Datasets:

Modalities:
Text
Formats:
text
Size:
< 1K
ArXiv:
Libraries:
Datasets
License:
WikiHow-taskset / replay_url.py
zdy023's picture
ver Aug22nd
d58c721 unverified
raw
history blame
21.3 kB
# Created by Danyang Zhang @X-Lance
from mitmproxy import http
from mitmproxy import ctx
import os.path
import functools
import classify_url
import load_response
import datetime
import locale
import gzip
import lxml.html
import lxml.cssselect
from pyserini.search.lucene import LuceneSearcher
import csv
import urllib.parse
from typing import List
from typing import Mapping
from mitmproxy.coretypes.multidict import MultiDict
import random
class Replayer:
# class `Replayer` {{{ #
def __init__(self, start_cache_index: int,
replay_path: str,
template_path: str,
index_path: str,
meta_path: str):
self.cache_index: int = start_cache_index
self.replay_path: str = replay_path
self.template_path: str = template_path
self.index_path: str = index_path
self.meta_path: str = meta_path
with open(meta_path) as f:
reader = csv.DictReader(f)
self.meta_database: Dict[str, Mapping[str, str]]\
= {itm["doc_id"]: itm for itm in reader}
self.doc_list: List[str] = list(self.meta_database.keys())
locale.setlocale(locale.LC_TIME, "en_US.utf8")
self.dateformat: str = "%a, %d %b %Y %H:%M:%S GMT"
self.searcher: LuceneSearcher = LuceneSearcher(self.index_path)
# CSS Selectors {{{ #
self.search_input_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("input#hs_query", translator="html")
self.result_list_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("div#searchresults_list.wh_block", translator="html")
self.result_list_anchor_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("div#search_adblock_bottom", translator="html")
self.result_footer_anchor_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("div#searchresults_footer>div.sr_text", translator="html")
self.result_thumb_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("div.result_thumb", translator="html")
self.result_title_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("div.result_title", translator="html")
self.result_view_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("li.sr_view", translator="html")
self.result_updated_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("li.sr_updated", translator="html")
self.result_verif_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("li.sp_verif", translator="html")
self.sha_index_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("input[name=\"sha_index\"]", translator="html")
self.sha_id_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("input[name=\"sha_id\"]", translator="html")
self.sha_title_selector: lxml.cssselect.CSSSelector\
= lxml.cssselect.CSSSelector("input[name=\"sha_title\"]", translator="html")
# }}} CSS Selectors #
self.search_page_capacity: int = 30
self.search_capacity: int = 300
def request(self, flow: http.HTTPFlow):
if flow.request.pretty_host=="www.wikihow.com":
url_path = flow.request.path
url_key = classify_url.classify(url_path)
if url_key in [ r"/x/collect\?t={first,later}&*"
, r"/x/collect\?t={exit,amp}&*"
, r"/x/amp-view\?*"
, r"/ev/*"
]:
# Control Flows {{{ #
headers = {}
response_time = datetime.datetime.utcnow()
headers["retry-after"] = "0"
headers["accept-ranges"] = "bytes"
headers["date"] = response_time.strftime(self.dateformat)
headers["x-timer"] = "S{:.6f},VS0,VE0".format(response_time.timestamp())
headers["x-c"] = "cache-tyo{:d}-TYO,M".format(self.cache_index)
headers["x-content-type-options"] = "nosniff"
headers["x-xss-protection"] = "1; mode=block"
headers["strict-transport-security"] = "max-age=31536000; includeSubDomains; preload"
headers["content-type"] = "text/plain; charset=UTF-8"
