i had a similar question combined with harvesting metadata and could solve it. I build a construction with Python like this:
1. Search and Export all Database-FeatureClasses, Shapefiles and Filegeodatabases and its Metadata as csv
2. List all Layers and the Datasources that are used in mxd-Files as csv
3. List all ArcGIS-Server Services and its mxd-Files as csv
4. Join the Tables 1. & 2. via FeatureClass-Name and join with 3. via Service-mxd-path
As a result you have a full Data inventory of your Geodata, Metadata and the used Services.
Very useful to search and filter your Data.