def get_detail_info(html):
item_additional = re.search(r'<div id="item_additional"(.+)</div>', html, re.DOTALL).group(0)
soup = BeautifulSoup(item_additional)
item_additional = soup.find(id='item_additional')
examine = soup.contents[0].contents[2][1:-1]
spans = soup.findAll('span')
min_price = int(rs_str_to_int(spans[0].contents[2][1:-1]))
price = int(rs_str_to_int(spans[1].contents[2][1:-1]))
max_price = int(rs_str_to_int(spans[2].contents[2][1:-1]))
return min_price, price, max_price, examine
def get_volume_info():
html = read_url(VOLUME_URL)
table = re.search(r'<table id="top100_table"(.+)</table>', html, re.DOTALL).group(0)
soup = BeautifulSoup(table)
raw_entries = soup.findAll('tr')[1:]
for raw_entry in raw_entries:
soup = BeautifulSoup(str(raw_entry))
columns = soup.findAll('td')
id = int(re.search(r'id=(\d+)', str(columns[0])).group(1))
volume = rs_str_to_int(columns[5].contents[0])
yield id, volume
def get_front_volume_info():
html = read_url(FRONT_VOLUME_URL)
top5_left = re.search(r'<div class="top5_left(.+)', html, re.DOTALL).group(0)
soup = BeautifulSoup(top5_left)
table = soup.find('table')
raw_entries = table.findAll('tr')[1:]
for raw_entry in raw_entries:
soup = BeautifulSoup(str(raw_entry))
columns = soup.findAll('td')
id = int(re.search(r'id=(\d+)', str(columns[0])).group(1))
volume = rs_str_to_int(columns[3].contents[0][:-1])
yield id, volume
def get_index_info(html):
table = re.search(r'<table id="search_results_table"(.+)</table>', html, re.DOTALL).group(0)
soup = BeautifulSoup(table)
raw_entries = soup.findAll('tr')[1:-1]
for raw_entry in raw_entries:
soup = BeautifulSoup(str(raw_entry))
columns = soup.findAll('td')
id = int(re.search(r'id=(\d+)', str(columns[0])).group(1))
name = soup.find('img')['alt']
price = rs_str_to_int(columns[2].contents[0])
members = 'star_members.png' in str(columns[4])
yield (id, name, members, price)
def get_volume_info():
html = read_url(VOLUME_URL)
table = re.search(r'<table id="top100_table"(.+)</table>', html,
re.DOTALL).group(0)
soup = BeautifulSoup(table)
raw_entries = soup.findAll('tr')[1:]
for raw_entry in raw_entries:
soup = BeautifulSoup(str(raw_entry))
columns = soup.findAll('td')
id = int(re.search(r'id=(\d+)', str(columns[0])).group(1))
volume = rs_str_to_int(columns[5].contents[0])
yield id, volume
def get_front_volume_info():
html = read_url(FRONT_VOLUME_URL)
top5_left = re.search(r'<div class="top5_left(.+)', html,
re.DOTALL).group(0)
soup = BeautifulSoup(top5_left)
table = soup.find('table')
raw_entries = table.findAll('tr')[1:]
for raw_entry in raw_entries:
soup = BeautifulSoup(str(raw_entry))
columns = soup.findAll('td')
id = int(re.search(r'id=(\d+)', str(columns[0])).group(1))
volume = rs_str_to_int(columns[3].contents[0][:-1])
yield id, volume