Additional IRC commands and bot tasks for EarwigBot https://en.wikipedia.org/wiki/User:EarwigBot
Du kan inte välja fler än 25 ämnen Ämnen måste starta med en bokstav eller siffra, kan innehålla bindestreck ('-') och vara max 35 tecken långa.

afc_statistics.py 32 KiB

11 år sedan
11 år sedan
10 år sedan
10 år sedan
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744
  1. # -*- coding: utf-8 -*-
  2. #
  3. # Copyright (C) 2009-2014 Ben Kurtovic <ben.kurtovic@gmail.com>
  4. #
  5. # Permission is hereby granted, free of charge, to any person obtaining a copy
  6. # of this software and associated documentation files (the "Software"), to deal
  7. # in the Software without restriction, including without limitation the rights
  8. # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  9. # copies of the Software, and to permit persons to whom the Software is
  10. # furnished to do so, subject to the following conditions:
  11. #
  12. # The above copyright notice and this permission notice shall be included in
  13. # all copies or substantial portions of the Software.
  14. #
  15. # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16. # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17. # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  18. # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19. # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  20. # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  21. # SOFTWARE.
  22. from datetime import datetime
  23. import re
  24. from os.path import expanduser
  25. from threading import Lock
  26. from time import sleep
  27. import mwparserfromhell
  28. import oursql
  29. from earwigbot import exceptions
  30. from earwigbot import wiki
  31. from earwigbot.tasks import Task
  32. class AFCStatistics(Task):
  33. """A task to generate statistics for WikiProject Articles for Creation.
  34. Statistics are stored in a MySQL database ("u_earwig_afc_statistics")
  35. accessed with oursql. Statistics are synchronied with the live database
  36. every four minutes and saved once an hour, on the hour, to self.pagename.
  37. In the live bot, this is "Template:AFC statistics".
  38. """
  39. name = "afc_statistics"
  40. number = 2
  41. # Chart status number constants:
  42. CHART_NONE = 0
  43. CHART_PEND = 1
  44. CHART_REVIEW = 3
  45. CHART_ACCEPT = 4
  46. CHART_DECLINE = 5
  47. CHART_MISPLACE = 6
  48. def setup(self):
  49. self.cfg = cfg = self.config.tasks.get(self.name, {})
  50. self.site = self.bot.wiki.get_site()
  51. self.revision_cache = {}
  52. # Set some wiki-related attributes:
  53. self.pagename = cfg.get("page", "Template:AFC statistics")
  54. self.pending_cat = cfg.get("pending", "Pending AfC submissions")
  55. self.ignore_list = cfg.get("ignoreList", [])
  56. default_summary = "Updating statistics for [[WP:WPAFC|WikiProject Articles for creation]]."
  57. self.summary = self.make_summary(cfg.get("summary", default_summary))
  58. # Templates used in chart generation:
  59. templates = cfg.get("templates", {})
  60. self.tl_header = templates.get("header", "AFC statistics/header")
  61. self.tl_row = templates.get("row", "#invoke:AfC|row")
  62. self.tl_footer = templates.get("footer", "AFC statistics/footer")
  63. # Connection data for our SQL database:
  64. kwargs = cfg.get("sql", {})
  65. kwargs["read_default_file"] = expanduser("~/.my.cnf")
  66. self.conn_data = kwargs
  67. self.db_access_lock = Lock()
  68. def run(self, **kwargs):
  69. """Entry point for a task event.
  70. Depending on the kwargs passed, we will either synchronize our local
  71. statistics database with the site (self.sync()) or save it to the wiki
  72. (self.save()). We will additionally create an SQL connection with our
  73. local database.
  74. """
  75. action = kwargs.get("action")
  76. if not self.db_access_lock.acquire(False): # Non-blocking
  77. if action == "sync":
  78. self.logger.info("A sync is already ongoing; aborting")
  79. return
  80. self.logger.info("Waiting for database access lock")
  81. self.db_access_lock.acquire()
  82. try:
  83. self.site = self.bot.wiki.get_site()
  84. self.conn = oursql.connect(**self.conn_data)
  85. self.revision_cache = {}
  86. try:
  87. if action == "save":
  88. self.save(kwargs)
  89. elif action == "sync":
  90. self.sync(kwargs)
  91. finally:
  92. self.conn.close()
  93. finally:
  94. self.db_access_lock.release()
  95. #################### CHART BUILDING AND SAVING METHODS ####################
  96. def save(self, kwargs):
  97. """Save our local statistics to the wiki.
