def TestIPNewsHtmlToText(self):
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '02011001'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': '<p>The story body line 1<br>Line 2</p>\
<p>abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi more</p>',
'word_count': '1',
'priority': '1'
}
with self.app.app_context():
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, item = f.format(article, subscriber)[0]
expected = '\r\nThe story body line 1 \r\nLine 2 \r\n\r\nabcdefghi abcdefghi abcdefghi abcdefghi ' \
'abcdefghi abcdefghi abcdefghi abcdefghi \r\nmore'
self.assertEqual(item['article_text'], expected)
def TestMultipleCategories(self):
article = {
'source': 'AAP',
'anpa_category': [{'name': 'Finance', 'qcode': 'F'},
{'name': 'Overseas Sport', 'qcode': 'S'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '04001005'}, {'qcode': '15011002'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': '1',
'task': {'desk': 1}
}
with self.app.app_context():
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
docs = f.format(article, subscriber)
self.assertEqual(len(docs), 2)
for seq, doc in docs:
if doc['category'] == 'S':
self.assertEqual(doc['subject_reference'], '15011002')
self.assertEqual(doc['subject_detail'], 'four-man sled')
if doc['category'] == 'F':
self.assertEqual(doc['subject_reference'], '04001005')
self.assertEqual(doc['subject_detail'], 'viniculture')
codes = set(doc['selector_codes'].split(' '))
expected_codes = set('cxx 0fh axx az and pxx 0ah 0ir 0px 0hw pnd pxd cnd cxd 0nl axd'.split(' '))
self.assertSetEqual(codes, expected_codes)
def testGeoBlockNotTwoStates(self):
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '04001005'}, {'qcode': '15011002'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': '1',
'task': {'desk': 1},
'urgency': 1,
'place': [{'qcode': 'VIC', 'name': 'VIC'}],
'targeted_for': [{'name': 'New South Wales', 'allow': False}, {'name': 'Victoria', 'allow': False}]
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, doc = f.format(article, subscriber)[0]
codes = set(doc['selector_codes'].split(' '))
expected_codes_str = 'an5 an4 an7 an6 ax5 ax6 ax7 an8 px6 ax4 ax8 px5 0ah 0px'
expected_codes_str += ' px8 0fh px7 px4 pn4 pn5 pn6 pn7 px0'
expected_codes = set(expected_codes_str.split(' '))
self.assertSetEqual(codes, expected_codes)
def testGeoBlockNotTwoStates(self):
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '04001005'}, {'qcode': '15011002'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {'desk': 1},
'urgency': 1,
'place': [{'qcode': 'VIC', 'name': 'VIC'}],
'targeted_for': [{'name': 'New South Wales', 'allow': False}, {'name': 'Victoria', 'allow': False}]
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, doc = f.format(article, subscriber)[0]
codes = set(doc['selector_codes'].split(' '))
expected_codes_str = 'an5 an4 an7 an6 ax5 ax6 ax7 an8 px6 ax4 ax8 px5 0ah 0px'
expected_codes_str += ' px8 0fh px7 px4 pn4 pn5 pn6 pn7 px0'
expected_codes = set(expected_codes_str.split(' '))
self.assertSetEqual(codes, expected_codes)
def testIPNewsHtmlToText(self):
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '02011001'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': '<p>The story body line 1<br>Line 2</p>\
<p>abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi abcdefghi more</p>',
'word_count': '1',
'priority': 1
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, item = f.format(article, subscriber)[0]
expected = '\r\nThe story body line 1 \r\nLine 2 \r\n\r\nabcdefghi abcdefghi abcdefghi abcdefghi ' \
'abcdefghi abcdefghi abcdefghi abcdefghi \r\nmore'
self.assertEqual(item['article_text'], expected)
def test_aap_ipnews_formatter_with_body_footer(self):
subscriber = self.app.data.find('subscribers', None, None)[0]
doc = self.article.copy()
doc['body_footer'] = 'call helpline 999 if you are planning to quit smoking'
f = AAPIpNewsFormatter()
seq, item = f.format(doc, subscriber)[0]
