def get_cities(data):
try:
d = {"country": data["country_code"], "city": "", "service": -1}
html = common.post_data(url, d)
except Exception:
print "Error occured in getting the list of countries: %s" % url
dump_data = {"level": 1, "time": common.format_time(), "data": {"data": url}, "brand_id": brand_id}
common.dump(dump_data)
return []
start = html.find(u'<select id="city" name="city">')
if start == -1:
return []
end = html.find(u"</select>", start)
html = html[start:end]
city_list = []
for m in re.findall(ur'<option value="(.+?)">', html):
if data["country_code"] == "GB" and "2 davies street" in m.lower():
continue
elif data["country_code"] == "RO" and "13 september street" in m.lower():
continue
elif "b1603daq" in m.lower():
continue
else:
city_list.append({"city_e": m, "country_e": data["country_e"], "country_code": data["country_code"]})
def fetch_stores(data):
url = data['url']
try:
body = cm.post_data(url)
except Exception, e:
cm.dump('Error in fetching stores: %s' % url, log_name)
return []
def fetch_cities(data):
"""
获得城市列表
:param data:
"""
url = data['post_url']
try:
html = cm.post_data(
url, {
'pid': data['country_id'],
'lang': 'en',
'action': 'popola_select_city'
})
except Exception:
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 []
city_list = []
for m in re.findall(ur'<option value="(\d+)".*?>(.+?)</option>', html):
if m[0] != '0':
d = data.copy()
d['city_id'] = string.atoi(m[0])
d['city_e'] = m[1].strip().upper()
city_list.append(d)
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')
def fetch_cities(data):
country_id = data['country_id']
try:
html = cm.post_data(url, {'country_id': country_id})
except Exception:
print 'Error occured: %s' % url
dump_data = {
'level': 2,
'time': cm.format_time(),
'data': data,
'brand_id': brand_id
}
cm.dump(dump_data)
return []
start = html.find('<select name="city_id" id="city_id">')
if start == -1:
return []
start += len('<select name="city_id" id="city_id">')
end = html.find('</select>', start)
html = html[start:end]
city_list = []
for m in re.findall(ur'<option\s.*?value="(\d+).*?">(.*?)</option>', html):
entry = {'city': m[1].strip().upper(), 'city_id': string.atoi(m[0])}
entry['country'] = data['country']
entry['country_id'] = data['country_id']
city_list.append(entry)
def get_cities(data):
try:
d = {'country': data['country_code'], 'city': '', 'service': -1}
html = common.post_data(url, d)
except Exception:
print 'Error occured in getting the list of countries: %s' % url
dump_data = {'level': 1, 'time': common.format_time(), 'data': {'data': url}, 'brand_id': brand_id}
common.dump(dump_data)
return []
start = html.find(u'<select id="city" name="city">')
if start == -1:
return []
end = html.find(u'</select>', start)
html = html[start:end]
city_list = []
for m in re.findall(ur'<option value="(.+?)">', html):
if data['country_code'] == 'GB' and '2 davies street' in m.lower():
continue
elif data['country_code'] == 'RO' and '13 september street' in m.lower():
continue
elif 'b1603daq' in m.lower():
continue
else:
city_list.append({'city_e': m, 'country_e': data['country_e'], 'country_code': data['country_code']})
def fetch_cities(data):
url = data['sel_url']
try:
body = cm.post_data(
url, {
'continent': data['continent'],
'country': data['country'],
'city': '',
'page': 0
})
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)
city_list = []
for c in raw['city']:
d = data.copy()
d['city'] = c
city_list.append(d)
return city_list
def fetch_store_list(data):
url = data['data_url']
param = {
'storelocator': 1,
'dofilter': 1,
'L': 0,
'map_sw': '-90.0%2C-180.0',
'map_ne': '90.0%2C180.0',
'slst': 'c',
'storetype': 1
}
# storelocator=1&dofilter=1&L=0&map_sw=-90.0%2C-180.0&map_ne=90.0%2C180.0&slst=c&storetype=1
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
results = []
for m in re.findall(ur'"id"\s*:\s*"([^"]+)"', body):
tmp = m.split(',')
for val in tmp:
d = data.copy()
d['store_id'] = string.atoi(val)
results.append(d)
def fetch_store_list(data):
url = data['url']
try:
body = cm.post_data(url, {
'cCode': data['country_code'],
'city': data['city'],
'postsearch': 1
})
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 []
results = []
for m in re.finditer(ur'<td class\s*=\s*"ftd"', body):
end = body.find('</tr>', m.start())
sub = body[m.start():end]
m1 = re.search(ur'<td class="ltd"><a href="(.+?)">', sub)
if m1 is None:
print 'Cannot find details: %s / %s' % (data['country_code'],
data['city'])
else:
d = data.copy()
d['url'] = data['host'] + m1.group(1)
results.append(d)
def get_countries(data):
"""
返回国家列表
:rtype : [{'country_id':**, 'country':**}, ...]
