def ops_published_data_crawl_handler(request):
"""Crawl published-data at OPS"""
# constituents: abstract, biblio and/or full-cycle
constituents = request.matchdict.get('constituents', 'full-cycle')
print('constituents:', constituents)
# CQL query string
query = request.params.get('expression', '')
log.info('query raw: ' + query)
# Transcode CQL query expression
search = cql_prepare_query(query)
# Propagate keywords to highlighting component
keywords_to_response(request, search=search)
log.info('query cql: ' + search.expression)
chunksize = int(request.params.get('chunksize', '100'))
try:
result = ops_published_data_crawl(constituents, search.expression,
chunksize)
return result
except Exception as ex:
log.error(
'OPS crawler error: query="{0}", reason={1}, Exception was:\n{2}'.
format(query, ex, _exception_traceback()))
request.errors.add('ops-published-data-crawl', 'query', str(ex))
def ops_published_data_search_handler(request):
"""Search for published-data at OPS"""
# Constituents: abstract, biblio and/or full-cycle
constituents = request.params.get('constituents', 'full-cycle')
# CQL query string
query = request.params.get('expression', '')
log.info('query raw: %s', query)
# Transcode CQL query expression
search = cql_prepare_query(query)
log.info('query cql: %s', search.expression)
# range: x-y, maximum delta is 100, default is 25
range = request.params.get('range')
range = range or '1-25'
# Search options
family_swap_default = asbool(request.params.get('family_swap_default'))
try:
if family_swap_default:
result = ops_published_data_search_swap_family(
constituents, search.expression, range)
else:
result = ops_published_data_search(constituents, search.expression,
range)
# Propagate keywords to highlighting component
keywords_to_response(request, search=search)
return result
except NoResultsException as ex:
# Forward response to let the frontend recognize zero hits
request.response.status = HTTPNotFound.code
return ex.data
except Exception as ex:
message = handle_generic_exception(request, ex, 'ops-search',
search.expression)
request.errors.add('ops-search', 'search', message)
log.info('query finished')
def prepare_search(request):
#pprint(request.params)
# CQL expression string
expression = request.params.get('expression', '').strip()
# Compute expression syntax
syntax_cql = asbool(
request.params.get('query_data[modifiers][syntax-cql]'))
syntax_ikofax = asbool(
request.params.get('query_data[modifiers][syntax-ikofax]'))
syntax = 'cql'
if syntax_ikofax or expression.startswith('ikofax:'):
expression = expression.replace('ikofax:', '')
syntax = 'ikofax'
log.info(u'DEPATISnet query: {}, syntax: {}'.format(expression, syntax))
# Compute query options, like
# - limit
# - sorting
# - whether to remove family members
options = {}
options.update({'syntax': syntax})
# propagate request parameters to search options parameters
request_to_options(request, options)
# Transcode query expression
if syntax == 'cql':
search = cql_prepare_query(expression)
elif syntax == 'ikofax':
search = ikofax_prepare_query(expression)
else:
request.errors.add('depatisnet-search', 'expression',
u'Unknown syntax {}'.format(syntax))
# Propagate keywords to highlighting component
keywords_to_response(request, search=search)
return search, options
def depatech_published_data_search_handler(request):
"""Search for published-data at MTC depa.tech"""
# Get hold of query expression and filter
expression = request.params.get('expression', '')
filter = request.params.get('filter', '')
query = SmartBunch({
'syntax': 'lucene',
'expression': expression,
'filter': filter,
})
if expression.startswith('DEPAROM V1.0') or expression.startswith('deparom:'):
query.syntax = 'deparom'
log.info('Query: {}'.format(query))
# Parse expression, extract and propagate keywords to user interface
if query.syntax == 'lucene':
parser = DepaTechParser(query.expression)
keywords_to_response(request, parser)
# TODO: Parse DEPAROM query expression and extract keywords
# Fixup query: wrap into quotes if cql string is a) unspecific, b) contains spaces and c) is still unquoted
if should_be_quoted(query.expression):
query.expression = '"%s"' % query.expression
# Lazy-fetch more entries
# TODO: get from patzilla.access.depatech
limit = 250
offset_local = int(request.params.get('range_begin', 0))
offset_remote = int(offset_local / limit) * limit
# Compute query options, like
# - limit
# - sorting
# - whether to remove family members
options = SmartBunch()
options.update({
'limit': limit,
'offset': offset_remote,
})
# Propagate request parameters to search options parameters
request_to_options(request, options)
try:
data = depatech_search(query, options)
#print data.prettify() # debugging
return data
except LoginException as ex:
request.errors.add('depatech-search', 'login', ex.details)
log.warn(request.errors)
except SyntaxError as ex:
request.errors.add('depatech-search', 'expression', str(ex.msg))
log.warn(request.errors)
except SearchException as ex:
message = ex.get_message()
request.errors.add('depatech-search', 'search', message)
log.warn(request.errors)
except NoResultsException as ex:
# Forward response to let the frontend recognize zero hits
request.response.status = HTTPNotFound.code
return ex.data
except OperationFailure as ex:
message = str(ex)
request.errors.add('depatech-search', 'internals', message)
log.error(request.errors)
except Exception as ex:
message = handle_generic_exception(request, ex, 'depatech-search', query)
request.errors.add('depatech-search', 'search', message)