def build_query_phrase_as_regex(phrase):
# Split into sub_phrases (remove whitespace, punctuation etc.)
sub_phrases = re.split(RE_SPLIT_WORD_UNICODE, phrase)
# If there are no sub_phrases - ignore this phrase
search_str_Q = Q()
if sub_phrases:
# Build regex - all sub_phrases we've found separated by 'non-word' characters
# and also allow VAV and/or HEI in front of each word.
# NOTE: regex syntax is DB dependent - this works on postgres
re_words = [u'\u05D5?\u05D4?' + word for word in sub_phrases]
regex = PG_RE_PHRASE_START + PG_RE_NON_WORD_CHARS.join(
re_words) + PG_RE_PHRASE_END
search_str_Q = Q(content__iregex=regex)
return search_str_Q
def build_query_phrase_as_regex(phrase):
# Split into sub_phrases (remove whitespace, punctuation etc.)
sub_phrases = re.split(RE_SPLIT_WORD_UNICODE, phrase)
# If there are no sub_phrases - ignore this phrase
search_str_Q = Q()
if sub_phrases:
# Build regex - all sub_phrases we've found separated by 'non-word' characters
# and also allow VAV and/or HEI in front of each word.
# NOTE: regex syntax is DB dependent - this works on postgres
re_words = [u'\u05D5?\u05D4?' + word for word in sub_phrases]
regex = PG_RE_PHRASE_START + PG_RE_NON_WORD_CHARS.join(
re_words) + PG_RE_PHRASE_END
search_str_Q = Q(content__iregex=regex)
return search_str_Q
def parse_to_q_object(get_params, params_dict):
member_query = MEMBER_MODEL.objects.filter(id__in=params_dict['members_ids'])
fixed_member_ids = [member.id for member in member_query]
if IS_ELECTIONS_MODE:
feeds = Facebook_Feed.objects.filter(persona__alt_object_id__in=fixed_member_ids)
else:
feeds = Facebook_Feed.objects.filter(persona__object_id__in=fixed_member_ids)
# all members asked for (through member search of party search), with OR between them.
members_OR_parties_Q = Q()
if feeds:
members_OR_parties_Q = Q(feed__id__in=[x.id for x in feeds])
# tags - search for all tags specified by their id
tags_Q = Q()
if params_dict['tags_ids']:
# | Q(tags__synonyms__proper_form_of_tag__id=tag_id)
tag_bundle_ids = [tag.id for tag in Tag.objects.filter_bundle(id__in=params_dict['tags_ids'])]
tags_to_queries = [Q(tags__id=tag_id) for tag_id in tag_bundle_ids]
if settings.DEBUG:
print 'tags_to_queries:', len(tags_to_queries)
for query_for_single_tag in tags_to_queries:
# tags_Q is empty for the first iteration
tags_Q = join_queries(query_for_single_tag, tags_Q, or_)
if settings.DEBUG:
print 'tags_Q:', tags_Q
# keywords - searched both in content and in tags of posts.
search_str_Q = Q()
# If regexes cause security / performance problem - switch this flag
# to False to use a (not as good) text search instead
use_regex = True
for phrase in params_dict['phrases']:
if use_regex:
# Split into words (remove whitespace, punctuation etc.)
words = re.split(RE_SPLIT_WORD_UNICODE, phrase)
# If there are no words - ignore this phrase
if words:
