def fetch_store_detail(s, data, isOfficial=False):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.name_e] = cm.html2plain(s['name']).strip()
entry[cm.country_e] = data['country']
val = cm.html2plain(s['city']).strip().upper()
entry[cm.city_e] = cm.extract_city(
val if val and val != '' else data['city'])[0]
entry[cm.addr_e] = cm.html2plain(s['address']).strip()
entry[cm.email] = s['email'].strip()
entry[cm.tel] = s['phone'].strip()
entry[cm.fax] = s['fax'].strip()
entry[cm.store_class] = 'Official Retailer' if isOfficial else 'Retailer'
try:
entry[cm.lat] = string.atof(s['lat']) if s['lat'] != '' else ''
except (ValueError, KeyError, TypeError) as e:
cm.dump('Error in fetching lat: %s' % str(e), log_name)
try:
entry[cm.lng] = string.atof(s['lng']) if s['lng'] != '' else ''
except (ValueError, KeyError, TypeError) as e:
cm.dump('Error in fetching lng: %s' % str(e), log_name)
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
return entry
def fetch_stores(data):
body = data['body']
start = body.find(u'<ul class="storelist storelist_%s' % data['code'])
if start == -1:
cm.dump('Error in finding stores for %s' % data['code'])
return []
body = cm.extract_closure(body[start:], ur'<ul\b', ur'</ul>')[0]
store_list = []
for m in re.findall(ur'<li class="sitem">(.+?)</li>', body, re.S):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
m1 = re.search(ur'<h3>(.+?)</h3>', m)
if m1 is not None:
entry[cm.name_c] = m1.group(1).strip()
m1 = re.search(ur'<div class="addr">(.+?)</div>', m)
if m1 is not None:
entry[cm.addr_e] = m1.group(1).replace(u'地址:', '').replace(u'地址：', '').strip()
m1 = re.search(ur'<div class="tel">(.+?)</div>', m)
if m1 is not None:
entry[cm.tel] = m1.group(1).replace(u'电话:', '').replace(u'电话：', '').strip()
entry[cm.city_c] = data['city']
ret = gs.look_up(data['city'], 3)
if ret is not None:
entry[cm.city_e] = ret['name_e']
entry[cm.city_c] = ret['name_c']
if ret['province'] != '':
entry[cm.province_e] = ret['province']['name_e']
entry[cm.country_e] = u'CHINA'
gs.field_sense(entry)
cm.dump('(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e]), 'canali_log.txt')
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_store_details(data):
url = data['host'] + data['url']
try:
body = cm.get_data(url)
except Exception:
cm.dump('Error in fetching store details: %s' % url, log_name)
return []
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
start = body.find(ur'<h3>available in store</h3>')
if start != -1:
type_sub = cm.extract_closure(body[start:], ur'<ul\b', ur'</ul>')[0]
entry[cm.store_type] = ', '.join(
cm.html2plain(tmp).strip()
for tmp in re.findall(ur'<li[^<>]*>(.+?)</li>', type_sub, re.S))
start = body.find(ur"<div class='gmap_info_box'")
if start == -1:
cm.dump('Error in fetching store details: %s' % url, log_name)
return []
body = cm.extract_closure(body[start:], ur'<div\b', ur'</div>')[0]
raw = json.loads(cm.extract_closure(body, ur'\{', ur'\}')[0])['table']
entry[cm.name_e] = cm.html2plain(raw['name'])
entry[cm.city_e] = data['city'].strip().upper()
entry[cm.country_e] = data['country'].strip().upper()
# entry[cm.store_type] = data['store_type']
entry[cm.addr_e] = cm.reformat_addr(raw['address'])
m = re.search(re.compile(ur'phone:(.*?)fax:(.*?)', re.I | re.S),
raw['phone'])
if m is not None:
entry[cm.tel] = m.group(1).strip()
entry[cm.fax] = m.group(2).strip()
else:
m = re.search(re.compile(ur'phone:(.*?)', re.I | re.S), raw['phone'])
if m is not None:
entry[cm.tel] = m.group(1).strip()
m = re.search(re.compile(ur'fax:(.*?)', re.I | re.S), raw['phone'])
if m is not None:
entry[cm.fax] = m.group(1).strip()
entry[cm.hours] = raw['hours']
if raw['lat'] is not None and raw['lat'] != '':
entry[cm.lat] = string.atof(raw['lat'])
if raw['lng'] is not None and raw['lng'] != '':
entry[cm.lat] = string.atof(raw['lng'])
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None:
entry[cm.province_e] = ret[1]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
log_name)
db.insert_record(entry, 'stores')
return [entry]
def fetch_stores(data):
url = data['url']
try:
body = cm.get_data(url)
except Exception:
cm.dump('Error in fetching stores: %s' % url, log_name)
return []
store_list = []
for m1 in re.finditer(ur'<lignecountry\s+titre\s*=\s*"([^"]+)"', body):
country = m1.group(1).strip().upper()
if country == 'U.S.A.':
country = 'US'
sub_country = cm.extract_closure(body[m1.start():], ur'<lignecountry\b', ur'</lignecountry>')[0]
for m2 in re.finditer(ur'<lignecity\s+titre\s*=\s*"([^"]+)"', sub_country):
city = m2.group(1).strip().upper()
sub_city = cm.extract_closure(sub_country[m2.start():], ur'<lignecity\b', ur'</lignecity>')[0]
m3 = re.search(ur'<!\[CDATA\[(.+?)\]\]>', sub_city, re.S)
if m3 is None:
continue
sub_city = m3.group(1)
store_subs = re.split(ur'<\s*h2\s*>\s*LANVIN BOUTIQUE\s*<\s*/h2\s*>', sub_city)
for s in store_subs:
if s.strip() == '':
continue
m4 = re.search(ur'<p>(.+?)</p>', s, re.S)
if m4 is None:
continue
