U2pipe

Users >> Rex_Gozar >> u2pipe

u2pipe / wininetd

Version 1.0.3
revised 2009-04-17 by Rex Gozar

Brief

You can connect a web server to a Universe database via u2pipe. Windows platforms use wininetd as the socket listener. Unix/Linux platforms can use xinetd or inetd. This document explains how to do it.

There are better, more sophisticated ways of doing this. You could use RedBack (or whatever its new name is), UniObjects, mvNet, or any other commercial solution. But I'm tired of the "barriers to entry" in the PICK programming world, and I wanted to provide a simple open source alternative for connecting web servers to IBM U2 databases.

Disclaimer

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Features

• Free (as in "free speech")
• Free (as in "free beer")
• Runs on Windows and unix.
• u2pipe - a single purpose utility to get the request from stdin, log into universe, and send the response to stdout.
• wininetd - an open source socket listening service
• Independent from web server; use any webserver/language you want -- ASP, JSP, PHP, ColdFusion, or whatever.
• Create as many socket ports as you want to access as many accounts as you want.
• Once a request is received, it can be handled by any Universe machine on your internal network.
• optional: You can restrict access to a list of "allowed" ip-addresses.
• optional: You can log every request and response.
• optional: You can (and should) run a filter program to validate TCL commands to execute.

Non-Features

• not tested on Unidata, but uses InterCall so it should work.
• no connection pooling, but only eats a user license for the duration of the request processing.
• no connection limit, but you can create your own (when using xinetd, you can specify the number of instances; pending requests are queued).
• no encryption, but you can (and should) do your own.
• no multi-byte character support is built in, but since raw bytes are sent back and forth there's nothing stopping you from passing multi-byte characters.
• not responsible for character mangling by transport protocol. For example, HTTP is not guaranteed to transport binary characters as-is, but you can use url encoding, base64, or some other ASCII-armoring techniques for binary data.

Binaries

• AIX: http://www.autopower.com/rgozar/u2pipe-1.0.3-aix.tgz
• Linux: http://www.autopower.com/rgozar/u2pipe-1.0.3-linux.tgz
• Windows: http://www.autopower.com/rgozar/u2pipe-1.0.3-setup.exe

Windows Installation

1. Download http://www.autopower.com/rgozar/u2pipe-1.0.3-setup.exe
2. Run it.

• Must be logged in as an administrator so the wininetd service can be installed.
• Requires Windows 2000, or a later version.
• Requires access to Universe, either on the same machine, or reachable via UniRPC.
• No reboot required, so you can install anytime.
• The following files are installed.
  • c:\windows\u2pipe.exe
  • c:\windows\u2pipe.ini
  • c:\windows\wininetd.exe
  • c:\windows\wininetd.conf
  • c:\windows\system32\unirpc32.dll
  • c:\windows\system32\uvclnt32.dll
  • c:\windows\system32\uvic32.dll

AIX and Linux Installation

1. You'll need to be savvy enough to install a new xinetd/inetd service (or at least know how to search Google.)
2. Download the binary for your platform and untar it. Copy (or move) the u2pipe binary into the directory of your choice (typically /usr/local/bin).
3. Edit /etc/services to add u2pipe with the port number of your choice.
   

   u2pipe   6789/tcp   # u2pipe Web Service

4. Edit your xinetd (or inetd) configuration file.
   
   AIX - sample inetd.conf

   u2pipe   stream   tcp   nowait   root   /usr/local/bin/u2pipe

   
   Linux - sample /etc/xinetd.d/u2pipe

   service u2pipe

{

       instances               = 3
       socket_type             = stream
       wait                    = no
       user                    = root
       server                  = /usr/local/bin/u2pipe
       nice                    = 10
       disable                 = no
       only_from               = localhost 10.0.0.0 192.168.1.132

}

1. Create the ini file. Use your favorite editor to create /etc/u2pipe.ini -- the format is described in "Configuring u2pipe.ini" below.
2. Optional: You can configure a specific ini file to use by specifying the the "server_args". For example:
   

     server_args = /etc/sales_account.ini

Configuring wininetd

You typically don't have to change wininetd.conf, but if you want to the format is:

port user:password full-u2pipe-path [full-inifile-path]

• space or tab separators
• comment lines start with any non-numeric character
• port
  Any available port number.
• user:password
  A specific user id and password separated by a colon (:) or the word none to use the local system account (default).
• full-u2pipe-path
  Typically c:\windows\u2pipe.exe
• full-inifile-path
  Optional: You can configure several u2pipe processes to run, each with their own ini file.
• sample: default configuration
  6789 none c:\windows\u2pipe.exe
• sample: using a different ini file
  9901 none c:\windows\u2pipe.exe c:\other.ini

In Windows Services, you must change wininetd properties from "Manual" to "Automatic" for the service to start at reboot.

You must restart the wininetd service for changes to wininetd.conf to take effect. This is easy to do from the DOS command line by typing:

net stop wininetd
net start wininetd

Configuring u2pipe.ini

Changes to u2pipe.ini take effect immediately (does not require restart).

You can create different ini files to log into different machines or accounts, and tie them to specific port numbers in wininetd.conf (or inetd.conf, or xinetd.d files -- you get the idea).

• [U2PIPE]
  required: section declaration.
• HOSTNAME
  required: The Universe machine's hostname or ip-address.
• USER
  required: The user id to log into the machine.
• PASSWORD
  required: The user's password.
• ACCOUNT
  required: The Universe account to log into.
• RUN
  optional: Prefix the request with the specified string. For example, you may want to run a program named "MYPROG" to parse the request string for processing, so you would set "RUN=MYPROG". The TCL command to execute will be "MYPROG whatever-the-request-string-is". You should always run your requests through a program to parse arguments and filter out malicious text.
• ALLOWIP
  optional: List of allowed client IP-addresses that can make requests, separated by semi-colons (;). To allow requests from the CMD shell, i.e. ECHO 000004DATE|U2PIPE, add an extra semi-colon -- for example, ALLOWIP=192.168.1.7; (note the trailing semi-colon).
• LOGFILE
  optional: Log the requests and responses to a file. This file can grow big fast, so it is recommended that you only enable it for debugging. Also note that the log file is not locked and multiple u2pipe processes may update it simultaneously.
• ;
  optional: Comments can be entered, or parameters no-op'ed, using a semicolon.

Sample u2pipe.ini with minimum configuration:

[U2PIPE]
hostname=localhost
user=bgates
password=iluvlinux!
account=SALES

Testing

Run both of these tests to make sure that u2pipe and wininetd are configured correctly. Error messages show up in the LOGFILE (specified in the ini file) or the Windows Event Viewer.

• From the DOS command line type: ECHO 000004DATE|U2PIPE
  • response: 000037Friday, September 29, 2006 03:18pm
  • Note that the response includes trailing CR and LF characters.

• From your browser, enter in the location: http://yourhost:6789/000004DATE
  • response: 000037Friday, September 29, 2006 03:18pm
  • try: http://yourhost:6789/000004TIME
  • try: http://yourhost:6789/000005USERS
  • try: http://yourhost:6789/000005LISTU

• Note: Any single word TCL command can be performed. We'll need to do more work to allow multi-word commands.

• Note: Multi-line responses may get mangled by HTTP. We'll need to use ASCII armoring (i.e. base64 encoding) to reliably pass data back and forth.

• Note: Browsers and web utilities don't handle spaces very well, so at the very least we should use some kind of URL encoding. Other socket clients don't have this problem -- we can send and receive binary data without worrying about data corruption or transformation.

Operation

u2pipe is designed to take a request from stdin, process it, and send the response to stdout.

REQUEST format:

123456command

where

• 123456 is a six-digit request length. For example, the command DATE contains four characters; the request length is 000004.

• command is the TCL command to run. Ideally, the command should be encoded or encrypted so you can pass arguments for processing.

