A Python robot that edits Wikipedia and interacts with people over IRC https://en.wikipedia.org/wiki/User:EarwigBot
選択できるのは25トピックまでです。 トピックは、先頭が英数字で、英数字とダッシュ('-')を使用した35文字以内のものにしてください。

341 行
14 KiB

  1. # -*- coding: utf-8 -*-
  2. from datetime import datetime
  3. import re
  4. from os.path import expanduser
  5. from threading import Lock
  6. import oursql
  7. from classes import BaseTask
  8. import config
  9. import wiki
  10. class Task(BaseTask):
  11. """A task to generate statistics for WikiProject Articles for Creation.
  12. Statistics are stored in a MySQL database ("u_earwig_afc_statistics")
  13. accessed with oursql. Statistics are updated live while watching the recent
  14. changes IRC feed and saved once an hour, on the hour, to self.pagename.
  15. In the live bot, this is "Template:AFC statistics".
  16. """
  17. name = "afc_statistics"
  18. number = 2
  19. def __init__(self):
  20. self.cfg = cfg = config.tasks.get(self.name, {})
  21. # Set some wiki-related attributes:
  22. self.pagename = cfg.get("page", "Template:AFC statistics")
  23. self.pending_cat = cfg.get("pending", "Pending AfC submissions")
  24. self.ignore_list = cfg.get("ignore_list", [])
  25. default_summary = "Updating statistics for [[WP:WPAFC|WikiProject Articles for creation]]."
  26. self.summary = self.make_summary(cfg.get("summary", default_summary))
  27. # Templates used in chart generation:
  28. templates = cfg.get("templates", {})
  29. self.tl_header = templates.get("header", "AFC statistics/header")
  30. self.tl_row = templates.get("row", "AFC statistics/row")
  31. self.tl_footer = templates.get("footer", "AFC statistics/footer")
  32. # Establish a connection with our SQL database:
  33. kwargs = cfg.get("sql", {})
  34. kwargs["read_default_file"] = expanduser("~/.my.cnf")
  35. self.conn = oursql.connect(**kwargs)
  36. self.db_access_lock = Lock()
  37. def run(self, **kwargs):
  38. self.site = wiki.get_site()
  39. action = kwargs.get("action")
  40. if not action:
  41. return
  42. methods = {
  43. "save": self.save,
  44. "sync", self.sync,
  45. "edit": self.process_edit,
  46. "move": self.process_move,
  47. "delete": self.process_delete,
  48. "restore": self.process_edit,
  49. }
  50. method = methods.get(action)
  51. if method:
  52. method(**kwargs)
  53. def save(self, **kwargs):
  54. if kwargs.get("fromIRC"):
  55. summary = " ".join((self.summary, "(!earwigbot)"))
  56. else:
  57. if self.shutoff_enabled():
  58. return
  59. summary = self.summary
  60. statistics = self.compile_charts()
  61. page = self.site.get_page(self.pagename)
  62. text = page.get()
  63. newtext = re.sub("(<!-- stat begin -->)(.*?)(<!-- stat end -->)",
  64. statistics.join(("\\1\n", "\n\\3")), text,
  65. flags=re.DOTALL)
  66. if newtext == text:
  67. return # Don't edit the page if we're not adding anything
  68. newtext = re.sub("(<!-- sig begin -->)(.*?)(<!-- sig end -->)",
  69. "\\1~~~ at ~~~~~\\3", newtext)
  70. page.edit(newtext, summary, minor=True, bot=True)
  71. def compile_charts(self):
  72. stats = ""
  73. with self.conn.cursor() as cursor, self.db_access_lock:
  74. cursor.execute("SELECT * FROM chart")
  75. for chart in cursor:
  76. stats += self.compile_chart(chart) + "\n"
  77. return stats[:-1] # Drop the last newline
  78. def compile_chart(self, chart_info):
  79. chart_id, chart_title, special_title = chart_info
  80. chart = "|".join((self.tl_header, chart_title))
  81. if special_title:
  82. chart += "".join(("|", special_title))
  83. chart = "".join(("{{", chart, "}}"))
  84. query = "SELECT * FROM page JOIN row ON page_id = row_id WHERE row_chart = ?"
