werc/bin/werc.rc
2008-12-28 11:51:03 +01:00

309 lines
8.9 KiB
Bash
Executable File

#!/usr/local/plan9/bin/rc
. ./cgilib.rc
cd ..
forbidden_uri_chars='[^a-zA-Z0-9_+\-\/\.]'
response_format=html
fn get_lib_file {
if(test -f $sitedir/_werc/lib/$1)
echo -n $sitedir/_werc/lib/$1
if not if(! ~ $#masterSite 0 && test -f $sitesdir/$masterSite/_werc/lib/$1)
echo -n $sitesdir/$masterSite/_werc/lib/$1
if not if(test -f lib/$1)
echo -n lib/$1
}
# Title
fn gentitle {
echo '<h1 class="headerTitle"><a href="/">' ^ $"siteTitle ^ ' <span id="headerSubTitle">' ^ $"siteSubTitle ^ '</span></a></h1>'
}
# Don't change var name or trailing ';', dirfilter might be changed from _werc/config!
# TODO: Specify the canonical path input format handled by dirfilter
dirfilter='s/\*$//; s,/+\./+,/,g; /\/[._]/d; /'^$forbidden_uri_chars^'/d; /^\/(robots|sitemap)\.txt$|\/index\.(md|html|txt|tpl)$/d; /_werc\/?$/d; s,^\./,,; s,\.(md|html|txt)$,,; '
# To be used from config files
fn hide_paths {
for(i in $*)
dirfilter=$dirfilter^'/^'$i'$/d; '
}
# Sidebar
fn gensidebar {
# Ignore stderr, last path element might be a file that doesn't exist (eg., foo for foo.md)
ls -F $sitedir^/./^$req_paths_list >[2]/dev/null | sed 's,^'$sitedir',,; '^$dirfilter | sort -u | awk -F/ '
function p(x, y, s) {
for(i=0; i < x-y; i+=1)
print s
}
# XXX Regexp to hide non-document/non-dir stuff, there must be a better way, if we really want this.
/^\/([a-zA-Z0-9+_\-]+[*\/]?)+$/ {
d = ""
if(match($0, "/$"))
d = "/"
sub("/$", "") # Strip trailing / for dirs so NF is consistent
p(NF, lNF, "<ul class=\"side-bar\">")
p(lNF, NF, "</ul>")
lNF = NF
bname = $NF d
path = $0 d
gsub("_", " ", bname)
if(index(ENVIRON["REQUEST_URI"] "/", path) == 1)
print "<li><a href=\"" path "\" class=\"thisPage\">&raquo;<i> " bname "</i></a>"
else
print "<li><a href=\"" path "\">&rsaquo; " bname "</a></li>"
}
END { p(lNF, 0, "</ul>") }
'
}
# Handlers
fn set_handler {
handler=$1
handler_args=$*(2-)
}
fn md_handler { cat $* | $formatter }
fn tpl_handler { template $1 }
fn html_handler {
# body states: 0 = no <body> found, 2 = after <body>, 1 = after <body></body>, -1 = after </body>
awk 'gsub(".*<[Bb][Oo][Dd][Yy][^>]*>", "") > 0 {body=2}
gsub("</ *[Bb][Oo][Dd][Yy][^>]*>.*", "") > 0 {print; body=body-1}
body==2 {print}
body==0 {buf=buf "\n" $0}
END {if(body<=0) {print buf}}' < $1
}
fn txt_handler {
echo '<pre>'
# XXX Insering a blank line between lines in input is good for fortunes.txt, but maybe not for other .txt files
# XXX Words are not broken, even if they are way beyond 82 chars long
cat $1 | sed 's/$/\n/g; s/</\&lt;/g; s/>/\&gt;/g' | fmt -l 82 -j
echo '</pre>'
}
fn dir_listing_handler {
d=`{basename -d $1}
echo $d|sed 's,.*//,,g; s,/$,,; s,/, / ,g; s,.*,<h1 class="dir-list-head">&</h1> <ul class="dir-list">,'
