def get_astro_changes(message):
"""This is a special hack to retrieve only astro papers"""
if message.getParam("last_recid"):
#last_recid = int(Integer.cast_(message.getParam("last_recid")).intValue())
last_recid = int(str(message.getParam("last_recid")))
if last_recid == -1:
api_calls.dispatch("create_collection_bibrec", "_astro_bibrec", "Astronomy")
message.setParam("table", "_astro_bibrec")
get_recids_changes(message)
def invenio_search(message):
'''Search Invenio using high-level API
@param kwargs: (Map<String, String[]>) list of parameters
This search is problematic becuase it can return different
types of values: None, list of ints, String
'''
params = message.getParam('kwargs')
if params is None:
return
params = HashMap.cast_(params).of_(String, List)
kwargs = {}
kset = params.keySet().toArray()
vset = params.values().toArray()
max_size = len(vset)
i = 0
while i < max_size:
v = list(ArrayList(vset[i]).of_(String))
if len(v) == 1:
kwargs[str(kset[i])] = v[0]
else:
kwargs[str(kset[i])] = v
i += 1
# versatile
#(wid, result) = api_calls.dispatch('invenio_search', kwargs)
# dumb, but fast
(wid, result) = api_calls.dispatch('invenio_search_xml2', kwargs)
if not (isinstance(result, str) or result is None):
raise Exception('Wrong arguments - I\'ll rather die than give you what you want!')
message.setResults(result)
def format_search_results(message):
'''Returns the citation summary for the given recids. This method
inserts the Invenio formatted citation summary directly into the
response object
@var message:
recids: Java int[]
req: SolrQueryRequest
rsp: SolrQueryResponse
'''
rsp = SolrQueryResponse.cast_(message.getParam('response'))
recids = message.getParamArray_int("recids")
start = time.time()
if config.MONTYSOLR_BUGDEBUG:
message.threadInfo("start: citation_summary")
c_time = time.time()
iset = intbitset(recids)
if config.MONTYSOLR_BUGDEBUG:
message.threadInfo("int[] converted to intbitset in: %s, size=%s" % (time.time() - c_time, len(iset)))
(wid, (output)) = api_calls.dispatch('citation_summary', iset, 'hcs', 'en', '', '')
if config.MONTYSOLR_BUGDEBUG:
message.threadInfo("end: citation_summary pid=%s, finished in %s" % (wid, time.time() - start))
rsp.add("inv_response", output)
def get_recids_changes(message):
"""Retrieves the recids of the last changed documents"""
last_recid = None
table = 'bibrec'
if message.getParam("table"):
table = str(message.getParam("table"))
if message.getParam("last_recid"):
#last_recid = int(Integer.cast_(message.getParam("last_recid")).intValue())
last_recid = int(str(message.getParam("last_recid")))
mod_date = None
if message.getParam("mod_date"):
mod_date = str(message.getParam("mod_date"))
max_records = 10000
if message.getParam('max_records'):
max_records = int(Integer.cast_(message.getParam("max_records")).intValue())
if last_recid and last_recid == -1:
mod_date = None
(wid, results) = api_calls.dispatch("get_recids_changes", last_recid, max_recs=max_records,
mod_date=mod_date, table=table)
if results:
data, last_recid, mod_date = results
out = HashMap().of_(String, JArray_int)
for k,v in data.items():
out.put(k, JArray_int(v))
message.setResults(out)
message.setParam('mod_date', mod_date)
message.setParam('last_recid', last_recid)
def parse_human_name(message):
input = unicode(message.getParam("input"))
(wid, results) = api_calls.dispatch("parse_human_name", input)
if results:
out = HashMap() #.of_(String, JArray_int)
for k,v in results.items():
out.put(k, v)
message.setResults(out)
def perform_request_search_bitset(message):
'''Search and use bitset for exchange of data
@param query: (str) query string
@param max_len: (int) how many max results to return
@param offset: (int) start from the xth element
'''
query = unicode(message.getParam("query")).encode("utf8")
max_len = message.getParam_int('max_len', 25)
offset = message.getParam_int('offset', 0)
#offset, hit_dump, total_matches, searcher_id = searching.multiprocess_search(query, 0)
(wid, (offset, hits, total_matches)) = api_calls.dispatch('search', query, max_len=max_len, offset=offset)
#message.threadInfo("query=%s, total_hits=%s" % (query, total_matches))
message.setResults(JArray_byte(intbitset(hits).fastdump()))
def get_citation_dict(message):
'''TODO: unittest'''
dictname = String.cast_(message.getParam('dictname'))
# we will call the local module (not dispatched remotely)
if hasattr(api_calls, '_dispatch'):
wid, cd = api_calls._dispatch("get_citation_dict", dictname)
else:
wid, cd = api_calls.dispatch("get_citation_dict", dictname)
if cd:
hm = HashMap().of_(String, JArray_int)
for k,v in cd.items():
j_array = JArray_int(v)
hm.put(k, j_array)
message.put('result', hm)
def perform_request_search_ints(message):
'''Search and use ints for exchange of data
@param query: (str) query string
@param max_len: (int) how many max results to return
@param offset: (int) start from the xth element
'''
query = unicode(message.getParam("query")).encode("utf8")
max_len = message.getParam_int('max_len', 25)
offset = message.getParam_int('offset', 0)
#offset, hit_list, total_matches, searcher_id = searching.multiprocess_search(query, 0)
(wid, (offset, hits, total_matches)) = api_calls.dispatch('search', query, max_len=max_len, offset=offset)
if len(hits):
message.setResults(JArray_int(hits))
else:
message.setResults(JArray_int([]))
message.setParam("total", total_matches)
def _dispatch_remote(func_name, args, kwargs):
"""This receives the data on the remote side and calls
the actual function that does the job and returns results.
"""
g = globals()
func_name_pre = '%s_remote_pre' % func_name
func_name_post = '%s_remote_post' % func_name
if func_name_pre in g:
args = list(args)
g[func_name_pre](args, kwargs)
(thread_id, result) = api_calls.dispatch(func_name, *args, **kwargs)
if func_name_post in g:
result = g[func_name_post](result)
return (os.getpid(), result)
def sort_and_format(message):
'''Calls sorting XOR formatting over the result set'''
start = time.time()
recids = intbitset(message.getParamArray_int("recids"))
kwargs = HashMap.cast_(message.getParam('kwargs'))
kws = {}
kset = kwargs.keySet().toArray()
vset = kwargs.values().toArray()
max_size = len(vset)
i = 0
while i < max_size:
v = str(vset[i])
if v[0:1] in ["'", '[', '{'] :
try:
v = eval(v)
except:
pass
kws[str(kset[i])] = v
i += 1
if config.MONTYSOLR_BUGDEBUG:
message.threadInfo("start: citation_summary")
message.threadInfo("int[] converted to intbitset in: %s, size=%s" % (time.time() - start, len(recids)))
(wid, (output)) = api_calls.dispatch('sort_and_format', recids, kws)
if config.MONTYSOLR_BUGDEBUG:
message.threadInfo("end: citation_summary pid=%s, finished in %s" % (wid, time.time() - start))
if isinstance(output, list):
message.setResults(JArray_int(output))
message.setParam("rtype", "int")
else:
message.setResults(output)
message.setParam("rtype", "string")
def _dispatch(func_name, *args, **kwargs):
return api_calls.dispatch(func_name, *args, **kwargs)
