building xdebug in linux machine

Though the instructions at xdebug.org were good, I still had a little hard time getting through the details. It took me a while to understand and get xdebug running in my linux machines. I tried with ubuntu gusty server (7.10) and fedora core 5, and had xdebug compiled and built in both.

You may also want to see the instruction at xdebug.

First download xdebug source from xdebug.org
I prefer wget.
#wget http://www.xdebug.org/files/xdebug-2.0.2.tgz

Untar tgz file and go into the xdebug-2.0.2 folder
#tar -xzf xdebug-2.0.2.tgz
#cd xdebug-2.0.2

To build the php extension, you need several binaries like phpize, php-config (part of php-dev package). The chances are your system may not have this particular package.

#apt-get install php5-dev (in ubuntu)
#yum install php5-dev (in fedora)
will automate the nitty-gritty job of solving dependencies. Some may even ask to upgrade mysql-server along with php depending upon the versions in your machine. Make sure your data are backed-up, before you say yes to install and upgrade dependencies packages.

After the installation, phpize should be available. phpize is a shell script to prepare PHP extension for compiling. Believe me I couldn't find the purpose of phpize anywhere in the internet. Maybe I haven't searched enough :(
Run phpize in your xdebug source directory. phpize generates several files, necessary for compiling, building the xdebug extension.

#phpize
will give the following output

Configuring for:
PHP Api Version:         20041225
Zend Module Api No:      20050922
Zend Extension Api No:   220051025

You will see that 'configure' script and several others are created.

#./configure --enable-xdebug

See the install instruction, if you get any errors. I didn't run into any. Hope you won't too.

#make
will build the extension xdebug.so in modules directory.

Move the extension to any folder you want. I moved to /opt/phpmodules/xdebug.so

You may either put the extension loading statement in php.ini or you may create a new ini file (xdebug.ini) in php.d directory. The configuration files will be automatically read from this directory (/etc/php5/conf.d in my case) and the extensions get loaded.

I wrote the following line to xdebug.ini

zend_extension="/opt/phpmodules/xdebug.so"

restart the apache server.

Load phpinfo (write a php script that calls phpinfo()) from the browser. If everything goes fine, you should see that xdebug module and zend logo in the browser.

#php -m
will also show xdebug under "PHP Modules" and "Zend Modules".

Congratulation. xDebug is successfully built and installed.

find and delete files at once... the shortcut

I was trying to delete a list of files with a character '-' in their filenames from a directory, containing lots of other files. Thanks to linux, you get everything you need here.

#find ./directorypath -regex '.*-.*'
lists all the files/directories containing '-' anywhere in the filename or directory name.
-regex allows to give any pattern to do powerful custom search.

To list only the files, add -type f
#find ./directorypath -type f -regex '.*-.*'

To delete all the files/directories with any pattern, just add command to the above line as shown below
#find ./directorypath -regex '.*-.*' -exec rm -rf {} \;

-exec rm -rf {} \;
executes rm -rf (remove forcefully and recursively in case of directory). The files/directories found by find command is passed to the rm -rf command via {}, which deletes them. \; marks the end of the command.

I suggest you first move all the files to some directory, before deleting them, just incase you don't accidentally delete some important files.

iptables... an easy way

I have been trying to open a trivial MySQL port in the fedora server box - just wanted to use desktop MySQL Client (like freeware EMS SQL Manager Lite for MySQL) rather than ubiquitous phpMyAdmin. I never understood this perplexing iptables command and its rules.

The easy way is to do a "Copy and paste" of a line from /etc/sysconfig/iptables and changing the port number (3306 for mysql - default).

-A RH-Firewall-1-INPUT -m state --state NEW -m tcp -p tcp --dport 3306 -j ACCEPT

Restarting the iptables service does the remaining job.

service iptables stop
service iptables start


Dekiwiki running along with other apps

I finally managed to install and run Dekiwiki in my Ubuntu 7.10 (gutsy) server machine. The problem with DekiWiki is that it has to be installed in the root directory. If I have other apps serving by the server, then DekiWiki will just claim the entire apache server, depriving me from accessing other apps - even phpMyAdmin.

After having Dekiwiki working, I made a tweak so that dekiwiki is run in port 8080 and port 80 is used to serve my default DocumentRoot (/var/www).