  98. After checking for emergency shutoff, the statistics chart is compiled,
  99. and then saved to self.pagename using self.summary iff it has changed
  100. since last save.
  101. """
  102. self.logger.info("Saving chart")
  103. if kwargs.get("fromIRC"):
  104. summary = self.summary + " (!earwigbot)"
  105. else:
  106. if self.shutoff_enabled():
  107. return
  108. summary = self.summary
  109. statistics = self._compile_charts()
  110. page = self.site.get_page(self.pagename)
  111. text = page.get()
  112. newtext = re.sub(u"<!-- stat begin -->(.*?)<!-- stat end -->",
  113. "<!-- stat begin -->\n" + statistics + "\n<!-- stat end -->",
  114. text, flags=re.DOTALL)
  115. if newtext == text:
  116. self.logger.info("Chart unchanged; not saving")
  117. return # Don't edit the page if we're not adding anything
  118. newtext = re.sub("<!-- sig begin -->(.*?)<!-- sig end -->",
  119. "<!-- sig begin -->~~~ at ~~~~~<!-- sig end -->",
  120. newtext)
  121. page.edit(newtext, summary, minor=True, bot=True)
  122. self.logger.info(u"Chart saved to [[{0}]]".format(page.title))
  123. def _compile_charts(self):
  124. """Compile and return all statistics information from our local db."""
  125. stats = ""
  126. with self.conn.cursor() as cursor:
  127. cursor.execute("SELECT * FROM chart")
  128. for chart in cursor:
  129. stats += self._compile_chart(chart) + "\n"
  130. return stats[:-1] # Drop the last newline
  131. def _compile_chart(self, chart_info):
  132. """Compile and return a single statistics chart."""
  133. chart_id, chart_title, special_title = chart_info
  134. chart = self.tl_header + "|" + chart_title
  135. if special_title:
  136. chart += "|" + special_title
  137. chart = "{{" + chart + "}}"
  138. query = "SELECT * FROM page JOIN row ON page_id = row_id WHERE row_chart = ?"
  139. with self.conn.cursor(oursql.DictCursor) as cursor:
  140. cursor.execute(query, (chart_id,))
  141. for page in cursor.fetchall():
  142. chart += "\n" + self._compile_chart_row(page)
  143. chart += "\n{{" + self.tl_footer + "}}"
  144. return chart
  145. def _compile_chart_row(self, page):
  146. """Compile and return a single chart row.
  147. 'page' is a dict of page information, taken as a row from the page
  148. table, where keys are column names and values are their cell contents.
  149. """
  150. row = u"{0}|s={page_status}|t={page_title}|z={page_size}|"
  151. if page["page_special_oldid"]:
  152. row += "sr={page_special_user}|sd={page_special_time}|si={page_special_oldid}|"
  153. row += "mr={page_modify_user}|md={page_modify_time}|mi={page_modify_oldid}"
  154. page["page_special_time"] = self._fmt_time(page["page_special_time"])
  155. page["page_modify_time"] = self._fmt_time(page["page_modify_time"])
  156. if page["page_notes"]:
  157. row += "|n=1{page_notes}"
  158. return "{{" + row.format(self.tl_row, **page) + "}}"
  159. def _fmt_time(self, date):
  160. """Format a datetime into the standard MediaWiki timestamp format."""
  161. return date.strftime("%H:%M, %d %b %Y")
  162. ######################## PRIMARY SYNC ENTRY POINTS ########################
  163. def sync(self, kwargs):
  164. """Synchronize our local statistics database with the site.
  165. Syncing involves, in order, updating tracked submissions that have
  166. been changed since last sync (self._update_tracked()), adding pending
  167. submissions that are not tracked (self._add_untracked()), and removing
  168. old submissions from the database (self._delete_old()).
