Search this blog

Sunday, 7 August 2011

Finding processes in the repository

I have a lot of processes and it's sometimes hard to remember the name of one containing something useful. A metadata view and search capability of the repository would be very useful.

While that is being invented, here's a process that scans through a repository and determines the name and location of the processes it finds. It also counts the number of operators used to give a sense of how big the process is. The process recursively scans files with extension ".rmp" from a starting folder. Various document operators including "extract information" and "cut documents" using Xpath are used to process the files and extract information from them. The "aggregate" operator is used to count operators.

One day I will make it into a report that can be viewed using RapidAnalytics.

Other data is also extracted and this could be used if desired.

The default location is the default Windows location of the samples processes; change this to the folder where your repository is.

Some gymnastics were required to get the loop files operator to work correctly. This seems to present zero sized directories when iterating so I was forced to use the "branch" operator to eliminate these. Pointing the loop files operator to the C:\ drive usually results in an error that I haven't got to the bottom of yet.

No comments:

Post a Comment