  85. with self.conn.cursor(oursql.DictCursor) as cursor:
  86. cursor.execute(query, (chart_id,))
  87. for page in cursor:
  88. chart += "\n" + self.compile_chart_row(page)
  89. chart += "".join(("\n{{", self.tl_footer, "}}"))
  90. return chart
  91. def compile_chart_row(self, page):
  92. row = "{0}|s={page_status}|t={page_title}|h={page_short}|z={page_size}|"
  93. row += "cr={page_create_user}|cd={page_create_time}|ci={page_create_oldid}|"
  94. row += "mr={page_modify_user}|md={page_modify_time}|mi={page_modify_oldid}|"
  95. page["page_create_time"] = self.format_time(page["page_create_time"])
  96. page["page_modify_time"] = self.format_time(page["page_modify_time"])
  97. if page["page_special_user"]:
  98. row += "sr={page_special_user}|sd={page_special_time}|si={page_special_oldid}|"
  99. page["page_special_time"] = self.format_time(page["page_special_time"])
  100. if page["page_notes"]:
  101. row += "n=1{page_notes}"
  102. return "".join(("{{", row.format(self.tl_row, **page), "}}"))
  103. def format_time(self, timestamp):
  104. return timestamp.strftime("%H:%M, %d %B %Y")
  105. def sync(self, **kwargs):
  106. with self.conn.cursor() as cursor, self.db_access_lock:
  107. self.sync_deleted(cursor) # Remove deleted subs
  108. self.sync_oldids(cursor) # Make sure all subs are up to date
  109. self.sync_pending(cursor) # Add missing pending subs
  110. self.sync_old(cursor) # Remove old declined and accepted subs
  111. def sync_deleted(self, cursor):
  112. query1 = "SELECT page_id FROM page"
  113. query2 = "SELECT page_id FROM page WHERE page_id = ?"
  114. cursor.execute(query1)
  115. for page in cursor:
  116. result = self.site.sql_query(query2, (page,))
  117. if not list(result)[0]:
  118. self.untrack_page(cursor, pageid=page)
  119. def sync_oldids(self, cursor):
  120. query1 = "SELECT page_id, page_title, page_modify_oldid FROM page"
  121. query2 = "SELECT page_latest FROM page WHERE page_id = ?"
  122. cursor.execute(query1)
  123. for page_id, title, oldid in cursor:
  124. result = self.site.sql_query(query2, (page_id,))
  125. try:
  126. real_oldid = list(result)[0][0]
  127. except IndexError: # Page doesn't exist!
  128. self.untrack_page(cursor, pageid=page_id)
  129. continue
  130. if real_oldid != oldid:
  131. self.update_page(cursor, title)
  132. def sync_pending(self, cursor):
  133. query = """SELECT page_title FROM page JOIN row ON page_id = row_id
  134. WHERE row_chart IN (1, 2, 3)"""
  135. cursor.execute(query)
  136. tracked = [i[0] for i in cursor.fetchall()]
  137. category = self.site.get_category(self.pending_cat)
  138. for page in category.members(limit=500):
  139. if page in self.ignore_list:
  140. continue
  141. if page not in tracked:
  142. self.track_page(cursor, page)
  143. def sync_old(self, cursor):
  144. query = """DELETE FROM page, row USING page JOIN row
  145. ON page_id = row_id WHERE row_chart IN (4, 5)
  146. AND ADDTIME(page_special_time, '36:00:00') < NOW()"""
  147. cursor.execute(query)
  148. def process_edit(self, page, **kwargs):
  149. if page in self.ignore_list:
  150. return
  151. with self.conn.cursor() as cursor, self.db_access_lock:
  152. self.sync_page(cursor, page)
  153. def process_move(self, page, **kwargs):
  154. query1 = "SELECT * FROM page WHERE page_title = ?"
  155. query2 = "SELECT page_latest FROM page WHERE page_title = ?"
  156. query3 = "UPDATE page SET page_title = ?, page_modify_oldid = ? WHERE page_title = ?"
  157. source, dest = page
  158. with self.conn.cursor() as cursor, self.db_access_lock:
  159. cursor.execute(query1, (source,))
  160. result = cursor.fetchall()
  161. if not result:
  162. self.track_page(cursor, dest)
  163. else:
  164. try:
  165. new_oldid = list(self.site.sql_query(query2, (dest,)))[0][0]
  166. except IndexError:
  167. new_oldid = result[11]
  168. cursor.execute(query3, (dest, new_oldid, source))
  169. def process_delete(self, page, **kwargs):
  170. query = "SELECT page_id FROM page WHERE page_title = ?"
  171. with self.conn.cursor() as cursor, self.db_access_lock:
  172. result = self.site.sql_query(query, (page,))
  173. if list(result)[0]:
  174. self.sync_page(cursor, page)
  175. else:
  176. self.untrack_page(cursor, title=page)
  177. def sync_page(self, cursor, page):
  178. query = "SELECT * FROM page WHERE page_title = ?"
  179. cursor.execute(query, (page,))
  180. result = cursor.fetchall()
  181. if result:
  182. self.update_page(cursor, page)
  183. else:
  184. self.track_page(cursor, page)
  185. def untrack_page(self, cursor, pageid=None, title=None):
  186. query = "DELETE FROM page JOIN row ON page_id = row_id WHERE ? = ?"
  187. if pageid:
  188. cursor.execute(query, ("page_id", pageid))
  189. elif title:
  190. cursor.execute(query, ("page_title", title))
  191. def track_page(self, cursor, title):
  192. """Update hook for when page is not in our database."""