headers["x-robots-tag"] = "noindex, nofollow"
flow.response = http.Response.make(204,
headers=headers)
self.cache_index += 1
# }}} Control Flows #
elif url_path.startswith("/wikiHowTo?search="):
# Search Pages {{{ #
url_items = urllib.parse.urlparse(url_path)
queries = urllib.parse.parse_qs(url_items.query)
search_keywords = queries["search"][0]
start_index = int(queries.get("start", ["0"])[0])
hits = self.searcher.search(search_keywords, k=self.search_capacity)
response_time = datetime.datetime.utcnow()
# build the webpage
page_body = lxml.html.parse(
os.path.join(self.template_path, "search-page.html.template")).getroot()
# 1. fill the parameters in the page body
search_input = self.search_input_selector(page_body)[0]
search_input.set("value", search_keywords)
# 2. prepare result list
#result_list = self.result_list_selector(page_body)[0]
result_list_bottom = self.result_list_anchor_selector(page_body)[0]
with open(os.path.join(self.template_path, "search-item.html.template")) as f:
result_item_html = "".join(f.readlines())
for i, h in zip(range(start_index, start_index+self.search_page_capacity),
hits[start_index:start_index+self.search_page_capacity]):
docid = h.docid
article_path = docid.replace("%2f", "/")
result_item = lxml.html.fromstring(result_item_html)
# Update Item Parameters {{{ #
result_item.set("href", "https://www.wikihow.com/{:}".format(article_path))
result_thumb = self.result_thumb_selector(result_item)[0]
thumb_url = self.meta_database[docid]["thumb_url"]
if thumb_url!="":
result_thumb.set("style",
"background-image: url(https://www.wikihow.com{:})".format(thumb_url.replace("%2f", "/")))
else:
result_thumb.set("style",
"background-image: url(https://www.wikihow.com/5/5f/{doc_id:}-Step-2.jpg/-crop-250-145-193px-{doc_id}-Step-2.jpg)".format(doc_id=article_path))
new_result_title = lxml.html.fromstring(
"<div class=\"result_title\">{:}</div>"\
.format(" ".join(
map( lambda w: "<b>" + w + "</b>"
, self.meta_database[docid]["title"].split()
)
)
)
)
result_title = self.result_title_selector(result_item)[0]
result_title.getparent().replace(result_title, new_result_title)
result_view = self.result_view_selector(result_item)[0]
view_counts = self.meta_database[docid]["sr_view"]
if view_counts!="":
result_view.text = "{:} views\t\t\t\t\t\t".format(view_counts)
else:
result_view.text = "0 views\t\t\t\t\t\t"
result_updated = self.result_updated_selector(result_item)[0]
updated_date = self.meta_database[docid]["sr_updated"]
if updated_date!="":
updated_date = datetime.datetime.strptime(updated_date,
"%B %d, %Y")
updating_duration = response_time - updated_date
# Calculate Time Diff String {{{ #
days = updating_duration.days
if days<7:
time_diff_str = "{:} day{:} ago".format(
days, "" if days==1 else "s")
elif days<30:
weeks = days // 7
time_diff_str = "{:} week{:} ago".format(
weeks, "" if weeks==1 else "s")
elif days<365:
months = days // 30
time_diff_str = "{:} month{:} ago".format(
months, "" if months==1 else "s")
else:
years = days // 365
time_diff_str = "{:} year{:} ago".format(
years, "" if years==1 else "s")
# }}} Calculate Time Diff String #
list(result_updated)[0].tail = time_diff_str + "\t\t\t\t\t\t"
else:
list(result_updated)[0].tail = "12 hours ago\t\t\t\t\t\t"
result_verif = self.result_verif_selector(result_item)[-1]
verif_type = self.meta_database[docid]["sr_verif"]
if verif_type=="E":
result_verif.text = "Expert Co-Authored\t\t\t\t\t\t\t"
elif verif_type=="Q":
result_verif.text = "Quality Tested\t\t\t\t\t\t\t"
else:
result_verif.getparent().remove(result_verif)
sha_index = self.sha_index_selector(result_item)[0]
sha_index.set("value", str(i+1))
sha_id = self.sha_id_selector(result_item)[0]
