regexxer: search/replace for the masses

March 29th, 2009 edited by Vicho

Article submitted by David A. Thompson. Guess what? We still need you to submit good articles about software you like!

This thing is great! If you manage a set of files (e.g., a code base) and end up needing to perform a search/replace across the files, or some subset thereof, maybe regexxer will end up being counted among your friends.

The regexxer web site summarizes regexxer as “a nifty GUI search/replace tool featuring Perl-style regular expressions”. After using regexxer, I have to agree…

If you have a code base that needs some search/replace time spent on it, regexxer really does seem to be the way to go. Again, the web site sums things up well: “If you need project-wide substitution and you’re tired of hacking sed command lines together, then you should definitely give it a try.”


regexxer main windowThe regexxer GUI is a straightforward, logically organized window with file management on the left and search/replace management on the right. The beauty of the thing is that the author(s?) seem to have added a lot of flexibility to the search/replace process while keeping the user interface relatively simple and clean.

File Management

One can select one or many files from a given folder. Recursion is optional and a pattern (e.g., *.lisp) can be defined to further limit files being acted on. A ‘find’ button then triggers execution of the search and the corresponding files are shown in the window below. Not much more to say…

The Search Replace process

The user interface makes it easy on the noggin with top-to-bottom arrangement of controls which follow the logic of the search replace. The perl-style search regex and the corresponding replacement are entered at the top of the window. Another ‘find’ button triggers the search and the corresponding matches are shown in context in the window below. At this point, I found the regexxer user interface close to ideal. The flexibility available to the user includes:

  • ‘On the fly’ editing of the file.
  • Performing the replace action one-at-a-time or per file or upon all files as a unit.
  • The ability to navigate through the matches (little green arrows at the bottom let the user ‘back up’ or ’skip’ matches as desired).


It’s likely I’m simply unaware of many of the alternatives ‘out there’. Typically, for this type of thing, many turn to sed, a powerful command-line tool. However, sed has a substantial learning curve. In contrast, if one has already mastered perl regex, regexxer requires very little in the way of additional investment of energy. Other command-line tools which provide some of sed’s functionality, such as modifile, rpl, and replace, perhaps have their place but they certainly don’t offer the warm fuzzies of the GUI interface of regexxer.

Emacs, which typically seems to already have an extension for just about any task except cooking breakfast, didn’t seem to shine here. I can’t help but suspect that there’s an emacs tool available for this sort of thing but I haven’t happened across it yet.

Among other GUI tools, several offer similar functionality. For example, the text editor jedit can act in much the same way as regexxer. Unfortunately, jedit relies on java and, thus, things get dicey on planet Debian. Another, more focused tool, kfilereplace, is very straightforward but does not seem to be under active development and has much more limited functionality (e.g., the option to perform replacements ‘one-at-a-time’ is greyed out in the configuration window). Other utilities which seem to be designed to perform this type of task (e.g., modifile or regexsearch) don’t seem to be available for Debian.


Apparently those on planet Debian and planet Ubuntu have Ross Burton and Daniel Holbach to thank for the packaging of this excelent tool.

Posted in Debian, Ubuntu | Comments Off

xmlstarlet: Command Line XML Toolkit

March 22nd, 2009 edited by Vicho

Article submitted by Vasily Faronov. Guess what? We still need you to submit good articles about software you like!

With the proliferation of XML-based formats, it is nice to have tools that manipulate XML documents in the traditional Unix-like fashion, as the good old grep(1), sed(1) and other tools do for plain text. xmlstarlet is one of such tools. In fact, it is an entire toolkit packed into one program: xmlstarlet can extract data from XML documents, alter them, validate them, and perform many other useful operations. xmlstarlet has been available in Debian since at least release 4.0 “etch”, and in Ubuntu since at least release 6.06 “Dapper Drake”.

Let’s look at a few features of xmlstarlet more closely.

Extracting Data from XML Documents

There is a flexible way of processing XML documents and extracting data from them — a language known as XSLT. Both Debian and Ubuntu provide utilities, such as xsltproc, to deal with XSLT. However, this language is not exactly terse, and it requires you to first compose a separate document defining the desired transformation, and then apply it to the original document. When all you want to do is extract a few values from a document, you’d like something more approachable.

xmlstarlet is your friend. It features a relatively simple command syntax for selecting data, based on an auxiliary language called XPath, which allows for addressing elements in XML documents in a style reminiscent of filesystem paths. Behind the scenes, xmlstarlet still generates XSLT code, and it helps to know the actual XSLT language, but simple queries can be done almost intuitively.

