Merge branch 'develop' into devsecops

aleks-ivanov · Jan 10, 2024 · 66fe498 · 66fe498
2 parents b0e0781 + 05d1d9c
commit 66fe498
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 11 deletions.
diff --git a/itext.tests/itext.kernel.tests/itext/kernel/pdf/canvas/parser/PdfTextExtractorTest.cs b/itext.tests/itext.kernel.tests/itext/kernel/pdf/canvas/parser/PdfTextExtractorTest.cs
@@ -122,5 +122,24 @@ public virtual void ShortOctalDataAsTextTest() {
                 NUnit.Framework.Assert.AreEqual("EC", PdfTextExtractor.GetTextFromPage(pdfDocument.GetPage(1)));
             }
         }
+
+        [NUnit.Framework.Test]
+        public virtual void NotDefaultCodespacesCyrillicTest() {
+            String inFile = sourceFolder + "notDefaultCodespacesCyrillic.pdf";
+            using (PdfDocument pdfDocument = new PdfDocument(new PdfReader(inFile))) {
+                String extractedText = PdfTextExtractor.GetTextFromPage(pdfDocument.GetPage(1));
+                NUnit.Framework.Assert.IsTrue(extractedText.Contains("бронирование"));
+                NUnit.Framework.Assert.IsTrue(extractedText.Contains("From"));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void NotDefaultCodespacesChineseTest() {
+            String inFile = sourceFolder + "notDefaultCodespacesChinese.pdf";
+            using (PdfDocument pdfDocument = new PdfDocument(new PdfReader(inFile))) {
+                String extractedText = PdfTextExtractor.GetTextFromPage(pdfDocument.GetPage(1));
+                NUnit.Framework.Assert.IsTrue(extractedText.Contains("L3B 廠： 新竹科學工業園區新竹市東區力行二路 1 號"));
+            }
+        }
     }
 }
diff --git a/....tests/resources/itext/kernel/parser/PdfTextExtractorTest/notDefaultCodespacesChinese.pdf b/....tests/resources/itext/kernel/parser/PdfTextExtractorTest/notDefaultCodespacesChinese.pdf
diff --git a/...tests/resources/itext/kernel/parser/PdfTextExtractorTest/notDefaultCodespacesCyrillic.pdf b/...tests/resources/itext/kernel/parser/PdfTextExtractorTest/notDefaultCodespacesCyrillic.pdf
diff --git a/itext/itext.kernel/itext/kernel/font/PdfType0Font.cs b/itext/itext.kernel/itext/kernel/font/PdfType0Font.cs
@@ -742,27 +742,31 @@ private static String GetOrdering(PdfDictionary cidFont) {
         }
 
         private static bool ContainsCodeInCodeSpaceRange(IList<byte[]> codeSpaceRanges, int code, int length) {
+            long unsignedCode = code & unchecked((int)(0xffffffff));
             for (int i = 0; i < codeSpaceRanges.Count; i += 2) {
                 if (length == codeSpaceRanges[i].Length) {
-                    int mask = 0xff;
-                    int totalShift = 0;
                     byte[] low = codeSpaceRanges[i];
                     byte[] high = codeSpaceRanges[i + 1];
-                    bool fitsIntoRange = true;
-                    for (int ind = length - 1; ind >= 0; ind--, totalShift += 8, mask <<= 8) {
-                        int actualByteValue = (code & mask) >> totalShift;
-                        if (!(actualByteValue >= (0xff & low[ind]) && actualByteValue <= (0xff & high[ind]))) {
-                            fitsIntoRange = false;
-                        }
-                    }
-                    if (fitsIntoRange) {
+                    long lowValue = BytesToLong(low);
+                    long highValue = BytesToLong(high);
+                    if (unsignedCode >= lowValue && unsignedCode <= highValue) {
                         return true;
                     }
                 }
             }
             return false;
         }
 
+        private static long BytesToLong(byte[] bytes) {
+            long res = 0;
+            int shift = 0;
+            for (int i = bytes.Length - 1; i >= 0; --i) {
+                res += (bytes[i] & 0xff) << shift;
+                shift += 8;
+            }
+            return res;
+        }
+
         private void FlushFontData() {
             if (cidFontType == CID_FONT_TYPE_0) {
                 GetPdfObject().Put(PdfName.Type, PdfName.Font);

diff --git a/port-hash b/port-hash
@@ -1 +1 @@
-bc6666faacd4ec5f4ac133a8d23fee1ee5e4884a
+1c6199fc40db8637ee9e95bbe09585289e71182f
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		bc6666faacd4ec5f4ac133a8d23fee1ee5e4884a
		1c6199fc40db8637ee9e95bbe09585289e71182f