     "findbarclose",
+    "finddiacriticmatchingchange",
   ];
   const handleEvent = function ({ type, detail }) {
     if (!PDFViewerApplication.initialized) {
@@ -236,6 +237,7 @@ class MozL10n {
       entireWord: !!detail.entireWord,
       highlightAll: !!detail.highlightAll,
       findPrevious: !!detail.findPrevious,
+      matchDiacritics: !!detail.matchDiacritics,
     });
   };
 

web/app.js

@@ -33,6 +33,7 @@ class PDFFindBar {
     this.highlightAll = options.highlightAllCheckbox;
     this.caseSensitive = options.caseSensitiveCheckbox;
     this.entireWord = options.entireWordCheckbox;
+    this.matchDiacritics = options.matchDiacriticsCheckbox;
     this.findMsg = options.findMsg;
     this.findResultsCount = options.findResultsCount;
     this.findPreviousButton = options.findPreviousButton;
@@ -82,6 +83,10 @@ class PDFFindBar {
       this.dispatchEvent("entirewordchange");
     });
 
+    this.matchDiacritics.addEventListener("click", () => {
+      this.dispatchEvent("diacriticmatchingchange");
+    });
+
     this.eventBus._on("resize", this._adjustWidth.bind(this));
   }
 
@@ -99,6 +104,7 @@ class PDFFindBar {
       entireWord: this.entireWord.checked,
       highlightAll: this.highlightAll.checked,
       findPrevious: findPrev,
+      matchDiacritics: this.matchDiacritics.checked,
     });
   }