- WIKIT_ADMIN
- Set this to ignore the protected-pages list, which normally prevents editing certain pages
- WIKIT_BASE
- Set this to define the URL of the wiki pages - this causes generated pages to always refer to wiki pages with this base location (no more cgi-bin URLs), see below for more info.
- WIKIT_CACHE
- Set this to a writable dir, and wikit will save new pages in HTML form in this area. This can be used with Apache to maintain a fast-access page cache so most accesses never launch CGI.
- WIKIT_CSS
- Set this to the URL of the style sheet to use, and wikit will become CSS-enabled.
- WIKIT_DUMP
- This is a debug facility, set this to a writable file and all accesses will append info to that file.
- WIKIT_EMAIL
- Set this to email address to use to report CGI failures (default goes nowhere).
- WIKIT_HIST
- Set this to a writable directory, and wikit will save every modified page in raw text form, under a unique name. It also appends entries to a ".index" text file in there. This is used by the Tcl'ers Wiki to move all page edits into a CVS repository (this is an independent cron script).
- WIKIT_REVS
- This is a URL prefix. When set, wikit will append "/<N>" to it as a link on each editable page, called "Revisions".
- WIKIT_SAVE
- If this is set to "0", and WIKIT_CACHE is set as well, then clear the cache on edit changes, but do not save changes back to it. This is needed in special cases, i.e. when accessing the same wiki from two different URLs, one using CSS and the other not. The format which should not be saved back in the cache should use WIKIT_SAVE=0 in the corresponding CGI script.
- WIKIT_WRU
- This variable is special-purpose, it allows the Tcler's Wiki to interact with the wru utility ("who are you, see [1]) to pick up cookie which add the initials of the poster to page edits and recent changes.
Static page access in ApacheThere is a way to make Apache use static page copies, based on an error-redirect trick. The idea is to make all pages static, and remove them from the cache when edited. An access to a page which is not in the cache forces a CGI call, which generates the page and updates the cache.Here's what I do to make this work:
- say the area you want to present is "mywiki"
- create a "mywiki/" dir, writable for CGI (usually user apache, www, or nobody)
- create a file "mywiki/.htaccess" containing:
ErrorDocument 404 /mywiki/nph-blah.cgi DirectoryIndex /mywiki/nph-blah.cgi/0 Options MultiViews
- create a file "mywiki/nph-blah.cgi" containing:
#!/bin/sh echo HTTP/1.0 200 OK cd /path/to/my/datafile WIKIT_BASE=http://<mysite>/mywiki/ export WIKIT_BASE WIKIT_CACHE=/path/to/mywiki export WIKIT_CACHE #WIKIT_CSS=http://url/of/my/style/sheet.css #export WIKIT_CSS # uncomment following as needed #WIKIT_DUMP=/path/to/mywikilogfile #export WIKIT_DUMP #WIKIT_HIST=/path/to/mywikihistdir #export WIKIT_HIST exec /path/to/tclkit wikit.kit mywikit.tkd
- (add "-readonly" at the end to make the wiki readonly)
24June03 CM - I found that the "DirectoryIndex /mywiki/nph-blah.cgi/0" option prevents all directory indexes to work. If you are using Apache, you can instead use this in 'httpd.conf':
# DirectoryIndex: Name of the file or files to use as a pre-written HTML # directory index. Separate multiple entries with spaces. # <IfModule mod_dir.c> DirectoryIndex index.html 0.html </IfModule>which will add "0.html" as a way to find a directory index, in the Wiki and elsewhere; but will not prevent the automatic index to work in non-wiki directories. This only works if a static page already exists in the cache, however... (it's usually the case for "0").
10apr03 rf - I had to use the "execCGI" option in my "mywiki/.htaccess" to get it working:
ErrorDocument 404 /mywiki/nph-blah.cgi DirectoryIndex /mywiki/nph-blah.cgi/0 Options MultiViews execCGI
Hmmm... is WIKIT_CACHE really available?? It doesn't seem to work at all. Also, just a wee small thingy, can WIKIT_CACHE be /path/to/my/htdocs? Will that work? I generally do not want people to be able to poke their browsers into my cgi area ;^)12nov02 jcw - Yes it is, that's how the Tclers' Wiki is set up. The trick is the redirect everything to a CGI. So what you do is create a dir just for the cache, make that the public one, i.e. the "real" public area, then redirect page misses to a CGI elsewhere, telling that CGI to update this area. When it does, the page will exist, i.e. not trigger an error from then on, hence be served statically. If you follow what I'm saying...10 Jun 2003 -WL- I finally got the cache to work, yay! One annoying thing is every search becomes a file not found.... I kind of wish the cache feature was a bit more straightforward. IE, if CACHE=blah, wikit knows to look in $CACHE for the file first, instead of bouncing around with 404 and WIKIT_BASE, etc. Now that the cache is there, I guess the next thing to do is to give wikit the ability to send out gzip encoded files...