  169. The sync will be canceled if SQL replication lag is greater than 600
  170. seconds, because this will lead to potential problems and outdated
  171. data, not to mention putting demand on an already overloaded server.
  172. Giving sync the kwarg "ignore_replag" will go around this restriction.
  173. """
  174. self.logger.info("Starting sync")
  175. replag = self.site.get_replag()
  176. self.logger.debug("Server replag is {0}".format(replag))
  177. if replag > 600 and not kwargs.get("ignore_replag"):
  178. msg = "Sync canceled as replag ({0} secs) is greater than ten minutes"
  179. self.logger.warn(msg.format(replag))
  180. return
  181. with self.conn.cursor() as cursor:
  182. self._update_tracked(cursor)
  183. self._add_untracked(cursor)
  184. self._update_stale(cursor)
  185. self._delete_old(cursor)
  186. self.logger.info("Sync completed")
  187. def _update_tracked(self, cursor):
  188. """Update tracked submissions that have been changed since last sync.
  189. This is done by iterating through every page in our database and
  190. comparing our stored latest revision ID with the actual latest revision
  191. ID from an SQL query. If they differ, we will update our information
  192. about the page (self._update_page()).
  193. If the page does not exist, we will remove it from our database with
  194. self._untrack_page().
  195. """
  196. self.logger.debug("Updating tracked submissions")
  197. query = """SELECT s.page_id, s.page_title, s.page_modify_oldid,
  198. r.page_latest, r.page_title, r.page_namespace
  199. FROM page AS s
  200. LEFT JOIN {0}_p.page AS r ON s.page_id = r.page_id
  201. WHERE s.page_modify_oldid != r.page_latest
  202. OR r.page_id IS NULL"""
  203. cursor.execute(query.format(self.site.name))
  204. for pageid, title, oldid, real_oldid, real_title, real_ns in cursor:
  205. if not real_oldid:
  206. self._untrack_page(cursor, pageid)
  207. continue
  208. msg = u"Updating page [[{0}]] (id: {1}) @ {2}"
  209. self.logger.debug(msg.format(title, pageid, oldid))
  210. msg = u" {0}: oldid: {1} -> {2}"
  211. self.logger.debug(msg.format(pageid, oldid, real_oldid))
  212. real_title = real_title.decode("utf8").replace("_", " ")
  213. ns = self.site.namespace_id_to_name(real_ns)
  214. if ns:
  215. real_title = u":".join((ns, real_title))
  216. try:
  217. self._update_page(cursor, pageid, real_title)
  218. except Exception:
  219. e = u"Error updating page [[{0}]] (id: {1})"
  220. self.logger.exception(e.format(real_title, pageid))
  221. def _add_untracked(self, cursor):
  222. """Add pending submissions that are not yet tracked.
  223. This is done by compiling a list of all currently tracked submissions
  224. and iterating through all members of self.pending_cat via SQL. If a
  225. page in the pending category is not tracked and is not in
  226. self.ignore_list, we will track it with self._track_page().
  227. """
  228. self.logger.debug("Adding untracked pending submissions")
  229. query = """SELECT r.page_id, r.page_title, r.page_namespace
  230. FROM {0}_p.page AS r
  231. INNER JOIN {0}_p.categorylinks AS c ON r.page_id = c.cl_from
  232. LEFT JOIN page AS s ON r.page_id = s.page_id
  233. WHERE s.page_id IS NULL AND c.cl_to = ?"""
  234. cursor.execute(query.format(self.site.name),
  235. (self.pending_cat.replace(" ", "_"),))
  236. for pageid, title, ns in cursor:
  237. title = title.decode("utf8").replace("_", " ")
  238. ns_name = self.site.namespace_id_to_name(ns)
  239. if ns_name:
  240. title = u":".join((ns_name, title))
  241. if title in self.ignore_list or ns == wiki.NS_CATEGORY:
  242. continue
  243. msg = u"Tracking page [[{0}]] (id: {1})".format(title, pageid)
  244. self.logger.debug(msg)
  245. try:
  246. self._track_page(cursor, pageid, title)
  247. except Exception:
  248. e = u"Error tracking page [[{0}]] (id: {1})"
  249. self.logger.exception(e.format(title, pageid))
  250. def _update_stale(self, cursor):
  251. """Update submissions that haven't been updated in a long time.
