package require http proc geturl {url} { if {![regexp {^http://.*} $url]} { set url "http://[set url]" } set token [http::geturl $url] set ncode [http::ncode $token] if {($ncode == "301") || ($ncode == "302")} { # an redirect upvar #0 $token state if {![info exists state(meta)]} { error } array set meta $state(meta) if {![info exists meta(Location)]} { error } return [geturl $meta(Location)] } set data [http::data $token] http::cleanup $token return $data } proc find_openid.server {input} { # <link rel="openid.server" href="http://openid.example.com/"> if {[regexp {<link rel="openid.server" href="([^"]*)"} $input dummy server]} { return $server } else { return {} } } proc find_openid.delegate {input} { # <link rel="openid.delegate" href="http://exampleuser.livejournal.com/"> if {[regexp {<link rel="openid.delegate" href="([^"]*)"} $input dummy delegate]} { return $delegate } else { return {} } } proc get_openid_server_and_delegate {url} { set stuff [geturl $arg(url)] set r(openid.server) [find_openid.server $stuff] set r(openid.delegate) [find_openid.delegater $stuff] return [array get r] } proc openid_assoc_request {identity_provider} {} proc openid_check_immediate {identity_provider identity} {} proc openid_check_setup {} {} proc openid_check_authentication {} {}
CJL - Which regexps are causing problems? I'm going to assume the first one is OK, but the ones in find_openid.server and find_openid.delegate aren't. Firstly, there's no need to match the text before and after the section of interest. Secondly '.*' is greedy - it will match as much as possible (i.e. it will match up to the LAST '>' in the input string, rather than just as far as the next one). To ask for the minimal match, use '.*?' instead, although I have found that doesn't always behave as I'd expect it to. In general, I'd say try to avoid the combination of '.*' if you can, and in this case it works if you match 'any number of non-quote characters'. In summary, I think your patterns should be more like:
regexp {<link rel="openid.server" href="([^"]*)"} $input dummy serverZarutian 21:02 5. januar 2006: Thank you, I am never entirely at ease with regexpes.