[api-minor] Don't normalize the text used in the text layer.

Some arabic chars like \ufe94 could be searched in a pdf, hence it must be normalized
when creating the search query. So to avoid to duplicate the normalization code,
everything is moved in the find controller.
The previous code to normalize text was using NFKC but with a hardcoded map, hence it
has been replaced by the use of normalize("NFKC") (it helps to reduce the bundle size
by 30kb).
In playing with this \ufe94 char, I noticed that the bidi algorithm wasn't taking into
account some RTL unicode ranges, the generated font wasn't embedding the mapping this
char and the unicode ranges in the OS/2 table weren't up-to-date.

When normalized some chars can be replaced by several ones and it induced to have
some extra chars in the text layer. To avoid any regression, when copying some text
from the text layer, a copied string is normalized (NFKC) before being put in the
clipboard (it works like this in either Acrobat or Chrome).

src/core/bidi.js

@@ -147,7 +147,11 @@ function bidi(str, startLevel = -1, vertical = false) {
       if (!charType) {
         warn("Bidi: invalid Unicode character " + charCode.toString(16));
       }
-    } else if (0x0700 <= charCode && charCode <= 0x08ac) {
+    } else if (
+      (0x0700 <= charCode && charCode <= 0x08ac) ||
+      (0xfb50 <= charCode && charCode <= 0xfdff) ||
+      (0xfe70 <= charCode && charCode <= 0xfeff)
+    ) {
       charType = "AL";
     }
     if (charType === "R" || charType === "AL" || charType === "AN") {

src/core/document.js

@@ -511,7 +511,13 @@ class Page {
     });
   }
 
-  extractTextContent({ handler, task, includeMarkedContent, sink }) {
+  extractTextContent({
+    handler,
+    task,
+    includeMarkedContent,
+    disableNormalization,
+    sink,
+  }) {
     const contentStreamPromise = this.getContentStream();
     const resourcesPromise = this.loadResources([
       "ExtGState",
@@ -539,6 +545,7 @@ class Page {
         task,
         resources: this.resources,
         includeMarkedContent,
+        disableNormalization,
         sink,
         viewBox: this.view,
       });

src/core/evaluator.js

@@ -24,6 +24,7 @@ import {
   IDENTITY_MATRIX,
   info,
   isArrayEqual,
+  normalizeUnicode,
   OPS,
   shadow,
   stringToPDFString,
@@ -2271,6 +2272,7 @@ class PartialEvaluator {
     seenStyles = new Set(),
     viewBox,
     markedContentData = null,
+    disableNormalization = false,
   }) {
     // Ensure that `resources`/`stateManager` is correctly initialized,
     // even if the provided parameter is e.g. `null`.
@@ -2524,7 +2526,10 @@ class PartialEvaluator {
     }
 
     function runBidiTransform(textChunk) {
-      const text = textChunk.str.join("");
+      let text = textChunk.str.join("");
+      if (!disableNormalization) {
+        text = normalizeUnicode(text);
+      }
       const bidiResult = bidi(text, -1, textChunk.vertical);
       return {
         str: bidiResult.str,
@@ -2859,7 +2864,7 @@ class PartialEvaluator {
           textChunk.prevTransform = getCurrentTextTransform();
         }
 
-        const glyphUnicode = glyph.normalizedUnicode;
+        const glyphUnicode = glyph.unicode;
         if (saveLastChar(glyphUnicode)) {
           // The two last chars are a non-whitespace followed by a whitespace
           // and then this non-whitespace, so we insert a whitespace here.
@@ -3242,6 +3247,7 @@ class PartialEvaluator {
                     seenStyles,
                     viewBox,
                     markedContentData,
+                    disableNormalization,
                   })
                   .then(function () {
                     if (!sinkWrapper.enqueueInvoked) {

