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

Add HistoryEntry class for responses to many commands. Store that info in the project.history_entry which then allows the test suite to interrogate via assertInResponse (much tidier).

This commit is contained in:
Paul Makepeace 2011-04-26 18:20:21 -04:00
parent 87903f7808
commit a34c8c72c9
5 changed files with 131 additions and 92 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
google/refine/history.py Normal file
View File

@ -0,0 +1,18 @@
#!/usr/bin/env python
"""
Google Refine history: parsing responses.
"""
# Copyright (c) 2011 Paul Makepeace, Real Programmers. All rights reserved.
import json
import re
class HistoryEntry(object):
# N.B. e.g. **response['historyEntry'] won't work as keys are unicode :-/
def __init__(self, id=None, time=None, description=None, **kwargs):
if id is None:
raise ValueError('History entry id must be set')
self.id = id
self.description = description
self.time = time

11
google/refine/refine.py
View File

@ -17,6 +17,7 @@ import urllib2
import urlparse import urlparse
from google.refine import facet from google.refine import facet
from google.refine import history
REFINE_HOST = os.environ.get('GOOGLE_REFINE_HOST', '127.0.0.1') REFINE_HOST = os.environ.get('GOOGLE_REFINE_HOST', '127.0.0.1')
REFINE_PORT = os.environ.get('GOOGLE_REFINE_PORT', '3333') REFINE_PORT = os.environ.get('GOOGLE_REFINE_PORT', '3333')
@ -209,6 +210,7 @@ class RefineProject:
self.project_id = project_id self.project_id = project_id
self.engine = facet.Engine() self.engine = facet.Engine()
self.sorting = facet.Sorting() self.sorting = facet.Sorting()
self.history_entry = None
# following filled in by get_models() # following filled in by get_models()
self.has_records = False self.has_records = False
self.column_order = {} # order of columns in UI self.column_order = {} # order of columns in UI
@ -228,8 +230,15 @@ class RefineProject:
if data is None: if data is None:
data = {} data = {}
data['engine'] = self.engine.as_json() data['engine'] = self.engine.as_json()
return self.server.urlopen_json(command, project_id=self.project_id, response = self.server.urlopen_json(command,
project_id=self.project_id,
data=data) data=data)
if 'historyEntry' in response:
# **response['historyEntry'] won't work as keys are unicode :-/
he = response['historyEntry']
self.history_entry = history.HistoryEntry(he['id'], he['time'],
he['description'])
return response
def get_models(self): def get_models(self):
"""Fill out column metadata. """Fill out column metadata.

7
google/test/refinetest.py
View File

@ -35,3 +35,10 @@ class RefineTestCase(unittest.TestCase):
if self.project: if self.project:
self.project.delete() self.project.delete()
self.project = None self.project = None
def assertInResponse(self, expect):
try:
desc = self.project.history_entry.description
self.assertTrue(expect in desc)
except AssertionError:
raise AssertionError('Expecting "%s" in "%s"' % (expect, desc))

31
google/test/test_history.py Normal file
View File

@ -0,0 +1,31 @@
#!/usr/bin/env python
"""
test_history.py
"""
# Copyright (c) 2011 Paul Makepeace, Real Programmers. All rights reserved.
import unittest
from google.refine.history import *
class HistoryTest(unittest.TestCase):
def test_init(self):
response = {
u"code":"ok",
u"historyEntry": {
u"id":1303851435223,
u"description":"Split 4 cells",
u"time":"2011-04-26T16:45:08Z"
}
}
he = response['historyEntry']
entry = HistoryEntry(he['id'], he['time'], he['description'])
self.assertEqual(entry.id, 1303851435223)
self.assertEqual(entry.description, 'Split 4 cells')
self.assertEqual(entry.time, '2011-04-26T16:45:08Z')
if __name__ == '__main__':
unittest.main()