:param data:
:return:
"""
url = data['url']
try:
html = cm.post_data(url, {'country': -1, 'city': -1, 'recordit': -1})
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 []
ret = []
for m in re.findall(
ur'<li>\s*?<a href=.+?country-(\d+).+?">(.+?)<\\/a><\\/li>', html,
re.S):
country_id = string.atoi(m[0].strip())
country = m[1].replace(r'\r', '').replace(r'\n', '').strip().upper()
ret.append({'country_id': country_id, 'country': country, 'url': url})
def get_store_list(data):
"""
获得城市中的商店列表
:param data:
:return:
"""
url = data['url']
try:
html = cm.post_data(url, {
'country': data['country_id'],
'city': data['city_id'],
'recordid': -1
})
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 []
pass
ret = []
for m in re.findall(ur'<a href=.+?store-(\d+).+?">', html, re.S):
store_id = string.atoi(m.strip())
entry = dict(data)
entry['store_id'] = store_id
ret.append(entry)
def fetch_cities(data, logger):
"""
获得城市列表
:param data:
"""
url = data['post_url']
try:
html = cm.post_data(
url, {
'pid': data['country_id'],
'lang': 'en',
'action': 'popola_select_city'
})
body = pq(html)
except Exception:
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 []
city_list = []
for item in body('option[value!="0"]'):
d = data.copy()
d['city_id'] = int(item.attrib['value'])
d['city_e'] = item.text.upper().strip()
city_list.append(d)
return city_list
def fetch_cities(data):
url = data['url']
try:
body = cm.post_data(
url, {
'searchtype': 'normal',
'reiter_selected': 'reiter1',
'country_id': data['country_code'],
'city_id': 0
})
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 []
m = re.search(ur'<option value="0"[^>]*>city</option>', body)
if m is None:
return []
end = body.find(u'</select>', m.end())
city_list = []
for c in re.findall(ur'<option value="(.+?)"[^>]*>.+?</option>',
body[m.end():end]):
d = data.copy()
d['city'] = c
city_list.append(d)
def fetch_stores(data):
url = data['url']
try:
body = cm.post_data(url)
except Exception, e:
cm.dump('Error in fetching stores: %s' % url, log_name)
return []
def fetch_stores(data):
param = {'action': 'getStoresFromAjax', 'country': data['country_code'],
'region': data['city'], 'collection': ''}
url = data['url']
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
store_list = []
for m1 in re.finditer(ur'<div class="shop-type-container">', body):
sub = cm.extract_closure(body[m1.start():], ur'<div\b', ur'</div>')[0]
store_class = ''
m2 = re.search(ur'<div class="shop-type-title">(.+?)</div>', sub, re.S)
if m2 is not None:
store_class = cm.reformat_addr(m2.group(1))
for m2 in re.finditer(ur'<div class="shop"', sub):
store_sub = cm.extract_closure(sub[m2.start():], ur'<div\b', ur'</div>')[0]
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.store_class] = store_class
entry[cm.country_e] = data['country_code']
entry[cm.city_e] = cm.extract_city(data['city'])[0]
m3 = re.search(ur'loadStore\((\d+)\s*,\s*(-?\d+\.\d+)\s*,\s*(-?\d+\.\d+)\)', store_sub)
if m3 is not None:
data['store_id'] = string.atoi(m3.group(1))
entry[cm.lat] = string.atof(m3.group(2))
entry[cm.lng] = string.atof(m3.group(3))
entry[cm.store_type] = ', '.join(get_detail(data))
m3 = re.search(ur'<div class="shop-name shop-details shop-main-name">([^<>]+)</div>', store_sub)
if m3 is not None:
entry[cm.name_e] = m3.group(1).strip()
addr_list = []
m3 = re.search(ur'<div class="shop-street shop-details">([^<>]+)</div>', store_sub)
if m3 is not None:
addr_list.append(cm.reformat_addr(m3.group(1)))
m3 = re.search(ur'<div class="shop-city shop-details">([^<>]+)</div>', store_sub)
if m3 is not None:
