Merge pull request #15701 from Snuffleupagus/move-string-helpers

Move some string helper functions to the worker-thread

src/core/annotation.js

@@ -23,26 +23,25 @@ import {
   AnnotationType,
   assert,
   BASELINE_FACTOR,
-  escapeString,
   FeatureTest,
   getModificationDate,
   IDENTITY_MATRIX,
-  isAscii,
   LINE_DESCENT_FACTOR,
   LINE_FACTOR,
   OPS,
   RenderingIntentFlag,
   shadow,
   stringToPDFString,
-  stringToUTF16BEString,
   unreachable,
   Util,
   warn,
 } from "../shared/util.js";
 import {
   collectActions,
+  escapeString,
   getInheritableProperty,
   getRotationMatrix,
+  isAscii,
   numberToString,
   stringToUTF16String,
 } from "./core_utils.js";
@@ -1879,7 +1878,11 @@ class WidgetAnnotation extends Annotation {
       value,
     };
 
-    const encoder = val => (isAscii(val) ? val : stringToUTF16BEString(val));
+    const encoder = val => {
+      return isAscii(val)
+        ? val
+        : stringToUTF16String(val, /* bigEndian = */ true);
+    };
     dict.set("V", Array.isArray(value) ? value.map(encoder) : encoder(value));
 
     const maybeMK = this._getMKDict(rotation);
@@ -3546,14 +3549,19 @@ class FreeTextAnnotation extends MarkupAnnotation {
     freetext.set("DA", da);
     freetext.set(
       "Contents",
-      isAscii(value) ? value : stringToUTF16BEString(value)
+      isAscii(value)
+        ? value
+        : stringToUTF16String(value, /* bigEndian = */ true)
     );
     freetext.set("F", 4);
     freetext.set("Border", [0, 0, 0]);
     freetext.set("Rotate", rotation);
 
     if (user) {
-      freetext.set("T", isAscii(user) ? user : stringToUTF16BEString(user));
+      freetext.set(
+        "T",
+        isAscii(user) ? user : stringToUTF16String(user, /* bigEndian = */ true)
+      );
     }
 
     if (apRef || ap) {

src/core/core_utils.js

@@ -313,6 +313,19 @@ function escapePDFName(str) {
   return buffer.join("");
 }
 
+// Replace "(", ")", "\n", "\r" and "\" by "\(", "\)", "\\n", "\\r" and "\\"
+// in order to write it in a PDF file.
+function escapeString(str) {
+  return str.replace(/([()\\\n\r])/g, match => {
+    if (match === "\n") {
+      return "\\n";
+    } else if (match === "\r") {
+      return "\\r";
+    }
+    return `\\${match}`;
+  });
+}
+
 function _collectJS(entry, xref, list, parents) {
   if (!entry) {
     return;
@@ -572,6 +585,10 @@ function getNewAnnotationsMap(annotationStorage) {
   return newAnnotationsByPage.size > 0 ? newAnnotationsByPage : null;
 }
 
+function isAscii(str) {
+  return /^[\x00-\x7F]*$/.test(str);
+}
+
 function stringToUTF16HexString(str) {
   const buf = [];
   for (let i = 0, ii = str.length; i < ii; i++) {
@@ -584,8 +601,11 @@ function stringToUTF16HexString(str) {
   return buf.join("");
 }
 
-function stringToUTF16String(str) {
+function stringToUTF16String(str, bigEndian = false) {
   const buf = [];
+  if (bigEndian) {
+    buf.push("\xFE\xFF");
+  }
   for (let i = 0, ii = str.length; i < ii; i++) {
     const char = str.charCodeAt(i);
     buf.push(
@@ -614,11 +634,13 @@ export {
   DocStats,
   encodeToXmlString,
   escapePDFName,
+  escapeString,
   getArrayLookupTableFactory,
   getInheritableProperty,
   getLookupTableFactory,
   getNewAnnotationsMap,
   getRotationMatrix,
+  isAscii,
   isWhiteSpace,
   log2,
   MissingDataException,