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.country_e] = country
entry[cm.city_e] = city
s = m4.group(1)
m4 = re.search(ur'(.+?)\n\s*\n', s, re.S)
if m4 is not None:
entry[cm.addr_e] = cm.reformat_addr(m4.group(1))
m4 = re.search(ur'Phone:(.+?)\n\s*\n', s, re.S)
if m4 is not None:
entry[cm.tel] = cm.reformat_addr(m4.group(1).strip())
m4 = re.search(ur'Boutique Hours:(.+?)\n\s*\n', s, re.S)
if m4 is not None:
entry[cm.hours] = cm.reformat_addr(m4.group(1).strip())
m4 = re.search(ur'Products available:(.+?)\n\s*\n', s, re.S)
if m4 is not None:
entry[cm.store_type] = m4.group(1).strip()
m4 = re.search(ur'Email:\s*<a href="mailto:([^"]+)">', s)
if m4 is not None:
entry[cm.email] = m4.group(1).strip()
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
gs.field_sense(entry)
cm.dump('(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e],
entry[cm.country_e],
entry[cm.continent_e]), log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_stores(db, data, logger):
"""
获得商店信息
:param data:
"""
url = data['post_url']
try:
html = cm.post_data(url, {
'pid': data['city_id'],
'lang': 'en',
'action': 'popola_box_DX'
})
if html.strip() == u'':
logger.error(
unicode.format(u'Failed to fetch stores for city {0}',
data['city_id']))
return []
body = pq(html)
except Exception as e:
print 'Error occured in getting city list: %s' % url
dump_data = {
'level': 2,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
store_list = []
for item in (pq(temp) for temp in body('a[href]')):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.url] = item[0].attrib['href']
entry[cm.name_e] = item('h3.titleShop')[0].text.strip()
# terms = cm.reformat_addr(item('div.txtBoxSingleStore p.lineHeight14')[0].text).split(',')
terms = cm.reformat_addr(
unicode(item('div.txtBoxSingleStore p.lineHeight14'))).split(',')
tel = cm.extract_tel(terms[-1])
if tel != '':
terms = terms[:-1]
entry[cm.tel] = tel
entry[cm.addr_e] = u', '.join([v.strip() for v in terms])
entry['country_e'] = data['country_e']
entry['city_e'] = data['city_e']
gs.field_sense(entry)
logger.info(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]))
store_list.append(entry)
cm.insert_record(db, entry, 'spider_stores.stores')
return store_list
def fetch(level=1, data=None, user='******', passwd=''):
db = cm.StoresDb()
db.connect_db(user=user, passwd=passwd)
db.execute(u'DELETE FROM %s WHERE brand_id=%d' % ('stores', brand_id))
try:
html = cm.get_data(url)
except Exception:
print 'Error occured: %s' % url
dump_data = {
'level': 1,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': brand_id
}
cm.dump(dump_data)
return []
js = json.loads(html)
store_list = []
for s in js['data']['list']:
entry = cm.init_store_entry(brand_id, brandname_e, brandname_c)
cm.update_entry(
entry, {
cm.lat: string.atof(s['geo']['lat']),
cm.lng: string.atof(s['geo']['lng'])
})
entry[cm.name_e] = s['contact']['title']
entry[cm.addr_e] = cm.reformat_addr(s['contact']['address'])
entry[cm.tel] = s['contact']['phone']
entry[cm.fax] = s['contact']['fax']
entry[cm.hours] = cm.reformat_addr(s['contact']['hours'])
entry[cm.store_type] = s['contact']['selling']
entry[cm.url] = host + s['link']
gs.update_city_map(s['city'], s['country'], s['continent'])
cm.update_entry(
entry, {
cm.continent_e: s['continent'],
cm.country_e: s['country'],
cm.city_e: s['city']
})
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (
brandname_e, brand_id, entry[cm.name_e], entry[cm.addr_e],
entry[cm.country_e], entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
db.disconnect_db()
gs.commit_maps(1)
gs.commit_maps(3)
return store_list
def fetch_stores(data):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
code = data['country_code']
if gs.look_up(code, 1) is None:
entry[cm.country_e] = cm.html2plain(data['country']).strip().upper()
else:
entry[cm.country_e] = code
entry[cm.name_e] = data['store_name']
entry[cm.city_e] = cm.extract_city(data['city'])[0]
entry[cm.lat] = data['lat'] if data['lat'] is not None else ''
entry[cm.lng] = data['lng'] if data['lng'] is not None else ''
m = re.search(ur'data-boutique\s*=\s*"%s"' % data['store_id'],
data['content'])
sub = data['content'][m.end():]
m1 = re.search(ur'<li class="isDistributeur[^<>]+>(.+?)</li>', sub)
if m1 is not None:
entry[cm.store_class] = cm.reformat_addr(m1.group(1))
m1 = re.search(ur'<li class="place-title[^<>]+>(.+?)</li>', sub, re.S)
if m1 is not None:
entry[cm.addr_e] = cm.reformat_addr(m1.group(1))
m1 = re.search(ur'<li class="contacts[^<>]+>(.+?)</li>', sub, re.S)
if m1 is not None:
m2 = re.search(ur'<a class="popupLaunch" href="([^"]+)"', m1.group(1))
if m2:
entry = fetch_details(data, m2.group(1), entry)
m2 = re.search(ur'<p>(.+?)</p>', m1.group(1), re.S)
if m2:
ct_list = tuple(
tmp.strip()
for tmp in cm.reformat_addr(m2.group(1)).split(','))
entry[cm.tel] = cm.extract_tel(ct_list[0])
if len(ct_list) > 1:
entry[cm.email] = ct_list[1].strip()
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
