opencultureconsulting
/
openrefine-client
mirror of https://github.com/opencultureconsulting/openrefine-client.git
2
0
Fork 0
Code Issues Releases Activity

Add NumericFacet & camelcase tests. Rename text_facet to compute_facets. Make FacetResponse more generic, with support for bins & baseBins.

This commit is contained in:
Paul Makepeace 2011-04-24 18:36:19 -04:00
parent 263c991804
commit da85e698dd
3 changed files with 78 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
google/refine.py
View File

@ -19,10 +19,17 @@ REFINE_HOST = os.environ.get('GOOGLE_REFINE_HOST', '127.0.0.1')
REFINE_PORT = os.environ.get('GOOGLE_REFINE_PORT', '3333')
def to_camel(attr):
"""convert this_attr_name to thisAttrName."""
return re.sub(r'_(.)', lambda x: x.group(1).upper(), attr)
# Do lower case first letter
return (attr[0].lower() +
re.sub(r'_(.)', lambda x: x.group(1).upper(), attr[1:]))
def from_camel(attr):
"""convert thisAttrName to this_attr_name."""
# Don't add an underscore for capitalized first letter
return re.sub(r'(?<=.)([A-Z])', lambda x: '_' + x.group(1), attr).lower()
class Facet(object):
def __init__(self, column, type, expression='value', **options):
@ -35,7 +42,8 @@ class Facet(object):
setattr(self, k, v)
def as_dict(self):
return dict([(to_camel(k), v) for k, v in self.__dict__.items()])
return dict([(to_camel(k), v) for k, v in self.__dict__.items()
if v is not None])
def include(self, selection):
for s in self.selection:
@ -62,9 +70,9 @@ class TextFacet(Facet):
invert=invert,
**options)
# Capitalize 'From' to get around python's reserved word.
class NumericFacet(Facet):
def __init__(self, column, select_blank=True, select_error=True, select_non_numeric=True, select_numeric=True, **options):
def __init__(self, column, From=None, to=None, select_blank=True, select_error=True, select_non_numeric=True, select_numeric=True, **options):
super(NumericFacet, self).__init__(
column,
type='range',
@ -72,27 +80,32 @@ class NumericFacet(Facet):
select_error=select_error,
select_non_numeric=select_non_numeric,
select_numeric=select_numeric,
From=From,
to=to,
**options)
class FacetResponse(object):
def __init__(self, facet):
self.name = facet['name']
self.column = self.name
self.expression = facet['expression']
self.invert = facet['invert']
for k, v in facet.items():
if isinstance(k, bool) or isinstance(k, basestring):
setattr(self, from_camel(k), v)
self.choices = {}
class FacetChoice(object):
def __init__(self, c):
self.count = c['c']
self.selected = c['s']
for choice in facet['choices']:
self.choices[choice['v']['v']] = FacetChoice(choice)
if 'blankChoice' in facet:
self.blank_choice = FacetChoice(facet['blankChoice'])
else:
self.blank_choice = None
if 'choices' in facet:
for choice in facet['choices']:
self.choices[choice['v']['v']] = FacetChoice(choice)
if 'blankChoice' in facet:
self.blank_choice = FacetChoice(facet['blankChoice'])
else:
self.blank_choice = None
if 'bins' in facet:
self.bins = facet['bins']
self.base_bins = facet['baseBins']
class FacetsResponse(object):
@ -138,6 +151,7 @@ class RefineServer(object):
if data is None:
data = {}
if project_id:
# XXX haven't figured out pattern on qs v body
if 'delete' in command:
data['project'] = project_id
else:
@ -362,7 +376,7 @@ class RefineProject:
response_json = self.do_json('delete-project')
return 'code' in response_json and response_json['code'] == 'ok'
def text_facet(self, facets=None):
def compute_facets(self, facets=None):
if facets:
self.engine = Engine(facets)
response = self.do_json('compute-facets',

