CGIservlet.html

   1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
   2 <HTML>
   3
   4 <HEAD>
   5
   6 <TITLE>CGIservlet manual</TITLE>
   7
   8
   9 </HEAD>
  10
  11 <BODY>
  12
  13 <H1 ALIGN="CENTER">CGIservlet</H1>
  14
  15 <P>
  16 A HTTPd "connector" for running CGI scripts on unix systems as WWW
  17 accessible Web sites. The servlet starts a true HTTP daemon that channels
  18 HTTP requests to forked daughter processes. CGIservlet.pl is NOT a
  19 full fledged server. Moreover, this servlet is definitely NOT intended
  20 as a replacement for a real server (e.g., Apache). It's design goal was
  21 SIMPLICITY, and not mileage.
  22 </P>
  23
  24 <P>
  25 Note that a HTTP server can be accessed on your local machine WITHOUT
  26 internet access (but WITH a DNS?):
  27 use "http://localhost[:port]/[path]" or "http://127.0.0.1[:port]/[path]"
  28 as the URL. It is also easy to restrict access to the servlet to localhost
  29 users (i.e., the computer running the servlet).
  30 </P>
  31
  32 <P>
  33 Suggested uses:
  34 </P>
  35
  36 <ul>
  37     <li>
  38
  39     <P>
  40     A testbed for CGI-scripts and document-trees outside the primary server.
  41     When developing new scripts and services, you don't want to mess up your
  42     current Web-site. CGIservlet is an easy way to start a temporary (private)
  43     server. CGIservlet allows to test separate HTTP server components, e.g.,
  44     user authentication, in isolation.
  45     </P>
  46
  47     <li>
  48
  49     <P>
  50     A special purpose temporary server (WWW everywhere/anytime).
  51     We run identification and other experiments over the inter-/intra-net using
  52     CGI-scripts. This means a lot of development and changes and only little
  53     actual run-time. The people doing this do not want "scripting" access to our
  54     departmental server with all its restrictions and security. So we need a
  55     small, lightweigth, easy-to-configure server that can be run by each
  56     investigator on her own account (and risk).
  57     </P>
  58
  59     <li>
  60
  61     <P>
  62     Interactive WWW presentations.
  63     Not everyone is content with the features of "standard" office presentation
  64     software. HTML and its associated browsers are an alternative (especially
  65     under Linux). However, you need a server to realize the full interactive
  66     nature of the WWW. CGIservlet with the necessary scripts can be run from
  67     a floppie (a Web server in 100 kB). The CGIservlet can actually run a
  68     (small) web site from RAM, without disk access (if you DONOT use the
  69     2>pid.log redirection on startup).
  70     With the "localhost" or "127.0.0.1" id in your browser you can use it
  71     standalone.
  72     </P>
  73
  74 </ul>
  75
  76 <P>
  77 When the servlet is started with the -r option, only requests from "localhost"
  78 or "127.0.0.1" are accepted (default) or from addresses indicated after the
  79 -r switch.
  80 </P>
  81
  82 <P>
  83 Running demo's and more information can be found at
  84 <A HREF="http://www.fon.hum.uva.nl/rob/OSS/OSS.html">
  85 http://www.fon.hum.uva.nl/rob/OSS/OSS.html</A>
  86 </P>
  87
  88 <H2 ALIGN="CENTER">Inner workings</H2>
  89
  90 <P>
  91 Whenever an HTTP request is received, the specified CGI script is
  92 started inside a child process as if it was inside a real server (e.g.,
  93 Apache). The evironment variables are set more or less as in Apache.
  94 Note that CGIservlet only uses a SINGLE script for ALL requests.
  95 No attemps for security are made, it is the script's responsibility to
  96 check access rights and the validity of the request.<br>
  97 When no scripts are given, CGIservlet runs as a bare bone WWW server
  98 configurable to execute scripts (the default setting is as a STATIC server).
  99 </P>
 100
 101 <PRE>
 102 Use:    CGIservlet.pl -&lt;switch&gt; &lt;argument&gt; 2&gt;pid.log &     (sh)
 103         CGIservlet.pl -&lt;switch&gt; &lt;argument&gt; &gt;&pid.log &     (csh)
 104         </PRE>
 105
 106         <P>
 107         The servlet prints out pid and port number on STDERR. It is
 108         adviced to store these in a separate file (this will become the
 109         error log). <BR>
 110         NOTE: When running CGIservlet from a Memmory Image (i.e. RAM),
 111         do NOT redirect the error output to a file, but use something
 112         like MAILTO!
 113         </P>
 114
 115         <PRE>
 116 Stop:   sh pid.log                      (kills the server process)
 117 </PRE>
 118
 119 <P>
 120 The first line in the file that receives STDERR output is a command
 121 to stop CGIservlet.
