JPS motors XML parser (September 2002)

Description

Takes XML motors output from Futureproof advertising system based at The News offices and transforms into pipe delimited format required for the www.motorstoday.co.uk web site built and hosted by Newmedia in Peterborough. Web based with front end for repairing parsing mistakes and emailing data files.

Tools

SQL Server
Coldfusion
Coldfusion Studio
Javascript

Technical

XML input is parsed via Java into a coldfusion data structure and held in memory. The pipe delimited output format requires seperate fields for car make and model. However, adverts are input as one large chunk of text. The parser carries out several passes through each advert to match up a manufacturer and then find a matching model. In some cases, a car manufacturer cannot be determined or the model cannot be successfully matched (or sometimes more than one model is found). In these cases, the user can manually pick make and model and submit the adverts. The parser also takes account of 'dealer adverts' where there will be many cars listed in one advert - the advert is split into individual cars and then each is worked on. Outputs in pipe delimited format are organised via publication codes, so that content can be matched to the source publication on the web site. The failure rate of the parser is extremely low and requires very little manual intervention.