152
google/test/test_tutorial.py
View File

@ -57,7 +57,7 @@ class TutorialTestFacets(refinetest.RefineTestCase):
ethnicity_facet.include('B') ethnicity_facet.include('B')
response = self.project.get_rows() response = self.project.get_rows()
self.assertEqual(response.filtered, 1255) self.assertEqual(response.filtered, 1255)
indexes = [r.index for r in response.rows] indexes = [row.index for row in response.rows]
self.assertEqual(indexes, [1, 2, 3, 4, 6, 12, 18, 26, 28, 32]) self.assertEqual(indexes, [1, 2, 3, 4, 6, 12, 18, 26, 28, 32])
# {8} # {8}
response = self.project.compute_facets() response = self.project.compute_facets()
@ -131,23 +131,22 @@ class TutorialTestEditing(refinetest.RefineTestCase):
# Section "3. Cell Editing": {1} # Section "3. Cell Editing": {1}
self.project.engine.remove_all() # redundant due to setUp self.project.engine.remove_all() # redundant due to setUp
# {2} # {2}
response = self.project.text_transform(column='Zip Code 2', self.project.text_transform(column='Zip Code 2',
expression='value.toString()[0, 5]') expression='value.toString()[0, 5]')
self.assertTrue('6067' in response['historyEntry']['description']) self.assertInResponse('transform on 6067 cells in column Zip Code 2')
# {3} - XXX history # {3} - XXX history
# {4} # {4}
office_title_facet = facet.TextFacet('Office Title') office_title_facet = facet.TextFacet('Office Title')
self.project.engine.add_facet(office_title_facet) self.project.engine.add_facet(office_title_facet)
response = self.project.compute_facets() response = self.project.compute_facets()
self.assertEqual(len(response.facets[0].choices), 76) self.assertEqual(len(response.facets[0].choices), 76)
response = self.project.text_transform('Office Title', 'value.trim()') self.project.text_transform('Office Title', 'value.trim()')
self.assertTrue('6895' in response['historyEntry']['description']) self.assertInResponse('6895')
response = self.project.compute_facets() response = self.project.compute_facets()
self.assertEqual(len(response.facets[0].choices), 67) self.assertEqual(len(response.facets[0].choices), 67)
# {5} # {5}
response = self.project.edit('Office Title', self.project.edit('Office Title', 'Councilmen', 'Councilman')
'Councilmen', 'Councilman') self.assertInResponse('13')
self.assertTrue('13' in response['historyEntry']['description'])
response = self.project.compute_facets() response = self.project.compute_facets()
self.assertEqual(len(response.facets[0].choices), 66) self.assertEqual(len(response.facets[0].choices), 66)
# {6} # {6}
@ -161,11 +160,11 @@ class TutorialTestEditing(refinetest.RefineTestCase):
self.assertEqual(clusters[0][0]['count'], 233) self.assertEqual(clusters[0][0]['count'], 233)
# Not strictly necessary to repeat 'Council Member' but a test # Not strictly necessary to repeat 'Council Member' but a test
# of mass_edit, and it's also what the front end sends. # of mass_edit, and it's also what the front end sends.
response = self.project.mass_edit('Office Title', [{ self.project.mass_edit('Office Title', [{
'from': ['Council Member', 'Councilmember'], 'from': ['Council Member', 'Councilmember'],
'to': 'Council Member' 'to': 'Council Member'
}]) }])
self.assertTrue('372' in response['historyEntry']['description']) self.assertInResponse('372')
response = self.project.compute_facets() response = self.project.compute_facets()
self.assertEqual(len(response.facets[0].choices), 65) self.assertEqual(len(response.facets[0].choices), 65)
@ -184,15 +183,14 @@ class TutorialTestEditing(refinetest.RefineTestCase):
facet.TextFacet('Candidate Name', match['value'])) facet.TextFacet('Candidate Name', match['value']))
self.assertEqual(len(response.rows), 1) self.assertEqual(len(response.rows), 1)
for row in response.rows: for row in response.rows:
response = self.project.star_row(row) self.project.star_row(row)
self.assertTrue(str(row.index + 1) in self.assertInResponse(str(row.index + 1))
response['historyEntry']['description'])
# {5}, {6}, {7} # {5}, {6}, {7}
response = self.project.compute_facets(facet.StarredFacet(True)) response = self.project.compute_facets(facet.StarredFacet(True))
self.assertEqual(len(response.facets[0].choices), 2) # true & false self.assertEqual(len(response.facets[0].choices), 2) # true & false