 122 </P>
 123
 124 <P>
 125 examples:
 126 </P>
 127
 128 <PRE>
 129 CGIservlet.pl -p 2345 -d /cgi-bin/CGIscriptor.pl -t /WWW 2>pid.log &
 130 CGIservlet.pl -p 8080 -b 'require "CGIscriptor.pl";' -t $PWD -e 'Handle_Request();' 2>pid.log &
 131 </PRE>
 132
 133 <P>
 134 The following example settings implement a static WWW server using 'cat'
 135 (and prohibiting Queries):
 136 <dl>
 137 <dt>-p 8080
 138 <dt>-t `pwd`
 139 <dt>-b ''
 140 <dt>-e
 141 'exit if $ENV{QUERY_STRING};$ENV{PATH_INFO}=~/\.([\w]+)$/; "Content-type: ".$mimeType{uc($1)}."\n\n";'
 142 <dt>-d 'cat -u -s'
 143 <dt>-w '/index.html'
 144 <dt>-c 32
 145 <dt>-l 512
 146 </dl>
 147
 148 This is identical to the (static) behaviour of CGIservlet when
 149 -e '' -d '' -x '' is used.
 150 <br>
 151 The CGIservlet command should be run from the intended server-root directory.
 152 </P>
 153
 154 <P>
 155 Another setting will use a package 'CGIscriptor.pl' with a function
 156 'HandleRequest()' to implement an interactive WWW server with inline Perl
 157 scripting:
 158 <dl>
 159 <dt>-p 8080
 160 <dt>-t `pwd`
 161 <dt>-b 'require "CGIscriptor.pl";'
 162 <dt>-e 'HandleRequest();'
 163 <dt>-d ''
 164 <dt>-w '/index.html'
 165 <dt>-c 32
 166 <dt>-l 32767
 167 </dl>
 168 </P>
 169
 170 <P>
 171 Look in the source code or in the CGIservletSETUP.pl file for the current
 172 default settings.
 173 </P>
 174
 175 <H2 ALIGN="CENTER">Command-line switches</H2>
 176
 177 <P>
 178 There are many switches to tailor the workings of CGIservlet.pl.
 179 Some are fairly esoteric and you should only look for them if you
 180 need something special urgently. When building a Web site,
 181 the specific options you need will "suggest" themselves (e.g., port
 182 number, script, or server-root directory). Most default settings
 183 should work fine.
 184 </P>
 185
 186 <P>
 187 You can add your own configuration in a file
 188 called 'CGIservletSETUP.pl'. This file will be executed ("eval"-ed)
 189 after the default setup, but before the command line options take
 190 effect. CGIservlet looks for the SETUP file in the startup directory
 191 and in the CGIscriptor subdirectory.<br>
 192 (Note that the $beginarg variable is evaluated AFTER the setup file).
 193 </P>
 194
 195 <P>
 196 In any case, it is best to change the default settings instead of
 197 using the option switches. All defaults are put in a single block.
 198 </P>
 199
 200 <P>
 201 <H3>switches and arguments</H3>
 202 </P>
 203
 204 <dl>
 205     <dt>Realy important
 206     <ul>
 207         <li>-p[ort] port number<br>
 208         For example -p 2345<br>
 209         Obviously the port CGIservlet listenes to. Default: -p 8080
 210
 211         <li>-a[lias] Alias1 RealURL1 ...<br>
 212         For example -a '/Stimulus.aifc' '/catAIFC.xmr'<br>
 213         Replaces the given Alias URL path by its real URL path. Accepts full
 214         regular expressions too (identified by NON-URL characters).<br>
 215         That is, on each request it performs (in order):<br>
 216         <pre>
 217 if($AliasTranslation{$Path})
 218 {
 219     $Path = $AliasTranslation{$Path};
 220 }
 221 elsif(@RegAliasTranslation)
 222 {
 223     my $i;
 224     for($i=0; $i&lt;scalar(@RegAliasTranslation); ++$i)
 225     {
 226         my $Alias   = $RegAliasTranslation[$i];
 227         my $RealURL = $RegURLTranslation[$i];
 228         last if ($Path =~ s#$Alias#$RealURL#g);
 229     };
 230 };
