XML Support For Tcl

TclXML

Steve Ball

Zveno Pty Ltd


  
    www.zveno.com
  

$Id: README.html,v 1.1 2005/11/04 07:27:00 balls Exp $

Revision History
Revision 3.0

Table of Contents

Installation
Dependencies
Pure-Tcl Installation
Compiled Installation
Usage
XPath

This package provides XML parsers for Tcl scripts. There is a generic front-end interface with plugin parser implementations. A number of parser implementations or wrappers are provided:

The package defines commands in the ::xml Tcl namespace:

xml::parserclass [info] [names]
xml::parser

Other packages are also supplied to provide support for related standards. In particular, work has commenced on the 'xpath' package for parsing XPath location paths.

See the website for more information on XML support for Tcl.

Contact Steve Ball for information about this release.

Installation

TclXML v3.0 features a pure-Tcl installer.

Note

Future work on the installer will include an 'uninstall' capability, as well as the capability to build the binary libraries in the package from the source.

Dependencies

Tcllib

http://dev.scriptics.com/software/tcllib/

In order for the Tcl-only parser to resolve external entities, the tcllib package must be installed.

Be sure to get a version which includes the uri package. Version 1.2 or better is recommended.

The latest CVS snapshot may be found at the SourceForge project page.

Expat

Expat is only required for the compiled version of the TclXML/expat package.

TclXML/expat requires expat v0.2, with modifications by Ajuba Solutions. This is now supplied in the expat subdirectory of this distribution and will maintained until TclXML has been resynchronised with the Expat SourceForge project (volunteer needed!).

libxml2

libxml2 is only required for the compiled version of the TclXML/libxml2 package. libiconv may also be required.

The source code for libxml2 and libiconv is not supplied with this package. Download libxml2 from xmlsoft.org separately, and libiconv from a GNU mirror site.

Version 2.6.9 (or better) is recommended.

Pure-Tcl Installation

no compilation required

The install.tcl script will install the pure-Tcl parser. Run it using wish, eg.

wish install.tcl

Note

If Tcl/Tk has been configured correctly on your system, you should be able to simply double-click the script.

If the pure-Tcl parser is good enough for you, then read no further.

Compiled Installation

Unix/Linux

You must have Tcl/Tk version 8.2 or better installed on your system. Tcl/Tk 8.3 or better is recommended.

  1. Make sure you have Tcllib 1.2 (or better) installed. Tcllib is still required, even for the compiled parser.

  2. If you wish to use TclXML/libxml2, make sure libxml2-2.6.9 (or better) is installed.

  3. Unpack the TclXML distribution and cd into the tclxml-3.0 directory.

  4. Run the configure script, with the [--prefix] and [--enable-threads] switches (the latter only if Tcl has been built with threads enabled). For example, on my system I have Tcl 8.4 installed in /usr/local/tcl8.4 so I would use the command:

    ./configure --prefix=/usr/local/tcl8.4 --enable-threads
  5. make
  6. Note

    Don't test the package using make test until all of the packages are installed (it is a current deficiency of the build system that the individual modules cannot be tested before installation - we hope to fix this soon!).
  7. make install

    You may need to do this as root. This installs the pure-Tcl parser class, as well as support files required for the next step.

  8. cd expat

    Now TclXML/expat is about to be built.

  9. Use exactly the same configure command from step (4) above.

  10. make
  11. make install

    You may need to do this as root. This installs the expat parser class.

  12. cd ../libxml2

    Now TclXML/libxml2 is about to be built. Remember, you must have built and installed libxml2 first.

  13. Use exactly the same configure command from step (4) above.

  14. make
  15. make install

    You may need to do this as root. This installs the libxml2 parser class.

  16. cd ..; make test

Windows (MS VC++)

Note

[Advice: ActiveTcl includes binaries for TclXML.]

You must have Tcl/Tk version 8.2 or better installed on your system. Tcl/Tk 8.3 or better is recommended.

