The fact that the URI regex we use to detect URIs in chat
conversations does not exclude double quotes causes it to pick up more
of the text than it should, in some cases.
Bug #598381
#define SCHEMES "(https?|s?ftps?|nntp|news|javascript|about|ghelp|apt|telnet|"\
"file|webcal|mailto)"
-#define BODY "([^\\ \\n]+)"
-#define END_BODY "([^\\ \\n]*[^,;\?><()\\ \"\\.\\n])"
+#define BODY "([^\\ \\n\"]+)"
+#define END_BODY "([^\\ \\n\"]*[^,;\?><()\\ \"\\.\\n])"
#define URI_REGEX "("SCHEMES"://"END_BODY")" \
"|((mailto:)?"BODY"@"BODY"\\."END_BODY")"\
"|((www|ftp)\\."END_BODY")"