def get_index(self, reference):
query = {
'rows': 1,
'q': 'name:"%s" OR id:"%s"' % (reference, reference),
'wt': 'json',
'fq': 'site_id:"%s"' % config.get('ckan.site_id')
}
try:
if query['q'].startswith('{!'):
raise SearchError('Local parameters are not supported.')
except KeyError:
pass
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError as e:
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e))
if solr_response.hits == 0:
raise SearchError('Dataset not found in the search index: %s' %
reference)
else:
return solr_response.docs[0]
def check_solr_schema_version(schema_file=None):
'''
Checks if the schema version of the SOLR server is compatible
with this CKAN version.
The schema will be retrieved from the SOLR server, using the
offset defined in SOLR_SCHEMA_FILE_OFFSET_MANAGED
('/schema?wt=schema.xml'). If SOLR is set to use the manually
edited `schema.xml`, the schema will be retrieved from the SOLR
server using the offset defined in
SOLR_SCHEMA_FILE_OFFSET_CLASSIC ('/admin/file/?file=schema.xml').
The schema_file parameter allows to override this pointing to
different schema file, but it should only be used for testing
purposes.
If the CKAN instance is configured to not use SOLR or the SOLR
server is not available, the function will return False, as the
version check does not apply. If the SOLR server is available,
a SearchError exception will be thrown if the version could not
be extracted or it is not included in the supported versions list.
:schema_file: Absolute path to an alternative schema file. Should
be only used for testing purposes (Default is None)
'''
if not is_available():
# Something is wrong with the SOLR server
log.warn('Problems were found while connecting to the SOLR server')
return False
# Try to get the schema XML file to extract the version
if not schema_file:
try:
# Try Managed Schema
res = _get_schema_from_solr(SOLR_SCHEMA_FILE_OFFSET_MANAGED)
res.raise_for_status()
except requests.HTTPError:
# Fallback to Manually Edited schema.xml
res = _get_schema_from_solr(SOLR_SCHEMA_FILE_OFFSET_CLASSIC)
schema_content = res.text
else:
with open(schema_file, 'rb') as f:
schema_content = f.read()
tree = xml.dom.minidom.parseString(schema_content)
version = tree.documentElement.getAttribute('version')
if not len(version):
msg = 'Could not extract version info from the SOLR schema'
if schema_file:
msg += ', using file {}'.format(schema_file)
raise SearchError(msg)
if not version in SUPPORTED_SCHEMA_VERSIONS:
raise SearchError('SOLR schema version not supported: %s. Supported'
' versions are [%s]'
% (version, ', '.join(SUPPORTED_SCHEMA_VERSIONS)))
return True
class PackageSearchQuery(SearchQuery):
def get_all_entity_ids(self, max_results=1000):
"""
Return a list of the IDs of all indexed packages.
"""
query = "*:*"
fq = "+site_id:\"%s\" " % config.get('ckan.site_id')
fq += "+state:active "
conn = make_connection()
try:
data = conn.query(query, fq=fq, rows=max_results, fields='id')
finally:
conn.close()
return [r.get('id') for r in data.results]
def get_index(self, reference):
'''
For a given package reference (ID or name), returns the record for it
from the SOLR index.
'''
query = {
'rows': 1,
'q': 'name:%s OR id:%s' % (reference, reference),
'wt': 'json',
'fq': 'site_id:"%s"' % config.get('ckan.site_id')
}
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))
try:
data = json.loads(solr_response)
if data['response']['numFound'] == 0:
raise SearchError('Dataset not found in the search index: %s' %
reference)
else:
return data['response']['docs'][0]
except Exception, e:
if not isinstance(e, SearchError):
log.exception(e)
raise SearchError(e)
class PackageSearchQuery(SearchQuery):
def get_all_entity_ids(self, max_results=1000):
"""
Return a list of the IDs of all indexed packages.
"""
query = "*:*"
fq = "+site_id:\"%s\" " % config.get('ckan.site_id')
fq += "+state:active "
conn = make_connection()
data = conn.search(query, fq=fq, rows=max_results, fields='id')
return [r.get('id') for r in data.docs]
def get_index(self, reference):
query = {
'rows': 1,
'q': 'name:"%s" OR id:"%s"' % (reference, reference),
'wt': 'json',
'fq': 'site_id:"%s"' % config.get('ckan.site_id')
}
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError, e:
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e))
if solr_response.hits == 0:
raise SearchError('Dataset not found in the search index: %s' %
reference)
else:
return solr_response.docs[0]
def run(self,
query=None,
terms=[],
fields={},
facet_by=[],
options=None,
**kwargs):
raise SearchError("SearchQuery.run() not implemented!")
