Kaynağa Gözat

Some updates for the new version of EarwigBot.

Still have to update bigrams to trigrams.
pull/24/head
Ben Kurtovic 12 yıl önce
ebeveyn
işleme
c368540fba
2 değiştirilmiş dosya ile 21 ekleme ve 15 silme
  1. +1
    -1
      .gitignore
  2. +20
    -14
      pages/copyvios.mako

+ 1
- 1
.gitignore Dosyayı Görüntüle

@@ -1,5 +1,5 @@
*.pyc *.pyc
temp/ temp/
www/ www/
config.ts-earwigbot.json
.earwigbot
rewrite.script rewrite.script

+ 20
- 14
pages/copyvios.mako Dosyayı Görüntüle

@@ -9,29 +9,35 @@
from time import time from time import time
from urlparse import parse_qs from urlparse import parse_qs


from earwigbot import bot, exceptions
import oursql import oursql


path.insert(0, "../earwigbot")

import earwigbot

def get_results(lang, project, title, query): def get_results(lang, project, title, query):
earwigbot.config.config.load("config.ts-earwigbot.json")
bot = bot.Bot(".earwigbot")
try: try:
site = earwigbot.wiki.get_site(lang=lang, project=project)
except earwigbot.wiki.SiteNotFoundError:
return None, None
site = bot.wiki.get_site(lang=lang, project=project)
except exceptions.SiteNotFoundError:
try:
site = bot.wiki.add_site(lang=lang, project=project)
except exceptions.APIError:
return None, None

page = site.get_page(title) page = site.get_page(title)
conn = open_sql_connection()
conn = open_sql_connection(bot)
if not query.get("nocache"): if not query.get("nocache"):
result = get_cached_results(page, conn) result = get_cached_results(page, conn)
if query.get("nocache") or not result: if query.get("nocache") or not result:
result = get_fresh_results(page, conn) result = get_fresh_results(page, conn)
return page, result return page, result


def open_sql_connection():
conn_args = earwigbot.config.config.wiki["_toolserverSQLCache"]
conn_args["read_default_file"] = expanduser("~/.my.cnf")
def open_sql_connection(bot):
conn_args = bot.config.wiki["_toolserverSQLCache"]
if "read_default_file" not in conn_args and "user" not in conn_args and "passwd" not in conn_args:
conn_args["read_default_file"] = expanduser("~/.my.cnf")
if "autoping" not in conn_args:
conn_args["autoping"] = True
if "autoreconnect" not in conn_args:
conn_args["autoreconnect"] = True
return oursql.connect(**conn_args) return oursql.connect(**conn_args)


def get_cached_results(page, conn): def get_cached_results(page, conn):
@@ -49,7 +55,7 @@
return None return None


url, cache_time, num_queries, original_tdiff = results[0] url, cache_time, num_queries, original_tdiff = results[0]
result = page.copyvio_compare(url, min_confidence=0.5)
result = page.copyvio_compare(url)
result.cached = True result.cached = True
result.queries = num_queries result.queries = num_queries
result.tdiff = time() - t_start result.tdiff = time() - t_start
@@ -68,7 +74,7 @@


def get_fresh_results(page, conn): def get_fresh_results(page, conn):
t_start = time() t_start = time()
result = page.copyvio_check(min_confidence=0.5, max_queries=10)
result = page.copyvio_check(max_queries=10)
result.cached = False result.cached = False
result.tdiff = time() - t_start result.tdiff = time() - t_start
cache_result(page, result, conn) cache_result(page, result, conn)


Yükleniyor…
İptal
Kaydet