src/core/fonts.js

@@ -33,11 +33,9 @@ import {
 } from "./fonts_utils.js";
 import {
   getCharUnicodeCategory,
-  getNormalizedUnicodes,
   getUnicodeForGlyph,
   getUnicodeRangeFor,
   mapSpecialUnicodeValues,
-  reverseIfRtl,
 } from "./unicode.js";
 import { getDingbatsGlyphsUnicode, getGlyphsUnicode } from "./glyphlist.js";
 import {
@@ -277,24 +275,6 @@ class Glyph {
       /* nonSerializable = */ true
     );
   }
-
-  /**
-   * This property, which is only used by `PartialEvaluator.getTextContent`,
-   * is purposely made non-serializable.
-   * @type {string}
-   */
-  get normalizedUnicode() {
-    return shadow(
-      this,
-      "normalizedUnicode",
-      reverseIfRtl(Glyph._NormalizedUnicodes[this.unicode] || this.unicode),
-      /* nonSerializable = */ true
-    );
-  }
-
-  static get _NormalizedUnicodes() {
-    return shadow(this, "_NormalizedUnicodes", getNormalizedUnicodes());
-  }
 }
 
 function int16(b0, b1) {
@@ -507,6 +487,9 @@ function adjustMapping(charCodeToGlyphId, hasGlyph, newGlyphZeroId, toUnicode) {
   const privateUseOffetStart = PRIVATE_USE_AREAS[privateUseAreaIndex][0];
   let nextAvailableFontCharCode = privateUseOffetStart;
   let privateUseOffetEnd = PRIVATE_USE_AREAS[privateUseAreaIndex][1];
+  const isInPrivateArea = code =>
+    (PRIVATE_USE_AREAS[0][0] <= code && code <= PRIVATE_USE_AREAS[0][1]) ||
+    (PRIVATE_USE_AREAS[1][0] <= code && code <= PRIVATE_USE_AREAS[1][1]);
   for (let originalCharCode in charCodeToGlyphId) {
     originalCharCode |= 0;
     let glyphId = charCodeToGlyphId[originalCharCode];
@@ -539,11 +522,7 @@ function adjustMapping(charCodeToGlyphId, hasGlyph, newGlyphZeroId, toUnicode) {
     if (typeof unicode === "string") {
       unicode = unicode.codePointAt(0);
     }
-    if (
+    if (unicode && !isInPrivateArea(unicode) && !usedGlyphIds.has(glyphId)) {
-      unicode &&
-      unicode < privateUseOffetStart &&
-      !usedGlyphIds.has(glyphId)
-    ) {
       toUnicodeExtraMap.set(unicode, glyphId);
       usedGlyphIds.add(glyphId);
     }
@@ -785,6 +764,7 @@ function createOS2Table(properties, charstrings, override) {
 
   let firstCharIndex = null;
   let lastCharIndex = 0;
+  let position = -1;
 
   if (charstrings) {
     for (let code in charstrings) {
@@ -796,7 +776,7 @@ function createOS2Table(properties, charstrings, override) {
         lastCharIndex = code;
       }
 
-      const position = getUnicodeRangeFor(code);
+      position = getUnicodeRangeFor(code, position);
       if (position < 32) {
         ulUnicodeRange1 |= 1 << position;
       } else if (position < 64) {

src/core/worker.js

@@ -745,7 +745,7 @@ class WorkerMessageHandler {
     });
 
     handler.on("GetTextContent", function (data, sink) {
-      const { pageIndex, includeMarkedContent } = data;
+      const { pageIndex, includeMarkedContent, disableNormalization } = data;
 
       pdfManager.getPage(pageIndex).then(function (page) {
         const task = new WorkerTask("GetTextContent: page " + pageIndex);
@@ -760,6 +760,7 @@ class WorkerMessageHandler {
             task,
             sink,
             includeMarkedContent,
+            disableNormalization,
           })
           .then(
             function () {

src/display/api.js

@@ -1122,6 +1122,8 @@ class PDFDocumentProxy {
  * @typedef {Object} getTextContentParameters
  * @property {boolean} [includeMarkedContent] - When true include marked
  *   content items in the items array of TextContent. The default is `false`.
+ * @property {boolean} [disableNormalization] - When true the text is *not*
+ *   normalized in the worker-thread. The default is `false`.
  */
 
