Optimize CPDF_CMapParser

In this parser, a vector of CodeRange values are built up during
operations that need to be sync'd with the CPDF_CMap that is being
initialized. In the existing implementation, the vector being built as
a member var for the parser, and copying the values over to the cmap
whenever there is a change. When profiling, this copy is where the code
spends most of its time.

The code has been rewritten to have the parser reference/modify the
instance of the vector in the cmap instead of having its own copy. This
removes all of the copies and significantly speeds things up.

BUG=chromium:881678

Change-Id: Ib8e75962507ca3d3b1ed066fd1faa4fbb7141122
Reviewed-on: https://pdfium-review.googlesource.com/42350
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>

1 files changed, 5 insertions, 2 deletions

diff --git a/core/fpdfapi/font/cpdf_cmap.h b/core/fpdfapi/font/cpdf_cmap.h
index 228c207876..68ad26cfa3 100644
--- a/core/fpdfapi/font/cpdf_cmap.h
+++ b/core/fpdfapi/font/cpdf_cmap.h
@@ -66,8 +66,11 @@ class CPDF_CMap final : public Retainable {
 
   void SetVertical(bool vert) { m_bVertical = vert; }
   void SetCodingScheme(CodingScheme scheme) { m_CodingScheme = scheme; }
-  void SetMixedFourByteLeadingRanges(std::vector<CodeRange> range) {
-    m_MixedFourByteLeadingRanges = range;
+  const std::vector<CodeRange>& GetMixedFourByteLeadingRanges() {
+    return m_MixedFourByteLeadingRanges;
+  }
+  void AppendMixedFourByteLeadingRanges(const CodeRange& range) {
+    m_MixedFourByteLeadingRanges.push_back(range);
   }
 
   int GetCoding() const { return m_Coding; }