log_name)
db.insert_record(entry, 'stores')
return tuple(entry)
def fetch_uk(body, data):
start = body.find(u'<div class="fableft">')
if start == -1:
print 'Error in finding %s stores' % data['name']
return []
body, start, end = cm.extract_closure(body[start:], ur'<div\b', ur'</div>')
if end == 0:
print 'Error in finding %s stores' % data['name']
return []
store_list = []
for m in re.findall(ur'<div>\s*(.+?)\s*</div>', body, re.S):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.country_e] = data['name']
addr_list = re.findall(ur'<p>\s*(.+?)\s*</p>', m)
tel = cm.extract_tel(addr_list[-1])
if tel != '':
entry[cm.tel] = tel
del addr_list[-1]
if data['name'] == 'AUSTRALIA':
country, province, city = gs.addr_sense(', '.join(addr_list),
data['name'])
if city is not None:
entry[cm.city_e] = city
if province is not None:
entry[cm.province_e] = province
else:
city = addr_list[-2].strip().upper()
entry[cm.city_e] = city
ret = gs.look_up(city, 3)
if ret is not None and ret['country']['name_e'] == gs.look_up(
'UK', 1)['name_e']:
entry[cm.city_e] = ret['name_e']
entry[cm.zip_code] = addr_list[-1].strip().upper()
entry[cm.addr_e] = ', '.join(addr_list)
entry[cm.city_e] = cm.extract_city(entry[cm.city_e])[0]
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s, %s)' % (
data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.city_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_stores(data):
url = data['url']
try:
body = cm.get_data(url)
except Exception:
cm.dump('Error occured in fetching stores: %s' % url, 'canali_log.txt')
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
store_list = []
for m in re.finditer(ur'<div class="storeInfo">', body):
sub = cm.extract_closure(body[m.start():], ur'<div\b', ur'</div>')[0]
m1 = re.search(ur'<span itemprop="streetAddress">(.+?)</span>', sub)
if m1 is None:
cm.dump('Error: failed to find the address: %s' % url,
'canali_log.txt')
continue
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.addr_e] = cm.reformat_addr(m1.group(1))
entry[cm.country_e] = data['country']
entry[cm.city_e] = data['city']
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None:
entry[cm.province_e] = ret[1]
m1 = re.search(ur'<span itemprop="telephone">(.+?)</span>', sub)
if m1 is not None:
entry[cm.tel] = m1.group(1).strip()
m1 = re.search(ur'data-latitude="(.+?)"', sub)
if m1 is not None:
entry[cm.lat] = string.atof(m1.group(1))
m1 = re.search(ur'data-longitude="(.+?)"', sub)
if m1 is not None:
entry[cm.lng] = string.atof(m1.group(1))
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
'canali_log.txt')
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_stores(data):
url = '%s/en/shops/%s.json' % (
data['host'], urllib.quote(data['country_id'].encode('utf-8')))
try:
body = cm.get_data(url)
raw = json.loads(body)
except Exception:
cm.dump('Error in fetching stores: %s' % url, 'unode50_log.txt')
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
store_items = []
for s in raw['distributors']:
s['store_class'] = 'distributor'
store_items.append(s)
for s in raw['shops']:
s['store_class'] = 'shop'
store_items.append(s)
store_list = []
for s in store_items:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.store_class] = s['store_class']
entry[cm.country_e] = data['country']
entry[cm.name_e] = s['title']
if s['lat'] is not None:
entry[cm.lat] = string.atof(str(s['lat']))
if s['lng'] is not None:
entry[cm.lng] = string.atof(str(s['lng']))
entry[cm.addr_e] = s['address']
entry[cm.city_e] = cm.extract_city(s['city'])[0]
entry[cm.tel] = s['phone']
entry[cm.zip_code] = s['postal_code']
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.city_e], entry[cm.country_e],
entry[cm.continent_e]), 'unode50_log.txt')
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_stores(data):
url = data['host'] + data['data_url']
if data['province_code'] != '':
param = {
'region': data['continent'],
'country': data['country_code'],
'state': data['province_code']
}
else:
param = {'region': data['continent'], 'country': data['country_code']}
try:
body = cm.get_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
raw = json.loads(body)['stores']
store_list = []
for s in raw:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.name_e] = s['name']
entry[cm.province_e] = data['province_code'].strip().upper()
entry[cm.country_e] = data['country_code'].strip().upper()
entry[cm.city_e] = s['city'].strip().upper()
addr_list = []
for tmp in ['address%d' % idx for idx in xrange(1, 4)]:
if tmp in s and s[tmp] is not None and s[tmp].strip() != '':
addr_list.append(s[tmp].strip())
entry[cm.addr_e] = ', '.join(addr_list)
entry[cm.tel] = s['phone']
entry[cm.zip_code] = s['postalCode']
hours = ''
for item in s['openingHours']:
hours += '%s %s' % (item['days'], item['openingHours'])
entry[cm.hours] = hours
entry[cm.url] = s['storePageUrl']
coords = s['coords']
entry[cm.lat] = string.atof(coords['lat'])
entry[cm.lng] = string.atof(coords['lng'])
