浏览代码

Move stats db to same server as enwiki so we can merge two queries using a join.

pull/15/head
Ben Kurtovic 11 年前
父节点
当前提交
08763664e0
共有 1 个文件被更改,包括 10 次插入15 次删除
  1. +10
    -15
      tasks/afc_statistics.py

+ 10
- 15
tasks/afc_statistics.py 查看文件

@@ -228,28 +228,23 @@ class AFCStatistics(Task):
self.untrack_page().
"""
self.logger.debug("Updating tracked submissions")
query1 = "SELECT page_id, page_title, page_modify_oldid FROM page"
query2 = """SELECT page_latest, page_title, page_namespace FROM page
WHERE page_id = ?"""
cursor.execute(query1)

for pageid, title, oldid in cursor:
result = list(self.site.sql_query(query2, (pageid,)))
if not result:
query = """SELECT s.page_id, s.page_title, s.page_modify_oldid,
r.page_latest, r.page_title, r.page_namespace FROM page AS s
LEFT JOIN {0}_p.page AS r ON s.page_id = r.page_id"""
cursor.execute(query.format(self.site.name))

for pageid, title, oldid, real_oldid, real_title, real_ns in cursor:
if not real_oldid:
self.untrack_page(cursor, pageid)
continue

real_oldid = result[0][0]
if oldid != real_oldid:
msg = u"Updating page [[{0}]] (id: {1}) @ {2}"
self.logger.debug(msg.format(title, pageid, oldid))
self.logger.debug(" {0} -> {1}".format(oldid, real_oldid))
base = result[0][1].decode("utf8").replace("_", " ")
ns = self.site.namespace_id_to_name(result[0][2])
real_title = real_title.decode("utf8").replace("_", " ")
ns = self.site.namespace_id_to_name(real_ns)
if ns:
real_title = u":".join((ns, base))
else:
real_title = base
real_title = u":".join((ns, real_title))
try:
self.update_page(cursor, pageid, real_title)
except Exception:


正在加载...
取消
保存