Before starting, download the binaries for libxml2 (or build them from source). xmlsoft has a link to the MS Windows binary distribution.

If you have a TEA build environment setup, just use the normal configure/make/make install pattern.

Alternatively, the win subdirectory contains a makefile.vc file for Visual Studio C++ v6.0. This makefile builds both the expat and libxml2 parser classes. In a Command Prompt window set up your environment so that nmake is on the path (by running VCVARS32.BAT), then type the following:

nmake -f makefile.vc TCLDIR=C:\Path\To\Tcl INSTALLDIR=C:\Path\To\Tcl LIBZDIR=C:\Path\To\libz LIBICONVDIR=C:\Path\To\libiconv LIBXML2DIR=C:\Path\To\libxml2

As an example, on my system I have Tcl installed in C:\Tcl and the libxml2 binaries unpacked in the directory C:\gnome. Accordingly, I would use the following command line:

nmake -f makefile.vc TCLDIR=C:\Tcl INSTALLDIR=C:\Tcl LIBZDIR=C:\gnome\zlib-1.1.4.win32 LIBICONVDIR=C:\gnome\libiconv-1.9.1.win32 LIBXML2DIR=C:\gnome\libxml2-2.6.9.win32

Install the package by appending 'install' to the command line used above, for example:

nmake -f makefile.vc TCLDIR=C:\Path\To\Tcl INSTALLDIR=C:\Path\To\Tcl LIBZDIR=C:\Path\To\libz LIBICONVDIR=C:\Path\To\libiconv LIBXML2DIR=C:\Path\To\libxml2 install

Macintosh OS X

There are two ways to build TclXML under Mac OS X:

  1. The usual Unix way, see above.

  2. As an embedded Framework using Xcode.

The macosx directory contains the Xcode files for building under OS X (Panther). TclXML has been tested on OS X 10.3 (or is that X.3?).

Start-up the project. Make sure that the references to the libxml2 and Tcl external frameworks are correct. Select the 'Make' target and build. This builds everything. The result is six Mac OS X Frameworks; three "normal" and three "embedded". The embedded frameworks will be in the embedded subdirectory of the Build Products directory. Copy tclxml.framework, TclXMLlibxml2.framework and expat.framework to any of the usual places for frameworks (~/Library/Frameworks, /Library/Frameworks, etc).

For earlier version of OS X using Project Builder, you will have to retrieve a previous version of the Project Builder files from the CVS repository.

Macintosh OS 9 (or earlier)

The installer script doesn't work properly on Macintosh yet, and the binary parsers have not been ported yet (sorry).

Since the distribution has an all-Tcl parser, you can get a working parser fairly easily.

Copy the library folder to the Tcl library folder in the Extensions folder of the System folder.

Rename the library folder to tclxml3.0. This isn't strictly necessary, but it helps keep things organised.

You're on your own as far as the demos, programs and documentation go.

Usage

See the website for links to tutorials and the reference manual.

In the meantime, here's a quick tutorial:

package require xml

set parser [xml::parser]
$parser configure -elementstartcommand EStart \
    -characterdatacommand PCData

proc EStart {tag attlist args} {
    array set attr $attlist
    puts "Element \"$tag\" started with [array size attr] attributes"
}

proc PCData text {
    incr ::count [string length $text]
}

set count 0
$parser parse [read stdin]

puts "The document contained $count characters"
exit 0

XPath

In addition to XML parsing packages, TclXML also provides a package for parsing XPath location paths. The XPath package only parsing the path's syntax, it does interpret the path. See TclDOM for a package that will interpret XPath location paths.

Note

This package is in its infancy, and does not support the full range of XPath features. Only a very limited subset of location paths are supported, of the form "/simple/example[2]". Paths within predicates will definitely fail.

To use the XPath package:

package require xpath

To parse a location path:

xpath::split {/simple/example}

This returns a Tcl list, each element of which is a three element sublist: {axis node-test {?predicate ...?}}.