tmp = cm.reformat_addr(m3.group(1))
m3 = re.search(ur'(\d{4,})', tmp)
if m3 is not None:
entry[cm.zip_code] = m3.group(1).strip()
addr_list.append(tmp)
entry[cm.addr_e] = ', '.join(addr_list)
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, %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)
def fetch_stores_eu(data):
url = data['data_url_eu']
param = {'query': data['city_code']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
def fetch_stores_eu(data):
url = data['data_url_eu']
param = {'query': data['city_code']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
def fetch_store_list(data):
url = data['data_url']
param = {'zone': data['zone_id']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching store list: %s, %s' % (url, param), log_name)
return []
def fetch_cities(data):
url = data['url']
param = {'ref': 'com', 'fld_Coun_Id': data['country_id'], 'search': data['search_type']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return ()
def fetch_countries(data):
url = data['data_url']
param = {'node_id': data['continent_id'], 'location_id': 0}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching countries: %s, %s' % (url, param), log_name)
return ()
def fetch_stores(data):
url = data['data_url']
param = {'countryId': data['country_id']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
def fetch_states(data):
url = data['data_url']
param = {'node_id': data['country_id'], 'location_id': 1}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching states: %s, %s' % (url, param), log_name)
return ()
def fetch_stores(data):
url = data['url']
param = {'searchinput': data['country']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
def fetch_stores(data):
url = data['data_url']
param = {'action': 'yoox_storelocator_change_city', 'city_id': data['city_id'], 'dataType': 'JSON'}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
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_stores(data):
url = data['url']
try:
body = cm.post_data(url, {}, {'Content-Type': 'application/json; charset=utf-8',
'Content-Length': 0, 'Connection': 'keep-alive',
'Pragma': 'no-cache', 'Cache-Control': 'no-cache'})
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, e), log_name)
return []
def fetch_store_list(data):
url = data['data_url']
param = {'zone': data['zone_id']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching store list: %s, %s' % (url, param),
log_name)
return []
def fetch_stores_beauty(data):
url = data['lst_url']
param = {'chkCat[0]': 'FRG', 'chkCat[1]': 'MKP', 'chkCat[2]': 'PRE', 'chkCat[3]': 'EXC', 'div': 'fnb',
'lstCountry': data['country_code'], 'lstCity': data['city']}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching countries: %s' % url, log_name)
return ()
def fetch_stores(data):
"""
获得商店信息
:param data:
:return:
"""
url = data["post_url"]
try:
js = json.loads(
cm.post_data(
url, {"country_id": data["country_id"], "retail_city": "", "retail_type": data["retail_type"]}
).decode("unicode_escape")
)
except Exception:
print "Error occured in getting country list: %s" % url
dump_data = {"level": 1, "time": cm.format_time(), "data": {"url": url}, "brand_id": data["brand_id"]}
cm.dump(dump_data)
return []
# country_id=108&retail_city=&retail_type=retail
# country_id=99&retail_city=&retail_type=service
store_list = []
for s in js:
entry = cm.init_store_entry(data["brand_id"], data["brandname_e"], data["brandname_c"])
tmp = s["retail_name"].strip()
if cm.is_chinese(tmp):
entry[cm.name_c] = tmp
else:
entry[cm.name_e] = tmp