self.assertEqual(response.facets[0].choices[True].count, 3) self.assertEqual(response.facets[0].choices[True].count, 3)
response = self.project.remove_rows() self.project.remove_rows()
self.assertTrue('3 rows' in response['historyEntry']['description']) self.assertInResponse('3 rows')
class TutorialTestDuplicateDetection(refinetest.RefineTestCase): class TutorialTestDuplicateDetection(refinetest.RefineTestCase):
@ -203,40 +201,36 @@ class TutorialTestDuplicateDetection(refinetest.RefineTestCase):
# Duplicate Row Detection and Deletion" # Duplicate Row Detection and Deletion"
# {7}, {8} # {7}, {8}
response = self.project.get_rows(sort_by='email') response = self.project.get_rows(sort_by='email')
indexes = [r.index for r in response.rows] indexes = [row.index for row in response.rows]
self.assertEqual(indexes, [4, 9, 8, 3, 0, 2, 5, 6, 1, 7]) self.assertEqual(indexes, [4, 9, 8, 3, 0, 2, 5, 6, 1, 7])
# {9} # {9}
response = self.project.reorder_rows() self.project.reorder_rows()
self.assertEqual('Reorder rows', self.assertInResponse('Reorder rows')
response['historyEntry']['description'])
response = self.project.get_rows() response = self.project.get_rows()
indexes = [r.index for r in response.rows] indexes = [row.index for row in response.rows]
self.assertEqual(indexes, range(10)) self.assertEqual(indexes, range(10))
# {10} # {10}
response = self.project.add_column('email', 'count', self.project.add_column('email', 'count',
'facetCount(value, "value", "email")') 'facetCount(value, "value", "email")')
self.assertTrue('column email by filling 10 rows' in self.assertInResponse('column email by filling 10 rows')
response['historyEntry']['description'])
response = self.project.get_rows() response = self.project.get_rows()
self.assertEqual(self.project.column_order['email'], 0) # i.e. 1st self.assertEqual(self.project.column_order['email'], 0) # i.e. 1st
self.assertEqual(self.project.column_order['count'], 1) # i.e. 2nd self.assertEqual(self.project.column_order['count'], 1) # i.e. 2nd
counts = [r['count'] for r in response.rows] counts = [row['count'] for row in response.rows]
self.assertEqual(counts, [2, 2, 1, 1, 3, 3, 3, 1, 2, 2]) self.assertEqual(counts, [2, 2, 1, 1, 3, 3, 3, 1, 2, 2])
# {11} # {11}
self.assertFalse(self.project.has_records) self.assertFalse(self.project.has_records)
response = self.project.blank_down('email') self.project.blank_down('email')
self.assertTrue('Blank down 4 cells' in self.assertInResponse('Blank down 4 cells')
response['historyEntry']['description'])
self.assertTrue(self.project.has_records) self.assertTrue(self.project.has_records)
response = self.project.get_rows() response = self.project.get_rows()
emails = [1 if r['email'] else 0 for r in response.rows] emails = [1 if row['email'] else 0 for row in response.rows]
self.assertEqual(emails, [1, 0, 1, 1, 1, 0, 0, 1, 1, 0]) self.assertEqual(emails, [1, 0, 1, 1, 1, 0, 0, 1, 1, 0])
# {12} # {12}
blank_facet = facet.BlankFacet('email', selection=True) blank_facet = facet.BlankFacet('email', selection=True)
# {13} # {13}
response = self.project.remove_rows(blank_facet) self.project.remove_rows(blank_facet)
self.assertTrue('Remove 4 rows' in self.assertInResponse('Remove 4 rows')
response['historyEntry']['description'])
self.project.engine.remove_all() self.project.engine.remove_all()
response = self.project.get_rows() response = self.project.get_rows()
email_counts = [(row['email'], row['count']) for row in response.rows] email_counts = [(row['email'], row['count']) for row in response.rows]
@ -256,31 +250,23 @@ class TutorialTestTransposeColumnsIntoRows(refinetest.RefineTestCase):
def test_transpose_columns_into_rows(self): def test_transpose_columns_into_rows(self):
# Section "5. Structural Editing, Transpose Columns into Rows" # Section "5. Structural Editing, Transpose Columns into Rows"
# {1}, {2}, {3} # {1}, {2}, {3}
response = self.project.transpose_columns_into_rows( self.project.transpose_columns_into_rows('FY1946', 64, 'pair')
'FY1946', 64, 'pair') self.assertInResponse('64 column(s) starting with FY1946')
self.assertTrue('64 column(s) starting with FY1946' in
response['historyEntry']['description'])
# {4} # {4}
response = self.project.add_column('pair', 'year', self.project.add_column('pair', 'year', 'value[2,6].toNumber()')
'value[2,6].toNumber()') self.assertInResponse('filling 26185 rows')
self.assertTrue('filling 26185 rows' in
response['historyEntry']['description'])
# {5} # {5}
response = self.project.text_transform(column='pair', self.project.text_transform(column='pair',
expression='value.substring(7).toNumber()') expression='value.substring(7).toNumber()')
self.assertTrue('transform on 26185 cells' in self.assertInResponse('transform on 26185 cells')
response['historyEntry']['description'])
# {6} # {6}
response = self.project.rename_column('pair', 'amount') self.project.rename_column('pair', 'amount')
self.assertTrue('Rename column pair to amount' in self.assertInResponse('Rename column pair to amount')
response['historyEntry']['description'])
# {7} # {7}
response = self.project.fill_down('country_name') self.project.fill_down('country_name')
self.assertTrue('Fill down 23805 cells' in self.assertInResponse('Fill down 23805 cells')
response['historyEntry']['description']) self.project.fill_down('program_name')
response = self.project.fill_down('program_name') self.assertInResponse('Fill down 23805 cells')
self.assertTrue('Fill down 23805 cells' in
response['historyEntry']['description'])
# spot check of last row for transforms and fill down # spot check of last row for transforms and fill down
response = self.project.get_rows() response = self.project.get_rows()
row10 = response.rows[9] row10 = response.rows[9]
@ -301,28 +287,26 @@ class TutorialTestTransposeFixedNumbeOfRowsIntoColumns(
# {1} # {1}
self.assertTrue('Column' in self.project.column_order) self.assertTrue('Column' in self.project.column_order)
# {8} # {8}
response = self.project.transpose_rows_into_columns('Column', 4) self.project.transpose_rows_into_columns('Column', 4)
self.assertTrue('Transpose every 4 cells in column Column' in self.assertInResponse('Transpose every 4 cells in column Column')
response['historyEntry']['description'])
# {9} - renaming column triggers a bug in Refine # {9} - renaming column triggers a bug in Refine
# {10} # {10}
response = self.project.add_column('Column 1', 'Transaction', self.project.add_column('Column 1', 'Transaction',
'if(value.contains(" sent "), "send", "receive")') 'if(value.contains(" sent "), "send", "receive")')
self.assertTrue('Column 1 by filling 4 rows' in self.assertInResponse('Column 1 by filling 4 rows')
response['historyEntry']['description'])
# {11} # {11}
transaction_facet = facet.TextFacet(column='Transaction', transaction_facet = facet.TextFacet(column='Transaction',
selection='send') selection='send')
self.project.engine.add_facet(transaction_facet) self.project.engine.add_facet(transaction_facet)
self.project.compute_facets() self.project.compute_facets()
# {12}, {13}, {14} # {12}, {13}, {14}
response = self.project.add_column('Column 1', 'Sender', self.project.add_column('Column 1', 'Sender',
'value.partition(" sent ")[0]') 'value.partition(" sent ")[0]')
# XXX resetting the facet shows data in rows with Transaction=receive # XXX resetting the facet shows data in rows with Transaction=receive
# which shouldn't have been possible with the facet. # which shouldn't have been possible with the facet.
response = self.project.add_column('Column 1', 'Recipient', self.project.add_column('Column 1', 'Recipient',
'value.partition(" to ")[2].partition(" on ")[0]') 'value.partition(" to ")[2].partition(" on ")[0]')
response = self.project.add_column('Column 1', 'Amount', self.project.add_column('Column 1', 'Amount',
'value.partition(" sent ")[2].partition(" to ")[0]') 'value.partition(" sent ")[2].partition(" to ")[0]')
# {15} # {15}
transaction_facet.reset().include('receive') transaction_facet.reset().include('receive')
@ -344,21 +328,17 @@ class TutorialTestTransposeFixedNumbeOfRowsIntoColumns(
'cells["Column 1"].value.partition(" received ")[2]' 'cells["Column 1"].value.partition(" received ")[2]'
'.partition(" from ")[0]') '.partition(" from ")[0]')
): ):
response = self.project.text_transform(column, expression) self.project.text_transform(column, expression)
self.assertTrue('2 cells' in self.assertInResponse('2 cells')
response['historyEntry']['description'])
# {17} # {17}
transaction_facet.reset() transaction_facet.reset()
# {18} # {18}
response = self.project.text_transform('Column 1', self.project.text_transform('Column 1', 'value.partition(" on ")[2]')
'value.partition(" on ")[2]') self.assertInResponse('4 cells')
self.assertTrue('4 cells' in
response['historyEntry']['description'])
# {19} # {19}
response = self.project.reorder_columns([ self.project.reorder_columns(['Transaction', 'Amount', 'Sender',
'Transaction', 'Amount', 'Sender', 'Recipient']) 'Recipient'])
self.assertEqual('Reorder columns', self.assertInResponse('Reorder columns')
response['historyEntry']['description'])
class TutorialTestTransposeVariableNumbeOfRowsIntoColumns(refinetest.RefineTestCase): class TutorialTestTransposeVariableNumbeOfRowsIntoColumns(refinetest.RefineTestCase):
@ -368,19 +348,17 @@ class TutorialTestTransposeVariableNumbeOfRowsIntoColumns(refinetest.RefineTestC
def test_transpose_variable_number_of_rows_into_columns(self): def test_transpose_variable_number_of_rows_into_columns(self):
# {20}, {21} # {20}, {21}
response = self.project.add_column('Column', 'First Line', self.project.add_column('Column', 'First Line',
'if(value.contains(" on "), value, null)') 'if(value.contains(" on "), value, null)')
self.assertTrue('Column by filling 4 rows' in self.assertInResponse('Column by filling 4 rows')
response['historyEntry']['description'])
response = self.project.get_rows() response = self.project.get_rows()
first_names = [row['First Line'][0:10] if row['First Line'] else None first_names = [row['First Line'][0:10] if row['First Line'] else None
for row in response.rows] for row in response.rows]
self.assertEqual(first_names, ['Tom Dalton', None, None, None, self.assertEqual(first_names, ['Tom Dalton', None, None, None,
'Morgan Law', None, None, None, None, 'Eric Batem']) 'Morgan Law', None, None, None, None, 'Eric Batem'])
# {22} # {22}
response = self.project.move_column('First Line', 0) self.project.move_column('First Line', 0)
self.assertTrue('Move column First Line to position 0' in self.assertInResponse('Move column First Line to position 0')
response['historyEntry']['description'])
self.assertEqual(self.project.column_order['First Line'], 0) self.assertEqual(self.project.column_order['First Line'], 0)
# {23} # {23}
self.project.engine.mode = 'record-based' self.project.engine.mode = 'record-based'
@ -388,28 +366,24 @@ class TutorialTestTransposeVariableNumbeOfRowsIntoColumns(refinetest.RefineTestC
self.assertEqual(response.mode, 'record-based') self.assertEqual(response.mode, 'record-based')
self.assertEqual(response.filtered, 4) self.assertEqual(response.filtered, 4)
# {24} # {24}
response = self.project.add_column('Column', 'Status', self.project.add_column('Column', 'Status',
'row.record.cells["Column"].value[-1]') 'row.record.cells["Column"].value[-1]')
self.assertTrue('filling 18 rows' in self.assertInResponse('filling 18 rows')
response['historyEntry']['description'])
# {25} # {25}
response = self.project.text_transform('Column', self.project.text_transform('Column',
'row.record.cells["Column"].value[1, -1].join("|")') 'row.record.cells["Column"].value[1, -1].join("|")')
self.assertTrue('18 cells' in self.assertInResponse('18 cells')
response['historyEntry']['description'])
# {26} # {26}
self.project.engine.mode = 'row-based' self.project.engine.mode = 'row-based'
# {27} # {27}
blank_facet = facet.BlankFacet('First Line', selection=True) blank_facet = facet.BlankFacet('First Line', selection=True)
response = self.project.remove_rows(blank_facet) self.project.remove_rows(blank_facet)
self.assertEqual('Remove 14 rows', self.assertInResponse('Remove 14 rows')
response['historyEntry']['description'])
self.project.engine.remove_all() self.project.engine.remove_all()
# {28} # {28}
'Split 4 cell(s) in column Column into several columns by separator' 'Split 4 cell(s) in column Column into several columns by separator'
response = self.project.split_column('Column', separator='|') self.project.split_column('Column', separator='|')
self.assertTrue('Split 4 cell(s) in column Column' in self.assertInResponse('Split 4 cell(s) in column Column')
response['historyEntry']['description'])
if __name__ == '__main__': if __name__ == '__main__':