# Build regex - all words we've found separated by 'non-word' characters
# and also allow VAV and/or HEI in front of each word.
# NOTE: regex syntax is DB dependent - this works on postgres
re_words = [u'\u05D5?\u05D4?' + word for word in words]
regex = PG_RE_PHRASE_START + PG_RE_NON_WORD_CHARS.join(re_words) + PG_RE_PHRASE_END
search_str_Q = join_queries(Q(content__iregex=regex), search_str_Q, or_)
else:
# Fallback code to use if we want to disable regex-based search
search_str_Q = join_queries(Q(content__icontains=phrase), search_str_Q, or_)
search_str_Q = Q(tags__name__contains=phrase) | search_str_Q
# tags query and keyword query concatenated. Logic is set according to request input
request_operator = get_params.get('tags_and_search_str_operator', DEFAULT_OPERATOR)
if settings.DEBUG:
print 'selected_operator:', request_operator
selected_operator = and_ if request_operator == 'and_operator' else or_
search_str_with_tags_Q = join_queries(tags_Q, search_str_Q, selected_operator)
if settings.DEBUG:
print 'search_str_with_tags_Q:', search_str_with_tags_Q
print '\n'
query_Q = join_queries(members_OR_parties_Q, search_str_with_tags_Q, and_)
if params_dict['excluded']:
excluded_query = Q(status_id__in=params_dict['excluded'])
excluded_query.negate()
query_Q = join_queries(query_Q, excluded_query, and_)
if settings.DEBUG:
from qserializer import QSerializer
qser = QSerializer()
print 'query to be executed:'
from pprint import pprint
pprint(qser.serialize(query_Q.clone()))
return query_Q
def parse_to_q_object(get_params, params_dict):
member_query = MEMBER_MODEL.objects.filter(
id__in=params_dict['members_ids'])
fixed_member_ids = [member.id for member in member_query]
if IS_ELECTIONS_MODE:
feeds = Facebook_Feed.objects.filter(
persona__alt_object_id__in=fixed_member_ids)
else:
feeds = Facebook_Feed.objects.filter(
persona__object_id__in=fixed_member_ids)
# all members asked for (through member search of party search), with OR between them.
members_OR_parties_Q = Q()
if feeds:
members_OR_parties_Q = Q(feed__id__in=[x.id for x in feeds])
# tags - search for all tags specified by their id
tags_Q = Q()
if params_dict['tags_ids']:
# | Q(tags__synonyms__proper_form_of_tag__id=tag_id)
tag_bundle_ids = [
tag.id for tag in Tag.objects.filter_bundle(
id__in=params_dict['tags_ids'])
]
tags_to_queries = [Q(tags__id=tag_id) for tag_id in tag_bundle_ids]
if settings.DEBUG:
print 'tags_to_queries:', len(tags_to_queries)
for query_for_single_tag in tags_to_queries:
# tags_Q is empty for the first iteration
tags_Q = join_queries(query_for_single_tag, tags_Q, or_)
if settings.DEBUG:
print 'tags_Q:', tags_Q
# keywords - searched both in content and in tags of posts.
search_str_Q = Q()
# If regexes cause security / performance problem - switch this flag
# to False to use a (not as good) text search instead
use_regex = True
for phrase in params_dict['phrases']:
if use_regex:
# Split into words (remove whitespace, punctuation etc.)
words = re.split(RE_SPLIT_WORD_UNICODE, phrase)
# If there are no words - ignore this phrase
if words:
# Build regex - all words we've found separated by 'non-word' characters
# and also allow VAV and/or HEI in front of each word.
# NOTE: regex syntax is DB dependent - this works on postgres
re_words = [u'\u05D5?\u05D4?' + word for word in words]
regex = PG_RE_PHRASE_START + PG_RE_NON_WORD_CHARS.join(
re_words) + PG_RE_PHRASE_END
search_str_Q = join_queries(Q(content__iregex=regex),
search_str_Q, or_)
else:
# Fallback code to use if we want to disable regex-based search
search_str_Q = join_queries(Q(content__icontains=phrase),
search_str_Q, or_)
search_str_Q = Q(tags__name__contains=phrase) | search_str_Q
# tags query and keyword query concatenated. Logic is set according to request input
request_operator = get_params.get('tags_and_search_str_operator',
DEFAULT_OPERATOR)
if settings.DEBUG:
print 'selected_operator:', request_operator
selected_operator = and_ if request_operator == 'and_operator' else or_
search_str_with_tags_Q = join_queries(tags_Q, search_str_Q,
selected_operator)
if settings.DEBUG:
print 'search_str_with_tags_Q:', search_str_with_tags_Q
print '\n'
query_Q = join_queries(members_OR_parties_Q, search_str_with_tags_Q, and_)
if params_dict['excluded']:
excluded_query = Q(status_id__in=params_dict['excluded'])
excluded_query.negate()
query_Q = join_queries(query_Q, excluded_query, and_)
if settings.DEBUG:
from qserializer import QSerializer
qser = QSerializer()
print 'query to be executed:'
from pprint import pprint
pprint(qser.serialize(query_Q.clone()))
return query_Q
