The Functional XML Parser

What's New?

February 18, 2000: Release of fxgrep, the Functional XML Querying Tool.

February 18, 2000: Release of version 1.4 with:

• Improved handling of wrong end-tags/invalid character data
• Slight change in the programming interface.

Contents

	About fxp
	fxp Versions and Changes
	Where to get fxp
	How to install fxp
	Example Applications: fxp, fxcanon, fxcopy, fxesis, and fxviz.

What is fxp?

	fxp, the pure parser. It parses a document and finds well-formedness errors, validity errors and other problems;
	fxcanon produces an equivalent canonical XML document. Canonical XML was invented by James Clark for testing XML parsers. It contains only the information a processor is required to pass to the application;
	fxcopy reproduces the document parsed by fxp. The copy can be generated in a different encoding than the input, and can be normalized in different ways concerning, e.g., expansion of entity references;
	fxesis adds a backend to fxp, producing an output similar to nsgmls's ESIS (Element Structure Information Set) output;
	fxviz is an XML tree visualizer. It produces a graph description suitable as input to Georg Sander's vcg.

Downloading fxp

Installation instructions

These are the steps for installing fxp under Unix:

1. Download the latest version of fxp;
2. Unpack the sources, and change to the fxp directory, e.g.:

       gunzip -c fxp-1.4.tar.gz | tar xf -
       cd fxp-1.4

3. Read the COPYRIGHT;
4. Edit the Makefile according to your needs. Probably you will only have to change the following:
   • INSTALL_PROGS is the list of programs to be installed. fxlib is only required if you want to develop applications with fxp.
   • FXP_BINDIR is where the executables are installed;
   • FXP_LIBDIR is where other files needed by fxp - the heap images and the library -- are installed;
   • SML_BINDIR is the directory where the SML executables are found. It must contain the .arch-n-opsys script from the SML-NJ distribution, so make sure that this is where SML-NJ is physically installed;
   • SML_EXEC is the name of the SML executable. This is the program that is called for generating the heap image and at execution of fxp. If sml will be in your PATH at execution time, you don't need the full path here.
   • SML_MAKEDEF is for defining the make command in SML. After version 110.0.3, SML-NJ changed the type of CM.make'. Therefore it is wrapped into the make defined by this variable. For working versions of SML-NJ, use the second variant of this definition.
5. Edit the file src/config.sml according to your needs. Currently only a single value can be configured here:
   • retrieveCommand is the command to be used by fxp for retrieving a remote URI from the internet and storing it in a temporary file on the local file system. It is a string value and should contain the strings %1 and %2, where
     • %1 is replaced by the URI;
     • %2 is replaced by the local filename.
     It is recommended that the command exits with failure in case the URI cannot be retrieved. If the command generates a HTML error message instead (like, e.g., "lynx -source %1 > %2"), this HTML file is considered to be XML and will probably cause a mess of parsing errors. If you don't need URI retrieval, use "exit 1" which always fails on Unix. Sensible values are, e.g:
     • "wget -qO %2 %1"
     • "got_it -o %2 %1"
     • "urlget -s -o %2 %1"
6. Compile fxp by typing make;
7. Install fxp by typing make install.
8. If you want to use fxviz, you should also install vcg.

fxp's Programming Interface