  252. This is intended to update notes that change without typical update
  253. triggers, like when submitters are blocked. It also resolves conflicts
  254. when pages are tracked during high replag, potentially causing data to
  255. be inaccurate (like a missed decline). It updates no more than the ten
  256. stalest pages that haven't been updated in two days.
  257. """
  258. self.logger.info("Updating stale submissions")
  259. query = """SELECT page_id, page_title, page_modify_oldid
  260. FROM page JOIN updatelog ON page_id = update_id
  261. WHERE ADDTIME(update_time, '48:00:00') < NOW()
  262. ORDER BY update_time ASC LIMIT 10"""
  263. cursor.execute(query)
  264. for pageid, title, oldid in cursor:
  265. msg = u"Updating page [[{0}]] (id: {1}) @ {2}"
  266. self.logger.debug(msg.format(title, pageid, oldid))
  267. try:
  268. self._update_page(cursor, pageid, title)
  269. except Exception:
  270. e = u"Error updating page [[{0}]] (id: {1})"
  271. self.logger.exception(e.format(title, pageid))
  272. def _delete_old(self, cursor):
  273. """Remove old submissions from the database.
  274. "Old" is defined as a submission that has been declined or accepted
  275. more than 36 hours ago. Pending submissions cannot be "old".
  276. """
  277. self.logger.debug("Removing old submissions from chart")
  278. query = """DELETE FROM page, row, updatelog USING page JOIN row
  279. ON page_id = row_id JOIN updatelog ON page_id = update_id
  280. WHERE row_chart IN (?, ?)
  281. AND ADDTIME(page_special_time, '36:00:00') < NOW()"""
  282. cursor.execute(query, (self.CHART_ACCEPT, self.CHART_DECLINE))
  283. ######################## PRIMARY PAGE ENTRY POINTS ########################
  284. def _untrack_page(self, cursor, pageid):
  285. """Remove a page, given by ID, from our database."""
  286. self.logger.debug("Untracking page (id: {0})".format(pageid))
  287. query = """DELETE FROM page, row, updatelog USING page JOIN row
  288. ON page_id = row_id JOIN updatelog ON page_id = update_id
  289. WHERE page_id = ?"""
  290. cursor.execute(query, (pageid,))
  291. def _track_page(self, cursor, pageid, title):
  292. """Update hook for when page is not in our database.
  293. A variety of SQL queries are used to gather information about the page,
  294. which is then saved to our database.
  295. """
  296. content = self._get_content(pageid)
  297. if content is None:
  298. msg = u"Could not get page content for [[{0}]]".format(title)
  299. self.logger.error(msg)
  300. return
  301. namespace = self.site.get_page(title).namespace
  302. status, chart = self._get_status_and_chart(content, namespace)
  303. if chart == self.CHART_NONE:
  304. msg = u"Could not find a status for [[{0}]]".format(title)
  305. self.logger.warn(msg)
  306. return
  307. m_user, m_time, m_id = self._get_modify(pageid)
  308. s_user, s_time, s_id = self._get_special(pageid, content, chart)
  309. notes = self._get_notes(chart, content, m_time, s_user)
  310. query1 = "INSERT INTO row VALUES (?, ?)"
  311. query2 = "INSERT INTO page VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)"
  312. query3 = "INSERT INTO updatelog VALUES (?, ?)"
  313. cursor.execute(query1, (pageid, chart))
  314. cursor.execute(query2, (pageid, status, title, len(content), notes,
  315. m_user, m_time, m_id, s_user, s_time, s_id))
  316. cursor.execute(query3, (pageid, datetime.utcnow()))
  317. def _update_page(self, cursor, pageid, title):
  318. """Update hook for when page is already in our database.
  319. A variety of SQL queries are used to gather information about the page,
  320. which is compared against our stored information. Differing information
  321. is then updated.