src/core/writer.js

@@ -13,9 +13,14 @@
  * limitations under the License.
  */
 
-import { bytesToString, escapeString, warn } from "../shared/util.js";
+import { bytesToString, warn } from "../shared/util.js";
 import { Dict, Name, Ref } from "./primitives.js";
-import { escapePDFName, numberToString, parseXFAPath } from "./core_utils.js";
+import {
+  escapePDFName,
+  escapeString,
+  numberToString,
+  parseXFAPath,
+} from "./core_utils.js";
 import { SimpleDOMNode, SimpleXMLParser } from "./xml_parser.js";
 import { BaseStream } from "./base_stream.js";
 import { calculateMD5 } from "./crypto.js";

src/shared/util.js

@@ -1037,36 +1037,6 @@ function stringToPDFString(str) {
   return strBuf.join("");
 }
 
-function escapeString(str) {
-  // replace "(", ")", "\n", "\r" and "\"
-  // by "\(", "\)", "\\n", "\\r" and "\\"
-  // in order to write it in a PDF file.
-  return str.replace(/([()\\\n\r])/g, match => {
-    if (match === "\n") {
-      return "\\n";
-    } else if (match === "\r") {
-      return "\\r";
-    }
-    return `\\${match}`;
-  });
-}
-
-function isAscii(str) {
-  return /^[\x00-\x7F]*$/.test(str);
-}
-
-function stringToUTF16BEString(str) {
-  const buf = ["\xFE\xFF"];
-  for (let i = 0, ii = str.length; i < ii; i++) {
-    const char = str.charCodeAt(i);
-    buf.push(
-      String.fromCharCode((char >> 8) & 0xff),
-      String.fromCharCode(char & 0xff)
-    );
-  }
-  return buf.join("");
-}
-
 function stringToUTF8String(str) {
   return decodeURIComponent(escape(str));
 }
@@ -1167,7 +1137,6 @@ export {
   createPromiseCapability,
   createValidAbsoluteUrl,
   DocumentActionEventType,
-  escapeString,
   FeatureTest,
   FONT_IDENTITY_MATRIX,
   FontType,
@@ -1180,7 +1149,6 @@ export {
   InvalidPDFException,
   isArrayBuffer,
   isArrayEqual,
-  isAscii,
   LINE_DESCENT_FACTOR,
   LINE_FACTOR,
   MissingPDFException,
@@ -1198,7 +1166,6 @@ export {
   string32,
   stringToBytes,
   stringToPDFString,
-  stringToUTF16BEString,
   stringToUTF8String,
   TextRenderingMode,
   UnexpectedResponseException,

test/unit/core_utils_spec.js

@@ -17,10 +17,14 @@ import { Dict, Ref } from "../../src/core/primitives.js";
 import {
   encodeToXmlString,
   escapePDFName,
+  escapeString,
   getInheritableProperty,
+  isAscii,
   isWhiteSpace,
   log2,
   parseXFAPath,
+  stringToUTF16HexString,
+  stringToUTF16String,
   toRomanNumerals,
   validateCSSFont,
 } from "../../src/core/core_utils.js";
@@ -221,6 +225,14 @@ describe("core_utils", function () {
     });
   });
 