entry[cm.addr_e] = s["retail_gmap"].strip()
entry[cm.zip_code] = s["retail_zipcode"].strip()
entry[cm.city_e] = s["retail_city"].strip().upper()
if s["retail_email"] is not None:
entry[cm.email] = s["retail_email"].strip()
if s["retail_website"] is not None:
entry[cm.url] = s["retail_website"].strip()
if data["retail_type"] == "retail":
entry[cm.store_class] = "Retail"
else:
entry[cm.store_class] = "Service Center"
entry[cm.country_e] = s["country_name"].strip().upper()
entry[cm.continent_e] = s["continent_name"].strip().upper()
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['url']
lat, lng = data['city_lat'], data['city_lng']
param = {'json': 1, 'lat': lat, 'lng': lng, 'latLow': lat - 0.25, 'latHigh': lat + 0.25, 'lngLow': lng - 0.25,
'lngHigh': lng + 0.25, 'includeResellers': 'true'}
try:
body = cm.post_data('%s?%s' % (url, urllib.urlencode(param)))
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
def fetch_stores(data):
"""
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'
})
except Exception:
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 m in re.finditer(ur'<a href="(.+?)".*?>', html):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'],
data['brandname_c'])
entry[cm.url] = m.group(1)
store_html, start, end = cm.extract_closure(html[m.start():],
ur'<a href', ur'</a>')
if end == 0:
continue
m1 = re.findall(ur'<h3 class="titleShop">(.+?)</h3>', store_html, re.S)
if len(m1) > 0:
entry[cm.name_e] = m1[0].strip()
m1 = re.findall(ur'<p\b.*?>(.+?)(?:</p>|</div>)', store_html, re.S)
if len(m1) > 0:
terms = cm.reformat_addr(m1[0]).split(',')
tel = cm.extract_tel(terms[-1])
if tel != '':
terms = terms[:-1]
entry[cm.tel] = tel
entry[cm.addr_e] = ', '.join([v.strip() for v in terms])
entry['country_e'] = data['country_e']
entry['city_e'] = data['city_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])
store_list.append(entry)
db.insert_record(entry, 'stores')
def fetch_stores(data):
url = data['store_url']
param = {'store_country': data['country_code'], 'store_city': data['city_code']}
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
store_list = []
for s in re.findall(ur'<marker\b([^<>]+)/\s*>', body):
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
m = re.search(ur'store_name="([^"]+)"', s)
if m is not None:
entry[cm.name_e] = cm.reformat_addr(m.group(1))
entry[cm.country_e] = data['country_code']
entry[cm.city_e] = data['city']
addr_list = []
for key in ['store_mall_name', 'store_address', 'store_zip_code']:
m = re.search(ur'%s="([^"]+)"' % key, s)
if m is not None:
tmp = cm.reformat_addr(m.group(1))
if tmp != '':
addr_list.append(tmp)
entry[cm.addr_e] = ', '.join(addr_list)
m = re.search(ur'store_zip_code="([^"]+)"', s)
if m is not None:
entry[cm.zip_code] = m.group(1).strip()
m = re.search(ur'store_telephone="([^"]+)"', s)
if m is not None:
entry[cm.tel] = m.group(1).strip()
m = re.search(ur'store_fax="([^"]+)"', s)
if m is not None:
entry[cm.fax] = m.group(1).strip()
m = re.search(ur'store_email="([^"]+)"', s)
if m is not None:
entry[cm.email] = m.group(1).strip()
m = re.search(ur'store_latitude="([^"]+)"', s)
if m is not None:
entry[cm.lat] = string.atof(m.group(1).strip())
m = re.search(ur'store_longitude="([^"]+)"', s)
if m is not None:
entry[cm.lng] = string.atof(m.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(data):
url = data['data_url']
state_id = data['state_id']
if state_id is None:
state_id = 0
param = {'request': 'setStores', 'c': data['country_id'], 's': state_id, 'l': data['city_id'], 'ca': ''}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s' % url, log_name)
