PDF: Use PDF metadata for the title instead of the last path element.

pdf/pdfium/pdfium_engine.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "pdf/pdfium/pdfium_engine.h"
 
 #include <math.h>
 
 #include "base/i18n/icu_encoding_detection.h"
 #include "base/i18n/icu_string_conversions.h"
@@ skipping 506 matching lines @@
 #endif
 }
 
 // Returns a VarDictionary (representing a bookmark), which in turn contains
 // child VarDictionaries (representing the child bookmarks).
 // If NULL is passed in as the bookmark then we traverse from the "root".
 // Note that the "root" bookmark contains no useful information.
 pp::VarDictionary TraverseBookmarks(FPDF_DOCUMENT doc, FPDF_BOOKMARK bookmark) {
   pp::VarDictionary dict;
   base::string16 title;
-  unsigned long buffer_size = FPDFBookmark_GetTitle(bookmark, NULL, 0);
-  size_t title_length = base::checked_cast<size_t>(buffer_size) /
-      sizeof(base::string16::value_type);
-  if (title_length > 0) {
-    PDFiumAPIStringBufferAdapter<base::string16> api_string_adapter(
-        &title, title_length, true);
-    void* data = api_string_adapter.GetData();
-    FPDFBookmark_GetTitle(bookmark, data, buffer_size);
-    api_string_adapter.Close(title_length);
+  size_t buffer_size = FPDFBookmark_GetTitle(bookmark, NULL, 0);

[raymes, 2015/08/26 06:33:15]

Should we keep this as an unsigned long since this is the type returned and
passed in?

[Sam McNally, 2015/08/26 08:23:10]

Done.

+  if (buffer_size > 0) {
+    PDFiumAPIStringBufferSizeInBytesAdapter<base::string16> api_string_adapter(
+        &title, buffer_size, true);
+    api_string_adapter.Close(FPDFBookmark_GetTitle(
+        bookmark, api_string_adapter.GetData(), buffer_size));
   }
   dict.Set(pp::Var("title"), pp::Var(base::UTF16ToUTF8(title)));
 
   FPDF_DEST dest = FPDFBookmark_GetDest(doc, bookmark);
   // Some bookmarks don't have a page to select.
   if (dest) {
     int page_index = FPDFDest_GetPageIndex(doc, dest);
     dict.Set(pp::Var("page"), pp::Var(page_index));
   }
 
@@ skipping 623 matching lines @@
 }
 
 void PDFiumEngine::SetScrollPosition(const pp::Point& position) {
   position_ = position;
 }
 
 bool PDFiumEngine::IsProgressiveLoad() {
   return doc_loader_.is_partial_document();
 }
 
+std::string PDFiumEngine::GetMetadata(const std::string& key) {
+  size_t size = FPDF_GetMetaText(doc(), key.c_str(), nullptr, 0);
+  if (size == 0)
+    return std::string();
+
+  base::string16 value;
+  PDFiumAPIStringBufferSizeInBytesAdapter<base::string16> string_adapter(
+      &value, size, false);
+  string_adapter.Close(
+      FPDF_GetMetaText(doc(), key.c_str(), string_adapter.GetData(), size));
+  return base::UTF16ToUTF8(value);
+}
+
 void PDFiumEngine::OnPartialDocumentLoaded() {
   file_access_.m_FileLen = doc_loader_.document_size();
   fpdf_availability_ = FPDFAvail_Create(&file_availability_, &file_access_);
   DCHECK(fpdf_availability_);
 
   // Currently engine does not deal efficiently with some non-linearized files.
   // See http://code.google.com/p/chromium/issues/detail?id=59400
   // To improve user experience we download entire file for non-linearized PDF.
   if (!FPDFAvail_IsLinearized(fpdf_availability_)) {
     doc_loader_.RequestData(0, doc_loader_.document_size());
@@ skipping 2735 matching lines @@
   IntersectClipRect(dc, settings.bounds.x(), settings.bounds.y(),
                     settings.bounds.x() + settings.bounds.width(),
                     settings.bounds.y() + settings.bounds.height());
 
   // A temporary hack. PDFs generated by Cairo (used by Chrome OS to generate
   // a PDF output from a webpage) result in very large metafiles and the
   // rendering using FPDF_RenderPage is incorrect. In this case, render as a
   // bitmap. Note that this code does not kick in for PDFs printed from Chrome
   // because in that case we create a temp PDF first before printing and this
   // temp PDF does not have a creator string that starts with "cairo".
-  base::string16 creator;
-  size_t buffer_bytes = FPDF_GetMetaText(doc, "Creator", NULL, 0);
-  if (buffer_bytes > 1) {
-    FPDF_GetMetaText(doc, "Creator",
-                     base::WriteInto(&creator, buffer_bytes + 1), buffer_bytes);
+  bool use_bitmap = false;
+  if (base::StartsWith(GetMetadata("Creator"), "cairo",
+                       base::CompareCase::INSENSITIVE_ASCII)) {
+    use_bitmap = true;
   }
-  bool use_bitmap = false;
-  if (base::StartsWith(creator, L"cairo", base::CompareCase::INSENSITIVE_ASCII))
-    use_bitmap = true;
 
   // Another temporary hack. Some PDFs seems to render very slowly if
   // FPDF_RenderPage is directly used on a printer DC. I suspect it is
   // because of the code to talk Postscript directly to the printer if
   // the printer supports this. Need to discuss this with PDFium. For now,
   // render to a bitmap and then blit the bitmap to the DC if we have been
   // supplied a printer DC.
   int device_type = GetDeviceCaps(dc, TECHNOLOGY);
   if (use_bitmap ||
       (device_type == DT_RASPRINTER) || (device_type == DT_PLOTTER)) {
@@ skipping 99 matching lines @@
     double* height) {
   FPDF_DOCUMENT doc = FPDF_LoadMemDocument(pdf_buffer, pdf_buffer_size, NULL);
   if (!doc)
     return false;
   bool success = FPDF_GetPageSizeByIndex(doc, page_number, width, height) != 0;
   FPDF_CloseDocument(doc);
   return success;
 }
 
 }  // namespace chrome_pdf