  322. """
  323. content = self._get_content(pageid)
  324. if content is None:
  325. msg = u"Could not get page content for [[{0}]]".format(title)
  326. self.logger.error(msg)
  327. return
  328. namespace = self.site.get_page(title).namespace
  329. status, chart = self._get_status_and_chart(content, namespace)
  330. if chart == self.CHART_NONE:
  331. self._untrack_page(cursor, pageid)
  332. return
  333. query = "SELECT * FROM page JOIN row ON page_id = row_id WHERE page_id = ?"
  334. with self.conn.cursor(oursql.DictCursor) as dict_cursor:
  335. dict_cursor.execute(query, (pageid,))
  336. result = dict_cursor.fetchall()[0]
  337. m_user, m_time, m_id = self._get_modify(pageid)
  338. if title != result["page_title"]:
  339. self._update_page_title(cursor, result, pageid, title)
  340. if m_id != result["page_modify_oldid"]:
  341. self._update_page_modify(cursor, result, pageid, len(content),
  342. m_user, m_time, m_id)
  343. if status != result["page_status"]:
  344. special = self._update_page_status(cursor, result, pageid, content,
  345. status, chart)
  346. s_user = special[0]
  347. else:
  348. s_user = result["page_special_user"]
  349. notes = self._get_notes(chart, content, m_time, s_user)
  350. if notes != result["page_notes"]:
  351. self._update_page_notes(cursor, result, pageid, notes)
  352. query = "UPDATE updatelog SET update_time = ? WHERE update_id = ?"
  353. cursor.execute(query, (datetime.utcnow(), pageid))
  354. ###################### PAGE ATTRIBUTE UPDATE METHODS ######################
  355. def _update_page_title(self, cursor, result, pageid, title):
  356. """Update the title of a page in our database."""
  357. query = "UPDATE page SET page_title = ? WHERE page_id = ?"
  358. cursor.execute(query, (title, pageid))
  359. msg = u" {0}: title: {1} -> {2}"
  360. self.logger.debug(msg.format(pageid, result["page_title"], title))
  361. def _update_page_modify(self, cursor, result, pageid, size, m_user, m_time, m_id):
  362. """Update the last modified information of a page in our database."""
  363. query = """UPDATE page SET page_size = ?, page_modify_user = ?,
  364. page_modify_time = ?, page_modify_oldid = ?
  365. WHERE page_id = ?"""
  366. cursor.execute(query, (size, m_user, m_time, m_id, pageid))
  367. msg = u" {0}: modify: {1} / {2} / {3} -> {4} / {5} / {6}"
  368. msg = msg.format(pageid, result["page_modify_user"],
  369. result["page_modify_time"],
  370. result["page_modify_oldid"], m_user, m_time, m_id)
  371. self.logger.debug(msg)
  372. def _update_page_status(self, cursor, result, pageid, content, status, chart):
  373. """Update the status and "specialed" information of a page."""
  374. query1 = """UPDATE page JOIN row ON page_id = row_id
  375. SET page_status = ?, row_chart = ? WHERE page_id = ?"""
  376. query2 = """UPDATE page SET page_special_user = ?,
  377. page_special_time = ?, page_special_oldid = ?
  378. WHERE page_id = ?"""
  379. cursor.execute(query1, (status, chart, pageid))
  380. msg = " {0}: status: {1} ({2}) -> {3} ({4})"
  381. self.logger.debug(msg.format(pageid, result["page_status"],
  382. result["row_chart"], status, chart))
  383. s_user, s_time, s_id = self._get_special(pageid, content, chart)
  384. if s_id != result["page_special_oldid"]:
  385. cursor.execute(query2, (s_user, s_time, s_id, pageid))
  386. msg = u" {0}: special: {1} / {2} / {3} -> {4} / {5} / {6}"
  387. msg = msg.format(pageid, result["page_special_user"],
  388. result["page_special_time"],
  389. result["page_special_oldid"], s_user, s_time, s_id)
  390. self.logger.debug(msg)
  391. return s_user, s_time, s_id
  392. def _update_page_notes(self, cursor, result, pageid, notes):
  393. """Update the notes (or warnings) of a page in our database."""