 /**
@@ -1598,7 +1600,10 @@ class PDFPageProxy {
    * @param {getTextContentParameters} params - getTextContent parameters.
    * @returns {ReadableStream} Stream for reading text content chunks.
    */
-  streamTextContent({ includeMarkedContent = false } = {}) {
+  streamTextContent({
+    includeMarkedContent = false,
+    disableNormalization = false,
+  } = {}) {
     const TEXT_CONTENT_CHUNK_SIZE = 100;
 
     return this._transport.messageHandler.sendWithStream(
@@ -1606,6 +1611,7 @@ class PDFPageProxy {
       {
         pageIndex: this._pageIndex,
         includeMarkedContent: includeMarkedContent === true,
+        disableNormalization: disableNormalization === true,
       },
       {
         highWaterMark: TEXT_CONTENT_CHUNK_SIZE,

src/pdf.js

@@ -35,6 +35,7 @@ import {
   FeatureTest,
   InvalidPDFException,
   MissingPDFException,
+  normalizeUnicode,
   OPS,
   PasswordResponses,
   PermissionFlag,
@@ -100,6 +101,7 @@ export {
   isPdfFile,
   loadScript,
   MissingPDFException,
+  normalizeUnicode,
   OPS,
   PasswordResponses,
   PDFDataRangeTransport,

src/shared/util.js

@@ -1026,6 +1026,25 @@ function createPromiseCapability() {
   return capability;
 }
 
+let NormalizeRegex = null;
+let NormalizationMap = null;
+function normalizeUnicode(str) {
+  if (!NormalizeRegex) {
+    // In order to generate the following regex:
+    //  - create a PDF containing all the chars in the range 0000-FFFF with
+    //    a NFKC which is different of the char.
+    //  - copy and paste all those chars and get the ones where NFKC is
+    //    required.
+    // It appears that most the chars here contain some ligatures.
+    NormalizeRegex =
+      /([\u00a0\u00b5\u037e\u0eb3\u2000-\u200a\u202f\u2126\ufb00-\ufb04\ufb06\ufb20-\ufb36\ufb38-\ufb3c\ufb3e\ufb40-\ufb41\ufb43-\ufb44\ufb46-\ufba1\ufba4-\ufba9\ufbae-\ufbb1\ufbd3-\ufbdc\ufbde-\ufbe7\ufbea-\ufbf8\ufbfc-\ufbfd\ufc00-\ufc5d\ufc64-\ufcf1\ufcf5-\ufd3d\ufd88\ufdf4\ufdfa-\ufdfb\ufe71\ufe77\ufe79\ufe7b\ufe7d]+)|(\ufb05+)/gu;
+    NormalizationMap = new Map([["ſt", "ſt"]]);
+  }
+  return str.replaceAll(NormalizeRegex, (_, p1, p2) => {
+    return p1 ? p1.normalize("NFKC") : NormalizationMap.get(p2);
+  });
+}
+
 export {
   AbortException,
   AnnotationActionEventType,
@@ -1064,6 +1083,7 @@ export {
   LINE_FACTOR,
   MAX_IMAGE_SIZE_TO_CACHE,
   MissingPDFException,
+  normalizeUnicode,
   objectFromMap,
   objectSize,
   OPS,

test/driver.js

@@ -693,6 +693,7 @@ class Driver {
               initPromise = page
                 .getTextContent({
                   includeMarkedContent: true,
+                  disableNormalization: true,
                 })
                 .then(function (textContent) {
                   return Rasterize.textLayer(

test/integration/copy_paste_spec.js

@@ -28,7 +28,7 @@ describe("Copy and paste", () => {
       await closePages(pages);
     });
 
