From 23f8e94e57462dd5cafd4ae10aa8c947fd0efcdb Mon Sep 17 00:00:00 2001 From: Paul Wise Date: Sat, 20 Apr 2013 13:46:50 +0800 Subject: [PATCH] Implement an incremental dump process --- MoinMoin/script/export/dump.py | 39 ++++++++++++++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) diff --git a/MoinMoin/script/export/dump.py b/MoinMoin/script/export/dump.py index d770e5b..91fa4ac 100644 --- a/MoinMoin/script/export/dump.py +++ b/MoinMoin/script/export/dump.py @@ -12,11 +12,16 @@ import sys, os, time, codecs, shutil, re, errno from MoinMoin import config, wikiutil, Page, user from MoinMoin import script from MoinMoin.action import AttachFile +from MoinMoin.logfile import editlog +from MoinMoin.util.filesys import touch url_prefix_static = "." logo_html = '' HTML_SUFFIX = ".html" +timestamp_text = u'''This is a MoinMoin timestamp file. +Please delete it to rebuild all pages. +''' page_template = u''' @@ -126,6 +131,21 @@ General syntax: moin [options] export dump [dump-options] help = "User the dump will be performed as (for ACL checks, etc)" ) + def changed_pages(self, request, timestamp): + pages = set() + old_pages = set() + unreadable_pages = set() + version = wikiutil.timestamp2version(timestamp) + log = editlog.EditLog(request) + for line in log: + if not request.user.may.read(line.pagename): + unreadable_pages.add(line.pagename) + elif line.ed_time_usecs < version: + old_pages.add(line.pagename) + else: + pages.add(line.pagename) + return (unreadable_pages, old_pages, pages) + def mainloop(self): """ moin-dump's main code. """ @@ -157,7 +177,24 @@ General syntax: moin [options] export dump [dump-options] # use this user for permissions checks request.user = user.User(request, name=self.options.dump_user) - pages = request.rootpage.getPageList(user='') # get list of all pages in wiki + timestamp = None + timestamp_file = os.path.join(outputdir, 'moin-last-update') + try: + timestamp = os.stat(timestamp_file).st_mtime + except OSError, err: + if err.errno == errno.EEXIST: + with open(timestamp_file, 'w') as f: + f.write(timestamp_text) + else: + script.fatal("Cannot check last update time of '%s'!" % timestamp_file) + if timestamp: + unreadable, old, pages = self.changed_pages(request, timestamp) + pages = list(pages) + touch(timestamp_file) + if unreadable or old: + script.log('Ignored %s old or unreadable pages' % (len(unreadable) + len(old))) + else: + pages = request.rootpage.getPageList(user='') # get list of all pages in wiki pages.sort() if self.options.page: # did user request a particular page or group of pages? try: -- 1.8.2.1