self.assertGreater(int(seq), 0)
self.assertEqual(seq, item['sequence'])
item.pop('sequence')
self.assertDictEqual(
item, {
'category': 'a',
'texttab': 't',
'fullStory': 1,
'ident': '0',
'headline': 'VIC:This is a test headline',
'service_level': 'a',
'originator': 'AAP',
'take_key': 'take_key',
'article_text':
'The story body<br>call helpline 999 if you are planning to quit smoking',
'priority': 'f',
'usn': '1',
'subject_matter': 'international law',
'news_item_type': 'News',
'subject_reference': '02011001',
'subject': 'crime, law and justice',
'wordcount': '1',
'subject_detail': 'international court or tribunal',
'genre': 'Current',
'keyword': 'slugline',
'author': 'joe'
})
def testIPNewsFormatter(self):
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, item = f.format(self.article, subscriber)[0]
item = json.loads(item)
self.assertGreater(int(seq), 0)
self.assertEqual(seq, item['sequence'])
item.pop('sequence')
self.assertDictEqual(
item, {
'category': 'a',
'texttab': 't',
'fullStory': 1,
'ident': '0',
'headline': 'VIC:This is a test headline',
'service_level': 'a',
'originator': 'AAP',
'take_key': 'take_key',
'article_text': 'The story body',
'priority': 'f',
'usn': '1',
'subject_matter': 'international law',
'news_item_type': 'News',
'subject_reference': '02011001',
'subject': 'crime, law and justice',
'wordcount': '1',
'subject_detail': 'international court or tribunal',
'genre': 'Current',
'keyword': 'slugline',
'author': 'joe'
})
def testIpNewsFormatterNoSubject(self):
article = {
'source': 'AAP',
'anpa_category': [{
'qcode': 'a'
}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {
'desk': 1
},
'urgency': 1,
'place': [{
'qcode': 'VIC',
'name': 'VIC'
}]
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, doc = f.format(article, subscriber)[0]
doc = json.loads(doc)
self.assertEqual(doc['subject_reference'], '00000000')
self.assertEqual(doc['headline'], 'VIC:This is a test headline')
article = {
'source': 'AAP',
'anpa_category': [{
'qcode': 'a'
}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': None,
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {
'desk': 1
},
'urgency': 1,
'place': None
}
seq, doc = f.format(article, subscriber)[0]
doc = json.loads(doc)
self.assertEqual(doc['subject_reference'], '00000000')
self.assertEqual(doc['headline'], 'This is a test headline')
def Test_is_in_subject(self):
article = {
'subject': [{'qcode': '04001005'}, {'qcode': '15011002'}],
}
f = AAPIpNewsFormatter()
self.assertTrue(f._is_in_subject(article, '150'))
self.assertFalse(f._is_in_subject(article, '151'))
self.assertTrue(f._is_in_subject(article, '04001'))
def Test_set_selector_codes(self):
article = {
'task': {'desk': 1},
'slugline': 'Test',
'urgency': 3
}
f = AAPIpNewsFormatter()
odbc_item = {}
with self.app.app_context():
f._set_selector_codes(article, 'ipnews', odbc_item, 'A')
self.assertSetEqual(set(odbc_item['selector_codes'].split()),
set('and axd pnd cxd 0fh 0ir 0px 0ah 0hw cxx axx cnd 0nl az pxd pxx'.split()))
def testIpNewsFormatterNoSubject(self):
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {'desk': 1},
'urgency': 1,
'place': [{'qcode': 'VIC', 'name': 'VIC'}]
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, doc = f.format(article, subscriber)[0]
doc = json.loads(doc)
self.assertEqual(doc['subject_reference'], '00000000')
self.assertEqual(doc['headline'], 'VIC:This is a test headline')
article = {
'source': 'AAP',
'anpa_category': [{'qcode': 'a'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': None,
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {'desk': 1},
'urgency': 1,
'place': None
}
seq, doc = f.format(article, subscriber)[0]
doc = json.loads(doc)
self.assertEqual(doc['subject_reference'], '00000000')
self.assertEqual(doc['headline'], 'This is a test headline')
def TestIPNewsFormatter(self):
with self.app.app_context():
output_channel = self.app.data.find('output_channels', None, None)[0]
f = AAPIpNewsFormatter()