• Example: 000004DATE

u2pipe reads in the request, strips off the six-digit request length, and prepends the optional RUN parameter (from the ini file above).

u2pipe uses InterCall functions ic_opensession(), ic_execute(), and ic_quit() to process the command.

• @TTY is set to "uvcs" (for Universe; probably "udcs" on Unidata).
• The LOGIN paragraph is not run.
• Only one command is allowed (no LF or @FM separating commands).
• All characters are executed -- backspaces and such are not interpreted.
• Program output to the screen (via DISPLAY or CRT) is captured for the response.
• WARNING: CHAR(3) (CTRL-C) output breaks your Universe BASIC program, possibly truncating your output. Binary output should be encoded.

u2pipe returns the length of the output and the output itself in the format of a response.

RESPONSE format:

123456output

where

• 123456 is a six-digit response length.

• output is the output from your TCL command, including both printing and non-printing characters. Ideally, your output should be encoded or encrypted to protect the data from transformations and prying eyes.

• Example: 000037Friday, September 29, 2006 03:18pm

Implementation

So you got the "DATE" command to work – Yahoo! break out the champagne (or maybe some of that "free beer")! No, wait a minute – I suppose you want to run something a little more complicated, huh?

• Maybe you want to run a subroutine, or a program that takes command line arguments.
• Maybe you want to run a TCL command that returns some XML.
• And some of you more security-minded folks out there want to know how encryption might fit into this.

I'm going to outline a simple implementation. You'll need to add your own features, like error return handling, beefier encryption, etc. The purpose of u2pipe is to provide a connection between a webserver and Universe; how to use that connection is up to you.

1. We'll devise our own data format for exchanging information between the web server and Universe.

2. We'll write a function on the frontend.

• to format our request
• send it to the backend
• wait for the backend's response
• parse the response.

3. We'll write a Universe BASIC program on the backend

• to parse the request
• filter/validate the command
• run the command
• format the response
• return the response

4. We'll set the RUN parameter in u2pipe.ini to run our backend program.

5. We'll test it.

1. Devising a data exchange format

This is a simple implementation, so I've decided that the webserver should be able to send TCL command strings with arguments and get the output in return. With that in mind, I've chosen to simply encode my command string and the output returned using base64. I've also decided to wrap the encoded output in parentheses "()" to help me find it in the response.

If I wanted to get fancy, I could encapsulate my request in XML (maybe like SOAP) and then encode, encrypt, or otherwise mangle the request before sending it to the backend. Conversely, I could also require the response to be XML formatted and encoded. Feel free to make it as simple or complicated as you wish.

2. Writing a frontend function.

ASP, ASP.NET, JSP, PHP, ColdFusion... it does not matter what frontend you use. As long as your frontend can either (a) communicate on a socket, or (b) retrieve the contents of a web page, you can write a function for interacting with u2pipe.

2a. Writing a frontend function: ColdFusion

I've already written a function for you. Change it as you need to implement your data exchange format.

<cffunction name="decryptString" output="false" returntype="any">
<cfargument name="text" type="string" required="true">
   <!--- *TODO* substitute your own encryption --->
   <cfreturn toString(toBinary(text))>
</cffunction>


<cffunction name="encryptString" output="false" returntype="string">
<cfargument name="text" type="any" required="true">
   <!--- *TODO* substitute your own encryption --->
   <cfreturn toBase64(text, "us-ascii")>
</cffunction>


<cffunction name="serviceRequest" output="false" returntype="string">
<cfargument name="myCommand" type="string" required="true">
   <!--- some constants --->
   <cfset var HEADERSZ = 6>
   <cfset var HOSTURL = "http://yourhost:6789">

   <!--- local variables --->
   <cfset var myLength = "">
   <cfset var myRequest = "">
   <cfset var myResponse = "">
   <cfset var myOutput = "">
   <cfset var bptr = 0>
   <cfset var eptr = 0>

   <!--- Encrypt the command --->
   <cfset myCommand = encryptString(myCommand)>

   <!--- Build the request from the command --->
   <cfset myLength = right(repeatString("0", HEADERSZ) & len(myCommand), HEADERSZ)>
   <cfset myRequest = HOSTURL & "/" & myLength & myCommand>
   <cftry>
      <cfhttp url="#myRequest#" timeout="60" />

      <!--- Parse the output from the response --->
      <cfset myResponse = cfhttp.fileContent>
      <cfset myLength = mid(myResponse, 1, HEADERSZ)>
      <cfset myOutput = mid(myResponse, (HEADERSZ + 1), myLength)>
      
      <!--- Parse output markers --->
      <!--- We've decided that output will be formatted: !(base64text) --->
      <cfset bptr = find("(", myOutput)>
      <cfset eptr = find(")", myOutput)>
      <cfif bptr lt eptr>
         <cfset myOutput = mid(myOutput, (bptr+1), (eptr-bptr-1))>
      <cfelse>
         <cfthrow message="missing or invalid markers!">
      </cfif>

      <!--- Decrypt the output --->
      <cfset myOutput = decryptString(myOutput)>

      <cfcatch type="any">
         <!--- *TODO* you'll need some exception handling --->
      </cfcatch>
   </cftry>
   <cfreturn myOutput>
</cffunction>


<!--- now get some data from Universe --->
<cfset command = "LIST VOC SAMPLE 3">
<cfset output = serviceRequest(command)>
<cfoutput>
<pre>
#output#
<nowiki>

</nowiki>

</cfoutput>

2b. Writing a frontend function: PHP

I've written a function for you PHP'ers too. Again, change it as you need to implement your data exchange format.

<?php

   function decryptString($text) {
      // *TODO* substitute your own encryption
      return base64_decode($text);
   }


   function encryptString($text) {
      // *TODO* substitute your own encryption
      return base64_encode($text);
   }



   function serviceRequest($myCommand) {

      // set constants
      define("HEADERSZ", 6);
      define("HOSTIP", "yourhost");
      define("HOSTPORT", 6789);

      if (($socket = socket_create([[AF_INET]], [[SOCK_STREAM]], [[SOL_TCP]])) < 0 ) {
         // *TODO* need to handle error"
         die("Could not create socket!\n");
         exit(1);
      }

      socket_set_option($socket, [[SOL_SOCKET]], [[SO_RCVTIMEO]], array('sec'=>20, 'usec'=>0));

      if (!socket_connect($socket, HOSTIP, HOSTPORT)) {
         // *TODO* need to handle error"
         die("Unable to connect!\n");
         exit(1);
      }

      $myCommand = encryptString($myCommand);

      $myLength = strlen($myCommand);
      $myRequest = sprintf("%0".HEADERSZ."d%s\n", $myLength, $myCommand);
      socket_write($socket, $myRequest);

      $myOutput = "";
      $myLength = socket_read($socket, HEADERSZ);
      while (FALSE != ($line = @socket_read($socket, $myLength, [[PHP_BINARY_READ]]))) {
         $myOutput = $myOutput . $line;
      }

      $bptr = strpos($myOutput, "(");
      $eptr = strpos($myOutput, ")");

      //assert($bptr !=0);
      //assert($eptr !=0);
      //assert($bptr < $eptr);

      $myOutput = substr($myOutput, ($bptr+1), ($eptr-$bptr-1));


      $myOutput = decryptString($myOutput);
      return ($myOutput);
   }


   $command = "LIST VOC SAMPLE 3";
   $output = serviceRequest($command);


   //header("Content-type: text/plain");
   ech<nowiki>o "<pre>".$output."

";</nowiki>

?>

3. Writing a backend Universe BASIC program

As I decided above, this backend program only needs to be able to base64 decode the command string and its arguments, and execute it. The output from the command is captured, cleaned up, and encoded to be returned to the frontend.

Note that all we have to do here is DISPLAY the response; u2pipe does the communication logic for us.

Let's name this program U2PIPERUN. Make sure you compile and catalog it in the appropriate account.

