Abstract: . . .  TagSoup: A SAX parser in Java for nasty, ugly HTML John Cowan ( cowan@ccil.org ) Page 2 Copyright 2002 John Cowan; licensed under the GNU GPL 35 Copyright This presentation is: • Copyright © 2004 John Cowan • . . .
. . .  confused if an attribute value does contain whitespace but has no quotation marks Page 31 A Few Other Points Page 32 Copyright 2002 John Cowan; licensed under the GNU GPL 35 Roll Your Own SAX2 properties let you: • Specify your own scanner object (if your surface syntax is not HTML ) • Specify your own schema object (if your elements, content models, attributes, and entities are not HTML ) • Specify your own auto-detector object (if you know how to . . .
. . .  Entities must resolve to a single character • You can clone the standard HTML schema, or start from an empty schema and build up Page 34 Copyright 2002 John Cowan; licensed under the GNU GPL 35 How Big Is TagSoup? • 7 classes: Parser, Schema, HTMLSchema, HTMLScanner, Element, ElementType, a private copy of AttributesImpl • 4 interfaces: Scanner, ScanHandler, AutoDetector, HTMLModels • 3 debug classes: PYXScanner, PYXWriter, a modified version of XMLWriter • About 4600 lines of Java 2000 . . .
. . .  35 State Table Markup Language • STML is another special-purpose language for describing arbitrary state machines • TagSoup's lexical analyzer is written in STML • A RELAX NG schema for TSSL is provided • An XSLT script converts the STML into tables for the Java HTMLScanner class Page 37 Copyright 2002 John Cowan; licensed under the GNU GPL 35 PYX Format • PYX format is a linearized syntax for XML, almost the same as SGML ESIS format • TagSoup provides support for PYX format on . . .
. . .  parsed Page 39 Improvements • Suggestions and patches are welcome • Subscribe to tagsoup-friends@yahoogroups.com Copyright 2002 John Cowan; licensed under the GNU GPL 35 Page 40 More Information http://tagsoup.info  . . .
--3000,5,300,2812,15676