def query_for(_type):
""" Get a SearchQuery instance sub-class suitable for the specified
type. """
try:
_type_n = _normalize_type(_type)
return _QUERIES[_type_n]()
except KeyError as ke:
raise SearchError("Unknown search type: %s" % _type)
def run(self,
query: Optional[Union[str, dict[str, Any]]] = None,
terms: Optional[list[str]] = None,
fields: Optional[dict[str, Any]] = None,
facet_by: Optional[list[str]] = None,
options: Optional[QueryOptions] = None,
**kwargs: Any) -> NoReturn:
raise SearchError("SearchQuery.run() not implemented!")
def get_index(self,reference):
query = {
'rows': 1,
'q': 'name:"%s" OR id:"%s"' % (reference,reference),
'wt': 'json',
'fq': 'site_id:"%s"' % config.get('ckan.site_id')}
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError, e:
raise SearchError('SOLR returned an error running query: %r Error: %r' %
(query, e))
def get_index(self, reference):
query = {
'rows': 1,
'q': 'name:%s OR id:%s' % (reference, reference),
'wt': 'json',
'fq': 'site_id:"%s"' % config.get('ckan.site_id')
}
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.
'''
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'
# this minimum match is explained
# http://wiki.apache.org/solr/DisMaxQParserPlugin#mm_.28Minimum_.27Should.27_Match.29
query['mm'] = '2<-1 5<80%'
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))
# #1683 Filter out the last row that is sometimes out of order
self.results = self.results[:rows_to_return]
# get any extras and add to 'extras' dict
for result in self.results:
extra_keys = filter(lambda x: x.startswith('extras_'),
result.keys())
extras = {}
for extra_key in extra_keys:
value = result.pop(extra_key)
extras[extra_key[len('extras_'):]] = value
if extra_keys:
result['extras'] = extras
# if just fetching the id or name, return a list instead of a dict
if query.get('fl') in ['id', 'name']:
self.results = [r.get(query.get('fl')) for r in self.results]
# get facets and convert facets list to a dict
self.facets = data.get('facet_counts', {}).get('facet_fields', {})
for field, values in self.facets.iteritems():
self.facets[field] = dict(zip(values[0::2], values[1::2]))
except Exception, e:
log.exception(e)
raise SearchError(e)
finally:
conn.close()
return {'results': self.results, 'count': self.count}
def run(self, query, permission_labels=None, **kwargs):
'''
Performs a dataset search using the given query.
:param query: dictionary with keys like: q, fq, sort, rows, facet
:type query: dict
:param permission_labels: filter results to those that include at
least one of these labels. None to not filter (return everything)
:type permission_labels: list of unicode strings; or None
:returns: dictionary with keys results and count
May raise SearchQueryError or SearchError.
'''
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
fq = []
if 'fq' in query:
fq.append(query['fq'])
fq.extend(query.get('fq_list', []))
# show only results from this CKAN instance
fq.append('+site_id:%s' % solr_literal(config.get('ckan.site_id')))
# filter for package status
if not '+state:' in query.get('fq', ''):
fq.append('+state:active')
# only return things we should be able to see
if permission_labels is not None:
fq.append('+permission_labels:(%s)' %
' OR '.join(solr_literal(p) for p in permission_labels))
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.
defType = query.get('defType', 'dismax')
if ':' not in query['q'] or defType == 'edismax':
query['defType'] = defType
query['tie'] = query.get('tie', '0.1')
# this minimum match is explained
# http://wiki.apache.org/solr/DisMaxQParserPlugin#mm_.28Minimum_.27Should.27_Match.29
query['mm'] = query.get('mm', '2<-1 5<80%')
query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError as e:
# Error with the sort parameter. You see slightly different
# error messages depending on whether the SOLR JSON comes back
# or Jetty gets in the way converting it to HTML - not sure why
#
if e.args and isinstance(e.args[0], str):
if "Can't determine a Sort Order" in e.args[0] or \
"Can't determine Sort Order" in e.args[0] or \
'Unknown sort order' in e.args[0]:
raise SearchQueryError('Invalid "sort" parameter')
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e))
self.count = solr_response.hits
self.results = solr_response.docs
# #1683 Filter out the last row that is sometimes out of order
self.results = self.results[:rows_to_return]
# get any extras and add to 'extras' dict
for result in self.results:
extra_keys = filter(lambda x: x.startswith('extras_'),
result.keys())
extras = {}
for extra_key in extra_keys:
value = result.pop(extra_key)
extras[extra_key[len('extras_'):]] = value
if extra_keys:
result['extras'] = extras
# if just fetching the id or name, return a list instead of a dict
if query.get('fl') in ['id', 'name']:
self.results = [r.get(query.get('fl')) for r in self.results]
# get facets and convert facets list to a dict
self.facets = solr_response.facets.get('facet_fields', {})
for field, values in six.iteritems(self.facets):
self.facets[field] = dict(zip(values[0::2], values[1::2]))
return {'results': self.results, 'count': self.count}
def run(self, query, permission_labels=None, **kwargs):
'''
Performs a dataset search using the given query.