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_stores(data):
"""
获得商店信息
:param data:
"""
url = data['url']
try:
info = json.loads(cm.get_data(url, {'tskay': data['key_term']}))
except Exception:
print 'Error occured: %s' % url
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
raw_list = info['shops']
store_list = []
for s in raw_list:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.city_e] = s['city'].strip().upper()
entry[cm.country_e] = data['country_e'].strip().upper()
entry[cm.name_e] = s['name'].strip()
addr = s['address']
entry[cm.addr_e] = addr
terms = addr.split(',')
if len(terms) > 1 and entry[cm.city_e] in terms[-1].strip().upper():
country = entry['country_e']
tmp = gs.look_up(country, 1)
if tmp is not None:
country = tmp['name_e']
if country == 'JAPAN':
# 日本邮编
m = re.search(ur'\d{3,}[ -\.]+?\d{3,}', terms[-1])
if m is not None:
entry[cm.zip_code] = m.group(0)
else:
m = re.search(ur'\d{4,}', terms[-1])
if m is not None:
entry[cm.zip_code] = m.group(0)
entry[cm.tel] = s['tel']
entry[cm.fax] = s['fax']
entry[cm.email] = s['email']
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
store_list.append(entry)
# db.insert_record(entry, 'stores')
return store_list
def fetch_stores(data):
url = data['store_url']
param = {'myid': data['key'], 'idioma': 'in'}
try:
body = cm.get_data(url, param)
except Exception:
cm.dump('Error in fetching countries: %s, %s' % (url, param), log_name)
return []
store_list = []
for s in json.loads(body):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.country_e] = data['country']
entry[cm.city_e] = cm.extract_city(data['city'])[0]
entry[cm.name_e] = cm.reformat_addr(s['title'])
m = re.search(ur'(.+?)-\s*<', s['key'])
addr_list = [entry[cm.name_e]]
if m is not None:
m1 = re.search(ur'-+', m.group(1))
if m1 is not None:
tmp = [m.group(1)[:m1.start()], m.group(1)[m1.end():]]
else:
tmp = [m.group(1)]
if len(tmp) > 1:
entry[cm.tel] = cm.extract_tel(tmp[1])
m1 = re.search(ur'\d{4,}', tmp[0])
if m1 is not None:
entry[cm.zip_code] = m1.group()
addr_list.append(tmp[0].strip())
entry[cm.addr_e] = ', '.join(addr_list)
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e])
if ret[0] is not None and entry[cm.country_e] == '':
entry[cm.country_e] = ret[0]
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_hk(data):
loc_list = ('Hong Kong', 'Kowloon', 'Macau', 'New Territories')
url = 'http://levi.com.hk/hk/storelocator'
store_list = []
for loc in loc_list:
param = {'loc': loc}
try:
body = cm.get_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s' % param, log_name)
continue
start = body.find(ur'<div id="addWrapper">')
if start == -1:
cm.dump('Error in fetching stores: %s' % param, log_name)
continue
sub = cm.extract_closure(body[start:], ur'<ul>', ur'</ul>')[0]
for s in re.findall(ur'<li>(.+?)</li>', sub, re.S):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.country_e] = 'MACAU' if loc == 'Macau' else 'HONG KONG'
entry[cm.city_e] = entry[cm.country_e]
m = re.search(ur'<div id="addStore">([^<>]+)', s)
entry[cm.addr_e] = cm.html2plain(m.group(1)) if m else ''
m = re.search(ur'<div id="addAddress">([^<>]+)', s)
tmp = cm.html2plain(m.group(1))
pat = re.compile(ur'business hours?\s*[:\.]?\s*', re.I)
if re.search(pat, tmp):
entry[cm.hours] = re.sub(pat, '', tmp).strip()
m = re.search(ur'<div id="addPhone">([^<>]+)', s)
tmp = cm.html2plain(m.group(1))
pat = re.compile(ur'(tel|phone|telephone)?\s*[:\.]?\s*', re.I)
if re.search(pat, tmp):
entry[cm.tel] = re.sub(pat, '', tmp).strip()
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump('(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e]), log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_details(data):
url = data[cm.url]
try:
html = cm.get_data(url)
except Exception:
print 'Error occured: %s' % url
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = data[cm.name_e]
start = html.find(ur'<div class="field-address">')
if start == -1:
return []
sub, start, end = cm.extract_closure(html[start:], ur'<div\b', ur'</div>')
if end == 0:
return []
m1 = re.search(ur'<div class="locality">(.+?)</div>', sub)
if m1 is not None:
entry[cm.city_e] = cm.extract_city(m1.group(1))[0]
m1 = re.search(ur'<div class="postal-code">(.+?)</div>', sub)
if m1 is not None:
entry[cm.zip_code] = m1.group(1).strip()
entry[cm.country_e] = data[cm.country_e]
pat = re.compile(ur'<[^<>]+?>', re.S)
entry[cm.addr_e] = cm.reformat_addr(re.sub(pat, u'\r\n', sub))
m1 = re.search(ur'<div class="field-telephone"><a href=".+?" class="tel">(.+?)</a></div>', html)
if m1 is not None:
entry[cm.tel] = m1.group(1).strip()
m1 = re.search(ur'<div class="field-opening-hours">\s*<p>(.+?)</p>\s*</div>', html, re.S)
if m1 is not None:
entry[cm.hours] = cm.reformat_addr(m1.group(1))
m1 = re.search(ur'"coordinates":\[(-?\d+\.\d{4,})\s*,\s*(-?\d+\.\d{4,})\]', html)
if m1 is not None:
lat = string.atof(m1.group(1))
lng = string.atof(m1.group(2))
cm.update_entry(entry, {cm.lat: lat, cm.lng: lng})