13dec2002 ps I wonder why the above .cgi has [echo HTTP/1.0 200 OK], my apache doesn't like it... It works fine without that statement. (I had similar problems so left it off until I realised that the "nph-" part of the script name is significant to Apache, and once you name the script thus, it needs to have the "echo" statement --MNO).And if you are having trouble with the parameters in the .htaccess file (like I had), it could well be that you need to set the options in a <Directory /mywiki> entry in your httpd.conf like this:
<VirtualHost *> ServerName www.mywiki.com AddHandler cgi-script .cgi <Directory /> ErrorDocument 404 /mywiki/wiki.cgi DirectoryIndex /mywiki/wiki.cgi/0 Options ExecCGI MultiViews </Directory> </VirtualHost>Because your httpd.conf may (quite rightly) have global AllowOverride None, which prevents .htaccess from working...
Wikit and AOLserverAlthough the Apache solution is the better one, with the possibility for static page access, I use AOLserver a lot so I want wikit to work in AOLserver too.The first thing that needs to be done is add a couple of lines to wikit/lib/wikit/web.tcl, near the beginning of [ProcessCGI] is a line:
set ::script_name $::env(SCRIPT_NAME)Right before that, insert:
if { [string first "AOLserver/3" $::env(SERVER_SOFTWARE)] > -1 } { #Aolserver does not provide a correct SCRIPT_NAME #fix this by parsing the WIKIT_BASE (which is mandatory anyway) catch { regexp -nocase {^http://[^/]+(/.+)/$} $::env(WIKIT_BASE) -> ::env(SCRIPT_NAME) } }This fixes the problem that the edit pages do not allow you to save the page (because the URL is wrong).I am working on a [ns_register_proc] to handle static page access, like the apache solution. Any chance someone has already done that? Should be simple enough:
if { the_static_page_exists } { ns_returnfile 200 text/html $static_file } else { ns_returnredirect /mywiki/wiki.cgi/$pagenumber }But might need some tweaks to wikit so two different types of urls are used for internal links.-- PS.06feb03 jcw - I've just added your first patch to the wikit code base, should now work with AOLserver - thx.
Static page access in Tclhttpd07mai03 I'd like to have this static-feature in Tclhttpd, too. Therefore I used the following tcl-code. Place it in the custom-directory of tclhttpd (if you have problems, that tclhttpd isn't reading the custom-directory, see [3].
set wikitCacheDir mywiki ;# above mentioned as WIKIT_CACHE set wikitCgiDir /cgi-bin/wikit.cgi Url_PrefixInstall /$wikitCacheDir [list wikit_handler $wikitCacheDir $wikitCgiDir] proc wikit_handler {wikitCacheDir wikitCgiDir sock suffix} { global Config upvar #0 Httpd$sock data # make suffix "relative" regsub {^/} $suffix {} suffix # redirect to "homepage" of wikit, if no "page"/suffix is given if {![string length $suffix]} { set suffix "0" } set cachefile [file join $Config(docRoot) $wikitCacheDir $suffix] # the url is http://xyz/0, but the cached file will be "0.html" # only append ".html" if request has no extension # (this avoids appending ".html" for other files in cache-dir, like # wikit.css if {[string equal [file extension $cachefile] ""]} { append cachefile ".html" } # check if file is in cache if {[file exists $cachefile]} { # file is in cache, so simply return cached file Httpd_ReturnFile $sock text/html $cachefile } else { # REDIRECT_URL must be set, because WIKIT checks it for caching # and tclhttpd doesn't set it itself set ::env(REDIRECT_URL) [file join $wikitCgiDir $suffix] Httpd_Redirect $::env(REDIRECT_URL) $sock } }I don't know if this is the preferred solution, but it works for me (at least for Tclhttpd 3.4.2).For running TclHttpd on Windows I've set the following Windows-System-Environment-variables (let's say my Tclhttpd is installed in
tclhttpd = D:/tclhttpd3.4.2)
WIKIT_CACHE = D:/tclhttpd3.4.2/mywiki/ WIKIT_BASE = http://localhost:8015/mywiki/I suppose the trailing / is important.When I have a CSS, I use, e.g.:
WIKIT_CSS = http://localhost:8015/mywiki/wikit.css-- Stefan Vogel.31July03 Jeff Smith I tried the above script on tclhttpd 3.4.2. It is redirecting to the /cgi-bin/wikit.cgi and working OK but no static pages are generated in the mywiki/ directory. My wikit.cgi script looks like
#!/bin/sh WIKIT_BASE=http://cobra:8015/mywiki/ export WIKIT_BASE WIKIT_CACHE=/usr/local/tclkit/mywiki export WIKIT_CACHE exec /usr/local/tclkit/tclkit /usr/local/tclkit/wikit.kitAny ideas?31July03 Jeff Smith Added
lappend Cgi(env-pass) REDIRECT_URLTo the beginning of the script above
set wikitCacheDir mywiki ;# above mentioned as WIKIT_CACHEand now static pages appear in mywiki/ directory. I think with cgi processing in tclhttpd on unix, not all the environment variables are passed, whereas with tclhttpd on windows they are.04 September 2003 Jeff Smith I had a go at making a Starkit of Wikit with inbuilt Tclhttpd including Stefan's script.June 2004 BBW Here is the URL domain handler I have for launching a wikit CGI script. This is similar in spirit to that provided by Stefan Vogel but I don't really care about the cache. I just wanted the URLs to not reference the CGI script.