Here's my directory structure
/var/www/
------------phpMyAdmin/...
------------dekiwiki/web/...
------------testapps/

I added the following lines in /var/www/ports.conf below with Listen 80

Listen 8080

Now the apache will also listen in a different port 8080

I changed the deki-apache.conf file, included with the dekiwiki application. I just made sure that DocumentRoot is different when served in port 8080.

Now dekiwiki is accessed from http://192.168.0.45:8080/
While other apps can be accessed from http://192.168.0.45/

My hours of labor in Dekiwiki seem to get paid off.

<VirtualHost *:80>
DocumentRoot "/var/www"
</VirtualHost>

<VirtualHost *:8080>
ServerName 192.168.0.45

ErrorLog /var/log/apache2/error.log
CustomLog /var/log/apache2/access.log common

DocumentRoot "/var/www/dekiwiki/web"

RewriteEngine On
RewriteCond %{REQUEST_URI} ^/$
RewriteRule ^/$ /index.php?title= [L,NE]

RewriteCond %{REQUEST_URI} !/(@api|editor|skins|config)/
RewriteCond %{REQUEST_URI} !/(redirect|texvc|index|Version).php
RewriteCond %{REQUEST_URI} !/error/(40(1|3|4)|500).html
RewriteCond %{REQUEST_URI} !/favicon.ico
RewriteCond %{REQUEST_URI} !/robots.txt
RewriteCond %{QUERY_STRING} ^$ [OR] %{REQUEST_URI} ^/Special:Search
RewriteRule ^/(.*)$ /index.php?title=$1 [L,QSA,NE]

# deki-api uses encoded slashes in query parameters so AllowEncodedSlashes must be On
AllowEncodedSlashes On

# mod_proxy rules
ProxyPass /@api http://localhost:8081 retry=1
ProxyPassReverse /@api http://localhost:8081
SetEnv force-proxy-request-1.0 1
SetEnv proxy-nokeepalive 1
</VirtualHost>

Python: HTML tables to Mediawiki converter

Some ideas and codes for the html2wiki are borrowed from html2csv converter. It reads any file and converts tables, if present, to wiki format. The code, in colored, can also be viewed from the code snippets.

import HTMLParser, re, sys
class html2wiki(HTMLParser.HTMLParser):
def __init__(self):
   HTMLParser.HTMLParser.__init__(self)
   self.wiki = ''      # The Wiki text
   self.wikirow = ''   # The current Wiki row of table being constructed from HTML
   self.inTD = 0      # Used to track if we are inside or outside a <TD>...</TD> tag.
   self.inTR = 0      # Used to track if we are inside or outside a <TR>...</TR> tag.
   self.re_multiplespaces = re.compile('\s+')  # regular expression used to remove spaces in excess
   self.rowCount = 0  # output row counter.
   self.rowspan = ''
   self.colspan = ''
   self.linebreak = '<br>'
   self.data = ''
   self.prop = ''

def handle_starttag(self, tag, attrs):
   if tag == 'table': self.start_table()
   elif   tag == 'tr': self.start_tr()
   elif tag == 'td': self.start_td(attrs)

def handle_endtag(self, tag):
   if tag == 'table': self.end_table();
   elif   tag == 'tr': self.end_tr()
   elif tag == 'td': self.end_td()

def start_table(self):
   self.wiki += '{| border=1' + self.linebreak
   self.wiki += '|-' + self.linebreak

def end_table(self):
   self.wiki += '|}' + self.linebreak

def start_tr(self):
   if self.inTR: self.end_tr()  # <TR> implies </TR>
   self.inTR = 1

def end_tr(self):
   if self.inTD: self.end_td()  # </TR> implies </TD>
   self.inTR = 0       
   if len(self.wikirow) > 0:
       self.wiki += self.wikirow
       self.wiki += '|-' + self.linebreak
       self.wikirow = ''
   self.rowCount += 1

def start_td(self, attrs):
   if not self.inTR: self.start_tr() # <TD> implies <TR>
   self.data = ''
   self.prop = ''
   self.rowspan = ''
   self.colspan = ''
   for key, value in attrs:
       if key == 'rowspan':
           self.rowspan = value
       elif key == 'colspan':
           self.colspan = value       
   self.inTD = 1