# Symlinks suck: '/.' forces ls to list the linked dir if $d is a symlink.
ls -F $d/. | sed $dirfilter' s,.*/([^/]+/?)$,<li><a href="\1">\1</a></li>,'
echo '</ul>'
}
fn 404_handler { template `{get_lib_file 404.tpl} }
fn blog_dir_handler {
blogDirs=$*
tpl_handler lib/feeds/html.tpl
}
fn blog_post_handler {
gen_blog_post_title $1 | $formatter
$formatter < $1
}
fn select_handler {
if(test -f $body.md) {
if(! ~ $#inBlog 0)
set_handler blog_post_handler $body.md
if not
set_handler md_handler $body.md
if(! ~ $#allowComments 0 && ~ $REQUEST_METHOD POST) {
get_post_args comment_text
d=`{date -n} # FIXME Obvious race
d=$body.md_werc/comments/$d/
u=$logged_user
if(~ $#logged_user 0) {
get_post_args comment_user_name comment_user_password
# TODO Should take this path too if the user can login but doesn't
# belong to required group
if(! login_user $comment_user_name $comment_user_password) {
u=$comment_user_name ':' $comment_user_password
d=$d^'_pending'
}
if not
u = $logged_user
}
umask 002
mkdir -m 775 -p $d
echo $u > $d/user
echo $comment_text > $d/body
}
}
if not if(test -f $body.tpl)
set_handler tpl_handler $body.tpl
if not if(test -f $body.html)
set_handler html_handler $body.html
# Explicit .html urls, unfortunately usually the web server will handle this as static files
if not if(~ $body *.html && test -f $body)
perm_redirect `{ echo $REQUEST_URI|sed 's/.html$//' }
# Rss feeds. TODO: we should check that the request is for a real blog dir
if not if(~ $REQUEST_URI */index.rss */index.atom) {
response_format=raw
uri=`{echo $uri | sed 's/index.(rss|atom)$//'}
if(~ $#blogDirs 0)
blogDirs=$sitedir^'/'$uri
uri=$baseuri$"uri
if(~ $REQUEST_URI */index.rss)
master_template=feeds/rss20.tpl
if not if(~ $REQUEST_URI */index.atom)
master_template=feeds/atom.tpl
}
# Blog handler
if not if(~ $body */[bB]log/index */[bB]log//index || ! ~ $#blogDirs 0) {
if(~ $#blogDirs 0)
blogDirs=`{basename -d $body}
u=`{cleanname $baseuri^`{basename -d '/'^$uri}|sed 's,:/,://,'} # Sed recovers '/' in 'http:/' stripped by cleanname
extraHeaders=$"extraHeaders ^ \
'<link rel="alternate" type="application/rss+xml" title="RSS" href="'$"u'/index.rss" />
<link rel="alternate" type="application/atom+xml" title="ATOM" href="'$"u'/index.atom" />'
set_handler tpl_handler lib/feeds/html.tpl
}
# Global tpl (eg sitemap.tpl)
if not if(test -f lib/^$uri^.tpl)
set_handler tpl_handler lib/^$uri^.tpl
if not if(test -f $body.txt)
set_handler txt_handler $body.txt
# Apps
if not if(! ~ $#enabled_apps 0 && select_apps)
set_handler app_handler
# Dir listing
if not if(~ $body */index)
set_handler dir_listing_handler $body
# File not found
if not {
set_handler 404_handler $body
dprint 'NOT FOUND: '$SERVER_NAME^$REQUEST_URI^' - '^$"HTTP_REFERER^' - '^$"HTTP_USER_AGENT
echo 'Status: 404 Not Found'
}
}
fn genbody { $handler $handler_args }
# Careful, the proper p9p path might not be set until initrc.local is sourced
path=(. $PLAN9/bin ./bin/ /bin/ /usr/bin)
site=$SERVER_NAME
headers=lib/headers.tpl
master_template=default_master.tpl
sidebar=sidebar
baseuri=http://$site/
current_date_time=`{date}
for(i in siteTitle siteSubTitle pageTitle extraHeaders)
$i = ''
. ./etc/initrc
if(test -f etc/initrc.local)
. ./etc/initrc.local
werc_root=`{pwd}
# Parse request URL
uri=`{echo -n $REQUEST_URI | sed 's/\?.*//; s/'^$forbidden_uri_chars^'//g; s/\.\.*/./g; 1q'}
ifs='/' { args=`{echo -n $uri} }
# Hack: preload post data so we can access it from templates where cgi's stdin is not accesible
if(~ $REQUEST_METHOD POST) {
get_post_args
login_user
}
if(! ~ $args '') {
if(~ $args($#args) 'index')
perm_redirect `{echo $REQUEST_URI | sed 's,/index$,/,'}
pageTitle=`{echo $args | sed -e 's/ / - /g' -e 's/_/ /g'}
req_path=$uri
}
if not {
req_path='/'
}
p=()
cd $sitedir
for(i in ('' $args)) {
p=($"p^/^$i)
req_paths_list=($req_paths_list $p)
# We don't want blog settings to cascade into posts, note that we are inBlog instead
if(! ~ $#blogDirs 0 && ! ~ $req_path */index.rss */index.atom */[bB]log */[bB]log/) {
inBlog = $blogDirs
blogDirs = ()
}
if(test -d $i) {
cd $i
if(test -f _werc/config)
. _werc/config
}
if(~ $#blogDirs 0 && ~ $#inBlog 0 && ~ $i [Bb]log)
inBlog = 'yes'
}
cd $werc_root
# Redirections and other preprocessing
if(~ $#redirectPermanent 1)
perm_redirect $"redirectPermanent
if not if(~ $#redirectPermanent 2 && {echo $SERVER_NAME^$REQUEST_URI|grep -s $redirectPermanent(1) }) {
# Experimental regexp sub-based redirect, probably should find a nicer interface
to=`{echo $SERVER_NAME^$REQUEST_URI|sed 's@'^$redirectPermanent(1)^'@'^$redirectPermanent(2)^'@'}
if(! ~ $to $REQUEST_URI)
perm_redirect $to
}
# Set Page title
if(~ $pageTitle '')
pageTitle=$siteTitle^' '^$siteSubTitle
if not
pageTitle=$"pageTitle^' | '^$"siteTitle^' '^$"siteSubTitle
body=$sitedir/$req_path
if(test -d $body) {
if(! ~ $body */)
perm_redirect $REQUEST_URI^'/'
body=$body/index
}
select_handler
if(! ~ $#debug 0)
dprint ' '$"SERVER_NAME^$"REQUEST_URI' - '$"HTTP_USER_AGENT' - '$"REQUEST_METHOD' - '$"handler
# Template/body selection
master_template=`{get_lib_file $master_template}
if(~ $response_format html) {
# Is awk_buffer really useful?
template $headers $master_template | awk_buffer
echo '</body></html>'
}
if not if(~ $response_format raw)
template < $master_template