= xamplr-pp-gem == INTRODUCTION This is the Ruby version of the xampl pull parser, xamplr-pp. The class name is xampl_PP. Both pull parsers and SAX parsers are stream based -- they parse the XML file bit by bit informing its client of interesting events as they are encountered. The whole XML document is not required to be in memory. The significant difference between pull parsers and SAX parsers is in where the 'main loop' is located: in the client for pull parsers, in the parser for SAX parsers. Clients call a method of the pull parser to get the next event. SAX parsers call methods of the client to notify it of events (so these are 'push parsers'). You can pass the pull parser around as an argument. As a way of illustrating the use of xamplr-pp a reasonably usable SAX-like parser (SAXish) is implemented as well. There is a saxdemo.rb file provided that provides statistics on the parsed XML file. It can be run using 'make sax' (look in the Makefile to see how it is used) == STATUS xamplr-pp has been in daily use in a SaaS content management system since about 2004. It is an integral component of xamplr. Unfortunately documentation is sparse. xamplr-pp has passed all unit tests using Ruby 1.6.7, 1.8.6, 1.8.7 on Linux and OS X. It has not even been tried on Ruby 1.9.1 at this point. The unit tests require Lapidary (remember that?). Run these by typing 'make ut' on the command line in the installation directory. xamplr-pp is not a validating parser, in fact, it doesn't enforce some wellformedness rules. DOCTYPE definitions are passed intact to the client program. No attempt is made to extract any kind of entity definitions from the DOCTYPE. Clever use of the DOCTYPE event and the resolver method can relieve this problem. I have some serious reservations about the way that xamplr-pp handles multi-byte characters, but it definitely works well with UTF8. Some attention has been paid to performance. It is reasonably quick, certainly on the fast side for pure ruby parsers. == DOCUMENTATION There isn't much. Fortunately, the API to xamplr-pp is very small and quite easy to understand. It turns out to be quite easy to build a SAX-like parser from a push parser. It also turns out that doing this pretty much uses the entire xamplr-pp api. So I've implemented a SAX-like parser (SAXish) and annotated it. This is pretty much all the documentation in this release. SAXish is, in itself, a pretty usable SAX-like parser (of course the limits to validation and wellformedness checking mentioned above apply). == CONFORMANCE Well, this is an interesting question. Right now, it doesn't do too well, but it does this in a 'good' way, or at least a manageable way. On the Oasis conformance tests xamplr-pp 435 tests pass that should pass, 188 fail that should fail, 3 failed that should pass, and 1188 passed that should fail. Generally speaking, xamplr-pp is 'permissive'. However to do this I had to *cheat*. I defined a resolver that returns a string rather than nil, this allows xamplr-pp to pretend that it is dealing with entities defined in the DOCTYPE. In other words, for you to achieve this level of conformance you'll need to supply an entity resolver. The three that failed that should have passed all involve apparently legal DOCTYPE declarations that xamplr-pp does not recognise as legal. It is unlikely that I do anything about these. The 1188 that passed that should not appear to be due to the non-enforcement of well-formedness rules. I have not looked at them all, I assure you of that, but it seems that this most involve entity definitions. Well, every one that I looked at did, can't say for what I didn't look at. == LICENCE xamplr-pp is licensed under the LGPLv3 (see LICENSE/COPYING) == CONTACT INFORMATION My email is hutch@xampl.com, feel free to contact me there, or you can use the facilities provided by Sourceforge. == Copyright Copyright (c) 2009 Bob Hutchison. See LICENSE for details.