def end_td(self):
   if self.inTD:           
       self.wikirow += '| ' + self.prop + self.re_multiplespaces.sub(' ',self.data.replace('\t',' ').replace(self.linebreak,'').replace('\r','').replace('"','""'))+ self.linebreak;
       self.data = ''
       self.inTD = 0

def handle_data(self, data):
   if self.inTD:
       if data.strip() != '':           
           self.prop = ''
           if self.rowspan != '':
               self.prop = ' rowspan = '+self.rowspan
           if self.colspan != '':
               self.prop += ' colspan = '+self.colspan
           if self.prop:
               self.prop += ' | '
           self.data += data

if __name__ == '__main__':           
parser = html2wiki()
if len(sys.argv) == 2:
   in_file = open(sys.argv[1],"r")
   text = in_file.read()
   parser.feed(text)
   in_file.close()
   print parser.wiki
else:
   print 'Argument - filename required'

Since I need a web interface for users, I don't want to create another similar app in php nor do I want to write cgi in python. So I wrote another tiny php script and exploit the python script. I have used tinymce so that I can now just copy and paste html tables directly to the edit box and do the conversion easily.

<?
if($_POST['submit']) {
 if(trim($_POST['html'])) {
     $input = stripslashes(trim($_POST['html']));
  
     $filename = 'uploads/'.date('Ymdhis').'.txt';
     $fp = fopen($filename, 'w');
     fwrite($fp,$input);
     fclose($fp);         
     $ret = exec("python html2wiki.py $filename", $output, $retval);
     $output = implode("\n",$output);
     unlink($filename);
 }
}
?>
<script language="javascript" type="text/javascript" src="/lib/tinymce/jscripts/tiny_mce/tiny_mce.js"></script>
<script language="javascript" type="text/javascript">
tinyMCE.init({
 theme:"simple",
 mode : "textareas"
});
</script>

<form name='converter' method='post'>
<input type='submit' value = 'Convert Html2Wiki >>' name='submit'><br>
<table>
<tr><td><textarea name='html' cols='50' rows='40'><?=$input?></textarea></td>
<td><textarea name='wiki' cols='50' rows='40'><?=$output?></textarea></td>
</tr></table>
</form>

The output is something like shown below.

Mediawiki: Reusing the same code for multiple Wikis

Sometimes you may need to install separate wikis for in the same server. One inefficient solution (in terms of time and space) is to install everytime you need Wiki. Other solution is to reuse the wiki code for all. But you need to install database separately for each -for this, you first need to dump fresh database right after first installation so that it can be used for other potential Wikis that may need to be installed for other purpose in the same server.



This has been tried for Mediawiki 1.6.7 only.



1. First install Mediawiki in any folder (say wikicode).

2. Dump/Export mysql database from phpMyAdmin.

3. Create a new folder in document root (say testwiki).

4. Copy index.php and LocalSettings.php from wiki installed directory (wikicode, in our case) to the newly created folder.

The folder structure should be like this.

public_html/

	wikicode/

		mediawiki folders/files

	testwiki/

		index.php

		LocalSettings.php

5. Make necessary changes in the LocalSettings.php, as shown below. The original and altered code excerpts are shown.



Original Localsettings.php right after installation.

[...]

if( defined( 'MW_INSTALL_PATH' ) ) {

	$IP = MW_INSTALL_PATH;

} else {

	$IP = dirname( __FILE__ );

}



$path = array( $IP, "$IP/includes", "$IP/languages" );

set_include_path( implode( PATH_SEPARATOR, $path ) );



[...]

$wgSitename         = "Demowiki";



$wgScriptPath	    = "/wikicode";

$wgScript           = "$wgScriptPath/index.php";

[...]

$wgStylePath        = "$wgScriptPath/skins";

$wgStyleDirectory   = "$IP/skins";

[...]

$wgUploadPath       = "$wgScriptPath/images";

$wgUploadDirectory  = "$IP/images";

[...]

Changed Localsettings.php in the new folder (testwiki, in our case)

[...]

if( defined( 'MW_INSTALL_PATH' ) ) {

	$IP = MW_INSTALL_PATH;

} else {

	$IP = dirname( __FILE__ );

}

$IP = $_SERVER['DOCUMENT_ROOT'].'/wikicode/';

$path = array( $IP, "$IP/includes", "$IP/languages" );

set_include_path( implode( PATH_SEPARATOR, $path ) );



[...]