0001:       DISPLAY "!":
0002:
0003:
0004:       DEFFUN DECODE64(TEXT)
0005:       DEFFUN ENCODE64(TEXT)
0006:       EQU LF TO CHAR(10)
0007:       EQU FF TO CHAR(12)
0008:       EQU CR TO CHAR(13)
0009:
0010:
0011:       COMMAND = FIELD(@SENTENCE, " ", 2, LEN(@SENTENCE))
0012:       COMMAND = DECODE64(COMMAND)
0013:
0014:
0015:       EXECUTE COMMAND CAPTURING OUTPUT RETURNING RETCODE
0016:
0017:
0018:       OUTPUT = CONVERT(@FM, LF, OUTPUT)
0019:       OUTPUT = CONVERT(CR, "", OUTPUT)
0020:       OUTPUT = TRIM(OUTPUT, FF, "B")
0021:       OUTPUT = TRIM(OUTPUT, LF, "B")
0022:       OUTPUT = ENCODE64(OUTPUT)
0023:
0024:
0025:       DISPLAY "(":OUTPUT:")"
0026:    END
Bottom at line 26.

4. Setting the RUN parameter in u2pipe.ini

Your u2pipe.ini file should look something like this:

[U2PIPE]
HOSTNAME=yourhost
USER=webuser
PASSWORD=webuser_password
ACCOUNT=WEBACCOUNT
RUN=U2PIPERUN

5. Testing your implementation

You should be ready to test your implementation. You may want to turn on the LOGFILE parameter in the u2pipe.ini file to see the data being passed back and forth. You may also want to add some kind of error logging to your U2PIPERUN program.

Other uses

• u2pipe could be used in windows BAT files (e.g. for scheduler routines). You could use bin\uvsh too, but uvsh cannot initiate jobs on other machines.

Security

Obviously, you don't want just anyone sending TCL commands to your Universe database.

• Configure your firewall to only allow specific machines to access yourhost:6789
• Set ALLOWIP in u2pipe.ini to only allow specific client ip-addresses.
• Encrypt your command in the request, and the output returned in the response. Universe has built-in encryption functions that you can use. You'll need some kind of encoding/encryption for all but the simplest commands anyway, so build it into your design/code.
• Create a unique user for logging in via u2pipe to your account, and restrict permissions.
• Lock down the Universe account accessed via u2pipe by:
  • removing all verbs
  • removing all paragraphs (procs, etc.)
  • removing all file pointers (use OPENPATH)
  • create an execution filter (set RUN in u2pipe.ini)
  • only run subroutines from your execution filter

ALSO, wininetd can be used for evil; googling on it reveals that it can be used as part of a Trojan-virus attack. Some anti-virus software may report it as such. Just like telnet and ftp, wininetd is only as safe as your firewall setup!

Design Decisions

Why write a socket-based solution instead of just using UniObjects? In my case, I needed to implement database connectivity for a remotely hosted web site; I couldn't "wire" UniObjects directly into the web server. And I really didn't want to.

Here are some decisions I made when writing this utility.

• I decided on a simple request/response model. I only wanted to create a pipe (stdin -> processing -> stdout). This is a requirement for wininetd.
• By using wininetd, I didn't have to worry about writing socket listener code, multi-threading, etc.
• Initially, I did not want a protocol (request and response length) but it made the memory allocation easier. It also allowed the client to easily check for the end of the response.
• I felt that encryption/decryption would have made the utility more complicated to write and would limit its flexibility. Most people have their own favorite crypto utilities anyway.
• I wanted the simplicity of Universe's InterCall API library. Most of the behavior is documented in the InterCall guide.

In designing your own web or socket solution, I think it is important to keep all your business logic on the database server. In some examples of UniObjects programming, I see business logic migrated or duplicated on the client - bad, bad, bad.

Whether you use UniObjects or u2pipe (or SQL for that matter), design your user interface to be database independent. Instead of embedding data access and updates directly in your web page (or GUI screen), put your database access logic in a separate function library. Then call those functions from your web page. This way, you'll be able to easily write test scaffolding to test your web page, then swap in the "real" functions that actually hit your database.

Possible Enhancements

• You could use this as a base for developing an actual web service.
• Rewrite to use UniObjects instead of InterCall.

Downloading wininetd source code

You can download source code from http://xmailserver.org/wininetd.html

u2pipe.c source code

• You must have UniDK installed on your development machine.
• Compiling notes
  • See the InterCall guide for the DLL's you'll need.
  • VC++ calling conventions need to be changed from __cdecl to __stdcall for the InterCall routines to link properly.
  • Additional include directories: c:\ibm\unidk\include
  • Additional lib directories: c:\ibm\unidk\lib
  • Additional libraries/modules: uvic32.lib
• In the process of porting the code to run in unix, I replaced the Windows ini functions with the iniparser.h and iniparser.c source code libraries.

/***
* u2pipe.c
* universe request pipe utility
* CVS $Revision: 1.3 $ $Date: 2007/05/11 19:33:27 $
*
* Copyright (C) 2006 Rex Gozar
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
* http://www.gnu.org/licenses/lgpl.html
*
* Rex Gozar
* [email protected]
***/

#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <intcall.h>
#include "iniparser.h"

#define MAXBUF (256)
#define MAXBIGBUF (8192)

#ifndef TRUE
   #define TRUE (1)
#endif
#ifndef FALSE
   #define FALSE (0)
#endif

#define [[APP_NAME]] ("[[U2PIPE]]")
#define [[DEFAULT_VALUE]] ("")
#define REQUIRED (TRUE)
#define [[NOT_REQUIRED]] (FALSE)



/***
* NOTE - GETSLASHSZ must be smaller than REQUESTLENSZ
* and (GETSLASHSZ + REQUESTLENSZ + 1) must be smaller than MAXBUF!
***/
#define GETSLASH ("GET /")
#define GETSLASHSZ (5)
#define REQUESTLENSZ (6)



/***
* helper functions
***/

char * get[[TimeStamp]](char *timestamp) {
   time_t seconds = time(NULL);
   struct tm *now = localtime(&seconds);
   strftime(timestamp, MAXBUF, "%Y%m%d %H%M%S", now);
   return timestamp;
}



/***
* main
*
* Note that the program should be compiled with __stdcall
* call convention so the intercall routines will properly link.
* Use __cdecl here to avoid a compiler warning.
***/

#ifndef WIN32
   #define __cdecl
#endif

int __cdecl main(int argc, char *argv[]) {

   time_t stime = time(NULL);
   time_t etime = 0;

   #ifdef WIN32
      const char *WINDIR = getenv("WINDIR");
   #else
      const char *ETCDIR = "/etc";
   #endif
   const char *[[U2PIPE_INI]] = "u2pipe.ini";

   long retcode = 0;
   char buffer[MAXBUF] = "";
   char timestamp[MAXBUF] = "";

   const char *[[CLIENT_IP]] = getenv("[[CLIENT_IP]]");
   const char *[[CLIENT_PORT]] = getenv("[[CLIENT_PORT]]");

   char ipAddress[MAXBUF] = "";
   char ipPort[MAXBUF] = "";
   char iniFile[MAXBUF] = "";

   dictionary *ini = NULL;
   char *iniHostname = NULL;
   char *iniUser = NULL;
   char *iniPassword = NULL;
   char *iniAccount = NULL;
   char *iniRun = NULL;
   char *iniLogfile = NULL;
   char *ini[[AllowIP]] = NULL;

   int logsw = FALSE;
   FILE *fpLog;

   int j = 0;

   char *request = NULL;
   long requestLen = 0;

   char *response = NULL;
   long responseLen = 0;
   long responseMax = 0;

   long sessionId = 0;
   long code = 0;
   long retcode2 = 0;

   char *command = NULL;
   long commandLen = 0;


   /* check for NULL pointers and initialize ip address and port strings */

   if (NULL != [[CLIENT_IP]]) {
      strncpy(ipAddress, [[CLIENT_IP]], strlen([[CLIENT_IP]]));
   }
   if (NULL != [[CLIENT_PORT]]) {
      strncpy(ipPort, [[CLIENT_PORT]], strlen([[CLIENT_PORT]]));
   }