-    it("must check that we've all the contents", async () => {
+    it("must check that we've all the contents on copy/paste", async () => {
       await Promise.all(
         pages.map(async ([browserName, page]) => {
           await page.keyboard.down("Control");
@@ -117,4 +117,47 @@ describe("Copy and paste", () => {
       );
     });
   });
+  describe("all text", () => {
+    let pages;
+
+    beforeAll(async () => {
+      pages = await loadAndWait("copy_paste_ligatures.pdf", ".textLayer");
+      await mockClipboard(pages);
+    });
+
+    afterAll(async () => {
+      await closePages(pages);
+    });
+
+    it("must check that we've all the contents on copy/paste", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          await page.keyboard.down("Control");
+          await page.keyboard.press("a");
+          await page.keyboard.up("Control");
+
+          await page.waitForTimeout(100);
+
+          await page.keyboard.down("Control");
+          await page.keyboard.press("c");
+          await page.keyboard.up("Control");
+
+          await page.waitForTimeout(100);
+
+          await page.waitForFunction(
+            `document.querySelector('#viewerContainer').style.cursor !== "wait"`
+          );
+
+          const text = await page.evaluate(() =>
+            navigator.clipboard.readText()
+          );
+
+          expect(!!text).withContext(`In ${browserName}`).toEqual(true);
+          expect(text)
+            .withContext(`In ${browserName}`)
+            .toEqual("abcdeffffiflffifflſtstghijklmno");
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -585,3 +585,4 @@
 !issue16221.pdf
 !issue16224.pdf
 !issue16278.pdf
+!copy_paste_ligatures.pdf

test/unit/api_spec.js

@@ -2340,7 +2340,9 @@ page 1 / 3`);
       );
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items, styles } = await pdfPage.getTextContent();
+      const { items, styles } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       expect(items.length).toEqual(1);
       // Font name will be a random object id.
       const fontName = items[0].fontName;
@@ -2376,7 +2378,9 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue13226.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(text).toEqual(
@@ -2394,7 +2398,9 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue16119.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2410,7 +2416,9 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue13201.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2436,7 +2444,9 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue11913.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2456,7 +2466,9 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue10900.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2475,11 +2487,27 @@ page 1 / 3`);
       const loadingTask = getDocument(buildGetDocumentParams("issue10640.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      let { items } = await pdfPage.getTextContent({
-      const text = mergeText(items);
+        disableNormalization: true,
+      });
+      let text = mergeText(items);
+      let expected = `Open Sans is a humanist sans serif typeface designed by Steve Matteson.
+Open Sans was designed with an upright stress, open forms and a neu-
+tral, yet friendly appearance. It was optimized for print, web, and mobile
+interfaces, and has excellent legibility characteristics in its letterforms (see
+figure \x81 on the following page). This font is available from the Google Font
+Directory [\x81] as TrueType files licensed under the Apache License version \x82.\x80.
+This package provides support for this font in LATEX. It includes Type \x81
+versions of the fonts, converted for this package using FontForge from its
+sources, for full support with Dvips.`;
 
-      expect(
+      expect(text.includes(expected)).toEqual(true);
-        text.includes(`Open Sans is a humanist sans serif typeface designed by Steve Matteson.
+
+      ({ items } = await pdfPage.getTextContent({
+        disableNormalization: false,
+      }));
+      text = mergeText(items);
+      expected = `Open Sans is a humanist sans serif typeface designed by Steve Matteson.
 Open Sans was designed with an upright stress, open forms and a neu-
 tral, yet friendly appearance. It was optimized for print, web, and mobile
 interfaces, and has excellent legibility characteristics in its letterforms (see
@@ -2487,8 +2515,8 @@ figure \x81 on the following page). This font is available from the Google Font
 Directory [\x81] as TrueType files licensed under the Apache License version \x82.\x80.
 This package provides support for this font in LATEX. It includes Type \x81
 versions of the fonts, converted for this package using FontForge from its
-sources, for full support with Dvips.`)
+sources, for full support with Dvips.`;
-      ).toEqual(true);
+      expect(text.includes(expected)).toEqual(true);
 
