results: make engines of results set

Previously, if a same result was returned from the
same engine, the name of the engine would be added
more than once. To eliminate engine name duplication
engine list was changed to set.
This commit is contained in:
Noémi Ványi 2017-11-03 19:52:47 +01:00
parent 7a9b18e9e9
commit 59c02e8e5f

View file

@ -197,7 +197,7 @@ class ResultContainer(object):
result['parsed_url'] = result['parsed_url']._replace(scheme="http") result['parsed_url'] = result['parsed_url']._replace(scheme="http")
result['url'] = result['parsed_url'].geturl() result['url'] = result['parsed_url'].geturl()
result['engines'] = [result['engine']] result['engines'] = set([result['engine']])
# strip multiple spaces and cariage returns from content # strip multiple spaces and cariage returns from content
if result.get('content'): if result.get('content'):
@ -227,7 +227,7 @@ class ResultContainer(object):
duplicated['positions'].append(position) duplicated['positions'].append(position)
# add engine to list of result-engines # add engine to list of result-engines
duplicated['engines'].append(result['engine']) duplicated['engines'].add(result['engine'])
# using https if possible # using https if possible
if duplicated['parsed_url'].scheme != 'https' and result['parsed_url'].scheme == 'https': if duplicated['parsed_url'].scheme != 'https' and result['parsed_url'].scheme == 'https':