return []
def fetch_cities(data):
url = data['data_url']
state_id = data['state_id']
if state_id is None:
param = {'request': 'countryChange', 'c': data['country_id']}
else:
param = {'request': 'stateChange', 'c': data['country_id'], 's': state_id}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
def fetch_cities(data):
url = data['url']
param = {
'ref': 'com',
'fld_Coun_Id': data['country_id'],
'search': data['search_type']
}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return ()
def fetch_stores(data):
url = data['data_url']
param = {'tx_iostorefinder_pi1[__referrer][extensionName]': 'IoStorefinder',
'tx_iostorefinder_pi1[__referrer][controllerName]': 'Store',
'tx_iostorefinder_pi1[__referrer][actionName]': 'search',
'tx_iostorefinder_pi1[countryid]': data['country_code'],
'tx_iostorefinder_pi1[zip]': 'POSTCODE', 'tx_iostorefinder_pi1[city]': 'Town'}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
def get_stores(data):
# data[StoreLocator][pays]=BO
url = data['url']
try:
html = cm.post_data(url, {'data[StoreLocator][pays]': data['country_code'],
'data[StoreLocator][ville]': '',
'data[StoreLocator][etat]': 0})
except Exception, e:
print 'Error occured: %s, %s' % (url, str(e))
dump_data = {'level': 1, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': brand_id}
cm.dump(dump_data)
return []
def fetch_cities(data):
url = data['data_url']
if data['no_region']:
param = {'node_id': data['country_id'], 'location_id': 2, 'no_region': 1}
else:
param = {'node_id': data['state_id'], 'location_id': 2}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return ()
def fetch_cities(data):
url = data['data_url']
param = {'DestinationURL': 'Worldwide-retailers', '(country)': data['country_code']}
try:
body = cm.post_data(url, param)
# m = re.search(ur'META HTTP-EQUIV="Location"\s+Content="([^"]+)"', body)
# if not m:
# raise IOError()
# body = cm.get_data(m.group(1))
except Exception, e:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return ()
def fetch_stores(data):
url = data['url']
try:
body = cm.post_data(url, {}, {
'Content-Type': 'application/json; charset=utf-8',
'Content-Length': 0,
'Connection': 'keep-alive',
'Pragma': 'no-cache',
'Cache-Control': 'no-cache'
})
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, e), log_name)
return []
def get_detail(data):
param = {'action': 'loadStoreFromAjax', 'id': data['store_id']}
url = data['url']
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
m = re.search(ur'<div class="lines">(.+?)</div>', body, re.S)
if m is None:
return ()
return tuple(term.strip() for term in re.findall(ur'<li>(.+?)</li>', m.group(1), re.S))
def fetch_stores(data):
url = data['post_shops']
param = {'city': data['city_e'], 'paulandjoe_women': 0, 'paulandjoe_man': 0,
'paulandjoe_sister': 0, 'paulandjoe_little': 0, 'paulandjoe_beauty': 0}
try:
html = cm.post_data(url, param)
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 []
store_list = []
try:
for store in (pq(tmp) for tmp in pq(html)('ul')):
try:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
entry[cm.name_e] = cm.html2plain(store('li.first')[0].text).strip()
entry[cm.country_e] = data[cm.country_e]
entry[cm.city_e] = data[cm.city_e]
addr_list = []
for term in (cm.reformat_addr(unicode(pq(tmp))) for tmp in store('li[class!="first"]')):
if term != '':
addr_list.append(term)
tel = cm.extract_tel(addr_list[-1])
if tel != '':
entry[cm.tel] = tel
del addr_list[-1]
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)
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')
