def test_open_export(self):
fp = refine.RefineProject(self.project.project_url()).export()
line = fp.next()
self.assertTrue('email' in line)
for line in fp:
self.assertTrue('M' in line or 'F' in line)
fp.close()
def test_open_export_csv(self):
fp = refine.RefineProject(self.project.project_url()).export()
csv_fp = csv.reader(fp, dialect='excel-tab')
row = csv_fp.next()
self.assertTrue(row[0] == 'email')
for row in csv_fp:
self.assertTrue(row[3] == 'F' or row[3] == 'M')
fp.close()
def compute_clusters(project_id,
column,
clusterer_type='binning',
function=None,
params=None):
# returns a list of cluters of {'value';..., 'count':...}
return refine.RefineProject(refine.RefineServer(),
project_id).compute_clusters(
column, clusterer_type, function, params)
def reconcile(project_id,
column,
service,
reconciliation_type=None,
reconciliation_config=None):
return refine.RefineProject(refine.RefineServer(),
project_id).reconcile(column, service,
reconciliation_type,
reconciliation_config)
def export_templating(project_id):
# export templating: return a fileobject of a project's data
return refine.RefineProject(refine.RefineServer(),
project_id).export_templating(
export_format='txt',
engine='',
prefix='',
template='',
rowSeparator='',
suffix='')
def text_transform(project_id,
column,
expression,
on_error='set-to-blank',
repeat=False,
repeat_count=10):
return refine.RefineProject(refine.RefineServer(),
project_id).text_transform(
column, expression, on_error, repeat,
repeat_count)
def split_column(project_id,
column,
separator=',',
mode='separator',
regex=False,
guess_cell_type=True,
remove_original_column=True):
return refine.RefineProject(refine.RefineServer(),
project_id).split_column(
column, separator, mode, regex,
guess_cell_type, remove_original_column)
def get_cell_value(project_id, columnIndex):
# split break.....
# nestedlist: [{u'cells': [{u'v': u'10136'},....{}]},{u 'cells': [...]}]
# The cellIndex is an index for that column's data into the list returned from get_rows().
# 'from': cell_value, 'to': new_cell_value
nested_list = refine.RefineProject(refine.RefineServer(),
project_id).get_cell_value()
AimCellValue = []
for inner_dicts in nested_list:
AimCellValue.append(inner_dicts['cells'][columnIndex])
return AimCellValue
def transpose_columns_into_rows(project_id,
start_column,
column_count,
combined_column_name,
separator=':',
prepend_column_name=True,
ignore_blank_cells=True):
return refine.RefineProject(refine.RefineServer(),
project_id).transpose_columns_into_rows(
start_column, column_count,
combined_column_name, separator,
prepend_column_name, ignore_blank_cells)
def get_split_cell_value(project_id, origin_column_length):
nested_list = refine.RefineProject(refine.RefineServer(),
project_id).get_cell_value()
print(len(nested_list))
AimcellValue = []
counter = []
for inner_dicts in nested_list:
if len(inner_dicts['cells']) > origin_column_length:
counter.append(
len(inner_dicts['cells']) - origin_column_length - 1)
AimcellValue.append(inner_dicts['cells'][origin_column_length +
1:])
return counter, AimcellValue
def do_json(project_id, command, data):
# do_json: issue a command to the server, parse & return encoded OR_JSON.
return refine.RefineProject(refine.RefineServer(),
project_id).do_json(command, data)
def do_raw(project_id, command, data):
# do_raw : issue a command to the server & return a response object
return refine.RefineProject(refine.RefineServer(),
project_id).do_raw(command, data)
def project_url(project_id):
# project_url
return refine.RefineProject(refine.RefineServer(),
project_id).project_url()
def project_name(project_id):
# functions in RefineProject:
# project_name
return refine.RefineProject(refine.RefineServer(),
project_id).project_name()
def get_operations(project_id):
return refine.RefineProject(refine.RefineServer(),
project_id).get_operations()
def flag_row(project_id, row, flagged=True):
return refine.RefineProject(refine.RefineServer(),
project_id).flag_row(row, flagged)
def reorder_columns(project_id, new_column_order):
return refine.RefineProject(refine.RefineServer(),
project_id).reorder_columns(new_column_order)
def annotate_one_row(project_id, row, annotation, state=True):
return refine.RefineProject(refine.RefineServer(),
project_id).annotate_one_row(
row, annotation, state)
def get_single_cell_value(project_id, cellIndex, rowIndex):
nested_list = refine.RefineProject(refine.RefineServer(),
project_id).get_cell_value()
aim_single_cell_value = nested_list[rowIndex]['cells'][cellIndex]
return aim_single_cell_value
def transpose_rows_into_columns(project_id, column, row_count):
return refine.RefineProject(refine.RefineServer(),
project_id).transpose_rows_into_columns(
column, row_count)
def fill_down(project_id, column):
return refine.RefineProject(refine.RefineServer(),
project_id).fill_down(column)
def blank_down(project_id, column):
return refine.RefineProject(refine.RefineServer(),
project_id).blank_down(column)
def move_column(project_id, column, index):
return refine.RefineProject(refine.RefineServer(),
project_id).move_column(column, index)
def get_models(project_id):
# get_models: fill out column metadata.
# Column structure is a list of columns in their order.
# The cellIndex is an index for that column's data into the list returned from get_rows().
return refine.RefineProject(refine.RefineServer(), project_id).get_models()
def add_column(project_id, column, new_column, expression='value'):
return refine.RefineProject(refine.RefineServer(),
project_id).add_column(column, new_column,
expression)
def rename_column(project_id, column, new_column):
return refine.RefineProject(refine.RefineServer(),
project_id).rename_column(column, new_column)
def get_reconciliation_services(project_id):
return refine.RefineProject(refine.RefineServer(),
project_id).get_reconciliation_services()
def get_reconciliation_service_by_name_or_url(project_id, name):
return refine.RefineProject(
refine.RefineServer(),
project_id).get_reconciliation_service_by_name_or_url(name)
def guess_types_of_column(project_id, column, service):
return refine.RefineProject(refine.RefineServer(),
project_id).guess_types_of_column(
column, service)
def star_row(project_id, row, starred=True):
return refine.RefineProject(refine.RefineServer(),
project_id).star_row(row, starred)
