def generate_features(input_data, mode, filename=None):
new_data = []
from umls import UMLS
from generate_features import Generator
cui_retriever = UMLS()
feature_generator = Generator(cui_retriever)
def save_one_line(doc_id, sentence_id, features, filename):
line = ','.join([doc_id, sentence_id, features]) + "\n"
# save data to files
with open(filename, 'a') as f:
f.writelines(line)
for i, line in enumerate(input_data):
import datetime
print(i, datetime.datetime.now(), mode, line)
sentence = line[-1]
sentence_id = line[-2]
doc_id = line[-3]
features = feature_generator.produce_features_from_sentence(
sentence, sentence_id, doc_id, mode)
new_data.append(features)
if filename is not None:
# save the data
save_one_line(doc_id, sentence_id, features, filename)
return new_data
def report_row(self):
""" Generates an HTML row for the report_row document.
"""
if self.criteria is None or len(self.criteria) < 1:
return ''
# collect criteria
rows = []
for crit in self.criteria:
# this criterium has been codified
if len(crit.snomed) > 0:
c_html = '<td rowspan="%d">%s</td><td rowspan="%d">%s</td>' % (len(crit.snomed), crit.text, len(crit.snomed), 'in' if crit.is_inclusion else 'ex')
for sno in crit.snomed:
rows.append(c_html + '<td>%s</td><td>%s</td>' % (sno, UMLS.lookup_snomed(sno)))
if len(c_html) > 0:
c_html = ''
# no codes for this criterium
else:
rows.append('<td>%s</td><td>%s</td><td></td>' % (crit.text, 'in' if crit.is_inclusion else 'ex'))
if len(rows) < 1:
return ''
# compose HTML
html = """<tr>
<td rowspan="%d">
<a href="http://clinicaltrials.gov/ct2/show/%s" target="_blank">%s</a>
</td>
<td rowspan="%d" onclick="toggle(this)">
<div style="display:none;">%s</div>
</td>
%s</tr>""" % (len(rows), self.nct, self.nct, len(rows), self.eligibility_formatted, rows[0])
rows.pop(0)
for row in rows:
html += "<tr>%s</tr>" % row
return html
sys.exit(1)
if _BG:
from settings_background import ENDPOINTS
else:
from settings import ENDPOINTS
def forever_alone():
with open('forever.txt') as handle:
return handle.read()
# if run as a script, we kick in here
if __name__ == "__main__":
UMLS.import_snomed_if_necessary()
if _BG:
print "-> Running as a background app"
# load all rules
rules = Rule.load_rules()
if len(rules) < 1:
print "There are no rules, no point in continuing"
print forever_alone()
sys.exit(1)
print "-> Did load %d %s" % (len(rules), 'rule' if 1 == len(rules) else 'rules')
# multiple endpoints? Ask which one to use
ep = None
if len(ENDPOINTS) > 1: