education aggregation

gistfile1.py

def _education_facet(company_id, text=None):
    return {
        "size": 0,
        "aggs": {
            "company_filtered": {
                "filter": {
                    "and": [
                        {
                            {
                                "type": {
                                    "value": "facebook_profile"
                                }
                            }
                        },
                        {
                            "has_parent": {
                                "type": "contact",
                                "filter": {
                                    "term": {
                                        "company_id.na": str(company_id)
                                    }
                                }
                            }
                        },
                        {
                            "query": {
                                "nested": {
                                    "query": {
                                        "match_phrase_prefix": {
                                            "education.name": text
                                        }
                                    },
                                    "path": "facebook_profile.education"
                                }
                            }
                        }
                    ]
                },
                "aggs": {
                    "educations": {
                        "scripted_metric": {
                            "init_script": "_agg['education_to_parents'] = [:];",
                            "map_script":
                                "def pid = doc['_parent'].join('');"
                                "def map = _agg.education_to_parents;"
                                "def text = '%s';" % (text if text else '') +
                                "def edu_getter = { "
                                "   e -> "
                                "   def education = e['name'].toLowerCase();            "
                                "   def include = false;"
                                "   if (text) {"
                                "       for (p in education.split(' ')*.trim()) {"
                                "           if (p.startsWith(text)) {"
                                "               include = true;"
                                "           }"
                                "       }"
                                "   } else {"
                                "       include = true;"
                                "   };"
                                "   if (include) {"
                                "       if (map[education]) {"
                                "           map[education].add(pid)"
                                "       } "
                                "       else {"
                                "           map[education] = [pid]    "
                                "       }        "
                                "   }"
                                "};"
                                "if (_source.containsKey('education')) {    "
                                "    for(e in _source['education']) { "
                                "        edu_getter(e) "
                                "    }        "
                                "};"
                                "if (_source.containsKey('organizations')) {    "
                                "    for(e in _source['organizations']) {        "
                                "        if (e['otype'] == 'school') {            "
                                "           edu_getter(e) "
                                "        }    "
                                "    }"
                                "};",
                            "combine_script":
                                "def map = [:]; "
                                "def top = [:]; "
                                "_agg.education_to_parents.each {"
                                "   e, parents -> map[e] = (parents as Set).size() "
                                "}; "
                                "map = map.sort {-it.value }; "
                                "map.eachWithIndex { val, idx ->    if (idx<100) {top[val.key]=val.value  }}; "
                                "return top",
                            "reduce_script":
                                "def top = [:]; "
                                "for (shard_top in _aggs) {"
                                "   shard_top.each {e, count -> top[e] = count}"
                                "}; return top"
                        }
                    }
                }
            }
        }
    }