+  describe("escapeString", function () {
+    it("should escape (, ), \\n, \\r, and \\", function () {
+      expect(escapeString("((a\\a))\n(b(b\\b)\rb)")).toEqual(
+        "\\(\\(a\\\\a\\)\\)\\n\\(b\\(b\\\\b\\)\\rb\\)"
+      );
+    });
+  });
+
   describe("encodeToXmlString", function () {
     it("should get a correctly encoded string with some entities", function () {
       const str = "\"\u0397ell😂' & <W😂rld>";
@@ -333,4 +345,50 @@ describe("core_utils", function () {
       expect(cssFontInfo.italicAngle).toEqual("2.718");
     });
   });
+
+  describe("isAscii", function () {
+    it("handles ascii/non-ascii strings", function () {
+      expect(isAscii("hello world")).toEqual(true);
+      expect(isAscii("こんにちは世界の")).toEqual(false);
+      expect(isAscii("hello world in Japanese is こんにちは世界の")).toEqual(
+        false
+      );
+    });
+  });
+
+  describe("stringToUTF16HexString", function () {
+    it("should encode a string in UTF16 hexadecimal format", function () {
+      expect(stringToUTF16HexString("hello world")).toEqual(
+        "00680065006c006c006f00200077006f0072006c0064"
+      );
+
+      expect(stringToUTF16HexString("こんにちは世界の")).toEqual(
+        "30533093306b3061306f4e16754c306e"
+      );
+    });
+  });
+
+  describe("stringToUTF16String", function () {
+    it("should encode a string in UTF16", function () {
+      expect(stringToUTF16String("hello world")).toEqual(
+        "\0h\0e\0l\0l\0o\0 \0w\0o\0r\0l\0d"
+      );
+
+      expect(stringToUTF16String("こんにちは世界の")).toEqual(
+        "\x30\x53\x30\x93\x30\x6b\x30\x61\x30\x6f\x4e\x16\x75\x4c\x30\x6e"
+      );
+    });
+
+    it("should encode a string in UTF16BE with a BOM", function () {
+      expect(
+        stringToUTF16String("hello world", /* bigEndian = */ true)
+      ).toEqual("\xfe\xff\0h\0e\0l\0l\0o\0 \0w\0o\0r\0l\0d");
+
+      expect(
+        stringToUTF16String("こんにちは世界の", /* bigEndian = */ true)
+      ).toEqual(
+        "\xfe\xff\x30\x53\x30\x93\x30\x6b\x30\x61\x30\x6f\x4e\x16\x75\x4c\x30\x6e"
+      );
+    });
+  });
 });

test/unit/util_spec.js

@@ -17,14 +17,11 @@ import {
   bytesToString,
   createPromiseCapability,
   createValidAbsoluteUrl,
-  escapeString,
   getModificationDate,
   isArrayBuffer,
-  isAscii,
   string32,
   stringToBytes,
   stringToPDFString,
-  stringToUTF16BEString,
 } from "../../src/shared/util.js";
 
 describe("util", function () {
@@ -246,40 +243,10 @@ describe("util", function () {
     });
   });
 
-  describe("escapeString", function () {
-    it("should escape (, ), \\n, \\r, and \\", function () {
-      expect(escapeString("((a\\a))\n(b(b\\b)\rb)")).toEqual(
-        "\\(\\(a\\\\a\\)\\)\\n\\(b\\(b\\\\b\\)\\rb\\)"
-      );
-    });
-  });
-
   describe("getModificationDate", function () {
     it("should get a correctly formatted date", function () {
       const date = new Date(Date.UTC(3141, 5, 9, 2, 6, 53));
       expect(getModificationDate(date)).toEqual("31410609020653");
     });
   });
-
-  describe("isAscii", function () {
-    it("handles ascii/non-ascii strings", function () {
-      expect(isAscii("hello world")).toEqual(true);
-      expect(isAscii("こんにちは世界の")).toEqual(false);
-      expect(isAscii("hello world in Japanese is こんにちは世界の")).toEqual(
-        false
-      );
-    });
-  });
-
-  describe("stringToUTF16BEString", function () {
-    it("should encode a string in UTF16BE with a BOM", function () {
-      expect(stringToUTF16BEString("hello world")).toEqual(
-        "\xfe\xff\0h\0e\0l\0l\0o\0 \0w\0o\0r\0l\0d"
-      );
-      expect(stringToUTF16BEString("こんにちは世界の")).toEqual(
-        "\xfe\xff\x30\x53\x30\x93\x30\x6b\x30\x61" +
-          "\x30\x6f\x4e\x16\x75\x4c\x30\x6e"
-      );
-    });
-  });
 });