       await loadingTask.destroy();
     });
@@ -2501,7 +2529,9 @@ sources, for full support with Dvips.`)
       const loadingTask = getDocument(buildGetDocumentParams("bug931481.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2529,7 +2559,9 @@ sozialökonomische Gerechtigkeit.`)
       const loadingTask = getDocument(buildGetDocumentParams("issue9186.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(
@@ -2550,7 +2582,9 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       );
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(text).toEqual(
@@ -2568,7 +2602,9 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       const loadingTask = getDocument(buildGetDocumentParams("bug1755201.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(6);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(/win aisle/.test(text)).toEqual(false);
@@ -2586,10 +2622,12 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       const pdfPage = await pdfDoc.getPage(568);
       let { items } = await pdfPage.getTextContent({
         includeMarkedContent: false,
+        disableNormalization: true,
       });
       const textWithoutMC = mergeText(items);
       ({ items } = await pdfPage.getTextContent({
         includeMarkedContent: true,
+        disableNormalization: true,
       }));
       const textWithMC = mergeText(items);
 
@@ -2607,7 +2645,9 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       );
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
       const text = mergeText(items);
 
       expect(text).toEqual("𠮷");
@@ -2619,7 +2659,9 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       const loadingTask = getDocument(buildGetDocumentParams("issue16221.pdf"));
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items } = await pdfPage.getTextContent();
+      const { items } = await pdfPage.getTextContent({
+        disableNormalization: true,
+      });
 
       expect(items.map(i => i.str)).toEqual(["Hello ", "World"]);
 

test/unit/pdf_find_controller_spec.js

@@ -542,7 +542,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 0,
         matchIndex: 0,
       },
-      pageMatches: [[2743]],
+      pageMatches: [[2734]],
       pageMatchesLength: [[14]],
     });
   });
@@ -561,7 +561,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 1,
         matchIndex: 0,
       },
-      pageMatches: [[], [1493]],
+      pageMatches: [[], [1486]],
       pageMatchesLength: [[], [11]],
     });
   });
@@ -594,7 +594,7 @@ describe("pdf_find_controller", function () {
         [],
         [],
         [],
-        [2087],
+        [2081],
       ],
       pageMatchesLength: [
         [24],
@@ -629,7 +629,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 0,
         matchIndex: 0,
       },
-      pageMatches: [[1501]],
+      pageMatches: [[1497]],
       pageMatchesLength: [[25]],
     });
   });
@@ -670,7 +670,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 0,
         matchIndex: 0,
       },
-      pageMatches: [[1946]],
+      pageMatches: [[1941]],
       pageMatchesLength: [[21]],
     });
   });
@@ -692,7 +692,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 0,
         matchIndex: 0,
       },
-      pageMatches: [[1946]],
+      pageMatches: [[1941]],
       pageMatchesLength: [[23]],
     });
   });
@@ -712,7 +712,7 @@ describe("pdf_find_controller", function () {
         pageIndex: 0,
         matchIndex: 0,
       },
-      pageMatches: [[1946]],
+      pageMatches: [[1941]],
       pageMatchesLength: [[23]],
     });
   });
@@ -976,4 +976,61 @@ describe("pdf_find_controller", function () {
       pageMatchesLength: [[5, 5]],
     });
   });
+
+  it("performs a search in a text with some arabic chars in different unicode ranges but with same normalized form", async function () {
+    const { eventBus, pdfFindController } = await initPdfFindController(
+      "ArabicCIDTrueType.pdf"
+    );
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      state: {
+        query: "\u0629",
+      },
+      matchesPerPage: [4],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[6, 25, 44, 63]],
+      pageMatchesLength: [[1, 1, 1, 1]],
+    });
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      state: {
+        query: "\ufe94",
+      },
+      matchesPerPage: [4],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[6, 25, 44, 63]],
+      pageMatchesLength: [[1, 1, 1, 1]],
+    });
+  });
+
+  it("performs a search in a text with some f ligatures", async function () {
+    const { eventBus, pdfFindController } = await initPdfFindController(
+      "copy_paste_ligatures.pdf"
+    );
+
+    await testSearch({
+      eventBus,
+      pdfFindController,
+      state: {
+        query: "f",
+      },
+      matchesPerPage: [9],
+      selectedMatch: {
+        pageIndex: 0,
+        matchIndex: 0,
+      },
+      pageMatches: [[5, 6, 6, 7, 8, 9, 9, 10, 10]],
+      pageMatchesLength: [[1, 1, 1, 1, 1, 1, 1, 1, 1]],
+    });
+  });
 });