 231          </pre>
 232          <p>
 233          The effects can be quite drastic, so be
 234          carefull. Note also, that entering many Regular Expression
 235          aliases could slow down your servlet. Checking stops after
 236          the first match.<br>
 237          Full regular expression alias translations are done in the
 238          order given! They are recognized as Aliases containing
 239          regexp's (i.e., non-URL) operator characters like '^' and
 240          '$'.<br>
 241          Note: The command line is NOT a good place for entering
 242          Aliases, change the code below or add aliases to
 243          CGIservletSETUP.pl.
 244         </p>
 245
 246         <li>--help<br>
 247         Prints the manual
 248
 249     </ul>
 250
 251     <dt>Script related
 252     <ul>
 253         <li>-b[egin] perl commands<br>
 254         For example -b 'require "CGIscriptor.pl";' or
 255         'require "/WWW/cgi-bin/XMLelement.pl";'<br>
 256         Perl commands evaluated at startup
 257
 258         <li>-d[o] perl script file<BR>
 259         For example -d '/WWW/cgi-bin/CGIscriptor.pl'<br>
 260         The actual CGI-script started as a perl {do "scriptfile"} command.
 261         The PATH_INFO and the QUERY are pushed on @ARGV.
 262
 263         <li>-x shell command
 264         <li>-qx shell command
 265         <li>-exec shell command<br>
 266         OS shell script or command, e.g., -x 'CGIscriptor.pl' or
 267         -x '/WWW/cgi-bin/my-script' <br>
 268         The actual CGI-script started as `my-script \'$Path\' \'$QueryString\'`.
 269         -qx and -exec[ute] are aliases of -x. For secutiry reasons, Paths or
 270         queries containing '-quotes are rejected.
 271
 272         <li>-e[val] perl commands<br>
 273         For example -e 'Handle_Request();'   <br>
 274         The argument is evaluated as perl code. The actual CGI-script
 275         can be loaded once with -b 'require module.pm' and you only have to
 276         call the central function(s).
 277     </ul>
 278
 279     <dt>WWW-tree related
 280     <ul>
 281         <li>-t[extroot] path
 282         For example -t "$PWD" or -t "/WWW/documents"<br>
 283         The root of the server hierachy. Defaults to the working directory
 284         at startup time (`pwd`)
 285
 286         <li>-w[elcome] filepath
 287         For example -w "/index.html"   (default)<br>
 288         The default welcome page used when no path is entered. Note that
 289         this path can point to anything (or nothing real at all).
 290     </ul>
 291     <dt>Security related<br>
 292     The following arguments supply some rudimentary security. It is the
 293     responsibility of the script to ensure that the requests are indeed
 294     "legal".
 295     <ul>
 296         <li>-c[hildren] maximum child processes<br>
 297         For example -c 32<br>
 298         The maximum number of subprocesses started. If there are more requests,
 299         the oldest requests are "killed". This should take care of "zombie"
 300         processes and server overloading. Note that new requests will be
 301         serviced irrespective of the success of killing it's older siblings.
 302
 303         <li>-xtime maximum running time of a child<br>
 304         For example -xtime 36000<br>
 305         The maximum time a child may run in seconds. After a new request has
 306         been servised, all children that have run for longer than this time
 307         will be killed. This stops runaway processes, often connected to
 308         web-crawlers.
 309
 310         <li>-l[ength] maximum length of request in bytes<br>
 311         For example -l 32768 <br>
 312         This prevents overloading the server with enormous queries. Reading of
 313         requests simply stops when this limit is reached. This DOES affect
 314         POST requests.  If the combined length of the COMPLETE HTTP request,
 315         including headers, exceeds this limit, the whole request is dropped.
 316
 317         <li>-r[estrict] [Remote-address [Remote-host]]<br>
 318         For example -r 127.0.0.1          (default of -r)<br>
 319         A space separated list of client IP addresses and/or domain names that
 320         should be serviced. Default, i.e., '-r' without any addresses or domain
 321         names, is the localhost IP address '127.0.0.1'.<br>
 322         When using CGIservlet for local purposes only (e.g., development or a
 323         presentation), it would be unsafe to allow others to access the servlet.
 324         If -r is used (or the corresponding @RemoteAddr or @RemoteHost lists are
 325         filled in the code below), all requests from clients whose Remote-address
 326         or Remote-host do not match the indicated addresses will be rejected.
 327         Partial addresses and domain names are allowed. Matching is done according
 328         to Remote-addr =~ /^\Q$pattern\E/ (front to back) and
 329         Remote-host =~ /\Q$pattern\E$/ (back to front)
 330
 331         <li>-m[emory]<br>
 332         No arguments.<br>
 333         Reads complete Web site into memory and runs from this image.
 334         Set $UseRAMimage = 1; to activate memory-only running.<br>
 335         Note that running osshellscripts from this
 336         image makes any "security" related claims very shaky.<br>
 337     </ul>
 338     <dt>Speedup
 339     <ul>
 340         <li>-n[oname]<br>
 341         No arguments. <br>
 342         Retrieving the domain name of the Client (i.e., Remote-host) is a
 343         very slow process and normally useless. To skip it, enter this
 344         option. Note that you cannot use '-r Remote-host' anymore after
 345         you enter -n, only IP addresses will work.
