def check_if_spam_json(json_data):
text_data = json.loads(json_data)["data"]
if text_data == "hb":
return False, None, ""
try:
data = json.loads(text_data)
except ValueError:
GlobalVars.charcoal_hq.send_message(
u"Encountered ValueError parsing the following:\n{0}".format(
json_data), False)
return False, None, ""
if "ownerUrl" not in data:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None, ""
title = data["titleEncodedFancy"]
title = unescape_title(title)
body = data["bodySummary"]
poster = data["ownerDisplayName"]
url = data["url"]
post_id = str(data["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii",
errors="replace")
site = data["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason, why = check_if_spam(title, body, poster, url, site,
post_id, False, True, 1, 0)
return is_spam, reason, why
def check_if_spam_json(json_data):
text_data = json.loads(json_data)["data"]
if text_data == "hb":
return False, None, ""
try:
data = json.loads(text_data)
except ValueError:
GlobalVars.charcoal_hq.send_message(u"Encountered ValueError parsing the following:\n{0}".format(json_data), False)
return False, None, ""
if "ownerUrl" not in data:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None, ""
title = data["titleEncodedFancy"]
title = parsing.unescape_title(title)
body = data["bodySummary"]
poster = data["ownerDisplayName"]
url = data["url"]
post_id = str(data["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii", errors="replace")
site = data["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason, why = check_if_spam(title=title,
body=body,
user_name=poster,
user_url=url,
post_site=site,
post_id=post_id,
is_answer=False,
body_is_summary=True,
owner_rep=1,
post_score=0)
return is_spam, reason, why
def check_if_spam_json(data):
d = json.loads(json.loads(data)["data"])
try:
_ = d["ownerUrl"] # noqa
except:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None, ""
title = d["titleEncodedFancy"]
title = unescape_title(title)
body = d["bodySummary"]
poster = d["ownerDisplayName"]
url = d["url"]
post_id = str(d["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii",
errors="replace")
quality_score = bayesian_score(title)
print quality_score
if quality_score < 0.3 and d["siteBaseHostAddress"] == "stackoverflow.com":
print GlobalVars.bayesian_testroom.send_message(
"[ SmokeDetector | BayesianBeta ] Quality score " +
str(quality_score * 100) + ": [" + title + "](" + url + ")")
site = d["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason, why = check_if_spam(title, body, poster, url, site,
post_id, False, True)
return is_spam, reason, why
def handle_spam_json(data, reason):
try:
d = json.loads(json.loads(data)["data"])
title = unescape_title(d["titleEncodedFancy"])
poster = d["ownerDisplayName"]
site = d["siteBaseHostAddress"]
url = d["url"]
poster_url = d["ownerUrl"]
post_id = str(d["id"])
handle_spam(title, poster, site, url, poster_url, post_id, reason, False)
except:
print "NOP"
def handle_spam_json(data, reason, why=""):
try:
d = json.loads(json.loads(data)["data"])
title = unescape_title(d["titleEncodedFancy"])
body = d["bodySummary"]
poster = d["ownerDisplayName"]
site = d["siteBaseHostAddress"]
url = d["url"]
poster_url = d["ownerUrl"]
post_id = str(d["id"])
handle_spam(title, body, poster, site, url, poster_url, post_id, reason, False, why)
except:
print "NOP"
def handle_spam_json(data, reason, why=""):
try:
d = json.loads(json.loads(data)["data"])
title = unescape_title(d["titleEncodedFancy"])
body = d["bodySummary"]
poster = d["ownerDisplayName"]
site = d["siteBaseHostAddress"]
url = d["url"]
poster_url = d["ownerUrl"]
post_id = str(d["id"])
handle_spam(title, body, poster, site, url, poster_url, post_id, reason, False, why)
except:
exc_type, exc_obj, exc_tb = sys.exc_info()
excepthook.uncaught_exception(exc_type, exc_obj, exc_tb)
def check_if_spam_json(data):
d = json.loads(json.loads(data)["data"])
try:
_ = d["ownerUrl"] # noqa
except:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None, ""
title = d["titleEncodedFancy"]
title = unescape_title(title)
body = d["bodySummary"]
poster = d["ownerDisplayName"]
url = d["url"]
post_id = str(d["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii", errors="replace")
site = d["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason, why = check_if_spam(title, body, poster, url, site, post_id, False, True, 1)
return is_spam, reason, why
def check_if_spam_json(data):
d = json.loads(json.loads(data)["data"])
try:
_ = d["ownerUrl"] # noqa
except:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None, ""
title = d["titleEncodedFancy"]
title = unescape_title(title)
body = d["bodySummary"]
poster = d["ownerDisplayName"]
url = d["url"]
post_id = str(d["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii",
errors="replace")
site = d["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason, why = check_if_spam(title, body, poster, url, site,
post_id, False, True, 1, 0)
return is_spam, reason, why
def check_if_spam_json(data):
d = json.loads(json.loads(data)["data"])
try:
_ = d["ownerUrl"] # noqa
except:
# owner's account doesn't exist anymore, no need to post it in chat:
# http://chat.stackexchange.com/transcript/message/18380776#18380776
return False, None
title = d["titleEncodedFancy"]
title = unescape_title(title)
body = d["bodySummary"]
poster = d["ownerDisplayName"]
url = d["url"]
post_id = str(d["id"])
print time.strftime("%Y-%m-%d %H:%M:%S"), title.encode("ascii", errors="replace")
quality_score = bayesian_score(title)
print quality_score
if quality_score < 0.3 and d["siteBaseHostAddress"] == "stackoverflow.com":
print GlobalVars.bayesian_testroom.send_message("[ SmokeDetector | BayesianBeta ] Quality score " + str(quality_score * 100) + ": [" + title + "](" + url + ")")
site = d["siteBaseHostAddress"]
site = site.encode("ascii", errors="replace")
sys.stdout.flush()
is_spam, reason = check_if_spam(title, body, poster, url, site, post_id, False, True)
return is_spam, reason