$wgSitename         = "Demowiki";



//$wgScriptPath	    = "/wikicode";

$wgScriptPath	    = "/testwiki";

$wgScript           = "$wgScriptPath/index.php";

[...]

//$wgStylePath        = "$wgScriptPath/skins";

$wgStylePath        = "http://localhost/wikicode/skins";

$wgStyleDirectory   = "$IP/skins";

[...]

$wgUploadPath       = "$wgScriptPath/images";

//$wgUploadDirectory  = "$IP/images";

$wgUploadDirectory  = $_SERVER['DOCUMENT_ROOT'].'/demowiki/images';

6. Change database information accordingly.



Blue color represents the added lines

Green color represents the commented lines in the modified LocalSettings.php

[...] represents some codes, those are removed for brevity.



Now you can copy the folder (testwiki) as many times as you like, install databases and make necessary changes in the LocalSettings.php. This should allow you to install Wiki with very little effort from your part.



Updates:

Sorry I didn't realize that there is also a small addition in index.php, if any of you have been trying to follow this. It's funny that as I was trying to follow this myself, i realize that it's not working yet.



Original index.php

require_once( './includes/Defines.php' );

Changed index.php

set_include_path(get_include_path().PATH_SEPARATOR.$_SERVER['DOCUMENT_ROOT'].'/wikicode/');

require_once( 'includes/Defines.php' );

jquery: tweaking auto-complete plugin for multiple words in a single input

If you have composed mails in gmail, then you would know that it allows to select multiple auto-complete email addresses separated by commas. That's what I want to implement in jquery auto-complete plugin. Currently it works like google suggest, allowing only one auto-complete word.

If you haven't used JQuery before, I suggest you to go over the site and see some examples and try some.

Well JQuery is the first javascript framework that I have ever used and got stuck with its beauty. Later I came to know about similar javascript frameworks like prototype, scriptulous. But now I am used to jquery. The another beauty is there are many third party plugins.

I extended the auto-complete plugin so that now it allows to choose the auto-complete mode - multiple or single. With multiple, it will work like gmail compose and with single it works like google suggest. Well single is what the original plugin provides. I will explain the changes done to achieve multiple mode.

$.fn.autocomplete = function(url, options) {
 ...
       options.mode = options.mode || "single";
       options.multipleSeparator =
  options.multipleSeparator || ",";
 ...
}

The default mode is single. If you pass 'multiple', now it should work like gmail compose. multipleSeparator separates the multiple words. The default is comma. Try to pass different parameters when you initialize the function.

    $(document).ready(function() {
             $("#suggest").autocomplete('search.php', { minChars:1,
   matchSubset:1, matchContains:1, cacheLength:10,
   onItemSelect:selectItem, formatItem:formatItem,
   selectOnly:1, mode:"multiple",
   multipleSeparator:"|" });
    });

When the item gets selected, you need to find whether it's multiple or single. If it's single, you don't need to do anything. But for multiple, you have to search for the separator and append the selected new value to the old one in the selectItem(...) function.

function selectItem(li) {
...
      if(options.mode == "multiple") {
 old_value = $input.val();
 if(old_value.lastIndexOf(options.multipleSeparator) >= 1) {
  sep_pos = old_value.lastIndexOf(options.multipleSeparator);
  value = old_value.substr(0,sep_pos+1);
  new_value = value + v +options.multipleSeparator;
 } else {
  new_value = v + options.multipleSeparator;
 }
} else {
 new_value = v;
}
...
}

The plugin use ajax to fetch the items from either file or database. For that it needs to construct url on its own. If you type 'a', then the url will be http://abc.com/search.php?q=a and it fetches data from that url. However when we are in multiple mode, the input will contain something like "ant,b", in this case, the constructed url will be http://abc.com/search.php?q=ant,b. Since only "b" needs to be passed , the text before comma should be excluded. So this needs some tweaks in the makeUrl(...) function.

function makeUrl(q) {
if(options.mode == "multiple") {
 if(q.lastIndexOf(options.multipleSeparator) >= 1) {
  sep_pos = q.lastIndexOf(options.multipleSeparator);
  q = q.substr(sep_pos+1);
 }
}
...
}

The package, containing jquery and modified auto-plugin code, can be downloaded from here.
Click here for the original auto-complete source.