def predict_case(docket_number, leaning=None):
log = logging.getLogger('PREDICT_CASE')
if not leaning:
leaning = scdb.get_case_info(docket_number)["decisionDirection"]
the_transcript = transcript.get_transcript_from_PDF("transcripts/" + docket_number + ".pdf")
petitioners, respondents = transcript.get_petitioners_and_respondents(the_transcript)
argument = transcript.get_argument(the_transcript)
arguments_by_advocate = transcript.get_arguments_by_advocate(petitioners, respondents, argument)
inputs = []
for petitioner, argument in arguments_by_advocate["petitioner"].iteritems():
statements = transcript.get_statements_in_argument(argument, petitioner)
number_of_words_per_speaker = scores.get_number_of_words_per_speaker(statements)
features = scores.get_features_from_statements(statements)
flat_features = scores.flatten_features(features)
normalized = scores.normalize_feature_list(flat_features)
myscores = scores.get_feature_vector(normalized)
inputs.extend(myscores)
for respondent, argument in arguments_by_advocate["respondent"].iteritems():
statements = transcript.get_statements_in_argument(argument, respondent)
number_of_words_per_speaker = scores.get_number_of_words_per_speaker(statements)
features = scores.get_features_from_statements(statements)
flat_features = scores.flatten_features(features)
normalized = scores.normalize_feature_list(flat_features)
myscores = scores.get_feature_vector(normalized)
inputs.extend(myscores)
if len(inputs) < 774:
# We were unable to match all arguments
# Not gonna help us, continue
print "Parsing error? Did not create enough features."
return 0
if len(inputs) > 774:
# More than one argument per advocate
# just get the first two
inputs = inputs[:774]
print "Parsing error? Had to truncate features."
inputs.append(leaning)
log.info(inputs[:25])
log.info("Querying network. . .")
return net.activate(inputs)[0]
def bar_chart_data(docket_number):
the_transcript = transcript.get_transcript_from_PDF("transcripts/" + docket_number + ".pdf")
petitioners, respondents = transcript.get_petitioners_and_respondents(the_transcript)
argument = transcript.get_argument(the_transcript)
arguments_by_advocate = transcript.get_arguments_by_advocate(petitioners, respondents, argument)
for petitioner, argument in arguments_by_advocate["petitioner"].iteritems():
statements = transcript.get_statements_in_argument(argument, petitioner)
return jsonify(scores.bar_chart_speaker(statements))
return {}
end_docket = file.find('_')
if file.find('q') != -1:
docket_number = file.find('q')
docket_number = file[slash+1:end_docket]
print "Checking database for docket #%s" % docket_number
print "Winner identified: %s" % scdb.get_winning_party(docket_number)
raw_input()
the_transcript = transcript.get_transcript_from_PDF(file)
petitioners, respondents = transcript.get_petitioners_and_respondents(the_transcript)
argument = transcript.get_argument(the_transcript)
arguments_by_advocate = transcript.get_arguments_by_advocate(petitioners, respondents, argument)
for petitioner, argument in arguments_by_advocate["petitioner"].iteritems():
statements = transcript.get_statements_in_argument(argument, petitioner)
#scores.get_statistics_from_statements(statements)
number_of_words_per_speaker = scores.get_number_of_words_per_speaker(statements)
features = scores.get_features_from_statements(statements)
flat_features = scores.flatten_features(features)
#import matplotlib.pyplot as plt
#from mpltools import style
#style.use('ggplot')
#plt.ion()
#D = flat_features
#plt.barh(range(20), D.values()[:20], align='center')
#plt.yticks(range(20), D.keys()[:20])