def _bing_api_ranking(url, keyword, server_info):
url_regex = re.compile("^" + url)
rank_infos = {}
counter = 0
base_api_url = 'https://api.datamarket.azure.com/Data.ashx/Bing/SearchWeb/v1/Web?'
# Set format
base_api_url += "$format=json&"
# Set Market (language-region)
base_api_url += "Market=%27" + server_info['market'] + "%27&"
# Set keyword
base_api_url += "Query=%27" + urllib.quote_plus(
keyword.encode("UTF-8")) + "%27&"
# Setup basic authentication
headers = {
"Authorization":
"Basic %s" % base64.b64encode(":" + config.azure_account_id)
}
logging.info(headers)
# Fetch 2x50 results and try to match url
num = 50
for i in range(0, 2):
start = i * num
api_url = base_api_url + "$skip=" + unicode(
start) + "&$top=" + unicode(num)
result = urlfetch_with_cache(api_url, 3600, 600, headers)
if result['status_code'] != 200:
raise Exception("Invalid response (url = <" + api_url +
">, status code = " +
unicode(result['status_code']) +
" and content = <" + unicode(result['content']) +
">")
response = json.loads(result['content'])
logging.info(response)
if response.get('d'):
results = response['d'].get('results')
if results:
# Total number of results is not available anymore via the API
rank_infos['total'] = None
for r in results:
logging.info(r)
counter += 1
if (url_regex.match(r['Url'])):
rank_infos['rank'] = counter
rank_infos['url'] = r['Url']
return rank_infos
return rank_infos
def _bing_api_ranking(url, keyword, server_info):
url_regex = re.compile("^" + url)
rank_infos = {}
counter = 0
base_api_url = "https://api.datamarket.azure.com/Data.ashx/Bing/SearchWeb/v1/Web?"
# Set format
base_api_url += "$format=json&"
# Set Market (language-region)
base_api_url += "Market=%27" + server_info["market"] + "%27&"
# Set keyword
base_api_url += "Query=%27" + urllib.quote_plus(keyword.encode("UTF-8")) + "%27&"
# Setup basic authentication
headers = {"Authorization": "Basic %s" % base64.b64encode(":" + config.azure_account_id)}
logging.info(headers)
# Fetch 2x50 results and try to match url
num = 50
for i in range(0, 2):
start = i * num
api_url = base_api_url + "$skip=" + unicode(start) + "&$top=" + unicode(num)
result = urlfetch_with_cache(api_url, 3600, 600, headers)
if result["status_code"] != 200:
raise Exception(
"Invalid response (url = <"
+ api_url
+ ">, status code = "
+ unicode(result["status_code"])
+ " and content = <"
+ unicode(result["content"])
+ ">"
)
response = json.loads(result["content"])
logging.info(response)
if response.get("d"):
results = response["d"].get("results")
if results:
# Total number of results is not available anymore via the API
rank_infos["total"] = None
for r in results:
logging.info(r)
counter += 1
if url_regex.match(r["Url"]):
rank_infos["rank"] = counter
rank_infos["url"] = r["Url"]
return rank_infos
return rank_infos
def _google_ajax_api_ranking(url, keyword, server_info):
url_regex = re.compile("^" + url)
rank_infos = {}
counter = 0
base_api_url = 'http://ajax.googleapis.com/ajax/services/search/web?v=1.0&'
# Set API Key if defined
try:
if config.google_ajax_api_key:
base_api_url += "key=" + urllib.quote_plus(
config.google_ajax_api_key) + "&"
except:
pass
# Set country param
base_api_url += "gl=" + server_info['gl'] + "&"
# Each call to google AJAX Search API returns 4 results.
# We are allowed to fetch 64 results (16 pages)
num = 4
for i in range(0, 16):
start = i * num
api_url = base_api_url + "start=" + unicode(
start) + "&q=" + urllib.quote_plus(keyword.encode("UTF-8"))
result = urlfetch_with_cache(api_url, 3600, 600)
if result['status_code'] != 200:
raise Exception("Invalid response (url = <" + google_url +
">, status code = " +
unicode(result['status_code']) +
" and content = <" + unicode(result['content']) +
">")
response = json.loads(result['content'])
if response.get('responseData'):
if i == 0:
try:
rank_infos['total'] = long(
response['responseData']['cursor'].get(
'estimatedResultCount'))
except:
rank_infos['total'] = None
results = response['responseData'].get('results')
if results:
for r in results:
counter += 1
if (url_regex.match(r['unescapedUrl'])):
rank_infos['rank'] = counter
rank_infos['url'] = r['unescapedUrl']
return rank_infos
return rank_infos
def _google_ajax_api_ranking(url, keyword, server_info):
url_regex = re.compile("^" + url)
rank_infos = {}
counter = 0
base_api_url = "http://ajax.googleapis.com/ajax/services/search/web?v=1.0&"
# Set API Key if defined
try:
if config.google_ajax_api_key:
base_api_url += "key=" + urllib.quote_plus(config.google_ajax_api_key) + "&"
except:
pass
# Set country param
base_api_url += "gl=" + server_info["gl"] + "&"
# Each call to google AJAX Search API returns 4 results.
# We are allowed to fetch 64 results (16 pages)
num = 4
for i in range(0, 16):
start = i * num
api_url = base_api_url + "start=" + unicode(start) + "&q=" + urllib.quote_plus(keyword.encode("UTF-8"))
result = urlfetch_with_cache(api_url, 3600, 600)
if result["status_code"] != 200:
raise Exception(
"Invalid response (url = <"
+ google_url
+ ">, status code = "
+ unicode(result["status_code"])
+ " and content = <"
+ unicode(result["content"])
+ ">"
)
response = json.loads(result["content"])
if response.get("responseData"):
if i == 0:
try:
rank_infos["total"] = long(response["responseData"]["cursor"].get("estimatedResultCount"))
except:
rank_infos["total"] = None
results = response["responseData"].get("results")
if results:
for r in results:
counter += 1
if url_regex.match(r["unescapedUrl"]):
rank_infos["rank"] = counter
rank_infos["url"] = r["unescapedUrl"]
return rank_infos
return rank_infos
