Hi Marc,
I would be the one configuring SOLR for @jfelkins . I have a customer configured similar to what you describe. They generate PDF attachments of MicroStation drawings and SOLR Index Search indexes the PDF attachments.
I'm not 100% confident in the behavior. Through limited experience, this is what I think is happening. Any clarification or corrections would be appreciated.
SOLR is indexing the object if the primary or secondary content file's extension is identified as indexable in the Data Formats (MIME types). SOLR doesn't actually care what the object is (CAD Document or Document).
The behavior needs to be validated. Before adding/enabling the drawing's MIME type for indexing, add the PDF of the drawing as an Attachment and index the object.
- If content sensitive search returns the object based on some content in the PDF, then no need to mess with the MIME types.
- If the object is not returned, then only the Primary content file's extension is being considered for indexing and we need to add/enable the Primary content's MIME type for indexing.
Note: When a content file is sent to SOLR Server for indexing, SOLR Server needs a Rosetta Stone to understand how to parse and index that particular file type. AutoCAD is understood, but most CAD packages are not. In this scenario, SOLR Server will return empty results on the the primary content file and move on to the PDF attachment. Since the PDF has the relevant information we want indexed, this is good enough for our needs.
Once indexed, Content Sensitive search in Windchill only returns the CAD Document object and doesn't distinguish between the primary and secondary content that was indexed.
The setup...
1. SOLR does not index Representation content.
According to PTC, it only indexes the primary (drawing) and secondary (attachment) content.
https://support.ptc.com/help/wnc/r12.1.1.0/en/index.html#page/Windchill_Help_Center/WCSysAdminIndexSearch/WCSysAdminIndexSearchIndexableTypes.html
We can configure Windchill to upload the Additional File content PDF generated by WVS as an as an Attachment.
https://www.ptc.com/en/support/article/CS348848
2. The PDF Attachment must be indexable.
As @buenosroas mentioned, a PDF of an image doesn't include any indexable text.
Note: In Creo Parametric you can stroke out text to line art to make the text on the PDF look correct (intf2d_out_pdf_stroke_text_font). The native text is still embedded in the PDF, though not displayed. Indexing will SOLR Indexing can't index the line art, but it will index the hidden text.
3. (Optional, based on testing above) SOLR must be told the MIME type should be indexed.
Get the current list of MIME types from your Windchill instance. I don't trust that the Data Formats list hasn't been altered previously.
java wt.content.DataFormatUtil -list>D:\PTC\dataFormats.txt
If the primary content file must be flagged for indexing, we need to add the appropriate CAD drawing file extensions (IDW, SLDDRW, etc.) before those CAD Document drawings will be indexed.
https://www.ptc.com/en/support/article/CS32720
For example:
The system where I have this configuration running indexes Document objects that represented legacy CAD formats (e.g. MicroStation primary content and PDF attachments). I had to add the MicroStation file extension to the MIME formats before those PDFs were indexed.
https://support.ptc.com/help/wnc/r12.0.2.0/en/index.html#page/Windchill_Help_Center%2FWCCG_Serv_DataFormats_AddUpdateDataFormats.html%23
formatName = Microstation
mimeType = image/vnd.dgn
description = DGN file
indexable = true
icon = netmarkets/images/microstation_3.png
extensions = DGN
SOLR Server probably still cannot index the native drawing format (SLDDRW, IDW, etc.). If interested, you can confirm what SOLR can index.
https://www.ptc.com/en/support/article/CS101260
Please let us know if step 3 is necessary. Hope this helps.
