def get_number_of_results(term, ajax=False, verbose=True):
if not ajax:
gs = GoogleSearch(term)
page = str(gs._get_results_page())
match = reg.search(page)
if match:
if verbose: print(term, match.groups()[0])
return int(match.groups()[0].replace(',',''))
else:
raw_input((term, page))
return int(search(term)['responseData']['cursor']['estimatedResultCount'])
def run(self, string):
query = "site:http://md5-database.org/md5 %s" % string
#if not thread:
# say("Querying Google: '%s'" % query)
gs = GoogleSearch(query)
gs.results_per_page = 10
results = gs._get_results_page()
texts = results.findAll(text=True)
texts = ''.join(texts)
results = re.findall(re.compile('MD5\}.*?MD5'), texts)
for line in results:
if string in line:
result = line[(line.find(',') + 1):line.find('.')].strip()
return result
return ''
def augmentedScoring(phrase) :
#TODO hit count ist nicht ganz korrekt
#=> more like: about .* results kann ueber die ganze seite gehen...
#=> investigate regex/google site!
hitScores = []
#without domains -> denominator
searchStr = "\"" + phrase + "\""
gs = GoogleSearch(searchStr)
gs.results_per_page = 50
sleep(searchSleepTime)
score = 0
matchStrings = ""
pageStrings = ""
try:
page = gs._get_results_page()
logging.info(gs.last_search_url)
pageStr = str(page)
pageStrings += pageStr
if pageStr.find("resultStats\">") != -1 or pageStr.find("No results found for") > -1:
m = re.search(r'resultStats\">.*bout (.*) results</div', pageStr)
if m is not None:
score = m.group(1)
matchStrings += m.group(0) + " - "
logging.info("score (" + searchStr + "): " + score)
score = score.replace(',','')
hitScores.append(int(score))
else:
logging.warning("No match! .. no google hits? (" + searchStr + ")")
hitScores.append(0)
for _ in domains: hitScores.append(0)
hitScores.append("first: no match")
hitScores.append(unicode(pageStrings, 'utf-8', "strict"))
logging.info(hitScores)
return hitScores
else:
logging.warning("No google hits! (" + searchStr + ")")
hitScores.append(0)
for _ in domains: hitScores.append(0)
hitScores.append("first: no match")
hitScores.append(unicode(pageStrings, 'utf-8', "strict"))
logging.info(hitScores)
return hitScores
except SearchError as se:
logging.warning("Search Error on: " + searchStr + " no results? " + str(se))
hitScores.append(0)
for _ in domains: hitScores.append(0)
hitScores.append("first: search error: " + str(se))
hitScores.append(unicode(pageStrings, 'utf-8', "strict"))
logging.info(hitScores)
return hitScores
denominationScore = float(score)
logging.info("denominator (" + searchStr + "): " + str(denominationScore))
#augmented with domains
for domain in domains :
searchStr = "\"" + phrase + "\" \"" + domain + "\""
gs = GoogleSearch(searchStr)
gs.results_per_page = 50
sleep(searchSleepTime)
score = 0
try:
page = gs._get_results_page()
logging.info(gs.last_search_url)
pageStr = str(page)
pageStrings += pageStr
if pageStr.find("resultStats\">") != -1 :
m = re.search(r'resultStats">.*bout (.*) results</div', pageStr)
if m is not None:
score = m.group(1)
matchStrings += m.group(0) + " - "
logging.info("score (" + searchStr + "): " + score)
score = score.replace(',','')
else: logging.warning("No match! .. no google hits? (" + searchStr + ")")
else: logging.warning("No google hits! (" + searchStr + ")")
except SearchError as se:
logging.warning("Search Error on: " + searchStr + " no results? " + str(se))
#relativating by general hit count
hitScores.append(float(score) / denominationScore)
hitScores.append(matchStrings)
hitScores.append(unicode(pageStrings, 'utf-8', "strict"))
logging.info(hitScores)
return hitScores
