ppa/archive/solr.py
import logging
from parasolr.django import AliasedSolrQuerySet, SolrQuerySet
logger = logging.getLogger(__name__)
class ArchiveSearchQuerySet(AliasedSolrQuerySet):
# search title query field syntax
# (query field configured in solr config; searches title & subtitle with
# boosting)
_title_search = (
"{!type=edismax qf=$search_title_qf " + "pf=$search_title_pf v=$title_query}"
)
_keyword_search = "{!type=edismax qf=$keyword_qf pf=$keyword_pf v=$keyword_query}"
# minimal set of fields to be returned from Solr for search page
return_fields = [
"id",
"author",
"pubdate",
"publisher",
"enumcron",
"order",
"source_id",
"label",
"title",
"subtitle",
"score",
"pub_date",
"collections",
"source_t",
"image_id_s",
"first_page_s",
"source_url",
"work_type_s",
"book_journal_s",
"group_id_s",
"cluster_id_s",
]
# aliases for any fields we want to rename for search and display
# (must also be included in return_fields list)
aliases = {
"source_t": "source",
"image_id_s": "image_id",
"first_page_s": "first_page",
"work_type_s": "work_type",
"book_journal_s": "book_journal",
"group_id_s": "group_id",
"cluster_id_s": "cluster_id",
}
keyword_query = None
within_cluster_id = None
def __init__(self, solr=None):
# field aliases: keys return the fields that will be returned
# from Solr for search page; values provide an aliased name if
# it should be different than solr index field.
# use alias if one is set, otherwise use field name
self.field_aliases = {
self.aliases.get(key, key): key for key in self.return_fields
}
self._workq = SolrQuerySet()
super().__init__(solr=solr)
def work_filter(self, *args, **kwargs):
"""Add filters to the work query"""
# filter out empty values to simplify view logic
# for checking whether query terms are present
kwargs = dict(
(opt, val) for opt, val in kwargs.items() if val not in [None, ""]
)
if args or kwargs:
self._workq = self._workq.filter(*args, **kwargs)
def work_title_search(self, title_query):
"""search works by title"""
if not title_query:
return
# include the edismax title search query in the filters
self.work_filter(self._title_search)
# add the actual query content as a query parameter
# FIXME: maybe all methods should return a new version for consistency
self.raw_params.update(title_query=title_query)
# return self.raw_query_parameters(title_query=title_query)
def keyword_search(self, query):
"""add keyword search"""
# *store* that there is a keyword present but don't do anything
# with it yet
self.keyword_query = query
def _clone(self):
# preserve local fields when cloning
qs_copy = super()._clone()
qs_copy.keyword_query = self.keyword_query
qs_copy.within_cluster_id = self.within_cluster_id
qs_copy._workq = self._workq
return qs_copy
def within_cluster(self, cluster_id):
"""Search within a group of reprints/editions"""
# filter both pages and works by cluster id
qs_copy = self.filter(cluster_id_s=cluster_id)
qs_copy.work_filter(cluster_id_s=cluster_id)
# store the cluster id since it impacts expand/collapse behavior
qs_copy.within_cluster_id = cluster_id
return qs_copy
def query_opts(self):
"""Extend default query options method to combine work and keyword
search options based on what filters are present."""
# create a queryset copy to update
qs_copy = self.all()
# for main archive search, by default we collapse on
# cluster id to collect all reprints/editions and their pages;
# when searching within a cluster, collapse on group id
collapse_on = "group_id_s" if self.within_cluster_id else "cluster_id_s"
# NOTE: Role of order here in separating works from pages (works < pages)
# may need to be revisited eventually.
collapse_filter = '{!collapse field=%s sort="order asc"}' % collapse_on
# We can apply collapse here since we need it for default search
# cluster id corresponds to index id for works not in a cluster,
# so collapsing by cluster id still includes works with no cluster id
qs_copy = qs_copy.filter(collapse_filter)
# if there is no keyword search present, only works should
# be returned; add item type filter and use filters from work queryset
if not self.keyword_query:
self.work_filter(item_type="work")
qs_copy.filter_qs.extend(self._workq.filter_qs)
# use set to ensure we don't duplicate a filter
qs_copy.filter_qs = list(set(qs_copy.filter_qs))
return qs_copy._base_query_opts()
# when there is a keyword query, add it & combine with any work filters
# combine all work filter queries into a single query
# search across keyword qf fields OR find works with pages that match
keyword_query = (
"((%s) OR ({!join from=group_id_s to=id v=$content_query}))"
% self._keyword_search
)
# by default, set combined query to keyword query (= no work filters)
combined_query = keyword_query
# if there are work filters, combine them with keyword
work_query = ""
if self._workq.filter_qs:
# convert filter queries to a single ANDed search query
work_query = "(%s)" % " AND ".join(self._workq.filter_qs)
# find works based on filter query but also restrict pages to those
# that match works with these filters
combined_query = (
"(%s) AND (%s OR {!join from=id to=group_id_s v=$work_query})"
% (keyword_query, work_query)
)
# pass combined workfilter query as a raw query parameter
qs_copy = qs_copy.raw_query_parameters(work_query=work_query)
content_query = "content:(%s)" % self.keyword_query
qs_copy = qs_copy.search(combined_query).raw_query_parameters(
content_query=content_query,
keyword_query=self.keyword_query,
work_query=work_query,
)
return qs_copy._base_query_opts()
def _base_query_opts(self):
# provide access to regular query opts logic, bypassing keyword/join
return super().query_opts()
class PageSearchQuerySet(AliasedSolrQuerySet):
# aliases for any fields we want to rename for search and display
# includes non-renamed fields to push them into the return
field_aliases = {
"id": "id",
"score": "score",
"order": "order",
"title": "title",
"label": "label",
"source_id": "source_id",
"image_id": "image_id_s",
"group_id": "group_id_s",
"cluster_id": "cluster_id_s",
}