Suppose we’d like to get a list of the recent headlines from the Debian Package of the Day website. We can use xmlstarlet to extract titles from the site’s RSS feed, because RSS is an application of XML. In RSS, entry titles are contained in title elements, in turn contained in item elements, which are in channel elements under the root rss element. The feed itself can be easily fetched with wget(1). Our pipeline would then look like this:

$ wget -O - 2>/dev/null | \
> xmlstarlet sel -t -m /rss/channel/item -v title -n
Fonty Python: manage your fonts
localepurge: Automagically remove unnecessary locale data
vnstat: a console-based network traffic monitor
rtpg-www: Please your dearest with rtorrent’s power
iftop - display bandwidth usage on an interface by host
atop: an ASCII full-screen performance monitor
dstat: versatile tool for generating system resource statistics
tellico: collection manager for books, videos, music, and a whole lot more
atool: handling archives without headaches
watch (from procps): execute a program at regular intervals, and show the output

The wget invocation is hopefully obvious (if it isn’t, just believe me that it downloads the feed and prints it to the standard output), so let’s dissect the xmlstarlet part.

  1. sel, which stands for “select”, is the subcommand to invoke within xmlstalrlet — since it’s a toolkit, it contains a number of such subcommands.
  2. The -t option designates the beginning of a template — roughly, a group of processing instructions.
  3. The -m option specifies a match, and /rss/channel/item is an XPath expression; together they translate to “for each item element found under a channel element under the rss element”.
  4. Then we specify what to do for that match: in our case, -v title prints out the value of the title element under the current item, and -n prints a newline separator.

The various options to xmlstarlet’s sel subcommand can be combined to produce fairly complex XSLT transformations. You can view the XSLT code generated by your command by adding the -C option.

Validating XML Documents

There are several ways of defining an XML document format, and the simplest of them is called document type definition, or DTD. A DTD defines which elements are allowed to appear in a document and what they can contain. DTDs for some popular formats, such as XHTML, are included in the Debian and Ubuntu archives. The val subcommand of xmlstarlet can validate documents against a DTD — that is, check if the documents comply with the formal requirements laid out in a DTD. In addition to DTD, xmlstarlet can also handle the more advanced XML Schema and RELAX NG languages.

As an example, let’s integrate xmlstarlet with gedit, the GNOME text editor, to enable easy validation of XHTML 1.0 Strict documents. We will need the w3c-dtd-xhtml package that contains the DTD files.

To validate against a DTD, xmlstarlet should be invoked with the val subcommand, the -d option (for “DTD”), and a path to the DTD file. As in the previous example, the document can be piped into xmlstarlet. We will integrate it into gedit by means of the latter’s “External Tools” plugin. Enable it by choosing “Edit” → “Preferences” → “Plugins” and marking “External Tools”. Then, in the “Tools” menu, select “External Tools” and click “New”. Name the tool as you wish, and optionally give it a description and a shortcut key. For “Command(s)”, enter this simple script:

#! /bin/sh
xmlstarlet val -d /usr/share/xml/xhtml/schema/dtd/1.0/xhtml1-strict.dtd -

xmlstarlet validating a remote XHTML document in geditChoose “Current document” for input, “Display in the bottom pane” for “Output”, and set “Applicability” to “All documents”. And that’s it. You can now validate any document you open in gedit from the “Tools” menu — even if the document comes from a remote location by way of the GNOME virtual filesystem.

With some more effort, you can write a script to validate any document for which you have the DTD installed. (Hint: you may use xmlcatalog(1) from the libxml2-utils package to locate DTD files by their public identifiers.)

Other Uses

This article cannot cover many other features of this toolkit, such as editing XML documents (ed), listing their element structure (el), or pretty-printing (fo). You may want to check out the examples that come with the xmlstarlet package (/usr/share/doc/xmlstarlet/examples), and to the reference help available by invoking xmlstarlet ‹COMMAND› --help.

Related Tools

The Debian archive also contains the python-4suite-xml package, which among other things provides some command tools for XML processing similar to xmlstarlet. However, 4suite seems to be intended more as a Python package, and consequently its command tools appear to be less feature-complete than xmlstarlet.

For solving specific XML-related problems, such as converting XML to and from other formats, you may want to have a look at the more specialized packages available in Debian and Ubuntu. The Debian Reference has an overview of some of those.

Posted in Debian, Ubuntu | 2 Comments »

fwbuilder: Manage Firewalls Professionally

March 15th, 2009 edited by Vicho

Article submitted by Vadim Kurland. Guess what? We still need you to submit good articles about software you like!