except (IndexError, TypeError) as e:
cm.dump(u'Error in parsing %s, %s' % (url, param), log_name)
print traceback.format_exc()
continue
except Exception, e:
print traceback.format_exc()
def parse_store(data, body=None):
if body is None:
url = data['url']
try:
body = cm.post_data(url)
except Exception:
cm.dump('Error in fetching stores: %s' % url, log_name)
return []
start = body.find(ur'jQuery.extend(Drupal.settings,')
latlng_map = {}
if start != -1:
for item in json.loads(cm.extract_closure(body[start:], ur'\{', ur'\}')[0])['getlocations']['key_1']['latlons']:
latlng_map[cm.reformat_addr(item[3])] = {'lat': string.atof(item[0]), 'lng': string.atof(item[1])}
def fetch_stores(data):
url = data['store_url']
param = {'univers[mode_femme]': 'pla_women', 'univers[mode_homme]': 'pla_dior_men',
'univers[baby_dior]': 'pla_baby_dior', 'univers[horlogerie]': 'pla_watches',
'univers[joaillerie]': 'pla_fine_jewelry', 'univers[dior_phone]': 'pla_dior_phone',
'continent': data['continent_id'], 'pays': data['country_id'], 'ville': data['city_id'],
'node_id': '581', 'search': 'SEARCH'}
if not data['no_region']:
param['region'] = data['state_id']
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
def get_detail(data):
param = {'action': 'loadStoreFromAjax', 'id': data['store_id']}
url = data['url']
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return ()
m = re.search(ur'<div class="lines">(.+?)</div>', body, re.S)
if m is None:
return ()
return tuple(term.strip()
for term in re.findall(ur'<li>(.+?)</li>', m.group(1), re.S))
def fetch_stores(data):
"""
获得商店信息
:param data:
:return:
"""
url = data['post_url']
try:
js = json.loads(cm.post_data(url, {'country_id': data['country_id'], 'retail_city': '',
'retail_type': data['retail_type']}).decode('unicode_escape'))
except Exception:
print 'Error occured in getting country list: %s' % url
dump_data = {'level': 1, 'time': cm.format_time(), 'data': {'url': url}, 'brand_id': data['brand_id']}
cm.dump(dump_data)
return []
# country_id=108&retail_city=&retail_type=retail
# country_id=99&retail_city=&retail_type=service
store_list = []
for s in js:
entry = cm.init_store_entry(data['brand_id'], data['brandname_e'], data['brandname_c'])
tmp = s['retail_name'].strip()
if cm.is_chinese(tmp):
entry[cm.name_c] = tmp
else:
entry[cm.name_e] = tmp
entry[cm.addr_e] = s['retail_gmap'].strip()
entry[cm.zip_code] = s['retail_zipcode'].strip()
entry[cm.city_e] = s['retail_city'].strip().upper()
if s['retail_email'] is not None:
entry[cm.email] = s['retail_email'].strip()
if s['retail_website'] is not None:
entry[cm.url] = s['retail_website'].strip()
if data['retail_type'] == 'retail':
entry[cm.store_class] = 'Retail'
else:
entry[cm.store_class] = 'Service Center'
entry[cm.country_e] = s['country_name'].strip().upper()
entry[cm.continent_e] = s['continent_name'].strip().upper()
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_cities(data):
param = {'action': 'getRegionsFromAjax', 'country': data['country_code']}
url = data['url']
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return []
results = []
for city in json.loads(body):
d = data.copy()
d['city'] = city
results.append(d)
return results
def fetch_cities(data):
param = {'action': 'getRegionsFromAjax', 'country': data['country_code']}
url = data['url']
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return []
results = []
for city in json.loads(body):
d = data.copy()
d['city'] = city
results.append(d)
return results
def fetch_cities(data):
url = data['data_url']
param = {'country': data['country_code']}
