Repeated starting sequences like <?, <!DECL or <![CDATA[ could
lead to quadratic behavior if no matching ending sequence was found.
Separate the inline HTML scanners. Remember if scanning the whole input
for a specific ending sequence failed and skip subsequent scans.
The basic idea is to remove suffixes >, ?> and ]]> from the
respective regex. Since these regexes are already constructed to match
lazily, they will stop before an ending sequence. To check whether an
ending sequence was found, we can simply test whether the input buffer
is large enough to hold the match plus a potential suffix. If the regex
doesn't find the ending sequence, it will match so many characters that
this test is guaranteed to fail. In this case, we set a flag to avoid
further attempts to execute the regex.
To check which inline HTML regex to use, we inspect the start of the
text buffer. This allows some fixed characters to be removed from the
start of some regexes. matchlen is adjusted with a single addition
that accounts for both the relevant prefix and suffix.
Repeated starting sequences like
<?,<!DECLor<![CDATA[could lead to quadratic behavior if no matching ending sequence was found. Separate the inline HTML scanners. Remember if scanning the whole input for a specific ending sequence failed and skip subsequent scans.The basic idea is to remove suffixes
>,?>and]]>from the respective regex. Since these regexes are already constructed to match lazily, they will stop before an ending sequence. To check whether an ending sequence was found, we can simply test whether the input buffer is large enough to hold the match plus a potential suffix. If the regex doesn't find the ending sequence, it will match so many characters that this test is guaranteed to fail. In this case, we set a flag to avoid further attempts to execute the regex.To check which inline HTML regex to use, we inspect the start of the text buffer. This allows some fixed characters to be removed from the start of some regexes.
matchlenis adjusted with a single addition that accounts for both the relevant prefix and suffix.Fixes #299.