# Create a tclhttpd domain that wires directly into the wiki cgi script # I deploy this as wiki/custom/wiki_startup.tcl # and I have the CGI script in wiki/bin # Start bin/httpd.tcl with the -library wiki/custom option set here [file dirname [info script]] set WikiCGI [file join $here ../bin/wiki.cgi] Url_PrefixRemove / ;# Remove normal doc domain Url_PrefixInstall / [list WikiDomain /] proc WikiDomain {prefix sock suffix} { global WikiCGI upvar #0 Httpd$sock data set extra /$suffix # er, extra == $data(url) WikiHandle $data(url) $extra $WikiCGI $sock } proc WikiHandle {url extra path sock} { global Doc env # We pass in / for the url argument because that turns # into the SCRIPT_NAME env var, which the wikit's web.tcl glues # onto $extra. # extra -> PATH_INFO # url -> SCRIPT_NAME Cgi_SetEnvAll $sock $path $extra / env CgiSpawn $sock $path } # We need a few files, like images and the stylesheet Doc_AddRoot /htdocs [file join $here ../htdocs] Doc_AddRoot /images [file join $here ../htdocs/images] # Set up logging file mkdir [file join $here ../log] set config::Config(LogFile) [file join $here ../log/log]MHo: Can't get caching to always work right. The problem is: WIKIT_BASE has to be specified full qualified. This means, it contains the servername and if I test on the same machine running the server on, the cached files contain something like <base href="http://localhost/wikiccsas/">, which is obviouseley wrong when accessing these cached html from another machine. The reason for this is I'm constructing WIKIT_BASE with Httpd_SelfUrl, which resolves in this situation to localhost. Heres my caching implementation:
# support more than one wiki Url_PrefixInstall /wikiccsas [list wikiHandler ccsas] Url_PrefixInstall /wikiccpm [list wikiHandler ccpm] proc wikiHandler {prefix sock suffix} { # additional vars for wikit set ::env(WIKIT_CACHE) [file join [Doc_Root] wikicache $prefix] set ::env(WIKIT_HIST) [file join [Doc_Root] wikihist $prefix] set ::env(WIKIT_BASE) [Httpd_SelfUrl /wiki$prefix/]; # scheint problematisch (localhost?) # prevent WIKIT_HIST from beeing overwritten late in patched cgi.tcl's (internal regarding to my own tclhttpd.exe..., not of interest here) set ::Config(WIKIT_HIST) $::env(WIKIT_HIST) global Config upvar #0 Httpd$sock data set wikitCgiDir /cgi-bin/wiki/${prefix}.exe; # PB-Wrapper to tclkitsh wikit.kit because with wikitsh.exe there are search-problems... # make suffix "relative" regsub {^/} $suffix {} suffix # redirect to "homepage" of wikit, if no "page"/suffix is given if {![string length $suffix]} { set suffix "0" } set cachefile [file join [Doc_Root] $::env(WIKIT_CACHE) $suffix] # the url is http://xyz/0, but the cached file will be "0.html" # only append ".html" if request has no extension # (this avoids appending ".html" for other files in cache-dir, like # wikit.css if {[string equal [file extension $cachefile] ""]} { append cachefile ".html" } # check if file is in cache if {[file exists $cachefile]} { # file is in cache, so simply return cached file Httpd_ReturnFile $sock text/html $cachefile } else { # REDIRECT_URL must be set, because WIKIT checks it for caching # and tclhttpd doesn't set it itself set ::env(REDIRECT_URL) [file join $wikitCgiDir $suffix] Httpd_RedirectSelf $::env(REDIRECT_URL) $sock } }
How do I configure in windows?19may03 Richard -- I dont get it. How do I set these variables (WIKIT_CSS for example) when using windows and the TCL-webserver? 17jun04 I'm running wikit.kit under tclhttpd on windows in CGI-mode successfully. See Matthias Hoffmann - Other Utilities for a possible solution.June 10, 2003 LV On Linux/Solaris what I do is invoke wikit from a CGI shell script. In that script, I set these variables and then invoke the wikit command itself. Would that be a possible solution for you as well?
9 October 2003 Stefan Vogel You might want to use the wikitool (from Tom Krehbiel) to admin your wikit inside tclhttpd. Have a look at wikitool
13-feb-2005 VK: is it possible to use other than utf-8 encoding in generated HTMLs? I want to use CP-1251 for editing my wiki pages, which I use on my own site (setting wikit was easy, thanks!).