diff options
| author | Jay Berkenbilt <ejb@ql.org> | 2020-10-27 15:51:41 +0100 |
|---|---|---|
| committer | Jay Berkenbilt <ejb@ql.org> | 2020-10-27 16:57:48 +0100 |
| commit | bcea54fcaa16a7d5feff0c4cd038fea51d1359ea (patch) | |
| tree | 2c0dc1076a7963c0d9ac77fc8bb87a57ef7ad964 /libqpdf/QPDFTokenizer.cc | |
| parent | 81d2c548dc619e21519651606970c21f720c8658 (diff) | |
| download | qpdf-bcea54fcaa16a7d5feff0c4cd038fea51d1359ea.tar.zst | |
Revert removal of unreadCh change for performance
Turns out unreadCh is much more efficient than seek(-1, SEEK_CUR).
Update comments and code to reflect this.
Diffstat (limited to 'libqpdf/QPDFTokenizer.cc')
| -rw-r--r-- | libqpdf/QPDFTokenizer.cc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/libqpdf/QPDFTokenizer.cc b/libqpdf/QPDFTokenizer.cc index 4f2e2153..4217575c 100644 --- a/libqpdf/QPDFTokenizer.cc +++ b/libqpdf/QPDFTokenizer.cc @@ -855,7 +855,7 @@ QPDFTokenizer::readToken(PointerHolder<InputSource> input, if (unread_char) { - input->seek(-1, SEEK_CUR); + input->unreadCh(char_to_unread); } if (token.getType() != tt_eof) |