whoosh_dir = os.path.join(os.getcwd(), "mase/query_logs/index")
config = {
#"proxyhost": "http://wwwcache.gla.ac.uk:8080", # <-- remove if not UGLW
"log_dir": log_dir,
"bing_api_key":
"", # Obtain key from https://datamarket.azure.com/dataset/5BA839F1-12CE-4CCE-BF57-A49D98D29A44
}
# create a ServiceManager
service = ServiceManager(config)
# create a Web SearchService
web_search_service = SearchService(service, "web_search")
web_search_service.search_engine = BingV3(web_search_service,
source='Web',
resultsPerPage=4)
service.add_search_service(web_search_service)
# Create a blacklist filter to block queries containing the terms below
query_black_list = BlackListModifier(terms="bad worse nasty filthy")
#Add our blacklist filter to the web search service
web_search_service.add_query_modifier(
query_black_list) # Add Web SearchService to ServiceManager
# create a file based QUERY LOGGER
web_search_service.query_logger = QueryLogger(web_search_service, log_mode=0)
web_search_service.postLogging = True
# create a index based QUERY LOGGER
config = {
"proxyhost": "http://wwwcache.gla.ac.uk:8080", # <-- remove if not UGLW
"log_dir": log_dir,
"bing_api_key": "/aROdM5Ck7fKHR4ge30r8W/K/D84GJkcl42lL8eNMSc=", # Obtain key from https://datamarket.azure.com/dataset/5BA839F1-12CE-4CCE-BF57-A49D98D29A44
}
# create a ServiceManager
service = ServiceManager(config)
# create a ServiceManager
service = ServiceManager(config)
# create a Web SearchService
web_search_service = SearchService(service, "web_search")
web_search_service.search_engine = BingV3(web_search_service, source="Web", resultsPerPage=4)
service.add_search_service(web_search_service)
# Create a blacklist filter to block queries containing the terms below
query_black_list = BlackListModifier(terms="bad worse nasty filthy")
# Add our blacklist filter to the web search service
web_search_service.add_query_modifier(query_black_list) # Add Web SearchService to ServiceManager
# create a file based QUERY LOGGER
web_search_service.query_logger = QueryLogger(web_search_service, log_mode=0)
web_search_service.postLogging = True
# create a index based QUERY LOGGER
whoosh_query_logger = WhooshQueryLogger(whoosh_query_index_dir=whoosh_dir, unique=True)
#print "page len" + str(pagelen)
results = []
reponse = []
with self.docIndex.searcher() as searcher:
results = searcher.search_page( query_terms, page, pagelen=pagelen )
# results = searcher.search( query_terms )
results.fragmenter = highlight.ContextFragmenter(maxchars=300, surround=300)
results.formatter = highlight.HtmlFormatter()
results.fragmenter.charlimit = 100000
print "WhooshTRECNewsEngine found: " + str(len(results)) + " results"
print "Page %d of %d - PageLength of %d" % (results.pagenum, results.pagecount, results.pagelen)
response = parse_whoosh_trec('WhooshTRECNewsEngine', query.search_terms, results)
return response
except:
print "Error in Search Service: Whoosh TREC News search failed"
config = { "log_dir": "treconomics/query_logs", }
# create a ServiceManager
service = ServiceManager(config)
# create a SearchService, choose search engine and enable query logging
trec_search_service = SearchService(service, "trec_search")
trec_search_service.search_engine = WhooshTRECNewsEngine(trec_search_service, my_whoosh_doc_index_dir)
# add SearchService to ServiceManager
service.add_search_service(trec_search_service)
log_dir = os.path.join(DOLLEMANPATH, log_dir)
else:
whoosh_dir = os.path.join(os.getcwd(), "mase/query_logs/index")
config = {
#"proxyhost": "http://wwwcache.gla.ac.uk:8080", # <-- remove if not UGLW
"log_dir": log_dir,
"bing_api_key": "", # Obtain key from https://datamarket.azure.com/dataset/5BA839F1-12CE-4CCE-BF57-A49D98D29A44
}
# create a ServiceManager
service = ServiceManager(config)
# create a Web SearchService
web_search_service = SearchService(service, "web_search")
web_search_service.search_engine = BingV3(web_search_service, source='Web', resultsPerPage = 4)
service.add_search_service(web_search_service)
# Create a blacklist filter to block queries containing the terms below
query_black_list = BlackListModifier(terms = "bad worse nasty filthy")
#Add our blacklist filter to the web search service
web_search_service.add_query_modifier(query_black_list)# Add Web SearchService to ServiceManager
# create a file based QUERY LOGGER
web_search_service.query_logger = QueryLogger(web_search_service, log_mode=0)
web_search_service.postLogging = True
# create a index based QUERY LOGGER
whoosh_query_logger = WhooshQueryLogger(whoosh_query_index_dir=whoosh_dir, unique=True)
results.fragmenter = highlight.ContextFragmenter(maxchars=300,
surround=300)
results.formatter = highlight.HtmlFormatter()
results.fragmenter.charlimit = 100000
print "WhooshTRECNewsEngine found: " + str(
len(results)) + " results"
print "Page %d of %d - PageLength of %d" % (
results.pagenum, results.pagecount, results.pagelen)
response = parse_whoosh_trec('WhooshTRECNewsEngine',
query.search_terms, results)
return response
except:
print "Error in Search Service: Whoosh TREC News search failed"
config = {
"log_dir": "treconomics/query_logs",
}
# create a ServiceManager
service = ServiceManager(config)
# create a SearchService, choose search engine and enable query logging
trec_search_service = SearchService(service, "trec_search")
trec_search_service.search_engine = WhooshTRECNewsEngine(
trec_search_service, my_whoosh_doc_index_dir)
# add SearchService to ServiceManager
service.add_search_service(trec_search_service)