   /* see if the ini file path is on the command line */   

   if (2 <= argc) {
      if (MAXBUF < ( 1 + strlen(argv[1]) )) {
         fprintf(stderr, "ini file pathname is longer than %i!\n", MAXBUF);
         exit(1);
      }
      sprintf(iniFile, "%s", argv[1]);
   } else {

      #ifdef WIN32

         if (NULL == WINDIR) {
            fprintf(stderr, "WINDIR is NULL!\n");
            exit(1);
         }

         /* build the default ini file path %WINDIR%\u2pipe.ini */
         if (MAXBUF < ( 1 + strlen(WINDIR) + 1 + strlen([[U2PIPE_INI]]) )) {
            fprintf(stderr, "%%WINDIR%%\\%s is longer than %i!\n", [[U2PIPE_INI]], MAXBUF);
            exit(1);
         }
         sprintf(iniFile, "%s\\%s", WINDIR, [[U2PIPE_INI]]);

      #else

         if (MAXBUF < ( 1 + strlen(ETCDIR) + 1 + strlen([[U2PIPE_INI]]) )) {
            fprintf(stderr, "%s/%s is longer than %i!\n", ETCDIR, [[U2PIPE_INI]], MAXBUF);
            exit(1);
         }
         sprintf(iniFile, "%s/%s", ETCDIR, [[U2PIPE_INI]]);

      #endif
   }

   /* parse config for user, password, account, etc */
   ini = iniparser_new(iniFile);
   iniHostname = iniparser_getstring(ini, "u2pipe:hostname", "");
   iniUser = iniparser_getstring(ini, "u2pipe:user", "");
   iniPassword = iniparser_getstring(ini, "u2pipe:password", "");
   iniAccount = iniparser_getstring(ini, "u2pipe:account", "");
   iniRun = iniparser_getstring(ini, "u2pipe:run", "");
   iniLogfile = iniparser_getstring(ini, "u2pipe:logfile", "");
   ini[[AllowIP]] = iniparser_getstring(ini, "u2pipe:allowIP", "");

   #ifdef _DEBUG
      fprintf(stderr, "iniFile=%s\n", iniFile);
      fprintf(stderr, "ipAddress=%s\n", ipAddress);
      fprintf(stderr, "ipPort=%s\n", ipPort);
      fprintf(stderr, "iniFile=%s\n", iniFile);
      fprintf(stderr, "iniHostname=%s\n", iniHostname);
      fprintf(stderr, "iniUser=%s\n", iniUser);
      fprintf(stderr, "iniPassword=%s\n", iniPassword);
      fprintf(stderr, "iniAccount=%s\n", iniAccount);
      fprintf(stderr, "iniRun=%s\n", iniRun);
      fprintf(stderr, "iniLogfile=%s\n", iniLogfile);
      fprintf(stderr, "ini[[AllowIP]]=%s\n", ini[[AllowIP]]);
   #endif

   
   /* open the log file */

   logsw = (0 != strlen(iniLogfile));
   if (logsw) {
      fpLog = fopen(iniLogfile, "a+");
      if (NULL == fpLog) {
         fprintf(stderr, "unable to open log file %s\n", iniLogfile);
         exit(1);
      }
      /* print a separator line for each request */
      fprintf(fpLog, "****************************************\n");
   }


   /* log the requester */

   if (logsw) {
      fprintf(fpLog, "%s ipAddress=%s\n", get[[TimeStamp]](timestamp), ipAddress);
      fprintf(fpLog, "%s ipPort=%s\n", get[[TimeStamp]](timestamp), ipPort);
   }
   

   /* check allowed ip-addresses */

   if (0 != strlen(ini[[AllowIP]])) {

      /* sandwich the list between semi-colons */
      if (MAXBUF < 3 + strlen(ini[[AllowIP]])) {
         if (logsw) {
            fprintf(fpLog, "%s ALLOWIP list is too long!\n", get[[TimeStamp]](timestamp));
         }
         exit(1);
      }
      sprintf(buffer, ";%s;", ini[[AllowIP]]);
      memset(ini[[AllowIP]], 0, MAXBUF);
      strncpy(ini[[AllowIP]], buffer, strlen(buffer));

      /* do the same to the ip address */
      if (MAXBUF < 3 + strlen(ipAddress)) {
         if (logsw) {
            fprintf(fpLog, "%s ipAddress is too long!\n", get[[TimeStamp]](timestamp));
         }
         exit(1);
      }
      sprintf(buffer, ";%s;", ipAddress);

      /* now look for the ipAddress within the allowed IP's */
      if (NULL == strstr(ini[[AllowIP]], buffer)) {
         if (logsw) {
            fprintf(fpLog, "%s ipAddress is not allowed!\n", get[[TimeStamp]](timestamp));
         }
         exit(1);
      }
   }


   /****
   *
   * requests are piped in via stdin, and must be formatted:
   *
   *    GET /000009something
   * or
   *    000009something
   *
   * where:
   *    "GET /" is an optional literal string (typically passed by a web request.)
   *    "000009" is a six-digit request length.
   *    "something" is the actual request.
   ***/


   /* look for "GET /" */

   memset(buffer, 0, MAXBUF);
   fread(buffer, sizeof(char), GETSLASHSZ, stdin);
   if (0 == strncmp(buffer, GETSLASH, GETSLASHSZ)) {
      /* discard the buffer contents and get REQUESTLENSZ more characters */
      memset(buffer, 0, MAXBUF);
      fread(buffer, sizeof(char), REQUESTLENSZ, stdin);
   } else {
      /* read in more characters to fill in the request length */
      j = GETSLASHSZ;
      while (j < REQUESTLENSZ) {
         int c = fgetc(stdin);
         if (EOF == c) {
            if (logsw) {
               fprintf(fpLog, "%s not enough characters in stdin!\n", get[[TimeStamp]](timestamp));
            }
            exit(1);
         }
         buffer[j] = c;
         j++;
      }
   }


   /* check the request length string */

   if (REQUESTLENSZ != strlen(buffer)) {
      if (logsw) {
         fprintf(fpLog, "%s request length is not %i characters!\n", get[[TimeStamp]](timestamp), REQUESTLENSZ);
      }
      exit(1);
   }

   for (j = 0; j < REQUESTLENSZ; j++) {
      if (!isdigit(buffer[j])) {
         if (logsw) {
            fprintf(fpLog, "%s request length is non-numeric!\n", get[[TimeStamp]](timestamp));
         }
         exit(1);
      }
   }

   requestLen = atoi(buffer);
   if (0 == requestLen) {
      if (logsw) {
         fprintf(fpLog, "%s request length is zero!\n", get[[TimeStamp]](timestamp));
      }
      exit(1);
   }


   /* get the memory to read in the rest of the request */

   request = (char *) malloc(1 + requestLen);
   if (NULL == request) {
      if (logsw) {
         fprintf(fpLog, "%s request malloc failed!\n", get[[TimeStamp]](timestamp));
      }
      exit(1);
   }


   /* read in the request */

   memset(request, 0, 1+requestLen);
   fread(request, sizeof(char), requestLen, stdin);


   /* log the request */

   if (logsw) {
      fprintf(fpLog, "%s REQUEST=%s\n", get[[TimeStamp]](timestamp), request);
   }