entry[cm.continent_e] = data[cm.continent_e]
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
return [entry]
def func(item):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.name_e] = cm.html2plain(item('h6')[0].text).strip()
addr_sub = unicode(pq(item('p')[0]))
addr_list = [
term.strip() for term in cm.reformat_addr(addr_sub).split(',')
]
tel = cm.extract_tel(addr_list[-1])
if tel != '':
entry[cm.tel] = tel
del addr_list[-1]
entry[cm.addr_e] = ', '.join(addr_list)
temp = item('a.track_map[href]')
m = hashlib.md5()
m.update(url)
if len(temp) > 0:
map_ref = temp[0].attrib['href']
m.update(map_ref)
m_query = re.search(r'q=([^;]+?)&', cm.html2plain(map_ref))
if m_query:
query_parm = m_query.group(1).replace('+', ' ')
entry['geo_query_param'] = query_parm
else:
m.update(entry[cm.addr_e])
fingerprint = m.hexdigest()
entry[cm.native_id] = fingerprint
if entry[cm.native_id] in data['store_list']:
return
entry[cm.country_e] = data['country']
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e])
if ret[0] is not None and entry[cm.country_e] == '':
entry[cm.country_e] = ret[0]
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
logger.info(
('(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e])))
cm.insert_record(db, entry, data['table'])
return entry
def fetch_stores(data):
url = data['home_url']
try:
body = cm.post_data(url, {'lz_sf': data['province'], 'lz_sx': data['city']})
except Exception:
cm.dump('Error in fetching stores: %s, %s, %s' % (url, data['province'], data['city']),
'samsonite_log.txt')
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
start = body.find(u'搜索结果')
if start == -1:
cm.dump('Error in fetching stores: %s, %s, %s' % (url, data['province'], data['city']),
'samsonite_log.txt')
return []
body = body[start + 4:]
store_list = []
for m in re.findall(ur'</script>\s*(\S+)\s*</span>', body, re.S):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = m.strip()
entry[cm.addr_e] = m.strip()
entry[cm.city_c] = data['city']
ret = gs.look_up(data['city'], 3)
if ret is not None:
entry[cm.city_e] = cm.extract_city(ret['name_e'])[0]
if ret['province'] != '':
entry[cm.province_e] = ret['province']['name_e']
entry[cm.province_c] = data['province']
ret = gs.look_up(data['province'], 2)
if ret is not None:
entry[cm.province_e] = ret['name_e']
entry[cm.country_e] = u'CHINA'
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump('(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e]), 'benetton_log.txt', False)
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_store_details(data):
# http://maps.oasis-stores.com/index-v2.php?coutnryISO=GB&brand=oasis&lat=51.42014&lng=-0.20954
url = data['store_url']
code = data['country_code']
city = data['city_e']
try:
html = cm.get_data(url, {'latitude': data['lat'], 'longitude': data['lng'], 'brand': 'oasis'})
except Exception:
print 'Error occured: %s' % url
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
raw = json.loads(html)
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = raw['name']
addr_list = []
for i in xrange(1, 4):
tmp = cm.html2plain(raw['address%d' % i]).strip()
if tmp != '':
addr_list.append(tmp)
entry[cm.addr_e] = ', '.join(addr_list)
state = raw['countryRegion']
if state is not None and state.strip() != '':
entry[cm.province_e] = state.strip().upper()
state = raw['state']
if state is not None and state.strip() != '':
entry[cm.province_e] = state.strip().upper()
state = raw['county']
if state is not None and state.strip() != '':
entry[cm.province_e] = state.strip().upper()
entry[cm.zip_code] = raw['postcode']
entry[cm.country_e] = data['country_e']
entry[cm.city_e] = cm.extract_city(data['city_e'])[0]
entry[cm.lat] = string.atof(data['lat'])
entry[cm.lng] = string.atof(data['lng'])
entry[cm.tel] = raw['phone']
entry[cm.email] = raw['email']
tmp = ['mon', 'tue', 'wed', 'thu', 'fri', 'sat', 'sun']
entry[cm.hours] = ', '.join([raw[d + '_open_times'] for d in tmp])
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
return [entry]
def fetch_stores(data):
url = data['home_url']
try:
body = cm.get_data(url)
except Exception:
print 'Error occured: %s' % url
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
raw = json.loads(body)['results']
store_list = []
for key in raw:
store = raw[key]
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = store['post_title']
entry[cm.url] = store['post_permalink'].replace(u'\\', '')
entry[cm.country_e] = store['country'].strip().upper()
entry[cm.city_e] = store['city'].strip().upper()
if '_yoox_store_latlong' in store:
m = re.findall(ur'-?\d+\.\d+', store['_yoox_store_latlong'])
if len(m) == 2:
entry[cm.lat] = string.atof(m[0])
entry[cm.lng] = string.atof(m[1])
if 'store_phone' in store:
entry[cm.tel] = store['store_phone'].replace('P:', '').replace('T:', '') \
.replace('P', '').replace('T', '').strip()
if 'store_email' in store:
entry[cm.email] = store['store_email']
if 'store_fax' in store:
