If you have control over a collection's data from the beginning and will be using a CSV, it is highly recommended that you name the CSV columns the same as the Solr schema's fields, underscores and all. Then all you need to do is run the post.rb script and you'll be home free!
Understandably, not all projects can ingest their CSVs into Solr without some tinkering, even if it's just that dates need to be reformatted.
Check out the example collection's CSV overrides in collections/example/scripts/overrides/file_csv.rb.
class FileCsv
def row_to_solr(doc, headers, row)
doc.add_child("<field name='id'>#{row['id']}</field>") if row['id']
doc.add_child("<field name='title'>#{row['title']}</field>") if row['title']
doc.add_child("<field name='category'>#{row['category']}</field>") if row['category']
doc.add_child("<field
[etc]
return doc
end
end
This method is looking at one single row of the CSV at a time. There are three arguments available in row_to_solr:
- doc (the xml document you are building for Solr)
- headers (an array of the column headers)
- row (a single row of the csv)
You may iterate through by header, or on an index of the CSV fields, but it's recommended that you use row['column_name'] to access values, for clarity.
Variables available from FileType are accessible from FileCsv, like @options and @file_location, if you should need to manipulate an item with more finesse.