  394. query = "UPDATE page SET page_notes = ? WHERE page_id = ?"
  395. cursor.execute(query, (notes, pageid))
  396. msg = " {0}: notes: {1} -> {2}"
  397. self.logger.debug(msg.format(pageid, result["page_notes"], notes))
  398. ###################### DATA RETRIEVAL HELPER METHODS ######################
  399. def _get_content(self, pageid):
  400. """Get the current content of a page by ID from the API.
  401. The page's current revision ID is retrieved from SQL, and then
  402. an API query is made to get its content. This is the only API query
  403. used in the task's code.
  404. """
  405. query = "SELECT page_latest FROM page WHERE page_id = ?"
  406. result = self.site.sql_query(query, (pageid,))
  407. try:
  408. revid = int(list(result)[0][0])
  409. except IndexError:
  410. return None
  411. return self._get_revision_content(revid)
  412. def _get_revision_content(self, revid, tries=1):
  413. """Get the content of a revision by ID from the API."""
  414. if revid in self.revision_cache:
  415. return self.revision_cache[revid]
  416. res = self.site.api_query(action="query", prop="revisions",
  417. revids=revid, rvprop="content")
  418. try:
  419. content = res["query"]["pages"].values()[0]["revisions"][0]["*"]
  420. except KeyError:
  421. if tries == 0:
  422. raise
  423. sleep(5)
  424. return self._get_revision_content(revid, tries=tries - 1)
  425. self.revision_cache[revid] = content
  426. return content
  427. def _get_status_and_chart(self, content, namespace):
  428. """Determine the status and chart number of an AFC submission.
  429. The methodology used here is the same one I've been using for years
  430. (see also commands.afc_report), but with the new draft system taken
  431. into account. The order here is important: if there is more than one
  432. {{AFC submission}} template on a page, we need to know which one to
  433. use (revision history search to find the most recent isn't a viable
  434. idea :P).
  435. """
  436. statuses = self.get_statuses(content)
  437. if namespace == wiki.NS_MAIN:
  438. if statuses:
  439. return None, self.CHART_MISPLACE
  440. return "a", self.CHART_ACCEPT
  441. elif "R" in statuses:
  442. return "r", self.CHART_REVIEW
  443. elif "P" in statuses:
  444. return "p", self.CHART_PEND
  445. elif "T" in statuses:
  446. return None, self.CHART_NONE
  447. elif "D" in statuses:
  448. return "d", self.CHART_DECLINE
  449. return None, self.CHART_NONE
  450. def get_statuses(self, content):
  451. """Return a list of all AFC submission statuses in a page's text."""
  452. valid = ["P", "R", "T", "D"]
  453. aliases = {
  454. "submit": "P",
  455. "afc submission/submit": "P",
  456. "afc submission/reviewing": "R",
  457. "afc submission/pending": "P",
  458. "afc submission/draft": "T",
  459. "afc submission/declined": "D"
  460. }
  461. statuses = []
  462. code = mwparserfromhell.parse(content)
  463. for template in code.filter_templates():
  464. name = template.name.strip().lower()
  465. if name == "afc submission":
  466. if template.has(1, ignore_empty=True):
  467. status = template.get(1).value.strip().upper()
  468. statuses.append(status if status in valid else "P")
  469. else:
  470. statuses.append("P")
  471. elif name in aliases:
  472. statuses.append(aliases[name])
  473. return statuses
  474. def _get_modify(self, pageid):
  475. """Return information about a page's last edit ("modification").
  476. This consists of the most recent editor, modification time, and the
  477. lastest revision ID.
  478. """
  479. query = """SELECT rev_user_text, rev_timestamp, rev_id FROM revision
  480. JOIN page ON rev_id = page_latest WHERE page_id = ?"""
  481. result = self.site.sql_query(query, (pageid,))
  482. m_user, m_time, m_id = list(result)[0]
  483. timestamp = datetime.strptime(m_time, "%Y%m%d%H%M%S")
  484. return m_user.decode("utf8"), timestamp, m_id
  485. def _get_special(self, pageid, content, chart):
  486. """Return information about a page's "special" edit.