test/unit/unicode_spec.js

@@ -15,11 +15,9 @@
 
 import {
   getCharUnicodeCategory,
-  getNormalizedUnicodes,
   getUnicodeForGlyph,
   getUnicodeRangeFor,
   mapSpecialUnicodeValues,
-  reverseIfRtl,
 } from "../../src/core/unicode.js";
 import {
   getDingbatsGlyphsUnicode,
@@ -152,69 +150,12 @@ describe("unicode", function () {
       expect(getUnicodeRangeFor(0x0041)).toEqual(0);
       // fi (Alphabetic Presentation Forms)
       expect(getUnicodeRangeFor(0xfb01)).toEqual(62);
+      // Combining diacritic (Cyrillic Extended-A)
+      expect(getUnicodeRangeFor(0x2dff)).toEqual(9);
     });
 
     it("should not get a Unicode range", function () {
-      expect(getUnicodeRangeFor(0x05ff)).toEqual(-1);
+      expect(getUnicodeRangeFor(0xaa60)).toEqual(-1);
-    });
-  });
-
-  describe("getNormalizedUnicodes", function () {
-    let NormalizedUnicodes;
-
-    beforeAll(function () {
-      NormalizedUnicodes = getNormalizedUnicodes();
-    });
-
-    afterAll(function () {
-      NormalizedUnicodes = null;
-    });
-
-    it("should get normalized Unicode values for ligatures", function () {
-      // fi => f + i
-      expect(NormalizedUnicodes["\uFB01"]).toEqual("fi");
-      // Arabic
-      expect(NormalizedUnicodes["\u0675"]).toEqual("\u0627\u0674");
-    });
-
-    it("should not normalize standard characters", function () {
-      expect(NormalizedUnicodes.A).toEqual(undefined);
-    });
-  });
-
-  describe("reverseIfRtl", function () {
-    let NormalizedUnicodes;
-
-    function getGlyphUnicode(char) {
-      if (NormalizedUnicodes[char] !== undefined) {
-        return NormalizedUnicodes[char];
-      }
-      return char;
-    }
-
-    beforeAll(function () {
-      NormalizedUnicodes = getNormalizedUnicodes();
-    });
-
-    afterAll(function () {
-      NormalizedUnicodes = null;
-    });
-
-    it("should not reverse LTR characters", function () {
-      const A = getGlyphUnicode("A");
-      expect(reverseIfRtl(A)).toEqual("A");
-
-      const fi = getGlyphUnicode("\uFB01");
-      expect(reverseIfRtl(fi)).toEqual("fi");
-    });
-
-    it("should reverse RTL characters", function () {
-      // Hebrew (no-op, since it's not a combined character)
-      const heAlef = getGlyphUnicode("\u05D0");
-      expect(reverseIfRtl(heAlef)).toEqual("\u05D0");
-      // Arabic
-      const arAlef = getGlyphUnicode("\u0675");
-      expect(reverseIfRtl(arAlef)).toEqual("\u0674\u0627");
     });
   });
 });

web/pdf_find_controller.js

@@ -18,8 +18,8 @@
 /** @typedef {import("./interfaces").IPDFLinkService} IPDFLinkService */
 
