def test_insert_article_db(self):
self.assertEquals(20, len(self.test.db_price))
a = Article()
a.title = 'test title'
a.date = datetime.utcnow()
a.price = '$1'
a.keywords = ['test', 'title']
a.source = 'test category 1'
self.test.insert_article_db(a)
self.test.print_db(self)
expected = '*** $1.0: 0|test category 1|test title 1|$1|test,keywords1 1|test category 1|test title|$1|test,title *** $2.0: 0|test category 2|test title 2|$2|test,keywords2 *** $3.0: 0|test category 3|test title 3|$3|test,keywords3 *** $4.0: 0|test category 4|test title 4|$4|test,keywords4 *** $5.0: 0|test category 5|test title 5|$5|test,keywords5 *** $6.0: 0|test category 6|test title 6|$6|test,keywords6 *** $7.0: 0|test category 7|test title 7|$7|test,keywords7 *** $8.0: 0|test category 8|test title 8|$8|test,keywords8 *** $9.0: 0|test category 9|test title 9|$9|test,keywords9 *** $10.0: 0|test category 10|test title 10|$10|test,keywords10 *** $11.0: 0|test category 11|test title 11|$11|test,keywords11 *** $12.0: 0|test category 12|test title 12|$12|test,keywords12 *** $13.0: 0|test category 13|test title 13|$13|test,keywords13 *** $14.0: 0|test category 14|test title 14|$14|test,keywords14 *** $15.0: 0|test category 15|test title 15|$15|test,keywords15 *** $16.0: 0|test category 16|test title 16|$16|test,keywords16 *** $17.0: 0|test category 17|test title 17|$17|test,keywords17 *** $18.0: 0|test category 18|test title 18|$18|test,keywords18 *** $19.0: 0|test category 19|test title 19|$19|test,keywords19 *** $20.0: 0|test category 20|test title 20|$20|test,keywords20 **************************'
self.assertEquals(expected, self.output.replace('\n', ' ').replace('\r', ''), '')#'price: ' + case[2] + ' not found in ' + case[0])
def test_get_not_posted_deals_w_children(self): self.assertEquals(20, len(self.test.db_price)) a = Article() a.title = 'test title' a.date = datetime.utcnow() a.price = '$1' a.keywords = ['test', 'keywords1'] a.source = 'test category 2' inserted = self.test.insert_article_db(a) self.assertFalse(inserted.posted) list = self.test.get_not_posted_deals_w_children() self.assertEquals(1, len(list)) for deal in list: deal.posted = True list = self.test.get_not_posted_deals_w_children() self.assertEquals(0, len(list))
def setUp(self): self.output = '' pathname = os.path.dirname(sys.argv[1]) orig_work_path = os.path.abspath(pathname) self.parser = DealParser(orig_work_path) self.test_work_path = os.path.join(orig_work_path, '') self.test = Engine(self.test_work_path, self.parser.get_adjusted_price) for i in range(1,21): u = datetime.utcnow() d = u if i <= 10 else u-timedelta(days=4) a = Article() a.title = 'test title ' + str(i) a.date = d a.price = '$' + str(i) a.keywords = ['test','keywords' + str(i)] a.source = 'test category ' + str(i) self.test.insert_article_db(a)
def articles_add(self, rawdog, config, articles, article_dates):
reImg = re.compile('<img src="http://feeds2\.feedburner\.com.*?>')
reAmazon = re.compile('&tag=.*?["> \Z]')
DB = 'pricentalk2'
HOST = 'localhost'#'pricentalk2.db.4240609.hostedresource.com'
DB_USER = '******'
DB_PASSWORD = '******'
conn = MySQLdb.Connection(db=DB, host=HOST, user=DB_USER, passwd=DB_PASSWORD)
cursor = conn.cursor()
articleCount = 0
for article in articles:
entry_info = article.entry_info
#check if title hashkey is exist
hashKey = ''
try:
#hashKey = hashlib.sha1(entry_info['title_raw']).hexdigest()
hashKey = sha.new(entry_info['title_raw']).hexdigest()
except:
continue
#hashKey = entry_info['title_raw']
sql = """select count(*) from tbl_article
where article_hashkey = '%s';
"""
cursor.execute(sql % hashKey)
result = cursor.fetchone()[0]
if int(result) > 0:
continue
a = Article()
a.title = entry_info['title_raw']
a.category = rawdog.feeds[article.feed].args['category']
a.source = rawdog.feeds[article.feed].args['name']
a.user_id = 'priceandtalk'
a.nickname = 'PriceAndTalk'
a.price = self.parser.get_price(a.title)
a.date = datetime.utcnow()
a.keywords = self.parser.extract_keywords(a.title)
if entry_info.has_key('content'):
for content in entry_info['content']:
a.content = content['value']
elif entry_info.has_key('summary_detail'):
a.content = entry_info['summary_detail']['value']
else:
a.content = ''
content = a.content
if rawdog.feeds[article.feed].args['category'] == 'Amazon':
content = re.sub(reAmazon, '&tag=pric048-20', content)
entry_info['link'] = re.sub(reAmazon, '&tag=pric048-20', entry_info['link'])
if entry_info.has_key('link'):
content += '<p>Source: <a href="' + entry_info['link'] + '">' + entry_info['link'] + '</a></p>'
sys.stderr.write(entry_info['link'] + '\n')
a.content = re.sub(reImg, '', content)
self.pt_poster.add_article(a)
try:
sql = """
INSERT INTO tbl_article(title, category, article_content, article_hashkey, keywords, date_added)
SELECT '%s','%s','%s','%s', '%s', now();
"""
#FROM DUAL
#WHERE (SELECT COUNT(*) FROM tbl_article WHERE article_hashkey = '%s')=0;
#"""
cursor.execute (sql % (a.title.replace("'", "''"), a.category.replace("'", "''"), a.content.replace("'", "''"), hashKey, ','.join(a.keywords).replace("'", "''")))
articleCount = articleCount + 1
#trace.write(sql %(a.content, hashKey, hashKey))
except:
pass
self.trace.write('Exception: %s. \n' % sys.exc_info()[0])
self.trace.write('Exception-Value: %s. \n' % sys.exc_info()[1])
self.trace.write('Exception-Stack: %s. \n' % sys.exc_info()[1])
if not a.price:
continue
self.recommend_engine.insert_article_db(a)
#trace.write('Key: %s: Content:%s' %(hashKey, a.content))
#trace.write(a.content)
#trace.wrtie()
cursor.close()
conn.commit()
conn.close()
self.trace.write('Total article: %d \n' % articleCount )
self.trace.write('--End of File--\n')