sha_id.set("value", self.meta_database[docid]["sha_id"])
sha_title = self.sha_title_selector(result_item)[0]
sha_title.set("value", article_path)
# }}} Update Item Parameters #
result_list_bottom.addprevious(result_item)
# 3. prepare footer
result_footer_bottom = self.result_footer_anchor_selector(page_body)[0]
with open(os.path.join(self.template_path, "search-button.html.template")) as f:
result_footer = lxml.html.fromstring("".join(f.readlines()))
next_button, previous_button, statistics_text = list(result_footer)
if start_index+self.search_page_capacity<self.search_capacity:
# has next
next_button.set("href", "/wikiHowTo?search={:}&start={:}&wh_an=1".format(
urllib.parse.quote_plus(search_keywords),
start_index+self.search_page_capacity))
result_footer_bottom.addprevious(next_button)
else:
disabled_next = lxml.html.fromstring(
"<span class=\"button buttonright primary disabled\">Next ></span>")
result_footer_bottom.addprevious(disabled_next)
if start_index>0:
# has previous
enabled_previous = lxml.html.fromstring(
"<a class=\"button buttonleft primary\">&lt; Previous</a>")
enabled_previous.set("href",
"/wikiHowTo?search={:}&start={:}&wh_an=1".format(
urllib.parse.quote_plus(search_keywords),
start_index-self.search_page_capacity))
result_footer_bottom.addprevious(enabled_previous)
else:
result_footer_bottom.addprevious(previous_button)
statistics_text.text = "{:} Results".format(self.search_capacity)
result_footer_bottom.addprevious(statistics_text)
# 4. return result
full_page = lxml.html.tostring(page_body)
headers = {}
headers["content-type"] = "text/html; charset=UTF-8"
headers["content-language"] = "en"
headers["x-frame-options"] = "SAMEORIGIN"
headers["x-p"] = "ma"
headers["expires"] = (response_time + datetime.timedelta(days=1))\
.strftime(self.dateformat)
headers["cache-control"] = "s-maxage=86400, must-revalidate, max-age=0"
headers["content-encoding"] = "gzip"
headers["accept-ranges"] = "bytes"
headers["date"] = response_time.strftime(self.dateformat)
headers["age"] = "0"
headers["x-timer"] = "S{:.6f},VS0,VE0".format(response_time.timestamp())
headers["x-c"] = "cache-tyo{:d}-TYO,M".format(self.cache_index)
headers["x-content-type-options"] = "nosniff"
headers["x-xss-protection"] = "1; mode=block"
headers["strict-transport-security"] = "max-age=31536000; includeSubDomains; preload"
headers["set-cookie"] =\
"whv=lbYVTnTp1cUHHoDGDOwR; expires={:}; domain=.wikihow.com; path=/; secure"\
.format((response_time + datetime.timedelta(days=3654)).strftime(self.dateformat))
headers["vary"] = "Cookie, Accept-Encoding"
flow.response = http.Response.make(200,
content=full_page,
headers=headers)
self.cache_index += 1
# }}} Search Pages #
elif url_path.startswith("/Special:Randomizer"):
# Random Page {{{ #
random_index = random.randrange(len(self.doc_list))
doc_id = self.doc_list[random_index]
article_path = doc_id.replace("%2f", "/")
headers = {}
response_time = datetime.datetime.utcnow()
headers["content-type"] = "text/html; charset=UTF-8"
headers["expires"] = "Thu, 01 Jan 1970 00:00:00 GMT"
headers["cache-control"] = "private, must-revalidate, max-age=0"
headers["x-p"] = "ma"
headers["location"] = "https://www.wikihow.com/{:}".format(article_path)
headers["content-encoding"] = "gzip"
headers["accept-ranges"] = "bytes"
headers["date"] = response_time.strftime(self.dateformat)
headers["x-timer"] = "S{:.6f},VS0,VE0".format(response_time.timestamp())
headers["x-c"] = "cache-tyo{:d}-TYO,M".format(self.cache_index)
headers["x-content-type-options"] = "nosniff"
headers["x-xss-protection"] = "1; mode=block"
headers["strict-transport-security"] = "max-age=31536000; includeSubDomains; preload"
headers["vary"] = "Cookie, Accept-Encoding"
flow.response = http.Response.make(302,
