def describe_data(tweets):
"""
Evaluate the Tweets provided and return the results object
"""
logger.info('Describing Tweets')
results = tweet_evaluation_module.setup_analysis(conversation=True,audience=True)
tweet_evaluation_module.analyze_tweets(tweets,results)
return results
def test_hashtag_count(self):
""" inject hashtags with a predetermined number of test tokens """
pass
# configure results structure
results = analysis.setup_analysis(do_conversation=True)
# use counter for verification
counter = 0
for tweet in self.tweets:
tweet['twitter_entities']['hashtags'].append({"text":"notarandomhashtag"})
analysis.analyze_tweet(tweet, results)
counter += 1
self.assertEqual(results['hashtags']['notarandomhashtag'], counter)
def test_conversation_length(self):
"""
Test count of Tweets in analysis code against known number
"""
# configure results structure
results = analysis.setup_analysis(do_conversation=True)
# is this id necessary for testing?
#results["unique_id"] = "TEST"
# run analysis code (including conversation)
analysis.analyze_tweets(self.tweets, results)
# ground truth from setUp()
self.assertEqual(results['tweet_count'],self.generator_length_truth)
def test_conversation_length(self):
"""
Test count of Tweets in analysis code against known number
"""
# configure results structure
results = analysis.setup_analysis(do_conversation=True)
# is this id necessary for testing?
#results["unique_id"] = "TEST"
# run analysis code (including conversation)
analysis.analyze_tweets(self.tweets, results)
# ground truth from setUp()
self.assertEqual(results['tweet_count'], self.generator_length_truth)
def test_hashtag_count(self):
""" inject hashtags with a predetermined number of test tokens """
pass
# configure results structure
results = analysis.setup_analysis(do_conversation=True)
# use counter for verification
counter = 0
for tweet in self.tweets:
tweet['twitter_entities']['hashtags'].append(
{"text": "notarandomhashtag"})
analysis.analyze_tweet(tweet, results)
counter += 1
self.assertEqual(results['hashtags']['notarandomhashtag'], counter)
def test_bio_term_count(self):
""" inject bio with a predetermined number of test tokens """
pass
# configure results structure
results = analysis.setup_analysis(do_audience=True)
# use counter for verification
counter = 1
for tweet in self.tweets:
addition = " test_term"*counter
tweet['actor']['summary'] += addition
analysis.analyze_tweet(tweet, results)
counter += 1
expected_test_count = int( next(results['bio_term_count'].get_tokens())[0] )
self.assertEqual(expected_test_count, sum(range(counter)))
def test_audience_length(self):
pass
# configure results structure
results = analysis.setup_analysis(do_audience=True)
# run analysis code (including audience) for user ids
analysis.analyze_tweets(self.tweets, results)
user_ids = results["tweets_per_user"].keys()
# get ground truth (# unique user ids) from test data file
p1 = subprocess.Popen(['cat', INPUT_FILE_NAME], stdout=subprocess.PIPE)
p2 = subprocess.Popen(['python', '-c', 'import sys; import json; print(len(set([json.loads(i)["actor"]["id"] for i in sys.stdin])))'], stdin=p1.stdout, stdout=subprocess.PIPE)
p1.stdout.close()
out, err = p2.communicate()
shell_user_count = int(out)
self.assertEqual(shell_user_count, len(user_ids))
def test_body_term_count(self):
""" inject body with a predetermined number of test tokens """
pass
# configure results structure
results = analysis.setup_analysis(do_conversation=True)
# use counter for verification
counter = 1
for tweet in self.tweets:
addition = " test_term" * counter
tweet['body'] += addition
analysis.analyze_tweet(tweet, results)
counter += 1
expected_test_count = int(
next(results['body_term_count'].get_tokens())[0])
self.assertEqual(expected_test_count, sum(range(counter)))
def test_audience_length(self):
pass
# configure results structure
results = analysis.setup_analysis(do_audience=True)
# run analysis code (including audience) for user ids
analysis.analyze_tweets(self.tweets, results)
user_ids = results["tweets_per_user"].keys()
# get ground truth (# unique user ids) from test data file
p1 = subprocess.Popen(['cat', INPUT_FILE_NAME], stdout=subprocess.PIPE)
p2 = subprocess.Popen([
'python', '-c',
'import sys; import json; print(len(set([json.loads(i)["actor"]["id"] for i in sys.stdin])))'
],
stdin=p1.stdout,
stdout=subprocess.PIPE)
p1.stdout.close()
out, err = p2.communicate()
shell_user_count = int(out)
self.assertEqual(shell_user_count, len(user_ids))
splitting_config = None
if args.splitting_config is not None:
