def runTest(self):
sample = setup.sample(100, 85, 'id', 'abstract', 'title')
self.assertEqual(len(sample), 2)
self.assertTrue(len(sample['train']) <= 85)
self.assertTrue(len(sample['test']) <= 15)
for f in sample['train']:
for arg in ['id', 'abstract', 'title']:
self.assertIn(arg, f.keys())
for f in sample['test']:
for arg in ['id', 'abstract', 'title']:
self.assertIn(arg, f.keys())
def runTest(self):
sample = setup.sample(100,85,'id','abstract','title')
self.assertEqual(len(sample),2)
self.assertTrue(len(sample['train']) <= 85)
self.assertTrue(len(sample['test']) <= 15)
for f in sample['train']:
for arg in ['id','abstract','title']:
self.assertIn(arg,f.keys())
for f in sample['test']:
for arg in ['id','abstract','title']:
self.assertIn(arg,f.keys())
#!/usr/bin/python
'''
Script to gather sample articles from solr and write them to json
'''
import plos_classification.setup as setup, json
from datetime import datetime
print datetime.now(), 'gathering sample articles'
s = setup.sample(15000, 10000, 'abstract', 'title', 'subject2_hierarchy',
'cross_published_journal_key')
print datetime.now(), 'writing sample to file'
o = open('data/sample.json', 'w')
json.dump(s, o, indent=2)
print datetime.now(), 'finished'
#!/usr/bin/python
'''
Script to gather sample articles from solr and write them to json
'''
import plos_classification.setup as setup, json
from datetime import datetime
print datetime.now(), 'gathering sample articles'
s = setup.sample(15000,10000,'abstract','title','subject2_hierarchy','cross_published_journal_key')
print datetime.now(), 'writing sample to file'
o = open('data/sample.json','w')
json.dump(s,o,indent=2)
print datetime.now(),'finished'