   /*********************
   * process the request
   **********************/

   sessionId = ic_opensession(iniHostname, iniUser, iniPassword, iniAccount, &code, NULL);
   if (0 != code) {
      if (logsw) {
         fprintf(fpLog, "%s ic_opensession failed! %i\n", get[[TimeStamp]](timestamp), code);
      }
      exit(1);
   }

   if (0 != strlen(iniRun)) {
      /* add one to the command length for the space between */
      commandLen = strlen(iniRun) + 1 + requestLen;
      command = (char *) malloc(1+commandLen);
      memset(command, 0, commandLen);
      sprintf(command, "%s %s", iniRun, request);
   } else {
      commandLen = requestLen;
      command = (char *) malloc(1+commandLen);
      memset(command, 0, commandLen);
      sprintf(command, "%s", request);
   }

   responseMax = MAXBIGBUF;
   response = (char *) malloc(responseMax);
   memset(response, 0, responseMax);


   ic_execute(command, &commandLen, response, &responseMax,
      &responseLen, &retcode, &retcode2, &code);

   if (logsw) {
      fprintf(fpLog, "%s [[IC_EXECUTE]]=%i\n", get[[TimeStamp]](timestamp), code);
      fprintf(fpLog, "%s responseLen=%i\n", get[[TimeStamp]](timestamp), responseLen);
   }

   
   /* see if we need to expand the response buffer to hold all the data */
   
   while ([[IE_BTS]] == code) {

      char *tmpbuf = NULL;
      long tmpmax = 0;
      long tmplen = 0;

      char *newbuf = NULL;
      long newmax = 0;

      /* create a temporary buffer to hold the next block of data */
      tmpmax = responseMax;
      tmpbuf = (char *) malloc(tmpmax);
      if (NULL == tmpbuf) {
         fprintf(stderr, "malloc failed when creating tmpbuf!\n");
      }
      memset(tmpbuf, 0, tmpmax);
      tmplen = 0;

      /* get the next block of data */
      ic_executecontinue(tmpbuf, &tmpmax, &tmplen, &retcode, &retcode2, &code);
      if (logsw) {
         fprintf(fpLog, "%s [[IC_EXECUTECONTINUE]]=%i\n", get[[TimeStamp]](timestamp), code);
      }

      /* create a new buffer to hold both the response and tmp buffers */
      /* note that the new buffer is double the size of the previous iteration */
      newmax = responseMax + tmpmax;
      newbuf = (char *) malloc(newmax);
      if (NULL == newbuf) {
         fprintf(stderr, "malloc failed when creating newbuf!\n");
         exit(1);
      }
      memset(newbuf, 0, newmax);

      /* copy both the response and tmpbuf into the new buffer */
      memcpy(newbuf, response, responseLen);
      memcpy(newbuf + responseLen, tmpbuf, tmpmax);

      /* release the memory for old data buffers and point response to the new data */
      free(tmpbuf);
      free(response);
      response = newbuf;
      responseMax = newmax;
      responseLen = responseLen + tmplen;
   }


   ic_quit(&code);


   /* log the response */
   if (logsw) {
      fprintf(fpLog, "%s RESPONSE=%s\n", get[[TimeStamp]](timestamp), response);
      fprintf(fpLog, "%s responseLen=%i\n", get[[TimeStamp]](timestamp), responseLen);
      fprintf(fpLog, "%s strlen(response)=%i\n", get[[TimeStamp]](timestamp), strlen(response));
   }

   /* return the response on stdout */

   fprintf(stdout, "%06i%s\n", responseLen, response);
   fflush(stdout);


   /* done */
   if (logsw) {
      etime = time(NULL) - stime;
      fprintf(fpLog, "%s COMPLETED (%lu)\n", get[[TimeStamp]](timestamp), etime);
      fflush(fpLog);
      fclose(fpLog);
   }
   exit(0);
}

/*
 Based upon libiniparser, by Nicolas Devillard
 Hacked into 1 file (m-iniparser) by Freek/2005
 Original terms following:

 -- -

 Copyright (c) 2000 by Nicolas Devillard (ndevilla AT free DOT fr).

 Written by Nicolas Devillard. Not derived from licensed software.

 Permission is granted to anyone to use this software for any
 purpose on any computer system, and to redistribute it freely,
 subject to the following restrictions:

 1. The author is not responsible for the consequences of use of
 this software, no matter how awful, even if they arise
 from defects in it.

 2. The origin of this software must not be misrepresented, either
 by explicit claim or by omission.

 3. Altered versions must be plainly marked as such, and must not
 be misrepresented as being the original software.

 4. This notice may not be removed or altered.

 */


#ifndef _[[INIPARSER_H_]]
#define _[[INIPARSER_H_]]
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/* #include <unistd.h> */
#include <ctype.h>

#ifdef __cplusplus
extern "C" {
#endif


typedef struct _dictionary_ {
   /** Number of entries in dictionary */
   int n;
   /** Storage size */
   int size;
   /** List of string values */
   char **val;
   /** List of string keys */
   char **key ;
   /** List of hash values for keys */
   unsigned *hash;
} dictionary ;


/* generated by genproto */

dictionary * iniparser_new(char *ininame);
void iniparser_free(dictionary *d);

int iniparser_getnsec(dictionary *d);
char * iniparser_getsecname(dictionary *d, int n);
void iniparser_dump(dictionary *d, FILE *f);
void iniparser_dump_ini(dictionary *d, FILE *f);
char * iniparser_getkey(dictionary *d, char *section, char *key);
char * iniparser_getstr(dictionary *d, char *key);
char * iniparser_getstring(dictionary *d, char *key, char *def);
int iniparser_getint(dictionary *d, char *key, int notfound);
double iniparser_getdouble(dictionary *d, char *key, double notfound);
int iniparser_getboolean(dictionary *d, char *key, int notfound);
int iniparser_find_entry(dictionary  *ini, char *entry);
int iniparser_setstr(dictionary *ini, char *entry, char *val);
void iniparser_unset(dictionary *ini, char *entry);

#ifdef __cplusplus
}
#endif

#endif

/*
 Based upon libiniparser, by Nicolas Devillard
 Hacked into 1 file (m-iniparser) by Freek/2005
 Original terms following:

 -- -

 Copyright (c) 2000 by Nicolas Devillard (ndevilla AT free DOT fr).

 Written by Nicolas Devillard. Not derived from licensed software.

 Permission is granted to anyone to use this software for any
 purpose on any computer system, and to redistribute it freely,
 subject to the following restrictions:

 1. The author is not responsible for the consequences of use of
 this software, no matter how awful, even if they arise
 from defects in it.

 2. The origin of this software must not be misrepresented, either
 by explicit claim or by omission.

 3. Altered versions must be plainly marked as such, and must not
 be misrepresented as being the original software.

 4. This notice may not be removed or altered.

 */


#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/* #include <unistd.h> */

#include "iniparser.h"

#ifdef __cplusplus
extern "C" {
#endif

/* strlib.c following */

#define ASCIILINESZ 1024
/*-------------------------------------------------------------------------*/
/**
  @brief    Convert a string to lowercase.
  @param    s   String to convert.
  @return   ptr to statically allocated string.

  This function returns a pointer to a statically allocated string
  containing a lowercased version of the input string. Do not free
  or modify the returned string! Since the returned string is statically
  allocated, it will be modified at each function call (not re-entrant).
 */
/*--------------------------------------------------------------------------*/

static char * strlwc(char *s)
{
    static char l[ASCIILINESZ+1];
    int i ;

    if (s==NULL) return NULL ;
    memset(l, 0, ASCIILINESZ+1);
    i=0 ;
    while (s[i] && i<ASCIILINESZ) {
        l[i] = (char)tolower((int)s[i]);
        i++ ;
    }
    l[ASCIILINESZ]=(char)0;
    return l ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Convert a string to uppercase.
  @param    s   String to convert.
  @return   ptr to statically allocated string.

  This function returns a pointer to a statically allocated string
  containing an uppercased version of the input string. Do not free
  or modify the returned string! Since the returned string is statically
  allocated, it will be modified at each function call (not re-entrant).
 */
/*--------------------------------------------------------------------------*/

static char * strupc(char *s)
{
    static char l[ASCIILINESZ+1];
    int i ;

    if (s==NULL) return NULL ;
    memset(l, 0, ASCIILINESZ+1);
    i=0 ;
    while (s[i] && i<ASCIILINESZ) {
        l[i] = (char)toupper((int)s[i]);
        i++ ;
    }
    l[ASCIILINESZ]=(char)0;
    return l ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Skip blanks until the first non-blank character.
  @param    s   String to parse.
  @return   Pointer to char inside given string.

