Re: [Wikitech-l] Parsing new XML dumps into the db?

John Grohol wrote: Well, you can do anything you like with it. ;) The XML source should be pretty self-explanatory about its contents; you should also find a description of the format at http://meta.wikimedia.org/wiki/Help:Export If you want to import it into a MediaWiki 1.5 instance, the currently canonical way would be: * Install a fresh wiki running MediaWiki 1.5rc4. * Set up AdminSettings.php if necessary; command-line access is required. * Pipe the file to importDump.php: gzip -dc pages_full.xml.gz | php importDump.php I've been working on a utility for filtering the XML backup dump files; it'll be used as part of our backup procedure for creating filtered dumps as well as the complete ones. For extra fun I've included an output mode to generate SQL statements in 1.4 or 1.5 schema which could be then read into an already-initialized database. At the moment this is command-line only, and the SQL mode is basically completely untested. In the future it should be better tested and I may add a GUI front-end and a helper for creating and initializing a database. The tool is written in C# and runs on the cross-platform Mono runtime, and should also run on .NET 1.1 if you're a Microsoft person. As it's new I don't have a binary release yet, but source and a makefile for compiling with Mono are in our CVS repository: http://cvs.sourceforge.net/viewcvs.py/wikipedia/mwdumper/ You might run it something like this: gzip -dc pages_full.xml.gz | mono mwdumper.exe --format=sql:1.4 > curold.sql Then import the SQL into an installed MediaWiki 1.4 database. (The cur and old tables should be _empty_ at this point; if using the installer there will be data in there, so delete it first.) -- brion vibber (brion @ pobox.com)