headers=headers)
self.cache_index += 1
# }}} Random Page #
elif url_path.startswith("/Special:RateItem"):
# Reting Flows {{{ #
headers = {}
response_time = datetime.datetime.utcnow()
headers["content-type"] = "text/html; charset=UTF-8"
headers["access-control-allow-credentials"] = "true"
headers["access-control-allow-origin"] = "https://www.wikihow.com"
headers["access-control-expose-headers"] = "AMP-Access-Control-Allow-Source-Origin"
headers["amp-access-control-allow-source-origin"] = "https://www.wikihow.com"
#headers["set-cookie"] =\
#"UseDC=master; expires={:}; Max-Age=10; path=/; domain=www.wikihow.com; secure; HttpOnly"
headers["content-language"] = "en"
headers["x-frame-options"] = "SAMEORIGIN"
headers["expires"] = "Thu, 01 Jan 1970 00:00:00 GMT"
headers["cache-control"] = "private, must-revalidate, max-age=0"
headers["x-p"] = "ck ma"
headers["content-encoding"] = "gzip"
headers["accept-ranges"] = "bytes"
headers["date"] = response_time.strftime(self.dateformat)
headers["x-timer"] = "S{:.6f},VS0,VE0".format(response_time.timestamp())
headers["x-c"] = "cache-tyo{:d}-TYO,M".format(self.cache_index)
headers["x-content-type-options"] = "nosniff"
headers["x-xss-protection"] = "1; mode=block"
headers["strict-transport-security"] = "max-age=31536000; includeSubDomains; preload"
headers["vary"] = "Cookie, Accept-Encoding"
flow.response = http.Response.make(200,
content=b'{"result":"true"}',
headers=headers)
cookies_attribute = MultiDict([
("expires", (response_time + datetime.timedelta(seconds=10))\
.strftime(self.dateformat)),
("Max-Age", "10"),
("path", "/"),
("domain", "www.wikihow.com"),
("secure", None),
("HttpOnly", None)
])
flow.response.cookies["UseDC"] = ("master", cookies_attribute)
flow.response.cookies["UseCDNCache"] = ("false", cookies_attribute.copy())
self.cache_index += 1
# }}} Reting Flows #
else:
# Normal Pages {{{ #
if url_key==r"/Special:RCWidget\?*":
filename = "%2fSpecial:RCWidget?function=WH.RCWidget.rcwOnLoadData&GuVHo&nabrequest=0&anonview=1"
else:
filename = url_path.replace("/", "%2f")
if len(filename)>100:
filename = filename[:100]
filename = os.path.join(self.replay_path, filename)
#ctx.log.info("Requesting {:}".format(filename))
if not os.path.exists(filename):
ctx.log.info("404: {:}".format(filename))
flow.response = http.Response.make(404)
else:
status_code, header, content = load_response.load_response(filename)
if "content-encoding" in header\
and header["content-encoding"]==b"gzip"\
and len(content)>0:
try:
content = gzip.decompress(content)
except Exception as e:
ctx.log.info(str(e))
response_time = datetime.datetime.utcnow()
#if "date" in header:
#header["date"] = response_time.strftime(self.dateformat)
#if "last-modified" in header:
#header["last-modified"] = (response_time - datetime.timedelta(days=1))\
#.strftime(self.dateformat)
#if "expires" in header:
#header["expires"] = (response_time + datetime.timedelta(days=3650))\
#.strftime(self.dateformat)
if "x-timer" in header:
header["x-timer"] = "S{:.6f},VS0,VE0".format(response_time.timestamp())
if "x-c" in header:
header["x-c"] = "cache-tyo{:d}-TYO,M".format(self.cache_index)
flow.response = http.Response.make(status_code,
content=content,
headers=header)
flow.response.refresh()
#ctx.log.info("WARN: {:}".format(content==flow.response.content))
self.cache_index += 1
# }}} Normal Pages #
#else:
#flow.response = http.Response.make(204)
#pass
# }}} class `Replayer` #
addons = [ Replayer( 15090 # just a positive integer is ok
, "dumps" # crawled data from WikiHow website
, "templates" # templates of the search result page
, "indices-t/indices" # indices for the search engine based on pyserini
, "indices-t/docs/doc_meta.csv" # doc meta
)
]