pythonsolrpysolr

Process facet.pivot Data in Solr/Python


I am querying a Solr core with this url:

select?q=*:*&rows=0&facet=on&facet.limit=-1&facet.mincount=0&facet.pivot=brand,series,sub_series

Using the Facet Pivot feature

Partial JSON response, notice how the fields are nested appropriately:

"facet_pivot":{
  "brand,series,sub_series":[{
      "field":"brand",
      "value":"A. Lange & Sohne",
      "count":69,
      "pivot":[{
          "field":"series",
          "value":"1815 Manual Wind",
          "count":1},
        {
          "field":"series",
          "value":"1815 Up Down",
          "count":1},
        {
          "field":"series",
          "value":"Datograph",
          "count":3,
          "pivot":[{
              "field":"sub_series",
              "value":"Perpetual",
              "count":2},
            {
              "field":"sub_series",
              "value":"Up Down",
              "count":1}]},

Now I replicate the same query with pySolr:

def nested_navbar():
    result = solr.search('*:*', **{
        'rows': '0',
        'facet': 'on',
        'facet.limit': '-1',
        'facet.mincount': '0',
        'facet.pivot': ['brand_exact', 'series_exact', 'sub_series_exact']
        })

    result = result.facets['facet_pivot']
    return result

For some reason, when I print the data (pdb, or Django templates), the dictionary in the result has 3 different arrays for brands, series and sub_series, without the original nesting. In other words, the pivot field is gone.


Solution

  • Per matslindh's advice, I setup logging for pySolr:

    1. Add export DEBUG_PYSOLR='true' to terminal.
    2. Run server and inspect log.

    Here is the link generated by pySolr:

    /?q=%2A%3A%2A&rows=0&facet=on&facet.limit=-1&facet.mincount=0&facet.pivot=brand&facet.pivot=series&facet.pivot=sub_series&wt=json
    

    The issue is that pySolr was creating a new facet.pivot field for each item in the array ('brand', 'series' and 'sub_series').

    My new code, which fixes the issue and generates a single facet.pivot field:

    'facet.pivot': 'brand,series,sub_series'