entry[cm.fax] = store['store_fax'].replace('F:', '').replace('F', '').strip()
if 'store_hours' in store:
entry[cm.hours] = cm.reformat_addr(store['store_hours'])
if 'store_address' in store:
entry[cm.addr_e] = cm.reformat_addr(store['store_address'])
if 'women' in store and 'men' in store:
entry[cm.store_type] = 'Women: %s, men: %s' % (', '.join(store['women']), ', '.join(store['men']))
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_stores_cn(data):
vals = ['Storenew%d.html' % idx for idx in xrange(1, 10)]
vals.append('Store.html')
store_list = []
for url in ('http://www.hushpuppies.com.cn/%s' % term for term in vals):
try:
body = cm.get_data(url)
except Exception, e:
cm.dump('Error in fetching stores: %s' % url, log_name)
return ()
city_map = dict((m[0].strip(), m[1].strip()) for m in re.findall(
ur'<a rel="([^"]+)" href="#"[^<>]*>([^<>]+)', body))
start = body.find(ur'<div id="all-list-wrap" style="float:left">')
if start == -1:
cm.dump('Error in fetching stores: %s' % url, log_name)
return ()
sub = cm.extract_closure(body[start:], ur'<div\b', ur'</div>')[0]
for m in re.findall(ur'<ul id="([^"]+)"[^<>]*>(.+?)</ul>', sub, re.S):
city = city_map[m[0].strip()]
for store in re.findall(ur'<li><a>([^<>]+)', m[1]):
entry = cm.init_store_entry(data['brand_id'],
data['brandname_e'],
data['brandname_c'])
entry[cm.country_e] = u'CHINA'
entry[cm.city_e] = city
entry[cm.addr_e] = store.strip()
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e]), log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_stores(db, data, logger):
q = pq(url='http://www.paulandjoe.com/en/ozcms/stores/list/?country_id=&postcode=')
store_list = []
# Country
country_a = q('#store_list>li>a')
country_b = q('#store_list>li>ul')
assert (len(country_a) == len(country_b))
for i in xrange(len(country_a)):
country = country_a[i].text.strip().upper()
store_a = pq(country_b[i])('a.marker-store')
store_b = pq(country_b[i])('span.store-infos')
assert (len(store_a) == len(store_b))
for j in xrange(len(store_a)):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
lat = store_a[j].attrib['data-latitude']
lat = float(lat) if lat else None
lng = store_a[j].attrib['data-longitude']
lng = float(lng) if lng else None
if lat and lng:
entry[cm.lat], entry[cm.lng] = lat, lng
entry[cm.name_e] = store_a[j].text
entry[cm.addr_e] = cm.reformat_addr(str(pq(store_b[j]))) # cm.reformat_addr(str(store_b[j]))
entry[cm.country_e] = country
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e])
if ret[0] is not None and entry[cm.country_e] == '':
entry[cm.country_e] = ret[0]
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
logger.info('(%s/%d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e],
entry[cm.country_e],
entry[cm.continent_e]))
store_list.append(entry)
cm.insert_record(db, entry, 'spider_stores.stores')
return tuple(store_list)
def fetch_stores(data):
url = data['url']
try:
body = cm.get_data(url, {'countryCode': data['country_code']})
except Exception:
print 'Error occured in fetching stores: %s, %s' % (url, data['country_code'])
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
raw = json.loads(body)['stores']
store_list = []
for key in raw:
store = raw[key]
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = store['name']
addr = store['address1']
if store['address2'] != '':
addr += ', ' + store['address2']
entry[cm.addr_e] = addr
entry[cm.zip_code] = store['postalCode']
entry[cm.city_e] = store['city'].strip().upper()
entry[cm.province_e] = store['stateCode'].strip().upper()
entry[cm.country_e] = data['country_code']
entry[cm.tel] = store['phone']
entry[cm.fax] = store['fax']
entry[cm.email] = store['email']
entry[cm.hours] = store['storeHours']
if store['latitude'] != '':
entry[cm.lat] = string.atof(store['latitude'])
if store['longitude'] != '':
entry[cm.lng] = string.atof(store['longitude'])
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_stores(data):
url = data['store_url']
try:
body = cm.get_data(url, {
'country': data['country'],
'city': data['city']
})
except Exception:
print 'Error occured: %s' % url
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
raw = json.loads(body)
store_list = []
for item in raw['items']:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.country_e] = data['country'].strip().upper()
tmp = cm.extract_city(data['city'])[0]
if entry[cm.country_e] == 'USA':
entry[cm.province_e] = tmp
else:
entry[cm.city_e] = tmp
gs.field_sense(entry)
addr = cm.reformat_addr(item['address'].replace(u'\\', ''))
addr_list = [tmp.strip() for tmp in addr.split(',')]
tel = cm.extract_tel(addr_list[-1])
if tel != '':
entry[cm.tel] = tel
del addr_list[-1]
entry[cm.addr_e] = ', '.join(addr_list)
entry[cm.store_type] = item['shop_type']
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e])
if ret[0] is not None and entry[cm.country_e] == '':
entry[cm.country_e] = ret[0]
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (
data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_cn(data):
url = 'http://www.lee.com.cn/xml/storefinder.xml'
store_list = []
for store in (pq(tmp) for tmp in pq(url=url)('NewDataSet Table')):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
tmp = store('shop_name')[0]
entry[cm.name_e] = tmp.text if tmp.text else ''
entry[cm.country_e] = 'CHINA'
tmp = store('city')[0]
entry[cm.city_e] = tmp.text if tmp.text else ''
tmp = store('province')[0]
entry[cm.province_e] = tmp.text if tmp.text else ''
tmp = store('district')[0]
entry[cm.district_e] = tmp.text if tmp.text else ''
tmp = store('address')[0]
entry[cm.addr_e] = tmp.text if tmp.text else ''
tmp = store('tel')[0]
entry[cm.tel] = tmp.text if tmp.text else ''
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e])
if ret[0] is not None and entry[cm.country_e] == '':
entry[cm.country_e] = ret[0]
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.city_e], entry[cm.country_e],
entry[cm.continent_e]), log_name)
db.insert_record(entry, 'stores')
store_list.append(entry)
return tuple(store_list)
def fetch_stores(data):
url = data['host'] + '/en/jil_store_stores.json'
try:
body = cm.get_data(url, {'city': data['city']})
except Exception:
print 'Error occured: %s' % url
dump_data = {'level': 0, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
raw = json.loads(body)
store_list = []
for store in raw:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = store['title']
entry[cm.country_e] = store['country'].strip().upper()
entry[cm.city_e] = store['city'].strip().upper()
entry[cm.lat] = string.atof(store['lat'])
entry[cm.lng] = string.atof(store['lng'])
entry[cm.tel] = store['telephone']
addr = store['address']
terms = addr.split(',')
m = re.search(ur'\d{5,}', terms[-1])
if m is not None:
entry[cm.zip_code] = m.group(0)
entry[cm.addr_e] = addr
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (data['brandname_e'], data['brand_id'],
entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
entry[cm.zip_code] = raw['postcode']
entry[cm.city_e] = cm.extract_city(raw['city'])[0]
entry[cm.province_e] = raw['region']
entry[cm.tel] = raw['phone']
entry[cm.fax] = raw['fax']
entry[cm.email] = raw['email']
entry[cm.hours] = cm.reformat_addr(raw['opening'])
entry[cm.lat] = string.atof(raw['latitude'])
entry[cm.lng] = string.atof(raw['longitude'])
entry[cm.url] = raw['link'].replace('\\', '')
raw = js['Country']
entry[cm.country_e] = raw['name']
raw = js['StoreLocatorType']
entry[cm.store_type] = raw['name']
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (
brandname_e, brand_id, entry[cm.name_e], entry[cm.addr_e],
entry[cm.country_e], entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch(level=1, data=None, user='******', passwd=''):
def func(data, level):
"""
:param data:
:param level: 0：国家；1：城市；2：商店列表
def get_stores(url, data):
"""
从json对象中获得商店信息
"""
opener = urllib2.build_opener()
opener.addheaders = [("User-Agent",
"Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko)"
"Chrome/27.0.1453.94 Safari/537.36"),
('Accept', '*/*'), ('X-Requested-With', 'XMLHttpRequest'), ('Connection', 'keep-alive')]
response = opener.open(url)
html = response.read().encode('utf-8')
jsonobj = json.loads(html)
stores = jsonobj[u'Stores'][u'Items']
region_list = jsonobj['Regions']
region_id = jsonobj['Region']
region = ''
if len(region_list) > 0 and region_id != 0:
for val in region_list:
if val['RegionId'] == region_id:
region = val['Name']
break
country = jsonobj['CurrentCountry']['Name']
store_list = []
for s in stores:
# print('Found store: %s, %s. Tel: %s, lat=%s, lng=%s' % (
# s['Name'], s['Address'], s['Phone'], s['Latitude'], s['Longitude']))
store_type = ['']
# Some stores may have varioius store types
if len(s['StoreTypes']) > 0:
store_type = list(val['Name'] for val in s['StoreTypes'])
if s['Url'] is not None:
url = s['Url']
else:
url = ''
if s['ZipCode'] is not None and not s['ZipCode'].__eq__(''):
zip = s['ZipCode']
else:
zip = ''
local_addr = s['Address']
if local_addr[-1] == '.':
local_addr = local_addr[:-1]
if not zip.__eq__(''):
addr = u'%s, %s, %s' % (local_addr, s['City'], zip)
else:
addr = u'%s, %s' % (local_addr, s['City'])
if region.__eq__(''):
addr = u'%s, %s' % (addr, country)
else:
addr = u'%s, %s, %s' % (addr, region, country)
for t in store_type:
entry = cm.init_store_entry(brand_id, brandname_e, brandname_c)
cm.update_entry(entry, {'addr_e': addr, 'country_e': country, 'city_e': s['City'],
'comments': s['Comments'],
'province_e': region.strip().upper(), 'zip': zip,
'email': s['Email'], 'fax': s['Fax'], 'lat': s['Latitude'], 'lng': s['Longitude'],
'name_e': s['Name'], 'tel': s['Phone'], 'store_type': t, 'url': url})
gs.field_sense(entry)
print '%s Found store: %s, %s (%s, %s)' % (
brandname_e, entry[cm.name_e], entry[cm.addr_e], entry[cm.country_e],
entry[cm.continent_e])
db.insert_record(entry, 'stores')
store_list.append(entry)
return store_list
def fetch_stores(data):