  487. I tend to use the term "special" as a verb a lot, which is bound to
  488. cause confusion. It is merely a short way of saying "the edit in which
  489. a declined submission was declined, an accepted submission was
  490. accepted, a submission in review was set as such, a pending submission
  491. was submitted, and a "misplaced" submission was created."
  492. This "information" consists of the special edit's editor, its time, and
  493. its revision ID. If the page's status is not something that involves
  494. "special"-ing, we will return None for all three. The same will be
  495. returned if we cannot determine when the page was "special"-ed.
  496. """
  497. charts = {
  498. self.CHART_NONE: (lambda pageid, content: None, None, None),
  499. self.CHART_MISPLACE: self.get_create,
  500. self.CHART_ACCEPT: self.get_accepted,
  501. self.CHART_REVIEW: self.get_reviewing,
  502. self.CHART_PEND: self.get_pending,
  503. self.CHART_DECLINE: self.get_decline
  504. }
  505. return charts[chart](pageid, content)
  506. def get_create(self, pageid, content=None):
  507. """Return (creator, create_ts, create_revid) for the given page."""
  508. query = """SELECT rev_user_text, rev_timestamp, rev_id
  509. FROM revision WHERE rev_id =
  510. (SELECT MIN(rev_id) FROM revision WHERE rev_page = ?)"""
  511. result = self.site.sql_query(query, (pageid,))
  512. c_user, c_time, c_id = list(result)[0]
  513. timestamp = datetime.strptime(c_time, "%Y%m%d%H%M%S")
  514. return c_user.decode("utf8"), timestamp, c_id
  515. def get_accepted(self, pageid, content=None):
  516. """Return (acceptor, accept_ts, accept_revid) for the given page."""
  517. query = """SELECT rev_user_text, rev_timestamp, rev_id FROM revision
  518. WHERE rev_comment LIKE "% moved page [[%]] to [[%]]%"
  519. AND rev_page = ? ORDER BY rev_timestamp DESC LIMIT 1"""
  520. result = self.site.sql_query(query, (pageid,))
  521. try:
  522. a_user, a_time, a_id = list(result)[0]
  523. except IndexError:
  524. return None, None, None
  525. timestamp = datetime.strptime(a_time, "%Y%m%d%H%M%S")
  526. return a_user.decode("utf8"), timestamp, a_id
  527. def get_reviewing(self, pageid, content=None):
  528. """Return (reviewer, review_ts, review_revid) for the given page."""
  529. return self._search_history(pageid, self.CHART_REVIEW, ["R"], [])
  530. def get_pending(self, pageid, content):
  531. """Return (submitter, submit_ts, submit_revid) for the given page."""
  532. res = self._get_status_helper(pageid, content, ("P", ""), ("u", "ts"))
  533. return res or self._search_history(pageid, self.CHART_PEND, ["P"], [])
  534. def get_decline(self, pageid, content):
  535. """Return (decliner, decline_ts, decline_revid) for the given page."""
  536. params = ("decliner", "declinets")
  537. res = self._get_status_helper(pageid, content, ("D"), params)
  538. return res or self._search_history(
  539. pageid, self.CHART_DECLINE, ["D"], ["R", "P", "T"])
  540. def _get_status_helper(self, pageid, content, statuses, params):
  541. """Helper function for get_pending() and get_decline()."""
  542. submits = []
  543. code = mwparserfromhell.parse(content)
  544. for tmpl in code.filter_templates():
  545. status = tmpl.get(1).value.strip().upper() if tmpl.has(1) else "P"
  546. if tmpl.name.strip().lower() == "afc submission":
  547. if all([tmpl.has(par, ignore_empty=True) for par in params]):
  548. if status in statuses:
  549. data = [unicode(tmpl.get(par).value) for par in params]
  550. submits.append(data)
  551. if not submits:
  552. return None
  553. user, stamp = max(submits, key=lambda pair: pair[1])
  554. query = """SELECT rev_id FROM revision WHERE rev_page = ?