:param query: dictionary with keys like: q, fq, sort, rows, facet
:type query: dict
:param permission_labels: filter results to those that include at
least one of these labels. None to not filter (return everything)
:type permission_labels: list of unicode strings; or None
:returns: dictionary with keys results and count
May raise SearchQueryError or SearchError.
'''
assert isinstance(query, (dict, MultiDict))
# check that query keys are valid
valid_solr_parameters = VALID_SOLR_PARAMETERS
for item in plugins.PluginImplementations(plugins.IPackageController):
if 'update_valid_solr_parameters' in dir(item):
valid_solr_parameters = item.update_valid_solr_parameters(
valid_solr_parameters)
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
fq = []
if 'fq' in query:
fq.append(query['fq'])
fq.extend(query.get('fq_list', []))
# show only results from this CKAN instance
fq.append('+site_id:%s' % solr_literal(config.get('ckan.site_id')))
# filter for package status
if not '+state:' in query.get('fq', ''):
fq.append('+state:active')
# only return things we should be able to see
if permission_labels is not None:
fq.append('+permission_labels:(%s)' %
' OR '.join(solr_literal(p) for p in permission_labels))
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.
defType = query.get('defType', 'dismax')
if ':' not in query['q'] or defType == 'edismax':
query['defType'] = defType
query['tie'] = query.get('tie', '0.1')
# this minimum match is explained
# http://wiki.apache.org/solr/DisMaxQParserPlugin#mm_.28Minimum_.27Should.27_Match.29
query['mm'] = query.get('mm', '2<-1 5<80%')
query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError, e:
# Error with the sort parameter. You see slightly different
# error messages depending on whether the SOLR JSON comes back
# or Jetty gets in the way converting it to HTML - not sure why
#
if e.args and isinstance(e.args[0], str):
if "Can't determine a Sort Order" in e.args[0] or \
"Can't determine Sort Order" in e.args[0] or \
'Unknown sort order' in e.args[0]:
raise SearchQueryError('Invalid "sort" parameter')
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e))
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.
'''
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
# 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]
fq_list = query.get('fq_list', [])
query['fq'].extend(fq_list)
# 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.
defType = query.get('defType', 'dismax')
if ':' not in query['q'] or defType == 'edismax':
query['defType'] = defType
query['tie'] = query.get('tie', '0.1')
# this minimum match is explained
# http://wiki.apache.org/solr/DisMaxQParserPlugin#mm_.28Minimum_.27Should.27_Match.29
query['mm'] = query.get('mm', '2<-1 5<80%')
query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection(decode_dates=False)
log.debug('Package query: %r' % query)
try:
solr_response = conn.search(**query)
except pysolr.SolrError, e:
# Error with the sort parameter. You see slightly different
# error messages depending on whether the SOLR JSON comes back
# or Jetty gets in the way converting it to HTML - not sure why
#
if e.args and isinstance(e.args[0], str):
if "Can't determine a Sort Order" in e.args[0] or \
"Can't determine Sort Order" in e.args[0] or \
'Unknown sort order' in e.args[0]:
raise SearchQueryError('Invalid "sort" parameter')
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e))
def _run(query):
'''
Custom final preparation of the solr query
and call to the solr api.
:param query:
:return:
'''
# 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 = query.get('rows', 0)
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
# 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]
fq_list = query.get('fq_list', [])
query['fq'].extend(fq_list)
# 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.
defType = query.get('defType', 'dismax')
boolean = query.get('extras', {}).get('ext_boolean', 'all')
if boolean not in ['all', 'any', 'exact']:
log.error('Ignoring unknown boolean search operator %r' % (boolean, ))
boolean = 'all'
if ':' not in query['q']:
query['defType'] = 'dismax'
query['tie'] = '0.1'
if boolean == 'any':
query['mm'] = '0'
elif boolean == 'all':
query['mm'] = '100%'
elif boolean == 'exact':
query['q'] = '"' + q.replace('"', '\\"') + '"'
query['qf'] = query.get('qf', QUERY_FIELDS)
conn = make_connection()
log.info('Package query: %r' % query)
try:
start_time = time.time()
solr_response = conn.raw_query(**query)
duration = time.time() - start_time
log.info("Solr returned the resilt after {0}".format(duration))
except SolrException, e:
raise SearchError(
'SOLR returned an error running query: %r Error: %r' %
(query, e.reason))
