Small optimizations for CBC_HighLevelEncoder::lookAheadTestchromium/3550

Currently ClusterFuzz is timing out when running cases that cause a
large number of calls to this method. Looking at the cases, I believe
these to be valid calls, so this CL attempts to lower the cost of
making each individual call.

Adds in pre-allocation of a vector that has a fixed size and uses a
const-ref for passing in |msg| to avoid copying.

BUG=chromium:881678

Change-Id: I61ec4dc96e79c84def5b10102cc58a96773ce07f
Reviewed-on: https://pdfium-review.googlesource.com/42230
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>

1 files changed, 1 insertions, 1 deletions

diff --git a/fxbarcode/datamatrix/BC_HighLevelEncoder.h b/fxbarcode/datamatrix/BC_HighLevelEncoder.h
index 50cfc8c592..9ad0d705be 100644
--- a/fxbarcode/datamatrix/BC_HighLevelEncoder.h
+++ b/fxbarcode/datamatrix/BC_HighLevelEncoder.h
@@ -29,7 +29,7 @@ class CBC_HighLevelEncoder {
                                     WideString ecLevel,
                                     bool allowRectangular,
                                     int32_t& e);
-  static int32_t lookAheadTest(WideString msg,
+  static int32_t lookAheadTest(const WideString& msg,
                                int32_t startpos,
                                int32_t currentMode);
   static bool isDigit(wchar_t ch);