7
google/test/test_engine.py
View File

@ -21,6 +21,9 @@ class FacetTest(unittest.TestCase):
engine = Engine(facet)
self.assertEqual(facet.selection, [])
self.assertTrue(str(engine))
facet = NumericFacet('column name', From=1, to=5)
self.assertEqual(facet.to, 5)
self.assertEqual(facet.From, 1)
def test_serialize(self):
engine = Engine()
@ -28,8 +31,8 @@ class FacetTest(unittest.TestCase):
self.assertEqual(engine_json, '{"facets": [], "mode": "row-based"}')
facet = TextFacet(column='column')
self.assertEqual(facet.as_dict(), {'selectError': False, 'name': 'column', 'selection': [], 'expression': 'value', 'invert': False, 'columnName': 'column', 'selectBlank': False, 'omitBlank': False, 'type': 'list', 'omitError': False})
facet = NumericFacet(column='column')
self.assertEqual(facet.as_dict(), {'selectBlank': True, 'name': 'column', 'selectError': True, 'expression': 'value', 'selection': [], 'selectNumeric': True, 'columnName': 'column', 'selectNonNumeric': True, 'type': 'range'})
facet = NumericFacet(column='column', From=1, to=5)
self.assertEqual(facet.as_dict(), {'from': 1, 'to': 5, 'selectBlank': True, 'name': 'column', 'selectError': True, 'expression': 'value', 'selection': [], 'selectNumeric': True, 'columnName': 'column', 'selectNonNumeric': True, 'type': 'range'})
def test_add_facet(self):
facet = TextFacet(column='Party Code')

49
google/test/test_refine.py
View File

@ -11,11 +11,35 @@ import sys
import os
import unittest
from google.refine import REFINE_HOST, REFINE_PORT
from google.refine import TextFacet, Engine
from google.refine import NumericFacet, TextFacet, Engine
from google.refine import RefineServer, Refine, RefineProject
from google.refine import to_camel, from_camel
PATH_TO_TEST_DATA = os.path.join('google', 'test', 'data')
class CamelTest(unittest.TestCase):
def test_to_camel(self):
pairs = (
('this', 'this'),
('this_attr', 'thisAttr'),
('From', 'from'),
)
for attr, camel_attr in pairs:
self.assertEqual(to_camel(attr), camel_attr)
def test_from_camel(self):
pairs = (
('this', 'this'),
('This', 'this'),
('thisAttr', 'this_attr'),
('ThisAttr', 'this_attr'),
('From', 'from'),
)
for camel_attr, attr in pairs:
self.assertEqual(from_camel(camel_attr), attr)
class RefineTestCase(unittest.TestCase):
project_file = None
project = None
@ -80,7 +104,7 @@ class TutorialTestFacets(RefineTestCase):
def test_basic_facet(self):
# {4}
party_code_facet = TextFacet(column='Party Code')
response = self.project.text_facet(party_code_facet)
response = self.project.compute_facets(party_code_facet)
pc = response.facets[0]
self.assertEqual(pc.name, 'Party Code')
self.assertEqual(pc.choices['D'].count, 3700)
@ -91,7 +115,7 @@ class TutorialTestFacets(RefineTestCase):
ethnicity_facet = TextFacet(column='Ethnicity')
engine.add_facet(ethnicity_facet)
self.project.engine = engine
response = self.project.text_facet()
response = self.project.compute_facets()
e = response.facets[1]
self.assertEqual(e.choices['B'].count, 1255)
self.assertEqual(e.choices['W'].count, 4469)
@ -102,7 +126,7 @@ class TutorialTestFacets(RefineTestCase):
indexes = [r.index for r in response.rows]
self.assertEqual(indexes, [1, 2, 3, 4, 6, 12, 18, 26, 28, 32])
# {8}
response = self.project.text_facet()
response = self.project.compute_facets()
pc = response.facets[0]
self.assertEqual(pc.name, 'Party Code')
self.assertEqual(pc.choices['D'].count, 1179)
@ -110,7 +134,7 @@ class TutorialTestFacets(RefineTestCase):
self.assertEqual(pc.blank_choice.count, 46)
# {9}
party_code_facet.include('R')
response = self.project.text_facet()
response = self.project.compute_facets()
e = response.facets[1]
self.assertEqual(e.choices['B'].count, 11)
# {10}
@ -121,8 +145,21 @@ class TutorialTestFacets(RefineTestCase):
# {11}
office_title_facet = TextFacet('Office Title')
self.project.engine.add_facet(office_title_facet)
response = self.project.text_facet()
response = self.project.compute_facets()
self.assertEqual(len(response.facets[2].choices), 76)
# {12} - XXX not sure how to interpret bins & baseBins yet
office_level_facet = NumericFacet('Office Level')
self.project.engine.add_facet(office_level_facet)
# {13}
office_level_facet.From = 300 # from reserved word
office_level_facet.to = 320
response = self.project.get_rows()
self.assertEqual(response.filtered, 1907)
response = self.project.compute_facets()
ot = response.facets[2] # Office Title
self.assertEqual(len(ot.choices), 21)
self.assertEqual(ot.choices['Chief of Police'].count, 2)
self.assertEqual(ot.choices['Chief of Police '].count, 211)
if __name__ == '__main__':