def __init__(self, petrGlobal={}, config_folder='data/config/', config_file='PETR_config.ini'):
# cli_args = petrarch2.parse_cli_args()
if not petrGlobal:
utilities.init_logger('PETRARCH.log')
logger = logging.getLogger('petr_log')
PETRglobals.RunTimeString = time.asctime()
logger.info('Using Config file: ' + config_file)
PETRreader.parse_Config(utilities._get_data(config_folder, config_file))
petrarch2.read_dictionaries()
print("SUCCESSFULL ON LOADING DICTIONARIES")
else:
print ("LOADING FROM MAP")
self.load(petrGlobal)
def run_pipeline(data,
out_file=None,
config=None,
write_output=True,
parsed=False):
# this is called externally
utilities.init_logger('PETRARCH.log')
logger = logging.getLogger('petr_log')
if config:
print('Using user-specified config: {}'.format(config))
logger.info('Using user-specified config: {}'.format(config))
PETRreader.parse_Config(config)
else:
logger.info('Using default config file.')
logger.info('Config path: {}'.format(
utilities._get_data('data/config/', 'PETR_config.ini')))
PETRreader.parse_Config(
utilities._get_data('data/config/', 'PETR_config.ini'))
read_dictionaries()
logger.info('Hitting read events...')
events = PETRreader.read_pipeline_input(data)
if parsed:
logger.info('Hitting do_coding')
updated_events = do_coding(events)
else:
events = utilities.stanford_parse(events)
updated_events = do_coding(events)
if not write_output:
output_events = PETRwriter.pipe_output(updated_events)
return output_events
elif write_output and not out_file:
print('Please specify an output file...')
logger.warning('Need an output file. ¯\_(ツ)_/¯')
sys.exit()
elif write_output and out_file:
PETRwriter.write_events(updated_events, out_file)
def main():
cli_args = parse_cli_args()
utilities.init_logger('PETRARCH.log')
logger = logging.getLogger('petr_log')
PETRglobals.RunTimeString = time.asctime()
print(cli_args)
if cli_args.config:
print('Using user-specified config: {}'.format(cli_args.config))
logger.info('Using user-specified config: {}'.format(cli_args.config))
PETRreader.parse_Config(cli_args.config)
else:
logger.info('Using default config file.')
PETRreader.parse_Config(
utilities._get_data('data/config/', 'PETR_config.ini'))
if cli_args.nullverbs:
print('Coding in null verbs mode; no events will be generated')
logger.info('Coding in null verbs mode; no events will be generated')
# Only get verb phrases that are not in the dictionary but are
# associated with coded noun phrases
PETRglobals.NullVerbs = True
elif cli_args.nullactors:
print('Coding in null actors mode; no events will be generated')
logger.info('Coding in null verbs mode; no events will be generated')
# Only get actor phrases that are not in the dictionary but
# associated with coded verb phrases
PETRglobals.NullActors = True
PETRglobals.NewActorLength = int(cli_args.nullactors)
read_dictionaries()
start_time = time.time()
print('\n\n')
paths = PETRglobals.TextFileList
if cli_args.inputs:
if os.path.isdir(cli_args.inputs):
if cli_args.inputs[-1] != '/':
paths = glob.glob(cli_args.inputs + '/*.xml')
else:
paths = glob.glob(cli_args.inputs + '*.xml')
elif os.path.isfile(cli_args.inputs):
paths = [cli_args.inputs]
else:
print(
'\nFatal runtime error:\n"' + cli_args.inputs +
'" could not be located\nPlease enter a valid directory or file of source texts.'
)
sys.exit()
out = "" # PETRglobals.EventFileName
if cli_args.outputs:
out = cli_args.outputs
if cli_args.command_name == 'parse':
run(paths, out, cli_args.parsed)
else:
run(paths, out, True) # <===
print("Coding time:", time.time() - start_time)
print("Finished")
from petrarch2 import petrarch2, PETRglobals, PETRreader, PETRtree, utilities
from ConfigParser import ConfigParser
from flask import jsonify, make_response
from flask.ext.httpauth import HTTPBasicAuth
from flask.ext.restful import Resource, reqparse
from flask.ext.restful.representations.json import output_json
import os
config = "/app/resources/PETR_config.ini"
PETRreader.parse_Config(config)
petrarch2.read_dictionaries()
class PhraseExtractAPI(Resource):
def __init__(self):
self.reqparse = reqparse.RequestParser()
self.reqparse.add_argument('text', type=unicode, location='json')
self.reqparse.add_argument('parse', type=unicode, location='json')
super(PhraseExtractAPI, self).__init__()
def get(self):
return """ This service expects a POST in the form '{"text":""Airstrikes
and artillery...", "parse" : "(ROOT (S (S (NP (NP (NNP Airstrikes))
(CC and) (NP (NN artillery)))..."}' It will return a list of nouns and verbs...TBD"""
def post(self):
args = self.reqparse.parse_args()
print args
text = args['text']
parse = args['parse']
output = self.get_phrases(text, parse)
return output
