What is libxml

Libxml is an XML toolkit:

• developped within the Gnome (gnome-xml) project, and also as one of W3C software
• available under LGPL and the W3C IPR
• present in Gnome and W3C CVS bases
• Web page is at http://xmlsoft.org/
• Packaged as libxml on most Linux distributions
• Extremely portable (Windows, Unices, Embedded targets, PSion, ...)
• Version 1.0 shipped beginning of 1999. Version 2.0 released beginning of April
• conformance tested against the OASis test suite

Libxml main interfaces

• The parser: does Well-Formeness and DTD validation, handle namespaces, can operate in pull or push modes
• SAX: callback based interface (opening, closing, characters ...)
• DOM tree: build a full in-memory tree followingly DOM interfaces
• HTML: an HTML parser, generating either SAX callbacks or a DOM tree
• Tree: routines to create/modify/save a DOM tree. Being able to save back parsed documents influenced a lot of the design of libxml
• URI module
• I/0 interface: modular with basic FTP and HTTP built-in
• I18N handlers: defaults UTF8 UTF16 and ISO-Latin, uses iconv if found
• XPath expression language to query XML documents
• Debug module (including a small shell) and memory usage checking

Libxml architecture overview

The DOM tree

Internationalization

XML requires support for the following encodings:

• UTF8: includes the ASCII range and use a variable lenght encoding for the rest of the full Unicode set
• UTF16: uses 16bits chars (and sometime surrogates)

Libxml v1 was very weak in that area. Libxml2 has native support for both encodings and ISO-Latin-x

If iconv library is found at compile time, it is used to add support to a large (and expandable) set of encodings including the most common Japanese encodings.

Libxml can save to a specific encoding, if a character cannot be encoded it is converted to a char ref on the fly like ሴ

On the workbench

Started but not completed:

• XPath: a sample and currently incomplete implementation

Todo:

• XPointer: built on top of XPath allow to address any portion of an XML document
• XLink detection: detect XML hypertext links and implement a simple link database
• Schemas: new version obsoleting DTDs inherited from SGML

Add-ons worked on:

• Gdome: the DOM interface
• Gtkhtml2: new version of the Gnome HTML widget

Existing use

Libxml is already in use in a number of environments:

• rpm2html/rpmfind: generation/maintenance/client for 300Megs of RDF encoded linux software package metadata
• configuration files for programs, metadata (gconf, glade, nautilus)
• data format for gnome programs (gill, gnumeric, gphoto, ...)
• embedded system (TV desktop set)
• handling of satellite images

However most users just don't give feedback, but patches are coming (and usually accepted !)

Future

Focuse on implementing future XML specifications

Would like a basic rendering widget (gtkhtml2)

Would like volunteers for an XSLT implementation