def extract_content():
DB = 'mysql+pymysql://homestead:[email protected]/public_opinion?charset=utf8'
session = db_session(DB)
M = db_model(DB, 'corpus')
query = session.query(M)
while True:
corpuses = query.filter(M.status == 'ready').order_by(M.id).limit(30).all()
if not corpuses:
break
for corpus in corpuses:
try:
summary_html = Document(corpus.html).summary(html_partial=True)
content = BS(summary_html).text.strip()
corpus.content = content
session.commit()
except:
corpus.content = '[extract_error]'
session.commit()
print('===> extract_content error, id: ', corpus.id)
corpus.status = 'extracted'
session.commit()
def export_csv():
DB = 'mysql+pymysql://homestead:[email protected]/public_opinion?charset=utf8'
session = db_session(DB)
M = db_model(DB, 'corpus')
query = session.query(M)
csv_file = open(csvfile_path, 'wb')
writer = csv.writer(csv_file)
table_head = ['url', 'website', 'published_at', 'word_freq', 'topic']
writer.writerow(table_head)
offset = 0
limit = 3000
while True:
corpuses = query.filter(M.status == 'marked').order_by(M.id).offset(offset).limit(limit).all()
if not corpuses:
break
for corpus in corpuses:
table_row = [corpus.url, corpus.website, corpus.published_at, corpus.word_freq, corpus.topic]
writer.writerow(table_row)
print('===> write id: ', corpus.id)
offset += limit
csv_file.close()
def mark_topic():
DB = 'mysql+pymysql://homestead:[email protected]/public_opinion?charset=utf8'
session = db_session(DB)
M = db_model(DB, 'corpus')
query = session.query(M)
while True:
corpuses = query.filter(M.status == 'segmented').order_by(M.id)
if not corpuses:
break
for corpus in corpuses:
try:
word_freq = json.loads(corpus.word_freq)
topic = 0
if check_topic1(word_freq):
topic = topic | TOPIC1
if check_topic2(word_freq):
topic = topic | TOPIC2
if check_topic3(word_freq):
topic = topic | TOPIC3
if check_topic4(word_freq):
topic = topic | TOPIC4
if check_topic5(word_freq):
topic = topic | TOPIC5
if check_topic6(word_freq):
topic = topic | TOPIC6
if check_topic7(word_freq):
topic = topic | TOPIC7
if check_topic8(word_freq):
topic = topic | TOPIC8
if check_topic9(word_freq):
topic = topic | TOPIC9
if check_topic10(word_freq):
topic = topic | TOPIC10
corpus.topic = topic
session.commit()
print('===> mark topic, id: ', corpus.id)
except:
print('===> mark topic error, id: ', corpus.id)
corpus.status = 'marked'
session.commit()
def mark_topic():
DB = 'mysql+pymysql://homestead:[email protected]/public_opinion?charset=utf8'
session = db_session(DB)
M = db_model(DB, 'corpus')
query = session.query(M)
while True:
corpuses = query.filter(M.status == 'segmented').order_by(M.id)
if not corpuses:
break
for corpus in corpuses:
try:
word_freq = json.loads(corpus.word_freq)
topic = 0
if check_topic1(word_freq):
topic = topic | TOPIC1
if check_topic2(word_freq):
topic = topic | TOPIC2
if check_topic3(word_freq):
topic = topic | TOPIC3
if check_topic4(word_freq):
topic = topic | TOPIC4
if check_topic5(word_freq):
topic = topic | TOPIC5
if check_topic6(word_freq):
topic = topic | TOPIC6
if check_topic7(word_freq):
topic = topic | TOPIC7
if check_topic8(word_freq):
topic = topic | TOPIC8
if check_topic9(word_freq):
topic = topic | TOPIC9
if check_topic10(word_freq):
topic = topic | TOPIC10
corpus.topic = topic
session.commit()
print('===> mark topic, id: ', corpus.id)
except:
print('===> mark topic error, id: ', corpus.id)
corpus.status = 'marked'
session.commit()
def segment():
DB = 'mysql+pymysql://homestead:[email protected]/public_opinion?charset=utf8'
session = db_session(DB)
M = db_model(DB, 'corpus')
query = session.query(M)
#while True:
corpuses = query.filter(M.status == 'extracted', M.content != '[extract_error]').order_by(M.id)
# if not corpuses:
# break
for corpus in corpuses:
try:
corpus.word_freq = segment_text(corpus.content)
session.commit()
print('===> segment id: ', corpus.id)
except:
print('===> segment error, id: ', corpus.id)
corpus.status = 'segmented'
session.commit()
def open_spider(self, spider):
self.session = session = db_session(DB)
self.model = db_model(DB, 'corpus')
self.query = self.session.query(self.model)
