Update HTML URL regexps to account for attributes before href/src

diff --git a/text.c b/text.c
index 031ab82b513b390120031845e6efd504282df669..b5c3d029a41add8f6e1767ec4586765525aec071 100644 (file)
--- a/text.c
+++ b/text.c
@@ -28,7 +28,7 @@ static const char *text_url_pattern =
        "(/[0-9a-z_=./+&%?-]*)?"
        "(?=\r?\n|[^0-9a-z_=./+&%?-])";
 static pcre *text_url_re;
-static const char *html_url_pattern = "<(a href|img src)=\"https?://[^\"]+(?=\")"; 
+static const char *html_url_pattern = "<(a ([^>]*\\s)?href|img ([^>]*\\s)?src)=\"https?://[^\"]+(?=\")"; 
 static pcre *html_url_re;
 
 static void
@@ -75,7 +75,7 @@ text_process(msgproc *m, char *buf, size_t size)
                if (re_stream_result(sr) == 1) {
                        char *s, *url = re_stream_getresult(sr);
                        if (m->mp_mod->mpm_type == MSGPROC_HTML)
-                               if ((s = strchr(url, '"'))) 
+                               if ((s = strrchr(url, '"'))) 
                                        url = s + 1;
                        if (m->callback) {
                                m->callback(m, url, m->call_data);