def dologsbackup(wikidb, outfile,
wikiconf, start, end, dryrun):
'''
do a logs xml dump one piece at a time, writing into uncompressed
temporary files and shovelling those into gzip's stdin for the
concatenated compressed output
'''
outfiles = {'logs': {'name': outfile}}
for filetype in outfiles:
outfiles[filetype]['temp'] = os.path.join(
wikiconf.temp_dir, os.path.basename(outfiles[filetype]['name']) + "_tmp")
if dryrun:
outfiles[filetype]['compr'] = None
else:
outfiles[filetype]['compr'] = gzippit(outfiles[filetype]['name'])
script_command = MultiVersion.mw_script_as_array(wikiconf, "dumpBackup.php")
command = [wikiconf.php] + script_command
command.extend(["--wiki=%s" % wikidb,
"--logs", "--report=1000",
"--output=file:%s" % outfiles['logs']['temp']
])
do_xml_stream(wikidb, outfiles, command, wikiconf,
start, end, dryrun, 'log_id', 'logging',
50000, 100000, '</logitem>\n')
def dostubsbackup(wikidb, history_file, current_file, articles_file,
wikiconf, start, end, dryrun):
'''
do a stubs xml dump one piece at a time, writing into uncompressed
temporary files and shovelling those into gzip's stdin for the
concatenated compressed output
'''
outfiles = {'history': {'name': history_file},
'current': {'name': current_file},
'articles': {'name': articles_file}}
for filetype in outfiles:
outfiles[filetype]['temp'] = os.path.join(
wikiconf.temp_dir, os.path.basename(outfiles[filetype]['name']) + "_tmp")
if dryrun:
outfiles[filetype]['compr'] = None
else:
outfiles[filetype]['compr'] = gzippit(outfiles[filetype]['name'])
script_command = MultiVersion.mw_script_as_array(wikiconf, "dumpBackup.php")
command = [wikiconf.php] + script_command
command.extend(["--wiki=%s" % wikidb,
"--full", "--stub", "--report=1000",
"--output=file:%s" % outfiles['history']['temp'],
"--output=file:%s" % outfiles['current']['temp'],
"--filter=latest",
"--output=file:%s" % outfiles['articles']['temp'],
"--filter=latest", "--filter=notalk",
"--filter=namespace:!NS_USER"])
if wikiconf.stubs_orderrevs:
command.append("--orderrevs")
callback = get_page_interval
else:
callback = None
do_xml_stream(wikidb, outfiles, command, wikiconf,
start, end, dryrun, 'page_id', 'page',
5000, 100000, '</page>\n', callback)