| author | Michael Krelin <hacker@klever.net> | 2007-09-14 22:52:21 (UTC) |
|---|---|---|
| committer | Michael Krelin <hacker@klever.net> | 2007-09-14 22:52:21 (UTC) |
| commit | 5f1d69ac7753243b93761944e9444f01d8a7e5dd (patch) (side-by-side diff) | |
| tree | 43ded5d2da27a54f3fa806ddc6f054970d3cb6fc | |
| parent | 429c48d3d08e6c2f6c385d8975f7b5bf5e67acf3 (diff) | |
| download | libopkele-5f1d69ac7753243b93761944e9444f01d8a7e5dd.zip libopkele-5f1d69ac7753243b93761944e9444f01d8a7e5dd.tar.gz libopkele-5f1d69ac7753243b93761944e9444f01d8a7e5dd.tar.bz2 | |
robustness improvements in link descovery mechanisms
accept stray spaces in link tags and inside relevant attribute values.
Thanks, Gen, for pointing it out.
Signed-off-by: Michael Krelin <hacker@klever.net>
| -rw-r--r-- | lib/consumer.cc | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/lib/consumer.cc b/lib/consumer.cc index 299b3bc..ff5da91 100644 --- a/lib/consumer.cc +++ b/lib/consumer.cc @@ -287,10 +287,10 @@ namespace opkele { if(r && r!=CURLE_WRITE_ERROR) throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r); // strip out everything past body - static const char *re_hdre = "<head[^>]*>", - *re_lre = "<link\\b([^>]+)>", - *re_rre = "\\brel=['\"]([^'\"]+)['\"]", - *re_hre = "\\bhref=['\"]([^'\"]+)['\"]"; + static const char *re_hdre = "<\\s*head[^>]*>", + *re_lre = "<\\s*link\\b([^>]+)>", + *re_rre = "\\brel\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]", + *re_hre = "\\bhref\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]"; #if defined(USE_LIBPCRECPP) static pcrecpp::RE_Options ro(PCRE_CASELESS|PCRE_DOTALL); static pcrecpp::RE |