  This function returns a pointer to the first non-blank character in the
  given string.
 */
/*--------------------------------------------------------------------------*/

static char * strskp(char *s)
{
    char * skip = s;
    if (s==NULL) return NULL ;
    while (isspace((int)*skip) && *skip) skip++;
    return skip ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Remove blanks at the end of a string.
  @param    s   String to parse.
  @return   ptr to statically allocated string.

  This function returns a pointer to a statically allocated string,
  which is identical to the input string, except that all blank
  characters at the end of the string have been removed.
  Do not free or modify the returned string! Since the returned string
  is statically allocated, it will be modified at each function call
  (not re-entrant).
 */
/*--------------------------------------------------------------------------*/

static char * strcrop(char *s)
{
    static char l[ASCIILINESZ+1];
    char * last ;

    if (s==NULL) return NULL ;
    memset(l, 0, ASCIILINESZ+1);
    strcpy(l, s);
    last = l + strlen(l);
    while (last > l) {
        if (!isspace((int)*(last-1)))
            break ;
        last -- ;
    }
    *last = (char)0;
    return l ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Remove blanks at the beginning and the end of a string.
  @param    s   String to parse.
  @return   ptr to statically allocated string.

  This function returns a pointer to a statically allocated string,
  which is identical to the input string, except that all blank
  characters at the end and the beg. of the string have been removed.
  Do not free or modify the returned string! Since the returned string
  is statically allocated, it will be modified at each function call
  (not re-entrant).
 */
/*--------------------------------------------------------------------------*/
static char * strstrip(char *s)
{
    static char l[ASCIILINESZ+1];
    char * last ;

    if (s==NULL) return NULL ;

    while (isspace((int)*s) && *s) s++;

    memset(l, 0, ASCIILINESZ+1);
    strcpy(l, s);
    last = l + strlen(l);
    while (last > l) {
        if (!isspace((int)*(last-1)))
            break ;
        last -- ;
    }
    *last = (char)0;

    return (char*)l ;
}


/* dictionary.c.c following */
/** Maximum value size for integers and doubles. */
#define MAXVALSZ    1024

/** Minimal allocated number of entries in a dictionary */
#define DICTMINSZ   128

/** Invalid key token */
#define [[DICT_INVALID_KEY]]    ((char*)-1)

/*
 Doubles the allocated size associated to a pointer
 'size' is the current allocated size.
 */
static void * mem_double(void *ptr, int size)
{
    void *newptr;

    newptr = calloc(2*size, 1);
    memcpy(newptr, ptr, size);
    free(ptr);
    return newptr ;
}


/*---------------------------------------------------------------------------
                            Function codes
 ---------------------------------------------------------------------------*/

/*-------------------------------------------------------------------------*/
/**
  @brief    Compute the hash key for a string.
  @param    key     Character string to use for key.
  @return   1 unsigned int on at least 32 bits.

  This hash function has been taken from an Article in Dr Dobbs Journal.
  This is normally a collision-free function, distributing keys evenly.
  The key is stored anyway in the struct so that collision can be avoided
  by comparing the key itself in last resort.
 */
/*--------------------------------------------------------------------------*/

static unsigned dictionary_hash(char *key)
{
    int         len ;
    unsigned    hash ;
    int         i ;

    len = strlen(key);
    for (hash=0, i=0 ; i<len ; i++) {
        hash += (unsigned)key[i] ;
        hash += (hash<<10);
        hash ^= (hash>>6) ;
    }
    hash += (hash <<3);
    hash ^= (hash >>11);
    hash += (hash <<15);
    return hash ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Create a new dictionary object.
  @param    size    Optional initial size of the dictionary.
  @return   1 newly allocated dictionary objet.

  This function allocates a new dictionary object of given size and returns
  it. If you do not know in advance (roughly) the number of entries in the
  dictionary, give size=0.
 */
/*--------------------------------------------------------------------------*/

static dictionary * dictionary_new(int size)
{
    dictionary *d ;

    /* If no size was specified, allocate space for DICTMINSZ */
    if (size<DICTMINSZ) size=DICTMINSZ ;

    d = (dictionary *)calloc(1, sizeof(dictionary));
    d->size = size ;
    d->val  = (char **)calloc(size, sizeof(char*));
    d->key  = (char **)calloc(size, sizeof(char*));
    d->hash = (unsigned int *)calloc(size, sizeof(unsigned));

    return d;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Delete a dictionary object
  @param    d   dictionary object to deallocate.
  @return   void

  Deallocate a dictionary object and all memory associated to it.
 */
/*--------------------------------------------------------------------------*/

static void dictionary_del(dictionary *d)
{
    int     i ;

    if (d==NULL) return ;
    for (i=0 ; i<d->size ; i++) {
        if (d->key[i]!=NULL)
            free(d->key[i]);
        if (d->val[i]!=NULL)
            free(d->val[i]);
    }
    free(d->val);
    free(d->key);
    free(d->hash);
    free(d);

    return;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Get a value from a dictionary.
  @param    d       dictionary object to search.
  @param    key     Key to look for in the dictionary.
  @param    def     Default value to return if key not found.
  @return   1 pointer to internally allocated character string.

  This function locates a key in a dictionary and returns a pointer to its
  value, or the passed 'def' pointer if no such key can be found in
  dictionary. The returned character pointer points to data internal to the
  dictionary object, you should not try to free it or modify it.
 */
/*--------------------------------------------------------------------------*/
static char * dictionary_get(dictionary *d, char *key, char *def)
{
    unsigned    hash ;
    int         i ;

    hash = dictionary_hash(key);
    for (i=0 ; i<d->size ; i++) {
        if (d->key==NULL)
            continue ;
        /* Compare hash */
        if (hash==d->hash[i]) {
            /* Compare string, to avoid hash collisions */
            if (!strcmp(key, d->key[i])) {
                return d->val[i] ;
            }
        }
    }
    return def ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Set a value in a dictionary.
  @param    d       dictionary object to modify.
  @param    key     Key to modify or add.
  @param    val     Value to add.
  @return   void

  If the given key is found in the dictionary, the associated value is
  replaced by the provided one. If the key cannot be found in the
  dictionary, it is added to it.

  It is Ok to provide a NULL value for val, but NULL values for the dictionary
  or the key are considered as errors: the function will return immediately
  in such a case.

  Notice that if you dictionary_set a variable to NULL, a call to
  dictionary_get will return a NULL value: the variable will be found, and
  its value (NULL) is returned. In other words, setting the variable
  content to NULL is equivalent to deleting the variable from the
  dictionary. It is not possible (in this implementation) to have a key in
  the dictionary without value.
 */
/*--------------------------------------------------------------------------*/

static void dictionary_set(dictionary *d, char *key, char *val)
{
    int         i ;
    unsigned    hash ;

    if (d==NULL || key==NULL) return ;

    /* Compute hash for this key */
    hash = dictionary_hash(key) ;
    /* Find if value is already in blackboard */
    if (d->n>0) {
        for (i=0 ; i<d->size ; i++) {
            if (d->key[i]==NULL)
                continue ;
            if (hash==d->hash[i]) { /* Same hash value */
                if (!strcmp(key, d->key[i])) {   /* Same key */
                    /* Found a value: modify and return */
                    if (d->val[i]!=NULL)
                        free(d->val[i]);
                    d->val[i] = val ? strdup(val) : NULL ;
                    /* Value has been modified: return */
                    return ;
                }
            }
        }
    }
    /* Add a new value */
    /* See if dictionary needs to grow */
    if (d->n==d->size) {

        /* Reached maximum size: reallocate blackboard */
        d->val  = (char **)mem_double(d->val,  d->size * sizeof(char*)) ;
        d->key  = (char **)mem_double(d->key,  d->size * sizeof(char*)) ;
        d->hash = (unsigned int *)mem_double(d->hash, d->size * sizeof(unsigned)) ;

        /* Double size */
        d->size *= 2 ;
    }

    /* Insert key in the first empty slot */
    for (i=0 ; i<d->size ; i++) {
        if (d->key[i]==NULL) {
            /* Add key here */
            break ;
        }
    }
    /* Copy key */
    d->key[i]  = strdup(key);
    d->val[i]  = val ? strdup(val) : NULL ;
    d->hash[i] = hash;
    d->n ++ ;
    return ;
}

/*-------------------------------------------------------------------------*/
/**
  @brief    Delete a key in a dictionary
  @param    d       dictionary object to modify.
  @param    key     Key to remove.
  @return   void

  This function deletes a key in a dictionary. Nothing is done if the
  key cannot be found.
 */
/*--------------------------------------------------------------------------*/
static void dictionary_unset(dictionary *d, char *key)
{
    unsigned    hash ;
    int         i ;

    hash = dictionary_hash(key);
    for (i=0 ; i<d->size ; i++) {
        if (d->key[i]==NULL)
            continue ;
        /* Compare hash */
        if (hash==d->hash[i]) {
            /* Compare string, to avoid hash collisions */
            if (!strcmp(key, d->key[i])) {
                /* Found key */
                break ;
            }
        }
    }
    if (i>=d->size)
        /* Key not found */
        return ;

    free(d->key[i]);
    d->key[i] = NULL ;
    if (d->val[i]!=NULL) {
        free(d->val[i]);
        d->val[i] = NULL ;
    }
    d->hash[i] = 0 ;
    d->n -- ;
    return ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Dump a dictionary to an opened file pointer.
  @param    d   Dictionary to dump
  @param    f   Opened file pointer.
  @return   void

  Dumps a dictionary onto an opened file pointer. Key pairs are printed out
  as @c [Key]=[Value], one per line. It is Ok to provide stdout or stderr as
  output file pointers.
 */
/*--------------------------------------------------------------------------*/

static void dictionary_dump(dictionary *d, FILE *f)
{
    int i;

    if (d==NULL || f==NULL) return;

    for (i=0; i<d->size; i++) {
        if (d->key[i] == NULL)
            continue ;
        if (d->val[i] != NULL) {
            fprintf(f, "[%s]=[%s]\n", d->key[i], d->val[i]);
        } else {
            fprintf(f, "[%s]=UNDEF\n", d->key[i]);
        }
    }

    return;
}


/* iniparser.c.c following */
#define ASCIILINESZ         1024
#define [[INI_INVALID_KEY]]     ((char*)-1)

/* Private: add an entry to the dictionary */
static void iniparser_add_entry(
    dictionary *d,
    char *sec,
    char *key,
    char *val)
{
    char longkey[2*ASCIILINESZ+1];

    /* Make a key as section:keyword */
    if (key!=NULL) {
        sprintf(longkey, "%s:%s", sec, key);
    } else {
        strcpy(longkey, sec);
    }

    /* Add (key,val) to dictionary */
    dictionary_set(d, longkey, val);
    return ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Get number of sections in a dictionary
  @param    d   Dictionary to examine
  @return   int Number of sections found in dictionary

  This function returns the number of sections found in a dictionary.
  The test to recognize sections is done on the string stored in the
  dictionary: a section name is given as "section" whereas a key is
  stored as "section:key", thus the test looks for entries that do not
  contain a colon.

  This clearly fails in the case a section name contains a colon, but
  this should simply be avoided.

  This function returns -1 in case of error.
 */
/*--------------------------------------------------------------------------*/

int iniparser_getnsec(dictionary *d)
{
    int i ;
    int nsec ;

    if (d==NULL) return -1 ;
    nsec=0 ;
    for (i=0 ; i<d->size ; i++) {
        if (d->key[i]==NULL)
            continue ;
        if (strchr(d->key[i], ':')==NULL) {
            nsec ++ ;
        }
    }
    return nsec ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Get name for section n in a dictionary.
  @param    d   Dictionary to examine
  @param    n   Section number (from 0 to nsec-1).
  @return   Pointer to char string

  This function locates the n-th section in a dictionary and returns
  its name as a pointer to a string statically allocated inside the
  dictionary. Do not free or modify the returned string!

  This function returns NULL in case of error.
 */
/*--------------------------------------------------------------------------*/

char * iniparser_getsecname(dictionary *d, int n)
{
    int i ;
    int foundsec ;

    if (d==NULL || n<0) return NULL ;
    foundsec=0 ;
    for (i=0 ; i<d->size ; i++) {
        if (d->key[i]==NULL)
            continue ;
        if (strchr(d->key[i], ':')==NULL) {
            foundsec++ ;
            if (foundsec>n)
                break ;
        }
    }
    if (foundsec<=n) {
        return NULL ;
    }
    return d->key[i] ;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Dump a dictionary to an opened file pointer.
  @param    d   Dictionary to dump.
  @param    f   Opened file pointer to dump to.
  @return   void

  This function prints out the contents of a dictionary, one element by
  line, onto the provided file pointer. It is OK to specify @c stderr
  or @c stdout as output files. This function is meant for debugging
  purposes mostly.
 */
/*--------------------------------------------------------------------------*/
void iniparser_dump(dictionary *d, FILE *f)
{
    dictionary_dump(d,f);
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Save a dictionary to a loadable ini file
  @param    d   Dictionary to dump
  @param    f   Opened file pointer to dump to
  @return   void

  This function dumps a given dictionary into a loadable ini file.
  It is Ok to specify @c stderr or @c stdout as output files.
 */
/*--------------------------------------------------------------------------*/

void iniparser_dump_ini(dictionary *d, FILE *f)
{
    int     i, j ;
    char    keym[ASCIILINESZ+1];
    int     nsec ;
    char *  secname ;
    int     seclen ;

    if (d==NULL || f==NULL) return ;

    nsec = iniparser_getnsec(d);
    if (nsec<1) {
        /* No section in file: dump all keys as they are */
        for (i=0 ; i<d->size ; i++) {
            if (d->key[i]==NULL)
                continue ;
            fprintf(f, "%s = %s\n", d->key[i], d->val[i]);
        }
        return ;
    }
    for (i=0 ; i<nsec ; i++) {
        secname = iniparser_getsecname(d, i) ;
        seclen  = (int)strlen(secname);
        fprintf(f, "\n[%s]\n", secname);
        sprintf(keym, "%s:", secname);
        for (j=0 ; j<d->size ; j++) {
            if (d->key[j]==NULL)
                continue ;
            if (!strncmp(d->key[j], keym, seclen+1)) {
                fprintf(f,
                        "%-30s = %s\n",
                        d->key[j]+seclen+1,
                        d->val[j] ? d->val[j] : "");
            }
        }
    }
    fprintf(f, "\n");
    return ;
}

/*-------------------------------------------------------------------------*/
/**
  @brief    Get the string associated to a key, return NULL if not found
  @param    d   Dictionary to search
  @param    key Key string to look for
  @return   pointer to statically allocated character string, or NULL.

  This function queries a dictionary for a key. A key as read from an
  ini file is given as "section:key". If the key cannot be found,
  NULL is returned.
  The returned char pointer is pointing to a string allocated in
  the dictionary, do not free or modify it.

