#!/usr/local/plan9/bin/rc . ./cgilib.rc . ./werclib.rc . ./wercconf.rc . ./corehandlers.rc . ./fltr_cache.rc cd .. forbidden_uri_chars='[^a-zA-Z0-9_+\-\/\.,:]' difs=$ifs # Used to restore default ifs when needed # Expected input: ls -F style, $sitedir/path/to/files/ # dirfilter='s/\*$//; s,/+\./+,/,g; s,^\./,,; /\/[._][^\/]/d; /'$forbidden_uri_chars'/d; /\/index\.(md|html|txt|tpl)$/d; /\/(robots|sitemap)\.txt$/d; /_werc\/?$/d; ' dirclean=' s/\.(md|html|txt)$//; ' # Careful, the proper p9p path might not be set until initrc.local is sourced path=(. $PLAN9/bin ./bin /bin /usr/bin) headers=lib/headers.tpl res_tail='' http_content_type='text/html' ll_add handlers_bar_left nav_tree werc_apps=( apps/* ) werc_root=`{pwd} sitesdir=sites . ./etc/initrc if(test -f etc/initrc.local) . ./etc/initrc.local for(a in $werc_apps) . ./$a/app.rc fn werc_exec_request { site=$SERVER_NAME base_url=http://$site sitedir=$sitesdir/$site master_template=`{get_lib_file default_master.tpl} current_date_time=`{date} # Note: $REQUEST_URI is not officially in CGI 1.1, but seems to be de-facto # Note: We only urldecode %5F->'_' because some sites (stackoverflow.com?) urlencode it in their links, # perhaps we should completel urldecode the whole url. req_path=`{echo -n $REQUEST_URI | sed 's/\?.*//; s!//+!/!g; s/%5[Ff]/_/g; s/'^$forbidden_uri_chars^'//g; s/\.\.*/./g; 1q'} req_url=$base_url^$req_path local_path=$sitedir$req_path local_file='' ifs='/' { args=`{echo -n $req_path} } # Preload post args for templates where cgi's stdin is not accessible if(~ $REQUEST_METHOD POST) { load_post_args login_user } if(~ $req_path */index) perm_redirect `{echo $req_path | sed 's,/index$,/,'} if(~ $local_path */) { if(test -d $local_path) local_path=$local_path^'index' # XXX: This redir might step on apps with synthetic dirs. if not if(ls `{basename -d $local_path}^* >/dev/null >[2]/dev/null) perm_redirect `{echo $req_path|sed 's,/+$,,'} } if not if(~ $req_path *'.' *',' *';' *':') perm_redirect `{echo $req_path | sed 's/[.,;:)]$//'} if not if(test -d $local_path) perm_redirect $req_path^'/' if(! ~ $#args 0) ifs=$NEW_LINE { pageTitle=`{ echo $args|sed -e 's/ / - /g' -e 's/([a-z])-([a-z])/\1 \2/g' -e 's/_/ /g' } } cd $sitedir req_paths_list='/' # Note: req_paths_list doesn't include 'stnythetic' dirs. conf_wd='/' # Used in config files to know where we are in the document tree. if(test -f _werc/config) . _werc/config for(i in $args) { conf_wd=$conf_wd^$i req_paths_list=($req_paths_list $conf_wd) if(test -d $i) { conf_wd=$conf_wd'/' cd $i if(test -f _werc/config) . _werc/config } } cd $werc_root if(~ $#perm_redir_to 1) perm_redirect $perm_redir_to for(l in $perm_redir_patterns) { p=$$l r=$p(1) # If target is absolute, then patern must match whole string if(~ $p(2) http://* https://) r='^'$r t=`{ echo $req_path | sed 's!'^$r^'!'^$p(2)^'!' } # Malicious danger! if(! ~ $"t '' $req_path) perm_redirect $t } setup_handlers # Set Page title if(! ~ $local_file '') { t=`{get_file_title $local_file} if(! ~ $"t '') pageTitle=$t } # XXX Is this never true? because we set pageTitle earlier based on url. if(~ $"pageTitle '') pageTitle=$"siteTitle' '$"siteSubTitle if not pageTitle=$"pageTitle' | '$"siteTitle' '$"siteSubTitle for(h in $extraHttpHeaders) echo $h echo 'Content-Type: '^$http_content_type echo # End of HTTP headers if(! ~ $#debug 0) dprint $"SERVER_NAME^$"REQUEST_URI - $"HTTP_USER_AGENT - $"REQUEST_METHOD - $"handler_body_main - $"master_template if(~ $REQUEST_METHOD HEAD) exit template $headers $master_template | awk_buffer echo $res_tail } werc_exec_request