# if file not in local directory, temporarily extend path to its location
config_file_full_path = args.config_file.split('/')
if len(config_file_full_path) > 1:
path = '/'.join(config_file_full_path[:-1])
sys.path.append(os.path.join(os.getcwd(), path))
else:
sys.path.append(os.getcwd())
splitting_config = importlib.import_module(
config_file_full_path[-1].rstrip('.py')).splitting_config
sys.path.pop()
results = analysis.setup_analysis(
conversation=args.do_conversation_analysis,
audience=args.do_audience_analysis,
identifier='analyzed',
input_results={})
results = analysis.setup_analysis(
conversation=args.do_conversation_analysis,
audience=args.do_audience_analysis,
identifier='baseline',
input_results=results)
else:
results = analysis.setup_analysis(
conversation=args.do_conversation_analysis,
audience=args.do_audience_analysis)
# manage input sources, file opening, and deserialization
if args.input_file_name is not None:
tweet_generator = analysis.deserialize_tweets(
parser.add_argument('-o','--output-dir',dest='output_directory',default=os.environ['HOME'] + '/tweet_evaluation/',
help='directory for output files; default is %(default)s')
parser.add_argument('-b','--baseline-input-file',dest='baseline_input_name',default=None,
help='Tweets against which to run a relative analysis')
parser.add_argument('--no-insights',dest='use_insights',action='store_false',default=True)
args = parser.parse_args()
# get the time right now, to use in output naming
time_now = datetime.datetime.now()
output_directory = '{0}/{1:04d}/{2:02d}/{3:02d}/'.format(args.output_directory.rstrip('/')
,time_now.year
,time_now.month
,time_now.day
)
# get the empty results object, which defines the measurements to be run
results = analysis.setup_analysis(do_conversation = args.do_conversation_analysis, do_audience = args.do_audience_analysis)
baseline_results = None
if args.baseline_input_name is not None:
baseline_results = analysis.setup_analysis(do_conversation = args.do_conversation_analysis, do_audience = args.do_audience_analysis)
if not args.use_insights:
results.pop('audience_api',None)
if args.baseline_input_name is not None:
baseline_results.pop('audience_api',None)
# manage input sources, file opening, and deserialization
if args.input_file_name is not None:
tweet_generator = analysis.deserialize_tweets(open(args.input_file_name))
else:
tweet_generator = analysis.deserialize_tweets(sys.stdin)
help="file containing tweets, tweet IDs, or user IDs; take input from stdin if not present")
parser.add_argument('-o','--output-dir',dest='output_directory',default=os.environ['HOME'] + '/tweet_evaluation/',
help='directory for output files; default is %(default)s')
args = parser.parse_args()
# get the time right now, to use in output naming
time_now = datetime.datetime.now()
time_string = time_now.isoformat().split(".")[0].translate(None,":")
output_directory = '{0}/{1:04d}/{2:02d}/{3:02d}/'.format(args.output_directory.rstrip('/')
,time_now.year
,time_now.month
,time_now.day
)
# create the output directory if it doesn't exist
###
results = analysis.setup_analysis(conversation = args.do_conversation_analysis, audience = args.do_audience_analysis)
# manage input source
if args.input_file_name is not None:
input_generator = open(args.input_file_name)
else:
input_generator = sys.stdin
# run analysis
run_analysis(input_generator, results)
# dump the output
output.dump_results(results, output_directory, args.unique_identifier)
help='directory for output files; default is %(default)s')
parser.add_argument('-b',
'--baseline-input-file',
dest='baseline_input_name',
default=None,
help='Tweets against which to run a relative analysis')
args = parser.parse_args()
# get the time right now, to use in output naming
time_now = datetime.datetime.now()
output_directory = '{0}/{1:04d}/{2:02d}/{3:02d}/'.format(
args.output_directory.rstrip('/'), time_now.year, time_now.month,
time_now.day)
# get the empty results object, which defines the measurements to be run
results = analysis.setup_analysis(
do_conversation=args.do_conversation_analysis,
do_audience=args.do_audience_analysis)
baseline_results = None
if args.baseline_input_name is not None:
baseline_results = analysis.setup_analysis(
do_conversation=args.do_conversation_analysis,
do_audience=args.do_audience_analysis)
# manage input sources, file opening, and deserialization
if args.input_file_name is not None:
tweet_generator = analysis.deserialize_tweets(
open(args.input_file_name))
else:
tweet_generator = analysis.deserialize_tweets(sys.stdin)