seq, item = f.format(self.article, output_channel, self.sel_codes)
self.assertGreater(int(seq), 0)
self.assertEquals(seq, item['sequence'])
item.pop('sequence')
self.assertDictEqual(item,
{'category': 'a', 'texttab': 't', 'fullStory': 1, 'ident': '0',
'headline': 'This is a test headline', 'selector_codes': 'aaa bbb',
'service_level': 'a', 'originator': 'AAP', 'take_key': 'take_key',
'article_text': 'The story body', 'priority': '1', 'usn': '1',
'subject_matter': 'international law', 'news_item_type': 'News',
'subject_reference': '02011001', 'subject': 'crime, law and justice',
'wordcount': '1', 'subject_detail': 'international court or tribunal',
'genre': 'Current', 'keyword': 'slugline', 'author': 'joe'})
def testIPNewsFormatter(self):
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
seq, item = f.format(self.article, subscriber)[0]
item = json.loads(item)
self.assertGreater(int(seq), 0)
self.assertEqual(seq, item['sequence'])
item.pop('sequence')
self.assertDictEqual(item,
{'category': 'a', 'texttab': 't', 'fullStory': 1, 'ident': '0',
'headline': 'VIC:This is a test headline', 'service_level': 'a', 'originator': 'AAP',
'take_key': 'take_key', 'article_text': 'The story body', 'priority': 'f', 'usn': '1',
'subject_matter': 'international law', 'news_item_type': 'News',
'subject_reference': '02011001', 'subject': 'crime, law and justice',
'wordcount': '1', 'subject_detail': 'international court or tribunal',
'genre': 'Current', 'keyword': 'slugline', 'author': 'joe'})
def test_aap_ipnews_formatter_with_body_footer(self):
subscriber = self.app.data.find('subscribers', None, None)[0]
doc = self.article.copy()
doc['body_footer'] = 'call helpline 999 if you are planning to quit smoking'
f = AAPIpNewsFormatter()
seq, item = f.format(doc, subscriber)[0]
self.assertGreater(int(seq), 0)
self.assertEqual(seq, item['sequence'])
item.pop('sequence')
self.assertDictEqual(item,
{'category': 'a', 'texttab': 't', 'fullStory': 1, 'ident': '0',
'headline': 'VIC:This is a test headline', 'service_level': 'a', 'originator': 'AAP',
'take_key': 'take_key',
'article_text': 'The story body<br>call helpline 999 if you are planning to quit smoking',
'priority': 'f', 'usn': '1',
'subject_matter': 'international law', 'news_item_type': 'News',
'subject_reference': '02011001', 'subject': 'crime, law and justice',
'wordcount': '1', 'subject_detail': 'international court or tribunal',
'genre': 'Current', 'keyword': 'slugline', 'author': 'joe'})
def testMultipleCategories(self):
article = {
'source': 'AAP',
'anpa_category': [{'name': 'Finance', 'qcode': 'F'},
{'name': 'Overseas Sport', 'qcode': 'S'}],
'headline': 'This is a test headline',
'byline': 'joe',
'slugline': 'slugline',
'subject': [{'qcode': '04001005'}, {'qcode': '15011002'}],
'anpa_take_key': 'take_key',
'unique_id': '1',
'type': 'text',
'body_html': 'body',
'word_count': '1',
'priority': 1,
'task': {'desk': 1},
'place': [{'qcode': 'VIC', 'name': 'VIC'}]
}
subscriber = self.app.data.find('subscribers', None, None)[0]
f = AAPIpNewsFormatter()
docs = f.format(article, subscriber)
self.assertEqual(len(docs), 2)
for seq, doc in docs:
if doc['category'] == 'S':
self.assertEqual(doc['subject_reference'], '15011002')
self.assertEqual(doc['subject_detail'], 'four-man sled')
self.assertEqual(doc['headline'], 'VIC:This is a test headline')
if doc['category'] == 'F':
self.assertEqual(doc['subject_reference'], '04001005')
self.assertEqual(doc['subject_detail'], 'viniculture')
self.assertEqual(doc['headline'], 'VIC:This is a test headline')
codes = set(doc['selector_codes'].split(' '))
expected_codes = set('cxx 0fh axx az and pxx 0ah 0ir 0px 0hw pnd pxd cnd cxd 0nl axd'.split(' '))
self.assertSetEqual(codes, expected_codes)
def Test_join_selector_codes(self):
f = AAPIpNewsFormatter()
result = f._join_selector_codes('ipnewS', 'newsi', 'cnewsi', 'cnewsi')
result_list = result.split()
self.assertEqual(len(result_list), 12)