  555. AND rev_user_text = ? AND ABS(rev_timestamp - ?) <= 60
  556. ORDER BY ABS(rev_timestamp - ?) ASC LIMIT 1"""
  557. result = self.site.sql_query(query, (pageid, user, stamp, stamp))
  558. try:
  559. dtime = datetime.strptime(stamp, "%Y%m%d%H%M%S")
  560. return user, dtime, list(result)[0][0]
  561. except (ValueError, IndexError):
  562. return None
  563. def _search_history(self, pageid, chart, search_with, search_without):
  564. """Search through a page's history to find when a status was set.
  565. Linear search backwards in time for the edit right after the most
  566. recent edit that fails the (pseudocode) test:
  567. ``status_set(any(search_with)) && !status_set(any(search_without))``
  568. """
  569. query = """SELECT rev_user_text, rev_timestamp, rev_id
  570. FROM revision WHERE rev_page = ? ORDER BY rev_id DESC"""
  571. result = self.site.sql_query(query, (pageid,))
  572. counter = 0
  573. last = (None, None, None)
  574. for user, ts, revid in result:
  575. counter += 1
  576. if counter > 50:
  577. msg = "Exceeded 50 content lookups while searching history of page (id: {0}, chart: {1})"
  578. self.logger.warn(msg.format(pageid, chart))
  579. return None, None, None
  580. try:
  581. content = self._get_revision_content(revid)
  582. except exceptions.APIError:
  583. msg = "API error interrupted SQL query in _search_history() for page (id: {0}, chart: {1})"
  584. self.logger.exception(msg.format(pageid, chart))
  585. return None, None, None
  586. statuses = self.get_statuses(content)
  587. req = search_with and not any([s in statuses for s in search_with])
  588. if any([s in statuses for s in search_without]) or req:
  589. return last
  590. timestamp = datetime.strptime(ts, "%Y%m%d%H%M%S")
  591. last = (user.decode("utf8"), timestamp, revid)
  592. return last
  593. def _get_notes(self, chart, content, m_time, s_user):
  594. """Return any special notes or warnings about this page.
  595. copyvio: submission is a suspected copyright violation
  596. unsourced: submission lacks references completely
  597. no-inline: submission has no inline citations
  598. short: submission is less than a kilobyte in length
  599. resubmit: submission was resubmitted after a previous decline
  600. old: submission has not been touched in > 4 days
  601. blocked: submitter is currently blocked
  602. """
  603. notes = ""
  604. ignored_charts = [self.CHART_NONE, self.CHART_ACCEPT, self.CHART_DECLINE]
  605. if chart in ignored_charts:
  606. return notes
  607. copyvios = self.config.tasks.get("afc_copyvios", {})
  608. regex = r"\{\{s*" + copyvios.get("template", "AfC suspected copyvio")
  609. if re.search(regex, content):
  610. notes += "|nc=1" # Submission is a suspected copyvio
  611. if not re.search(r"\<ref\s*(.*?)\>(.*?)\</ref\>", content, re.I|re.S):
  612. regex = r"(https?:)|\[//(?!{0})([^ \]\t\n\r\f\v]+?)"
  613. sitedomain = re.escape(self.site.domain)
  614. if re.search(regex.format(sitedomain), content, re.I | re.S):
  615. notes += "|ni=1" # Submission has no inline citations
  616. else:
  617. notes += "|nu=1" # Submission is completely unsourced
  618. if len(content) < 1000:
  619. notes += "|ns=1" # Submission is short
  620. statuses = self.get_statuses(content)
  621. if "D" in statuses and chart != self.CHART_MISPLACE:
  622. notes += "|nr=1" # Submission was resubmitted
  623. time_since_modify = (datetime.utcnow() - m_time).total_seconds()
  624. max_time = 4 * 24 * 60 * 60
  625. if time_since_modify > max_time:
  626. notes += "|no=1" # Submission hasn't been touched in over 4 days
  627. if chart == self.CHART_PEND and s_user:
  628. submitter = self.site.get_user(s_user)
  629. try:
  630. if submitter.blockinfo:
  631. notes += "|nb=1" # Submitter is blocked
  632. except exceptions.UserNotFoundError: # Likely an IP
  633. pass
  634. return notes