Firewall Builder is available from the libfwbuilder and fwbuilder packages in both Debian and Ubuntu in Universe. Packages for the current development builds are available from the project download area on SourceForge.

Eveyone knows about netfilter/iptables, a powerful firewall framework and command line tool that is part of every Linux distribution. Unfortunately, managing a security policy with it remains a non-trivial task for several reasons. Partially this is because of the complex syntax of the command line interface and the vast amount of available options and parameters. Another reason is that the administrator has to understand the internal path of the packet inside the Linux kernel and its interaction with different parts of netfilter in order to build rules correctly. This is not a specific problem of iptables though, other popular Open Source firewall platforms, such as OpenBSD PF, ipfilter and ipfw present similar challenges.

What is needed is a tool that lets an administrator define the security policy on a higher level of abstraction and hide the internal structure of the target firewall platform. For example, such a tool should decide which iptables chain is right for each generated iptables rule automatically, without the administrator’s input. It should also pick the right iptables targets for both policy and NAT rules as well as properly use most popular iptables modules, all automatically. Such tool should also implement best practices in policy design and help administrator deploy and activate generated policy on the firewall.

Firewall Builder does just that.


Firewall Builder is a GUI firewall configuration and management tool that supports iptables (netfilter), ipfilter, pf, ipfw, Cisco PIX (FWSM, ASA) and Cisco routers extended access lists. It presents all supported firewalls to the administrator in terms of unified abstract firewall that takes the best features from all of them and hides their specifics and inconveniences. Firewall Builder is more complex than many basic firewall configuration GUIs such as Firestarter, but on the other hand one can build very complex policies with Firewall Builder and fully utilize flexibility and power of iptables and other supported firewalls.

The general idea should be familiar to anyone who has ever worked with commercial firewall management systems. All configuration management operations can be performed from one central place: the Firewall Builder GUI. You create and manage collection of objects that describe network addresses, hosts and firewalls, as well as services, and then build firewall policy and NAT rules using these objects. Policy rules are defined in terms of “Source” and “Destination” addresses and “Service” and can have additional parameters such as interface association, direction, time interval and optional platform-dependent attributes. NAT rules are defined by addresses and services before and after translation.

Example of a policy rule

Rules are built with simple drag and drop operations and then firewall configuration can be generated with one click of a mouse. In the end, Firewall Builder produces a script or configuration file in the language of the target firewall. For iptables, it creates shell script that loads iptables rules, while for other platforms it creates a configuration file suitable for them. This makes it simple to deploy and activate the generated policy and also helps integrate Firewall Builder with existing automation scripts.

Fragment of the standard TCP objects library
The program comes with a collection of over 100 standard objects that can be used to describe popular TCP, UDP and ICMP services.

Firewall Builder implements many best practices in firewall policy design and firewall management procedures. Here are just a few examples:

  • It enforces a policy structure that denies all traffic by default and only permits what is necessary.
  • The administrator can easily define IP address of the management workstation and Firewall Builder will automatically add a rule to ensure that ssh access from it to the firewall is always permitted. This rule is designed to assure that ssh session over which the installer activates a new policy does not break or hang. This helps to avoid accidents when errors in the policy rules cut off remote access to the firewall in the middle of an activation, making it impossible to fix the error and causing prolonged network outage.
  • For Cisco PIX (ASA) and IOS access lists, where each access-list commands are immediately activated as they are entered, Firewall Builder can optionally create temporary access lists to ensure uninterrupted ssh access from the management workstation to the firewall for the duration of the policy reload session. This method provides the best protection against outages caused by loss of contact with the firewall because of errors in policy.
  • For iptables, Firewall Builder can generate a script using iptables-restore for atomic activation. If iptables-resore detects an error in the script and refuses to load the policy, script leaves the firewall in the state it was in before. For other firewall platforms it uses appropriate activation methods to achieve the same goal.
  • The built-in policy installer supports “test” install mode with automatic roll-back. This is another safety mechanism that helps minimize outages in case of errors in the policy. These measures are available for all supported systems, such as Linux/iptables, *BSD/pf, Cisco PIX and Cisco IOS.

Quick Tour

Main window

The main window of the program includes objects tree on the left (1), brief information about object selected in the tree (2), current firewall policy view (3) and a dialog panel where you can edit objects parameters (4).

As all Open Source projects, Firewall Builder depends on the user community who provide testing, bug reports and other forms of feedback. You can file bug reports and feature requests using the bug tracking system. The mailing list is a great place to ask for help and discuss the program with other users.

This was just a brief introduction to the Firewall Builder package. If you are interested in the program, you can find more information on the project web site at The slideshows Introduction to Firewall Builder 3.0 for the impatient and Getting starter with Firewall Builder can help you get more familiar with the program.