 import { binarySearchFirstItem, scrollIntoView } from "./ui_utils.js";
+import { getCharacterType, getNormalizeWithNFKC } from "./pdf_find_utils.js";
 import { createPromiseCapability } from "pdfjs-lib";
-import { getCharacterType } from "./pdf_find_utils.js";
 
 const FindState = {
   FOUND: 0,
@@ -126,12 +126,7 @@ function normalize(text) {
   } else {
     // Compile the regular expression for text normalization once.
     const replace = Object.keys(CHARACTERS_TO_NORMALIZE).join("");
-    const toNormalizeWithNFKC =
+    const toNormalizeWithNFKC = getNormalizeWithNFKC();
-      "\u2460-\u2473" + // Circled numbers.
-      "\u24b6-\u24ff" + // Circled letters/numbers.
-      "\u3244-\u32bf" + // Circled ideograms/numbers.
-      "\u32d0-\u32fe" + // Circled ideograms.
-      "\uff00-\uffef"; // Halfwidth, fullwidth forms.
 
     // 3040-309F: Hiragana
     // 30A0-30FF: Katakana
@@ -840,6 +835,7 @@ class PDFFindController {
     }
 
     let promise = Promise.resolve();
+    const textOptions = { disableNormalization: true };
     for (let i = 0, ii = this._linkService.pagesCount; i < ii; i++) {
       const extractTextCapability = createPromiseCapability();
       this._extractTextPromises[i] = extractTextCapability.promise;
@@ -848,7 +844,7 @@ class PDFFindController {
         return this._pdfDocument
           .getPage(i + 1)
           .then(pdfPage => {
-            return pdfPage.getTextContent();
+            return pdfPage.getTextContent(textOptions);
           })
           .then(
             textContent => {

web/pdf_find_utils.js

@@ -112,4 +112,46 @@ function getCharacterType(charCode) {
   return CharacterType.ALPHA_LETTER;
 }
 
-export { CharacterType, getCharacterType };
+let NormalizeWithNFKC;
+function getNormalizeWithNFKC() {
+  /* eslint-disable no-irregular-whitespace */
+  NormalizeWithNFKC ||= ` ¨ª¯²-µ¸-º¼-¾IJ-ijĿ-ŀʼnſDŽ-njDZ-dzʰ-ʸ˘-˝ˠ-ˤʹͺ;΄-΅·ϐ-ϖϰ-ϲϴ-ϵϹևٵ-ٸक़-य़ড়-ঢ়য়ਲ਼ਸ਼ਖ਼-ਜ਼ਫ਼ଡ଼-ଢ଼ำຳໜ-ໝ༌གྷཌྷདྷབྷཛྷཀྵჼᴬ-ᴮᴰ-ᴺᴼ-ᵍᵏ-ᵪᵸᶛ-ᶿẚ-ẛάέήίόύώΆ᾽-῁ΈΉ῍-῏ΐΊ῝-῟ΰΎ῭-`ΌΏ´-῾ - ‑‗․-… ″-‴‶-‷‼‾⁇-⁉⁗ ⁰-ⁱ⁴-₎ₐ-ₜ₨℀-℃℅-ℇ℉-ℓℕ-№ℙ-ℝ℠-™ℤΩℨK-ℭℯ-ℱℳ-ℹ℻-⅀ⅅ-ⅉ⅐-ⅿ↉∬-∭∯-∰〈-〉①-⓪⨌⩴-⩶⫝̸ⱼ-ⱽⵯ⺟⻳⼀-⿕　〶〸-〺゛-゜ゟヿㄱ-ㆎ㆒-㆟㈀-㈞㈠-㉇㉐-㉾㊀-㏿ꚜ-ꚝꝰꟲ-ꟴꟸ-ꟹꭜ-ꭟꭩ豈-嗀塚晴凞-羽蘒諸逸-都飯-舘並-龎ff-stﬓ-ﬗיִײַ-זּטּ-לּמּנּ-סּףּ-פּצּ-ﮱﯓ-ﴽﵐ-ﶏﶒ-ﷇﷰ-﷼︐-︙︰-﹄﹇-﹒﹔-﹦﹨-﹫ﹰ-ﹲﹴﹶ-ﻼ！-ﾾￂ-ￇￊ-ￏￒ-ￗￚ-ￜ￠-￦`;
+
+  if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
+    const ranges = [];
+    const range = [];
+    const diacriticsRegex = /^\p{M}$/u;
+    // Some chars must be replaced by their NFKC counterpart during a search.
+    for (let i = 0; i < 65536; i++) {
+      const c = String.fromCharCode(i);
+      if (c.normalize("NFKC") !== c && !diacriticsRegex.test(c)) {
+        if (range.length !== 2) {
+          range[0] = range[1] = i;
+          continue;
+        }
+        if (range[1] + 1 !== i) {
+          if (range[0] === range[1]) {
+            ranges.push(String.fromCharCode(range[0]));
+          } else {
+            ranges.push(
+              `${String.fromCharCode(range[0])}-${String.fromCharCode(
+                range[1]
+              )}`
+            );
+          }
+          range[0] = range[1] = i;
+        } else {
+          range[1] = i;
+        }
+      }
+    }
+    if (ranges.join("") !== NormalizeWithNFKC) {
+      throw new Error(
+        "getNormalizeWithNFKC - update the `NormalizeWithNFKC` string."
+      );
+    }
+  }
+  return NormalizeWithNFKC;
+}
+
+export { CharacterType, getCharacterType, getNormalizeWithNFKC };