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return []
results = []
for m in re.findall(ur'<option value="([^"]+)">([^<>]+)', body):
d = data.copy()
d['city'] = m[1].strip().upper()
d['city_code'] = m[0]
results.append(d)
def fetch_cities(data):
url = data['data_url']
param = {
'DestinationURL': 'Worldwide-retailers',
'(country)': data['country_code']
}
try:
body = cm.post_data(url, param)
# m = re.search(ur'META HTTP-EQUIV="Location"\s+Content="([^"]+)"', body)
# if not m:
# raise IOError()
# body = cm.get_data(m.group(1))
except Exception, e:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return ()
def fetch_cities(data):
url = data['data_url']
param = {'country': data['country_code']}
try:
body = cm.post_data(url, param)
except Exception:
cm.dump('Error in fetching cities: %s, %s' % (url, param), log_name)
return []
results = []
for m in re.findall(ur'<option value="([^"]+)">([^<>]+)', body):
d = data.copy()
d['city'] = m[1].strip().upper()
d['city_code'] = m[0]
results.append(d)
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 get_store_details(data):
url = data['url']
try:
html = cm.post_data(url, {'country': data['country_id'], 'city': data['city_id'], 'recordid': data['store_id']})
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 []
entry = cm.init_store_entry(brand_id, brandname_e, brandname_c)
info = json.loads(html)['elements']
addr = cm.reformat_addr(info['address'].replace('\\', '').replace('<p>', ',').replace('</p>', ','))
# 第一行为商店名称
terms = addr.split(',')
if len(terms) > 0:
entry[cm.name_e] = cm.reformat_addr(terms[0])
entry[cm.addr_e] = addr
gmap_url = info['gmap']
m = re.findall(ur'(-?\d+\.\d+),(-?\d+\.\d+)', gmap_url)
if len(m) > 0:
cm.update_entry(entry, {cm.lat: string.atof(m[0][0]), cm.lng: string.atof(m[0][1])})
entry[cm.url] = info['shareurl'].replace('\\', '')
entry[cm.hours] = info['openingtimes']
entry[cm.comments] = info['other']
# Geo
country = data['country']
city = data['city']
cm.update_entry(entry, {cm.country_e: country, cm.city_e: city})
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)' % (
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')
return entry
def fetch_cities(data):
url = data['data_url']
state_id = data['state_id']
if state_id is None:
param = {'request': 'countryChange', 'c': data['country_id']}
else:
param = {
'request': 'stateChange',
'c': data['country_id'],
's': state_id
}
try:
body = cm.post_data(url, param)
except Exception, e:
cm.dump('Error in fetching stores: %s, %s' % (url, param), log_name)
return []
def fetch_stores(data):
"""
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"})
except Exception:
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 m in re.finditer(ur'<a href="(.+?)".*?>', html):
entry = cm.init_store_entry(data["brand_id"], data["brandname_e"], data["brandname_c"])
entry[cm.url] = m.group(1)
store_html, start, end = cm.extract_closure(html[m.start() :], ur"<a href", ur"</a>")
if end == 0:
continue
m1 = re.findall(ur'<h3 class="titleShop">(.+?)</h3>', store_html, re.S)
if len(m1) > 0:
entry[cm.name_e] = m1[0].strip()
m1 = re.findall(ur"<p\b.*?>(.+?)(?:</p>|</div>)", store_html, re.S)
if len(m1) > 0:
terms = cm.reformat_addr(m1[0]).split(",")
tel = cm.extract_tel(terms[-1])
if tel != "":
terms = terms[:-1]
entry[cm.tel] = tel
entry[cm.addr_e] = ", ".join([v.strip() for v in terms])
entry["country_e"] = data["country_e"]
entry["city_e"] = data["city_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],
)
store_list.append(entry)
db.insert_record(entry, "stores")