 346     </ul>
 347 </dl>
 348
 349 <H3 ALIGN="CENTER">Configuration with the <a href="/CGIservletSETUP.pl">CGIservletSETUP.pl</a> file</H3>
 350
 351 <P>
 352 You can add your own configuration in a file
 353 called '<a href="/CGIservletSETUP.pl">CGIservletSETUP.pl</a>'.
 354 This file will be executed ("eval"-ed)
 355 after the default setup, but before the command line options take
 356 effect. CGIservlet looks for the SETUP file in the startup directory
 357 and in the CGIservlet and CGIscriptor subdirectories.
 358 (Note that the $beginarg variable is evaluated even later).
 359 </P>
 360
 361
 362 <H3 ALIGN="CENTER">Changing POST to GET requests</H3>
 363
 364 <P>
 365 CGIservlet normally only handles requests with the GET method. Processing
 366 the input from POST requests is left to the reading application. POST
 367 requests add some extra complexity to processing requests. Sometimes,
 368 the reading application doesn't handle POST requests. CGIservlet
 369 already has to manage the HTTP request. Therefore, it can easily
 370 handle the POST request. If the variable $POSTtoGET is set to any
 371 non-false value, the content of whole POST request is added to the
 372 QUERY_STRING environment variable (preceeded by a '&' if necessary).
 373 The content-length is set to 0. If $POSTtoGET equals 'GET', the method
 374 will also be changed to 'GET'.
 375 </P>
 376
 377 <H3>remarks:</H3>
 378
 379 <P>
 380 All of the arguments of -d, -x, and -e are processed sequentially
 381 in this order. This might not be what you want so you should be
 382 carefull when using multiple executable arguments.
 383 If none of the executable arguments is DEFINED (i.e., they are entered
 384 as -d '' -e '' -x ''), each request is treated as a simple
 385 text-retrieval. THIS CAN BE A SECURITY RISK!
 386 </P>
 387
 388 <P>
 389 The wiring of an interactive web-server, which also calls shell
 390 scripts with the extension '.cgi', is in place. You can
 391 "activate" it by changing the "$ExecuteOSshell = 0;" line to
 392 "$ExecuteOSshell = 1;".<br>
 393 If you have trouble doing this, it might be a good idea
 394 to reconsider using a dynamic web server. Executing shell
 395 scripts inside a web server is a rather dangerous practise.
 396 </P>
 397
 398 <P>
 399 CGIservlet can run its "standard" web server from memory.
 400 At startup, all files are read into a hash table.  Upon
 401 request, the contents of the file are placed in the
 402 environment variable: CGI_FILE_CONTENTS.<br>
 403 No further disk access is necessary. This means that:
 404 <ol>
 405     <li> CGIservlet can run a WWW site from a removable disk,
 406     e.g., a floppy
 407     <li> The web servlet can run without any read or write privilege.
 408     <li> The integrity of the Web-site contents can be secured at the
 409     level you want
 410 </ol>
 411 To compres the memory (RAM) immage, you should hook the
 412 compression function to <br>
 413 $CompressRAMimage = sub { return shift;}; <br>
 414 and the decompression function to <br>
 415 $DecompressRAMimage = sub { return shift;}; <br>
 416 </P>
 417
 418 <H2 ALIGN="CENTER">license</H2>
 419
 420 <P>
 421 This program is free software; you can redistribute it and/or
 422 modify it under the terms of the GNU General Public License
 423 as published by the Free Software Foundation; either version 2
 424 of the License, or (at your option) any later version.
 425 </P>
 426
 427 <P>
 428 This program is distributed in the hope that it will be useful,
 429 but WITHOUT ANY WARRANTY; without even the implied warranty of
 430 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 431 GNU General Public License for more details.
 432 </P>
 433
 434 <P>
 435 You should have received a copy of the GNU General Public License
 436 along with this program; if not, write to the Free Software
 437 Foundation, Inc., 59 Temple Place - Suite 330,
 438 Boston, MA  02111-1307, USA.
 439 </P>
 440
 441 <PRE>
 442 Author: Rob van Son
 443         email:
 444         R.J.J.H.vanSon@gmail.com
 445         Institute of Phonetic Sciences/ACLC
 446         University of Amsterdam
 447
 448         copying freely from the mhttpd server by Jerry LeVan (levan@eagle.eku.edu)
 449 Date:   15 Jan 2002
 450 Ver:    1.3
 451 Env:    Perl 5.002 and later
 452
 453 Note:   CGIservlet.pl was directly inspired by Jerry LeVan's
 454         (levan@eagle.eku.edu) simple mhttpd server which again was
 455         inspired by work of others. CGIservlet is used as a bare bones
 456         socket server for a single CGI script at a time.
 457 </PRE>
 458
 459 </BODY>
 460
 461 </HTML>