web/pdf_page_view.js

@@ -368,6 +368,7 @@ class PDFPageView {
       if (!textLayer.renderingDone) {
         const readableStream = pdfPage.streamTextContent({
           includeMarkedContent: true,
+          disableNormalization: true,
         });
         textLayer.setTextContentSource(readableStream);
       }

web/pdf_viewer.js

@@ -665,6 +665,8 @@ class PDFViewer {
       }
       buffer.length = 0;
       const page = await this.pdfDocument.getPage(pageNum);
+      // By default getTextContent pass disableNormalization equals to false
+      // which is fine because we want a normalized string.
       const { items } = await page.getTextContent();
       for (const item of items) {
         if (item.str) {

web/text_highlighter.js

@@ -208,9 +208,20 @@ class TextHighlighter {
       return;
     }
 
+    let lastDivIdx = -1;
+    let lastOffset = -1;
     for (let i = i0; i < i1; i++) {
       const match = matches[i];
       const begin = match.begin;
+      if (begin.divIdx === lastDivIdx && begin.offset === lastOffset) {
+        // It's possible to be in this situation if we searched for a 'f' and we
+        // have a ligature 'ff' in the text. The 'ff' has to be highlighted two
+        // times.
+        continue;
+      }
+      lastDivIdx = begin.divIdx;
+      lastOffset = begin.offset;
+
       const end = match.end;
       const isSelected = isSelectedPage && i === selectedMatchIdx;
       const highlightSuffix = isSelected ? " selected" : "";

web/text_layer_builder.js

@@ -20,7 +20,8 @@
 // eslint-disable-next-line max-len
 /** @typedef {import("./text_accessibility.js").TextAccessibilityManager} TextAccessibilityManager */
 
-import { renderTextLayer, updateTextLayer } from "pdfjs-lib";
+import { normalizeUnicode, renderTextLayer, updateTextLayer } from "pdfjs-lib";
+import { removeNullCharacters } from "./ui_utils.js";
 
 /**
  * @typedef {Object} TextLayerBuilderOptions
@@ -212,6 +213,16 @@ class TextLayerBuilder {
       }
       end.classList.remove("active");
     });
+
+    div.addEventListener("copy", event => {
+      const selection = document.getSelection();
+      event.clipboardData.setData(
+        "text/plain",
+        removeNullCharacters(normalizeUnicode(selection.toString()))
+      );
+      event.preventDefault();
+      event.stopPropagation();
+    });
   }
 }