  193. page = self.site.get_page(title)
  194. status, chart = self.get_status_and_chart(page)
  195. if not status or status in ("accept", "decline"):
  196. return
  197. pageid = page.pageid()
  198. title = page.title()
  199. short = self.get_short_title(title)
  200. size = len(page.get())
  201. notes = self.get_notes(page)
  202. c_user, c_time, c_id = self.get_create(pageid)
  203. m_user, m_time, m_id = self.get_modify(pageid)
  204. s_user, s_time, s_id = self.get_special(page, status)
  205. query1 = "INSERT INTO row VALUES (?, ?)"
  206. query2 = "INSERT INTO page VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)"
  207. cursor.execute(query1, (pageid, chart))
  208. cursor.execute(query2, (pageid, status, title, short, size, notes,
  209. c_user, c_time, c_id, m_user, m_time, m_id,
  210. s_user, s_time, s_id))
  211. def update_page(self, cursor, title):
  212. """Update hook for when page is in our database."""
  213. page = self.site.get_page(title)
  214. status, chart = self.get_status_and_chart(page)
  215. if not status:
  216. self.untrack_page(cursor, title=title)
  217. pageid = page.pageid()
  218. title = page.title()
  219. size = len(page.get())
  220. notes = self.get_notes(page)
  221. m_user, m_time, m_id = self.get_modify(pageid)
  222. query = "SELECT * FROM page JOIN row ON page_id = row_id WHERE page_id = ?"
  223. with self.conn.cursor(oursql.DictCursor) as dict_cursor:
  224. dict_cursor.execute(query, (pageid,))
  225. result = dict_cursor.fetchall()[0]
  226. if title != result["page_title"]:
  227. query = "UPDATE page SET page_title = ?, page_short = ? WHERE page_id = ?"
  228. short = self.get_short_title(title)
  229. cursor.execute(query, (title, short, pageid))
  230. if m_id != result["page_modify_oldid"]:
  231. query = """UPDATE page SET page_size = ?, page_modify_user = ?,
  232. page_modify_time = ?, page_modify_oldid = ?
  233. WHERE page_id = ?"""
  234. cursor.execute(query, (size, m_user, m_time, m_id, pageid))
  235. if status != result["page_status"]:
  236. query1 = """UPDATE page JOIN row ON page_id = row_id
  237. SET page_status = ?, row_chart = ? WHERE page_id = ?"""
  238. query2 = """UPDATE page SET page_special_user = ?,
  239. page_special_time = ?, page_special_oldid = ?
  240. WHERE page_id = ?"""
  241. cursor.execute(query1, (status, chart, pageid))
  242. s_user, s_time, s_id = self.get_special(page, status)
  243. if s_id != result["page_special_oldid"]:
  244. cursor.execute(query2, (s_user, s_time, s_id, pageid))
  245. if notes != result["page_notes"]:
  246. query = "UPDATE page SET page_notes = ? WHERE page_id = ?"
  247. cursor.execute(query, (notes, pageid))
  248. def get_status_and_chart(self, page):
  249. content = page.get()
  250. if page.is_redirect():
  251. target = page.get_redirect_target()
  252. if self.site.get_page(target).namespace() == 0:
  253. return "accept", 4
  254. return None, 0
  255. elif re.search("\{\{afc submission\|r\|(.*?)\}\}", content, re.I):
  256. return "review", 3
  257. elif re.search("\{\{afc submission\|h\|(.*?)\}\}", content, re.I):
  258. return "pend", 2
  259. elif re.search("\{\{afc submission\|\|(.*?)\}\}", content, re.I):
  260. return "pend", 1
  261. elif re.search("\{\{afc submission\|t\|(.*?)\}\}", content, re.I):
  262. return None, 0
  263. elif re.search("\{\{afc submission\|d\|(.*?)\}\}", content, re.I):
  264. return "decline", 5
  265. return None, 0
  266. def get_short_title(self, title):
  267. return re.sub("Wikipedia(\s*talk)?\:Articles\sfor\screation\/", "", title)
  268. def get_create(self, pageid):
  269. query1 = "SELECT MIN(rev_id) FROM revision WHERE rev_page = ?"
  270. query1 = "SELECT rev_user_text, rev_timestamp, rev_id FROM revision WHERE rev_id = ?"
  271. result1 = self.site.sql_query(query1, (pageid,))
  272. rev_id = list(result1)[0][0]
  273. result2 = self.site.sql_query(query2, (rev_id,))
  274. m_user, m_time, m_id = list(result2)[0]
  275. return m_user, datetime.strptime(m_time, "%Y%m%d%H%M%S"), m_id
  276. def get_modify(self, pageid):
  277. query = """SELECT rev_user_text, rev_timestamp, rev_id FROM revision
  278. JOIN page ON rev_id = page_latest WHERE page_id = ?"""
  279. result = self.site.sql_query(query, (pageid,))
  280. m_user, m_time, m_id = list(result)[0]
  281. return m_user, datetime.strptime(m_time, "%Y%m%d%H%M%S"), m_id
  282. def get_special(self, page, status):
  283. return None, None, None
  284. def get_notes(self, page):
  285. return None