def validate(self):
for key, value in self.items():
if key in self.BOOLEAN_OPTIONS:
try:
value = asbool(value)
except ValueError:
raise SearchQueryError('Value for search option %r must be True or False (1 or 0) but received %r' % (key, value))
elif key in self.INTEGER_OPTIONS:
try:
value = int(value)
except ValueError:
raise SearchQueryError('Value for search option %r must be an integer but received %r' % (key, value))
elif key in self.UNSUPPORTED_OPTIONS:
raise SearchQueryError('Search option %r is not supported' % key)
self[key] = value
def convert_legacy_parameters_to_solr(legacy_params):
'''API v1 and v2 allowed search params that the SOLR syntax does not
support, so use this function to convert those to SOLR syntax.
See tests for examples.
raises SearchQueryError on invalid params.
'''
options = QueryOptions(**legacy_params)
options.validate()
solr_params = legacy_params.copy()
solr_q_list = []
if solr_params.get('q'):
solr_q_list.append(solr_params['q'].replace('+', ' '))
non_solr_params = set(legacy_params.keys()) - VALID_SOLR_PARAMETERS
for search_key in non_solr_params:
value_obj = legacy_params[search_key]
value = value_obj.replace('+', ' ') if isinstance(
value_obj, basestring) else value_obj
if search_key == 'all_fields':
if value:
solr_params['fl'] = '*'
elif search_key == 'offset':
solr_params['start'] = value
elif search_key == 'limit':
solr_params['rows'] = value
elif search_key == 'order_by':
solr_params['sort'] = '%s asc' % value
elif search_key == 'tags':
if isinstance(value_obj, list):
tag_list = value_obj
elif isinstance(value_obj, basestring):
tag_list = [value_obj]
else:
raise SearchQueryError(
'Was expecting either a string or JSON list for the tags parameter: %r'
% value)
solr_q_list.extend([
'tags:"%s"' % escape_legacy_argument(tag) for tag in tag_list
])
else:
if len(value.strip()):
value = escape_legacy_argument(value)
if ' ' in value:
value = '"%s"' % value
solr_q_list.append('%s:%s' % (search_key, value))
del solr_params[search_key]
solr_params['q'] = ' '.join(solr_q_list)
if non_solr_params:
log.debug('Converted legacy search params from %r to %r',
legacy_params, solr_params)
return solr_params
def run(self, query):
'''
Performs a dataset search using the given query.
@param query - dictionary with keys like: q, fq, sort, rows, facet
@return - dictionary with keys results and count
May raise SearchQueryError or SearchError.
'''
from solr import SolrException
assert isinstance(query, (dict, MultiDict))
# check that query keys are valid
if not set(query.keys()) <= VALID_SOLR_PARAMETERS:
invalid_params = [
s for s in set(query.keys()) - VALID_SOLR_PARAMETERS
]
raise SearchQueryError("Invalid search parameters: %s" %
invalid_params)
# default query is to return all documents
q = query.get('q')
if not q or q == '""' or q == "''":
query['q'] = "*:*"
# number of results
query['rows'] = min(1000, int(query.get('rows', 10)))
# order by score if no 'sort' term given
order_by = query.get('sort')
if order_by == 'rank' or order_by is None:
query['sort'] = 'score desc, name asc'
# show only results from this CKAN instance
fq = query.get('fq', '')
if not '+site_id:' in fq:
fq += ' +site_id:"%s"' % config.get('ckan.site_id')
# filter for package status
if not '+state:' in fq:
fq += " +state:active"
query['fq'] = fq
# faceting
query['facet'] = query.get('facet', 'true')
query['facet.limit'] = query.get(
'facet.limit', config.get('search.facets.limit', '50'))
query['facet.mincount'] = query.get('facet.mincount', 1)
# return the package ID and search scores
query['fl'] = query.get('fl', 'name')
# return results as json encoded string
query['wt'] = query.get('wt', 'json')
# query field weighting: disabled for now as solr 3.* is required for
# the 'edismax' query parser, our current Ubuntu version only has
# packages for 1.4
#
# query['defType'] = 'edismax'
# query['tie'] = '0.5'
# query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection()
log.debug('Package query: %r' % query)
try:
solr_response = conn.raw_query(**query)
except SolrException, e:
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e.reason))
def run(self, query):
'''
Performs a dataset search using the given query.
@param query - dictionary with keys like: q, fq, sort, rows, facet
@return - dictionary with keys results and count
May raise SearchQueryError or SearchError.
'''
from solr import SolrException
assert isinstance(query, (dict, MultiDict))
# check that query keys are valid
if not set(query.keys()) <= VALID_SOLR_PARAMETERS:
invalid_params = [
s for s in set(query.keys()) - VALID_SOLR_PARAMETERS
]
raise SearchQueryError("Invalid search parameters: %s" %
invalid_params)
# default query is to return all documents
q = query.get('q')
if not q or q == '""' or q == "''":
query['q'] = "*:*"
# number of results
rows_to_return = min(1000, int(query.get('rows', 10)))
if rows_to_return > 0:
# #1683 Work around problem of last result being out of order
# in SOLR 1.4
rows_to_query = rows_to_return + 1
else:
rows_to_query = rows_to_return
query['rows'] = rows_to_query
# order by score if no 'sort' term given
order_by = query.get('sort')
if order_by == 'rank' or order_by is None:
query['sort'] = 'score desc, name asc'
# show only results from this CKAN instance
fq = query.get('fq', '')
if not '+site_id:' in fq:
fq += ' +site_id:"%s"' % config.get('ckan.site_id')
# filter for package status
if not '+state:' in fq:
fq += " +state:active"
query['fq'] = fq
# faceting
query['facet'] = query.get('facet', 'true')
query['facet.limit'] = query.get(
'facet.limit', config.get('search.facets.limit', '50'))
query['facet.mincount'] = query.get('facet.mincount', 1)
# return the package ID and search scores
query['fl'] = query.get('fl', 'name')
# return results as json encoded string
query['wt'] = query.get('wt', 'json')
# If the query has a colon in it then consider it a fielded search and do use dismax.
if ':' not in query['q']:
query['defType'] = 'dismax'
query['tie'] = '0.1'
query['mm'] = '1'
query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection()
log.debug('Package query: %r' % query)
try:
solr_response = conn.raw_query(**query)
except SolrException, e:
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e.reason))