def create_query_object():
"""
Nested method which returns a Query object for the given query string, page number and page length attributes from the parent method.
"""
query_object = Query(query_text)
query_object.skip = page
query_object.top = page_len
response = self._search_interface.issue_query(query_object)
query_object.response = response
return query_object
def create_query_object():
"""
Nested method which returns a Query object for the given query string, page number and page length attributes from the parent method.
"""
query_object = Query(query_text)
query_object.skip = page
query_object.top = page_len
response = self.__search_interface.issue_query(query_object)
query_object.response = response
return query_object
def issue_query(self, text, page=1, pagelen=100):
""" Creates a Query object, issues query to the search engine, attaches response to query object
:param text: query string
:param page: integer
:param pagelen: integer
:return: ifind.search.Query
"""
q = Query(text)
q.skip = page
q.top = pagelen
response = self.si.issue_query(q)
q.response = response
return q
def cache_queries():
execution_time = timeit.default_timer()
query_list = read_query_terms()
for query in query_list:
if not (query.isspace() or query == ""):
query_start_time = timeit.default_timer()
print "-" * 80
print "> {0}".format(query)
query = Query(terms=query)
query.top = 10
query.skip = 1
ENGINE.search(query)
print " >> Elapsed time: {0:.2f} second(s)".format(timeit.default_timer() - query_start_time)
print "=" * 80
print "> Total execution time: {0:.2f} seconds".format(timeit.default_timer() - execution_time)
print "> Page caching thread will die shortly, or just kill the Python process."
def run_queries(engine, query_file, result_file):
infile = open(query_file, "r")
outfile = open(result_file, "w")
while infile:
line = infile.readline()
parts = line.partition(' ')
query_num = parts[0]
query_str = unicode(parts[2])
max_limit = 1000
print query_num, query_str
def buildQueryParts(term_list, op):
qp = ''
for t in term_list:
if t:
if qp:
qp = qp + " " + op + " " + t
else:
qp = t
return qp
or_query = buildQueryParts(query_str.split(' '), 'OR')
query = Query(terms=or_query, top=1000)
#query = Query(terms=query_str,top=1000)
query.skip = 1
response = engine.search(query)
if response:
print query_num + " " + str(len(response.results))
rank = 0
for r in response.results:
rank = rank + 1
trec_line = query_num + " Q0 " + r.docid + " " + str(
rank) + " " + str(1000 - rank) + " Exp\n"
outfile.write(trec_line)
if not line:
break
def run_queries(engine, query_file, result_file):
infile = open(query_file,"r")
outfile = open(result_file,"w")
while infile:
line = infile.readline()
parts = line.partition(' ')
query_num = parts[0]
query_str = unicode( parts[2] )
max_limit = 1000
print query_num, query_str
def buildQueryParts(term_list, op):
qp = ''
for t in term_list:
if t:
if qp:
qp = qp + " "+ op +" " + t
else:
qp = t
return qp
or_query= buildQueryParts(query_str.split(' '), 'OR')
query = Query(terms=or_query,top=1000)
#query = Query(terms=query_str,top=1000)
query.skip = 1
response = engine.search(query)
if response:
print query_num + " " + str(len(response.results))
rank = 0
for r in response.results:
rank = rank + 1
trec_line = query_num + " Q0 " + r.docid + " " + str( rank) + " " + str(1000-rank) + " Exp\n"
outfile.write(trec_line)
if not line:
break
def cache_queries():
execution_time = timeit.default_timer()
query_list = read_query_terms()
for query in query_list:
if not (query.isspace() or query == ""):
query_start_time = timeit.default_timer()
print "-" * 80
print "> {0}".format(query)
query = Query(terms=query)
query.top = 10
query.skip = 1
ENGINE.search(query)
print " >> Elapsed time: {0:.2f} second(s)".format(
timeit.default_timer() - query_start_time)
print "=" * 80
print "> Total execution time: {0:.2f} seconds".format(
timeit.default_timer() - execution_time)
print "> Page caching thread will die shortly, or just kill the Python process."
query = query.strip()
query = query.lower()
query = query.replace('"', '')
query = query.split()
new_query = ""
for term in query:
if term not in ignore_terms:
new_query = new_query + term + ' '
new_query = new_query.strip()
query_list.append(new_query)
input_file.close()
return query_list
query_list = get_query_list()
for query in query_list:
query_obj = Query(terms=query, top=top)
query_obj.skip = skip
start_time = timeit.default_timer()
results = engine.search(query_obj)
print
print "Query '{0}' executed in {1:.2f} seconds".format(query, (timeit.default_timer() - start_time))
print "Got {0} result(s)".format(len(results))
print
query = query.replace('"', '')
query = query.split()
new_query = ""
for term in query:
if term not in ignore_terms:
new_query = new_query + term + ' '
new_query = new_query.strip()
query_list.append(new_query)
input_file.close()
return query_list
query_list = get_query_list()
for query in query_list:
query_obj = Query(terms=query, top=top)
query_obj.skip = skip
start_time = timeit.default_timer()
results = engine.search(query_obj)
print
print "Query '{0}' executed in {1:.2f} seconds".format(
query, (timeit.default_timer() - start_time))
print "Got {0} result(s)".format(len(results))
print
