Better escaping for PDF title; fixes #636

tesseract-ocr · Apr 2, 2017 · 9038faf · 9038faf
1 parent c38eeda
commit 9038faf
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 20 deletions.
diff --git a/api/pdfrenderer.cpp b/api/pdfrenderer.cpp
@@ -174,6 +174,11 @@ const int kBasicBufSize = 2048;
 // If the font is 10 pts, nominal character width is 5 pts
 const int kCharWidth = 2;
 
+// Used for memory allocation. A codepoint must take no more than this
+// many bytes, when written in the PDF way. e.g. "<0063>" for the
+// letter 'c'
+const int kMaxBytesPerCodepoint = 20;
+
 /**********************************************************************
  * PDF Renderer interface implementation
  **********************************************************************/
@@ -304,6 +309,22 @@ void ClipBaseline(int ppi, int x1, int y1, int x2, int y2,
     *line_y1 = *line_y2 = (y1 + y2) / 2;
 }
 
+bool CodepointToUtf16be(int code, char *utf16) {
+  if ((code > 0xD7FF && code < 0xE000) || code > 0x10FFFF) {
+    tprintf("Dropping invalid codepoint %d\n", code);
+    return false;
+  }
+  if (code < 0x10000) {
+    snprintf(utf16, sizeof(utf16), "%04X", code);
+  } else {
+    int a = code - 0x010000;
+    int high_surrogate = (0x03FF & (a >> 10)) + 0xD800;
+    int low_surrogate = (0x03FF & a) + 0xDC00;
+    snprintf(utf16, sizeof(utf16), "%04X%04X", high_surrogate, low_surrogate);
+  }
+  return true;
+}
+
 char* TessPDFRenderer::GetPDFTextObjects(TessBaseAPI* api,
                                          double width, double height) {
   STRING pdf_str("");
@@ -442,25 +463,13 @@ char* TessPDFRenderer::GetPDFTextObjects(TessBaseAPI* api,
       if (grapheme && grapheme[0] != '\0') {
         GenericVector<int> unicodes;
         UNICHAR::UTF8ToUnicode(grapheme, &unicodes);
-        char utf16[20];
+        char utf16[kMaxBytesPerCodepoint];
         for (int i = 0; i < unicodes.length(); i++) {
           int code = unicodes[i];
-          // Convert to UTF-16BE https://en.wikipedia.org/wiki/UTF-16
-          if ((code > 0xD7FF && code < 0xE000) || code > 0x10FFFF) {
-            tprintf("Dropping invalid codepoint %d\n", code);
-            continue;
-          }
-          if (code < 0x10000) {
-            snprintf(utf16, sizeof(utf16), "<%04X>", code);
-          } else {
-            int a = code - 0x010000;
-            int high_surrogate = (0x03FF & (a >> 10)) + 0xD800;
-            int low_surrogate = (0x03FF & a) + 0xDC00;
-            snprintf(utf16, sizeof(utf16), "<%04X%04X>",
-                     high_surrogate, low_surrogate);
+          if (CodepointToUtf16be(code, utf16)) {
+            pdf_word += utf16;
+            pdf_word_len++;
           }
-          pdf_word += utf16;
-          pdf_word_len++;
         }
       }
       delete []grapheme;
@@ -471,9 +480,9 @@ char* TessPDFRenderer::GetPDFTextObjects(TessBaseAPI* api,
           kCharWidth * prec(100.0 * word_length / (fontsize * pdf_word_len));
       pdf_str.add_str_double("", h_stretch);
       pdf_str += " Tz";          // horizontal stretch
-      pdf_str += " [ ";
+      pdf_str += " [ <";
       pdf_str += pdf_word;       // UTF-16BE representation
-      pdf_str += " ] TJ";        // show the text
+      pdf_str += "> ] TJ";       // show the text
     }
     if (last_word_in_line) {
       pdf_str += " \n";
@@ -960,15 +969,27 @@ bool TessPDFRenderer::EndDocumentHandler() {
   offsets_.back() += pages_objsize;    // manipulation #2
 
   // INFO
+  STRING utf16_title = "FEFF";  // byte_order_marker
+  GenericVector<int> unicodes;
+  UNICHAR::UTF8ToUnicode(title(), &unicodes);
+  char utf16[kMaxBytesPerCodepoint];
+  for (int i = 0; i < unicodes.length(); i++) {
+    int code = unicodes[i];
+    if (CodepointToUtf16be(code, utf16)) {
+      utf16_title += utf16;
+    }
+  }
+
   char* datestr = l_getFormattedDate();
   n = snprintf(buf, sizeof(buf),
                "%ld 0 obj\n"
                "<<\n"
                "  /Producer (Tesseract %s)\n"
                "  /CreationDate (D:%s)\n"
-               "  /Title (%s)"
+               "  /Title <%s>\n"
                ">>\n"
-               "endobj\n", obj_, TESSERACT_VERSION_STR, datestr, title());
+               "endobj\n",
+               obj_, TESSERACT_VERSION_STR, datestr, utf16_title.c_str());
   lept_free(datestr);
   if (n >= sizeof(buf)) return false;
   AppendPDFObject(buf);

diff --git a/api/renderer.h b/api/renderer.h
@@ -57,6 +57,7 @@ class TESS_API TessResultRenderer {
     /**
      * Starts a new document with the given title.
      * This clears the contents of the output data.
+     * Title should use UTF-8 encoding.
      */
     bool BeginDocument(const char* title);