  This function is only provided for backwards compatibility with
  previous versions of iniparser. It is recommended to use
  iniparser_getstring() instead.
 */
/*--------------------------------------------------------------------------*/
char * iniparser_getstr(dictionary *d, char *key)
{
    return iniparser_getstring(d, key, NULL);
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Get the string associated to a key
  @param    d       Dictionary to search
  @param    key     Key string to look for
  @param    def     Default value to return if key not found.
  @return   pointer to statically allocated character string

  This function queries a dictionary for a key. A key as read from an
  ini file is given as "section:key". If the key cannot be found,
  the pointer passed as 'def' is returned.
  The returned char pointer is pointing to a string allocated in
  the dictionary, do not free or modify it.
 */
/*--------------------------------------------------------------------------*/
char * iniparser_getstring(dictionary *d, char *key, char *def)
{
    char * lc_key ;
    char * sval ;

    if (d==NULL || key==NULL)
        return def ;

    lc_key = strdup(strlwc(key));
    sval = dictionary_get(d, lc_key, def);
    free(lc_key);
    return sval ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Get the string associated to a key, convert to an int
  @param    d Dictionary to search
  @param    key Key string to look for
  @param    notfound Value to return in case of error
  @return   integer

  This function queries a dictionary for a key. A key as read from an
  ini file is given as "section:key". If the key cannot be found,
  the notfound value is returned.
 */
/*--------------------------------------------------------------------------*/
int iniparser_getint(dictionary *d, char *key, int notfound)
{
    char *str ;

    str = iniparser_getstring(d, key, [[INI_INVALID_KEY]]);
    if (str==[[INI_INVALID_KEY]]) return notfound ;
    return atoi(str);
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Get the string associated to a key, convert to a double
  @param    d Dictionary to search
  @param    key Key string to look for
  @param    notfound Value to return in case of error
  @return   double

  This function queries a dictionary for a key. A key as read from an
  ini file is given as "section:key". If the key cannot be found,
  the notfound value is returned.
 */
/*--------------------------------------------------------------------------*/
double iniparser_getdouble(dictionary *d, char *key, double notfound)
{
    char * str ;

    str = iniparser_getstring(d, key, [[INI_INVALID_KEY]]);
    if (str==[[INI_INVALID_KEY]]) return notfound ;
    return atof(str);
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Get the string associated to a key, convert to a boolean
  @param    d Dictionary to search
  @param    key Key string to look for
  @param    notfound Value to return in case of error
  @return   integer

  This function queries a dictionary for a key. A key as read from an
  ini file is given as "section:key". If the key cannot be found,
  the notfound value is returned.

  A true boolean is found if one of the following is matched:

  - A string starting with 'y'
  - A string starting with 'Y'
  - A string starting with 't'
  - A string starting with 'T'
  - A string starting with '1'

  A false boolean is found if one of the following is matched:

  - A string starting with 'n'
  - A string starting with 'N'
  - A string starting with 'f'
  - A string starting with 'F'
  - A string starting with '0'

  The notfound value returned if no boolean is identified, does not
  necessarily have to be 0 or 1.
 */
/*--------------------------------------------------------------------------*/
int iniparser_getboolean(dictionary *d, char *key, int notfound)
{
    char *c ;
    int ret ;

    c = iniparser_getstring(d, key, [[INI_INVALID_KEY]]);
    if (c==[[INI_INVALID_KEY]]) return notfound ;
    if (c[0]=='y' || c[0]=='Y' || c[0]=='1' || c[0]=='t' || c[0]=='T') {
        ret = 1 ;
    } else if (c[0]=='n' || c[0]=='N' || c[0]=='0' || c[0]=='f' || c[0]=='F') {
        ret = 0 ;
    } else {
        ret = notfound ;
    }
    return ret;
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Finds out if a given entry exists in a dictionary
  @param    ini     Dictionary to search
  @param    entry   Name of the entry to look for
  @return   integer 1 if entry exists, 0 otherwise

  Finds out if a given entry exists in the dictionary. Since sections
  are stored as keys with NULL associated values, this is the only way
  of querying for the presence of sections in a dictionary.
 */
/*--------------------------------------------------------------------------*/

int iniparser_find_entry(
    dictionary  *ini,
    char        *entry
)
{
    int found=0 ;
    if (iniparser_getstring(ini, entry, [[INI_INVALID_KEY]])!=[[INI_INVALID_KEY]]) {
        found = 1 ;
    }
    return found ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Set an entry in a dictionary.
  @param    ini     Dictionary to modify.
  @param    entry   Entry to modify (entry name)
  @param    val     New value to associate to the entry.
  @return   int 0 if Ok, -1 otherwise.

  If the given entry can be found in the dictionary, it is modified to
  contain the provided value. If it cannot be found, -1 is returned.
  It is Ok to set val to NULL.
 */
/*--------------------------------------------------------------------------*/

int iniparser_setstr(dictionary *ini, char *entry, char *val)
{
    dictionary_set(ini, strlwc(entry), val);
    return 0 ;
}

/*-------------------------------------------------------------------------*/
/**
  @brief    Delete an entry in a dictionary
  @param    ini     Dictionary to modify
  @param    entry   Entry to delete (entry name)
  @return   void

  If the given entry can be found, it is deleted from the dictionary.
 */
/*--------------------------------------------------------------------------*/
void iniparser_unset(dictionary *ini, char *entry)
{
    dictionary_unset(ini, strlwc(entry));
}


/*-------------------------------------------------------------------------*/
/**
  @brief    Parse an ini file and return an allocated dictionary object
  @param    ininame Name of the ini file to read.
  @return   Pointer to newly allocated dictionary

  This is the parser for ini files. This function is called, providing
  the name of the file to be read. It returns a dictionary object that
  should not be accessed directly, but through accessor functions
  instead.

  The returned dictionary must be freed using iniparser_free().
 */
/*--------------------------------------------------------------------------*/

dictionary * iniparser_new(char *ininame)
{
    dictionary *d ;
    char lin[ASCIILINESZ+1];
    char sec[ASCIILINESZ+1];
    char key[ASCIILINESZ+1];
    char val[ASCIILINESZ+1];
    char *where ;
    FILE *ini ;
    int lineno ;

    if ((ini=fopen(ininame, "r"))==NULL) {
        return NULL ;
    }

    sec[0]=0;

    /*
     * Initialize a new dictionary entry
     */
    d = dictionary_new(0);
    lineno = 0 ;
    while (fgets(lin, ASCIILINESZ, ini)!=NULL) {
        lineno++ ;
        where = strskp(lin); /* Skip leading spaces */
        if (*where==';' || *where=='#' || *where==0)
            continue ; /* Comment lines */
        else {
            if (sscanf(where, "[%[^]]", sec)==1) {
                /* Valid section name */
                strcpy(sec, strlwc(sec));
                iniparser_add_entry(d, sec, NULL, NULL);
            } else if (sscanf (where, "%[^=] = \"%[^\"]\"", key, val) == 2
                   ||  sscanf (where, "%[^=] = '%[^\']'",   key, val) == 2
                   ||  sscanf (where, "%[^=] = %[^;#]",     key, val) == 2) {
                strcpy(key, strlwc(strcrop(key)));
                /*
                 * sscanf cannot handle "" or '' as empty value,
                 * this is done here
                 */
                if (!strcmp(val, "\"\"") || !strcmp(val, "''")) {
                    val[0] = (char)0;
                } else {
                    strcpy(val, strcrop(val));
                }
                iniparser_add_entry(d, sec, key, val);
            }
        }
    }
    fclose(ini);
    return d ;
}



/*-------------------------------------------------------------------------*/
/**
  @brief    Free all memory associated to an ini dictionary
  @param    d Dictionary to free
  @return   void

  Free all memory associated to an ini dictionary.
  It is mandatory to call this function before the dictionary object
  gets out of the current context.
 */
/*--------------------------------------------------------------------------*/

void iniparser_free(dictionary *d)
{
    dictionary_del(d);
}

#ifdef __cplusplus
}
#endif

LGPL License

http://www.gnu.org/licenses/lgpl.html

Revisions

• 2006-10-05: Initial version 1.0.0.
• 2007-03-13: Version 1.0.1 fixes an installer issue where wininetd would not be installed as a service when the install path did not default to "c:\windows" for the user. Wininetd can still be installed manually by typing wininetd --install.
• 2007-04-14: http://www.autopower.com/rgozar/u2pipe-1.0.1-setup.exe was hanging when clicked. Now fixed.
• 2007-08-27: Version 1.0.3 now ported for unix (AIX and RedHat Linux binaries hyperlinked above).
• 2009-04-17: Revised linux installation notes.