def test_merge_differing_capitalization(self):
# this tests #19
insert_rows(self.scratch_db, 'brand', [
dict(brand='CardScan',
company='Newell Rubbermaid',
scraper_id='sr.campaign.hrc'),
dict(brand='Cardscan',
company='Newell Rubbermaid',
scraper_id='sr.campaign.hrc'),
])
insert_row(self.output_db, 'scraper_company_map', dict(
company='Newell Rubbermaid',
scraper_company='Newell Rubbermaid',
scraper_id='sr.campaign.hrc')
)
build_scraper_brand_map_table(self.output_db, self.scratch_db)
self.assertEqual(
select_all(self.output_db, 'scraper_brand_map'),
[dict(brand='CardScan',
company='Newell Rubbermaid',
scraper_brand='CardScan',
scraper_company='Newell Rubbermaid',
scraper_id='sr.campaign.hrc'),
dict(brand='CardScan',
company='Newell Rubbermaid',
scraper_brand='Cardscan',
scraper_company='Newell Rubbermaid',
scraper_id='sr.campaign.hrc'),
])
def test_discard_null_judgment(self):
# this tests issue #22
insert_row(self.scratch_db, 'claim', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='',
claim='>80% of variables are metasyntactic'))
build_claim_table(self.output_db, self.scratch_db)
self.assertEqual(select_all(self.output_db, 'claim'), [])
def test_one_row_group(self):
ONE_ROW = dict(
company='Foo',
scraper_company='Foo',
scraper_id='sr.campaign.bar')
insert_row(self.output_db, 'scraper_company_map', ONE_ROW)
self.assertEqual(
list(select_groups(
self.output_db, 'scraper_company_map', ['company'])),
[(('Foo',), [ONE_ROW])])
def test_discard_null_judgment(self):
# this tests issue #22
insert_row(self.scratch_db, 'rating', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='',
scope=''))
build_rating_table(self.output_db, self.scratch_db)
self.assertEqual(select_all(self.output_db, 'rating'), [])
def setUp(self):
super().setUp()
# have company/brand maps ready
insert_row(self.output_db, 'scraper_company_map', dict(
scraper_id='sr.campaign.qux',
company='Foo',
scraper_company='Foo & Co.'))
insert_row(self.output_db, 'scraper_brand_map', dict(
scraper_id='sr.campaign.qux',
company='Foo',
brand='Bar',
scraper_company='Foo & Co.',
scraper_brand='BAR™'))
def test_map_brand(self):
insert_row(self.scratch_db, 'rating', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='BAR™',
judgment=-1))
build_rating_table(self.output_db, self.scratch_db)
self.assertEqual(
[strip_null(row) for row in select_all(self.output_db, 'rating')],
[dict(campaign_id='qux',
company='Foo',
brand='Bar',
scope='',
judgment=-1)])
def test_map_brand(self):
insert_row(self.scratch_db, 'claim', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='BAR™',
claim='code of conduct mandates unreadable code',
judgment=-1))
build_claim_table(self.output_db, self.scratch_db)
self.assertEqual(
[strip_null(row) for row in select_all(self.output_db, 'claim')],
[dict(campaign_id='qux',
company='Foo',
brand='Bar',
scope='',
claim='code of conduct mandates unreadable code',
judgment=-1)])
def test_map_company(self):
insert_row(self.scratch_db, 'claim', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='',
claim='>80% of variables are metasyntactic',
judgment=1))
build_claim_table(self.output_db, self.scratch_db)
self.assertEqual(
[strip_null(row) for row in select_all(self.output_db, 'claim')],
[dict(campaign_id='qux',
company='Foo',
brand='',
scope='',
claim='>80% of variables are metasyntactic',
judgment=1)])
def test_infer_judgment_from_grade(self):
# this tests issue #23
insert_row(self.scratch_db, 'rating', dict(
scraper_id='sr.campaign.qux',
campaign_id='qux',
company='Foo & Co.',
brand='',
scope='',
grade='B'))
build_rating_table(self.output_db, self.scratch_db)
self.assertEqual(
[strip_null(row) for row in select_all(self.output_db, 'rating')],
[dict(campaign_id='qux',
company='Foo',
brand='',
scope='',
grade='B',
judgment=1)])
def insert_rows(db, table_name, rows):
"""Call insert_row() multiple times. Would want to use multi-row
insert syntax before putting this in msd.db (but we don't actually
need it)."""
for row in rows:
insert_row(db, table_name, row)