Posted in Debian, Ubuntu | Comments Off

PIDA: the Python Integrated Development Application

March 8th, 2009 edited by Tincho

Article submitted by Javier Derderian. Guess what? We still need you to submit good articles about software you like!

PIDA screenshotPIDA is an IDE (integrated development environment) written in Python and the pygtk graphical toolkit. It is slightly different from other IDEs: rather than attempting to write a set of development tools of its own, PIDA reuses available tools. In this regards PIDA can be used as a framework for putting together your own customized IDE.

Although still a young application, PIDA already boasts a huge number of features because of the power of some of the tools it integrates. For example features such as code completion and syntax highlighting are well implemented in PIDA’s integrated editors far better than any editor built for a commercial IDE. PIDA currently features many code editing helpers: syntax highlighting, code completion, automatic indenting, block commenting, etc; project management, version control management, Python debugger and profiler, GTK+ GUI building and rapid application design.

Among the already integrated components you can find:

  • VIM and Emacs as embedded editors with full support of each one’s features:
    • Syntax Higlighting
    • Code completion
    • Plugins
  • Bazaar, Git, Subversion (and more) as version control systems.

It’s actually designed to program in any language, but it has some Python specific features like a Python shell. You can program you own plugins, and there’s a very nice API documentation to help you go trought the plugin development path.

Some already available plugins are:

Send code to a pastebin service
PdfTex preview
PdfTex preview compiles and displays pdf documents every time the buffer is saved.
Show class/function from python file, and show compilation errors
Python Debugger
Python Debugger based on RPDB2 the WinPDB Back End
Unit Tester
Perform unit tests
Docbook browser
Browse local docbook
Todo manager
Manage a personnal todo list per project
RFC Viewer
Download RFC index, search and view RFC pages inside PIDA
This plugin, developed outside of the project, integrates lots of Bazaar function that are not included in the base version control integration

PIDA is a great way of keep using Vim and have a nice GUI around to help you work faster with the file browser, the project manager and the internal shells. You can get more info on using and developing PIDA in the handbook

There are official packages available in both Debian and Ubuntu for a long time now.

And remember: PIDA LOVES YOU!

Posted in Debian, Ubuntu | 10 Comments »

bash-completion: the greatest things since bash completion

March 1st, 2009 edited by Tincho

Article submitted by Andre Masella. Guess what? We still need you to submit good articles about software you like!

Pressing the tab key in bash to auto-complete a file name is one of the most time saving tricks especially when dealing with very long file names. Unfortunately, file name completion is not always the right behavior. Take Subversion for example. The first argument to svn is the sub-command to use. The file name is also restricted: svn add only takes files not under revision control and svn rm only takes files that are under revision control.

This is where the bash-completion package steps in. After installing it with a quick apt-get install bash-completion, a few lines need to be uncommented in sudo vim /etc/bash.bashrc and the shell restarted. After that, try this:

$ svn <TAB><TAB>
add         cl          diff        list        move        propdel     rename      unlock
annotate    cleanup     export      lock        mv          propedit    resolve     update
blame       co          -h          log         pdel        propget     resolved    --version
cat         commit      help        ls          pedit       proplist    revert
changelist  copy        --help      merge       pget        propset     rm
checkout    cp          import      mergeinfo   plist       pset        status
ci          delete      info        mkdir       praise      remove      switch

Ta-da! Smarter completion for subversion.

bash-completion will alter the behavior of most commands to limit the display to relevant files. For example, mpg321 will only display MP3 files in the list. Programs like rmmod, iwconfig, ifup, and lvm will display relevant choices that are not files at all. Even bash’s fg and bg will now tab-complete with job identifiers.Completion for man is useful as it will auto-complete only man pages that exist and allows you to incrementally narrow your search by providing the beginning of the man page name, just like with regular files.

Occasionally, it doesn’t behave as expected. Particularly, sometimes a file with the wrong extension will be filtered out by bash-completion. For example, if you save an image for certain Internet forums, the file will sometime lack an extension. bash-completion will then filter out that file because it does not have the right extension. This can also happen if the capitalization is unusual. For example, bash-completion will suggest files that end in .mp3 or .MP3 for mpg321, but not .Mp3. In that case, either rename the file or insert # at the beginning of the line. The # makes bash think this line is a comment and bash-completion returns to regular file name completion. Once finished, remove the # and run the command.

bash-completion is available in Debian and Ubuntu. If it isn’t available, it is very easy to install from source.

Posted in Debian, Ubuntu | 18 Comments »