# country=Greece&city=ATHENS&adutl=+01&kids=+02&undercolor=+06&togetmap=mapdata
url = data['data_url']
param = {
'country': data['country'],
'city': data['city'],
'adutl': ' 01',
'kids': ' 02',
'undercolor': ' 06',
'togetmap': 'mapdata'
}
try:
body = cm.get_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param),
'benetton_log.txt', False)
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
store_list = []
for m in re.findall(ur'<marker (.+?)>', body):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
m1 = re.search(ur'name=\\"(.+?)\\"', m)
if m1 is not None:
entry[cm.name_e] = cm.html2plain(
m1.group(1).strip().replace(u'\\', ''))
m1 = re.search(ur'address=\\"(.+?)\\"', m)
if m1 is not None:
addr = cm.reformat_addr(
cm.html2plain(m1.group(1)).replace(u'\\', ''))
tel = cm.extract_tel(addr)
if tel != '':
entry[cm.tel] = tel
addr = addr.replace(tel, '')
entry[cm.addr_e] = cm.reformat_addr(addr)
m1 = re.search(ur'lat=\\"(.+?)\\"', m)
if m1 is not None:
entry[cm.lat] = string.atof(m1.group(1))
m1 = re.search(ur'lng=\\"(.+?)\\"', m)
if m1 is not None:
entry[cm.lng] = string.atof(m1.group(1))
entry[cm.country_e] = data['country'].strip().upper()
entry[cm.city_e] = cm.extract_city(data['city'])[0]
gs.field_sense(entry)
cm.dump(
'(%s / %d) Found store: %s, %s (%s, %s)' %
(data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e]),
'benetton_log.txt', False)
db.insert_record(entry, 'stores')
store_list.append(entry)
def fetch_store_details(data):
url = data['url']
try:
body = cm.get_data(url)
except Exception:
print 'Error occured: %s' % url
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
ret = gs.look_up(data['country'], 1)
if ret is not None:
entry[cm.country_e] = ret['name_e']
m = re.search(ur'<span class="type">Address</span>\s*<p>(.+?)</p>', body,
re.S)
if m is not None:
addr = cm.reformat_addr(m.group(1))
country, province, city = gs.addr_sense(addr)
if country is not None and entry[cm.country_e] == '':
entry[cm.country_e] = country
if province is not None:
entry[cm.province_e] = province
if city is not None:
entry[cm.city_e] = city
entry[cm.addr_e] = addr
m = re.search(ur'<span class="type">Phone</span>\s*<p>(.+?)</p>', body,
re.S)
if m is not None:
entry[cm.tel] = m.group(1)
m = re.search(ur'<span class="type">Opening hours</span>\s*<p>(.+?)</p>',
body, re.S)
if m is not None:
entry[cm.hours] = cm.reformat_addr(m.group(1))
m = re.search(ur'<span class="type">You can find</span>\s*<p>(.+?)</p>',
body, re.S)
if m is not None:
entry[cm.store_type] = cm.reformat_addr(m.group(1))
m = re.search(ur'google.maps.LatLng\(\s*(-?\d+\.\d+)\s*,\s*(-?\d+\.\d+)',
body, re.S)
entry[cm.lat] = string.atof(m.group(1))
entry[cm.lng] = string.atof(m.group(2))
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s)' % (
data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.country_e], entry[cm.continent_e])
db.insert_record(entry, 'stores')
return [entry]
def fetch_stores(data):
url = data['url']
try:
body = cm.post_data(url, {
'rsp': 'json',
'country': data['country_code']
})
except Exception:
print 'Error occured: %s' % url
dump_data = {
'level': 0,
'time': cm.format_time(),
'data': {
'url': url
},
'brand_id': data['brand_id']
}
cm.dump(dump_data)
return []
raw = json.loads(body)
store_list = []
for s in raw['stores']:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.name_e] = cm.html2plain(s['name']).strip()
addr_list = []
for key in ['address1', 'address2']:
if s[key].strip() != '':
addr_list.append(cm.reformat_addr(s[key]))
entry[cm.addr_e] = ' '.join(addr_list)
# r=s['region'].strip().upper()
# m = re.search(ur'\b([A-Z]{2})\b', r)
# if data[cm.country_e]=='UNITED STATES' and m is not None:
# # 美国
# ret = gs.look_up(m.group(1), 2)
# if ret is not None:
# r = ret['name_e']
# entry[cm.province_e] = r
entry[cm.city_e] = cm.extract_city(s['city'])[0]
entry[cm.zip_code] = s['zip'].strip()
entry[cm.country_e] = data[cm.country_e]
entry[cm.lat] = string.atof(s['lat'])
entry[cm.lng] = string.atof(s['lng'])
entry[cm.tel] = s['phone'].strip()
entry[cm.fax] = s['fax'].strip()
entry[cm.email] = s['emailaddress'].strip()
entry[cm.url] = s['website'].strip()
days = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
opening = []
if 'openingHours' in s and s['openingHours'] is not None:
for m in re.finditer(ur'i:(\d);s:\d+:\\?"([^\\"]+?)\\?"',
s['openingHours']):
opening.append(
'%s: %s' %
(days[string.atoi(m.group(1))], m.group(2).strip()))
entry[cm.hours] = ', '.join(opening)
gs.field_sense(entry)
ret = gs.addr_sense(entry[cm.addr_e], entry[cm.country_e])
if ret[1] is not None and entry[cm.province_e] == '':
entry[cm.province_e] = ret[1]
if ret[2] is not None and entry[cm.city_e] == '':
entry[cm.city_e] = ret[2]
gs.field_sense(entry)
print '(%s / %d) Found store: %s, %s (%s, %s, %s)' % (
data['brandname_e'], data['brand_id'], entry[cm.name_e],
entry[cm.addr_e], entry[cm.city_e], entry[cm.country_e],
entry[cm.continent_e])
store_list.append(entry)
db.insert_record(entry, 'stores')
