Exemple #1
0
print >> sys.stderr, "Columns in", var_names
data_out = {}
keep_names = set([
    'remove', 'uri', 'title', 'number', 'pub_date', 'author', 'start_page',
    'end_page', 'type', 'journal', 'volume', 'doi'
])
for row, data in data_in.items():
    new_data = dict(data)

    # Add these columns

    new_data['remove'] = ''
    new_data['uri'] = ''
    new_data['title'] = improve_title(new_data['title'])
    [new_data['start_page'],
     new_data['end_page']] = parse_pages(new_data['pages'])
    new_data['pub_date'] = parse_date_parts(new_data['month'],
                                            new_data['year'])

    # Delete everything not in the keep_names set

    for name in new_data.keys():
        if name not in keep_names:
            del new_data[name]

    data_out[row] = new_data
var_names = data_out[
    data_out.keys()[1]].keys()  # create a list of var_names from the first row
print >> sys.stderr, "Columns out", var_names
write_csv_fp(sys.stdout, data_out)
data_in = read_csv_fp(sys.stdin)
var_names = data_in[data_in.keys()[1]].keys()  # create a list of var_names from the first row
print >>sys.stderr, "Columns in", var_names
data_out = {}
keep_names = set(['remove', 'uri', 'title', 'number', 'pub_date', 'author', 'start_page', 'end_page', 'type',
                  'journal', 'volume', 'doi'])
for row, data in data_in.items():
    new_data =dict(data)

    # Add these columns

    new_data['remove'] = ''
    new_data['uri'] = ''
    new_data['title'] = improve_title(new_data['title'])
    [new_data['start_page'], new_data['end_page']] = parse_pages(new_data['pages'])
    new_data['pub_date'] = parse_date_parts(new_data['month'], new_data['year'])

    # Delete everything not in the keep_names set

    for name in new_data.keys():
        if name not in keep_names:
            del new_data[name]

    data_out[row] = new_data
var_names = data_out[data_out.keys()[1]].keys()  # create a list of var_names from the first row
print >>sys.stderr, "Columns out", var_names
write_csv_fp(sys.stdout, data_out)



Exemple #3
0
 def test_parse_date_parts_with_day(self):
     date = parse_date_parts('AUG 15', '2014')
     print date
     self.assertEqual(date, '2014-08-15T00:00:00')
Exemple #4
0
 def test_parse_date_parts_with_months(self):
     date = parse_date_parts('JUL-AUG', '2014')
     print date
     self.assertEqual(date, '2014-07-01T00:00:00')
Exemple #5
0
 def test_parse_date_parts_default(self):
     date = parse_date_parts('AUG', '2014')
     print date
     self.assertEqual(date, '2014-08-01T00:00:00')