office-gobmx/bin/get-bugzilla-attachments-by-mimetype
Xisco Fauli 8b8e9d3126 bin: Add script to get attachments from OO forums
Testing it locally, I could download 52.000 documents
Reuse mimetypes dictionary from get-bugzilla-attachments-by-mimetype
by putting it into an external file

Change-Id: I875d90f6119c3c3bdfea6a0efd3bbc8c5be1eb63
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/135457
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
2022-06-07 20:26:13 +02:00

430 lines
15 KiB
Python
Executable file

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# This file is part of the LibreOffice project.
#
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
#
# This digs through a pile of bugzilla's and populates the cwd with a big
# collection of bug-docs in per-filetype dirs with bug-ids as names with
# prefixes to indicate which bug-tracker, e.g.
#
# fdo-bugid-X.suffix
# rhbz-bugid-X.suffix
# moz-bugid-X.suffix
#
# where X is the n'th attachment of that type in the bug
#
# The results are stored in the current directory, categorized by the
# extension of the downloaded file. When a file already exists, it is assumed
# it is already downloaded by a previous run, and up-to-date.
from __future__ import print_function
import feedparser
import base64
import datetime
import glob
import re
import os, os.path
import stat
import sys
import threading
try:
import queue
except:
import Queue as queue
try:
from urllib.request import urlopen
except:
from urllib import urlopen
try:
import xmlrpc.client as xmlrpclib
except:
import xmlrpclib
from xml.dom import minidom
from xml.sax.saxutils import escape
from attachment_mimetypes import mimetypes
def urlopen_retry(url):
maxretries = 3
for i in range(maxretries + 1):
try:
return urlopen(url)
except IOError as e:
print("caught IOError: " + str(e))
if maxretries == i:
raise
print("retrying...")
def get_from_bug_url_via_xml(url, mimetype, prefix, suffix):
id = url.rsplit('=', 2)[1]
print("id is " + prefix + id + " " + suffix)
print("parsing " + id)
sock = urlopen_retry(url+"&ctype=xml")
dom = minidom.parse(sock)
sock.close()
attachmentid=0
for attachment in dom.getElementsByTagName('attachment'):
attachmentid += 1
print(" mimetype is", end=' ')
for node in attachment.childNodes:
if node.nodeName == 'type':
# check if attachment is deleted
if not node.firstChild:
print('deleted attachment, skipping')
continue
print(node.firstChild.nodeValue, end=' ')
if node.firstChild.nodeValue.lower() != mimetype.lower():
print('skipping')
break
elif node.nodeName == 'data':
# check if attachment is deleted (i.e. https://bugs.kde.org/show_bug.cgi?id=53343&ctype=xml)
if not node.firstChild:
print('deleted attachment, skipping')
continue
download = suffix + '/' +prefix + id + '-' + str(attachmentid) + '.' + suffix
if os.path.isfile(download):
print("assuming " + download + " is up to date")
continue
# prevent re-downloading FDO attachments from TDF
if prefix == "tdf" and int(id) < 88776:
fdodownload = download.replace("tdf", "fdo")
if os.path.isfile(fdodownload):
print("assuming FDO " + fdodownload + " is up to date")
continue
print('downloading as ' + download)
tmpfile = download + ".tmp"
f = open(tmpfile, 'wb')
f.write(base64.b64decode(node.firstChild.nodeValue))
f.close()
os.rename(tmpfile, download)
break
def get_novell_bug_via_xml(url, mimetype, prefix, suffix):
id = url.rsplit('=', 2)[1]
print("id is " + prefix + id + " " + suffix)
print("parsing " + id)
sock = urlopen_retry(url+"&ctype=xml")
dom = minidom.parse(sock)
sock.close()
attachmentid=0
for comment in dom.getElementsByTagName('thetext'):
commentText = comment.firstChild.nodeValue
match = re.search(r".*Created an attachment \(id=([0-9]+)\)", commentText)
if not match:
continue
attachmentid += 1
download = suffix + '/' + prefix + id + '-' + str(attachmentid) + '.' + suffix
if os.path.isfile(download):
print("assuming " + download + " is up to date")
continue
realAttachmentId = match.group(1)
handle = urlopen_retry(novellattach + realAttachmentId)
if not handle:
print("attachment %s is not accessible" % realAttachmentId)
continue
print(" mimetype is", end=' ')
info = handle.info()
if info.get_content_type:
remoteMime = info.get_content_type()
else:
remoteMime = info.gettype()
print(remoteMime, end=' ')
if remoteMime != mimetype:
print("skipping")
continue
print('downloading as ' + download)
tmpfile = download + ".tmp"
f = open(tmpfile, 'wb')
f.write(handle.read())
f.close()
os.rename(tmpfile, download)
def create_query(mimetype):
query = dict()
query['query_format']='advanced'
query['field0-0-0']='attachments.mimetype'
query['type0-0-0']='equals'
query['value0-0-0']=mimetype
return query
def get_downloaded_files(prefix, suffix):
return glob.glob(os.path.join(suffix, '%s*.%s' % (prefix, suffix)))
def get_file_bz_ids(files, prefix):
return set([os.path.basename(f).split('-')[0].replace(prefix, '', 1) for f in files])
def get_changed_date(files):
newest = max([os.stat(f)[stat.ST_MTIME] for f in files])
# Subtract a day to avoid timezone differences. The worst thing that
# can happen is that we are going to process more bugs than necessary.
return datetime.date.fromtimestamp(newest - 24 * 60 * 60)
def get_through_rpc_query(rpcurl, showurl, mimetype, prefix, suffix):
try:
os.mkdir(suffix)
except:
pass
def process(query, full, have=[]):
try:
proxy = xmlrpclib.ServerProxy(rpcurl)
result = proxy.Bug.search(query)
bugs = result['bugs']
print(str(len(bugs)) + ' bugs to process')
if full:
available = set([str(bug['id']) for bug in bugs])
# we already have files from all available bugs
if available.difference(set(have)) == set():
print("assuming all downloaded files are up to date")
return
for bug in bugs:
url = showurl + str(bug['id'])
get_from_bug_url_via_xml(url, mimetype, prefix, suffix)
except xmlrpclib.Fault as err:
print("A fault occurred")
print("Fault code: %s" % err.faultCode)
print(err.faultString)
query = create_query(mimetype)
query['column_list']='bug_id'
files = get_downloaded_files(prefix, suffix)
if files != []:
print('looking for updated bugs having %s attachment(s)' % mimetype)
query_changed = query.copy()
query_changed['field0-1-0'] = 'days_elapsed'
query_changed['type0-1-0'] = 'lessthaneq'
query_changed['value0-1-0'] = str((datetime.date.today() - get_changed_date(files)).days)
process(query_changed, False)
print('looking for all bugs having %s attachment(s)' % mimetype)
process(query, True, get_file_bz_ids(files, prefix))
def get_through_rss_query(queryurl, mimetype, prefix, suffix):
try:
os.mkdir(suffix)
except:
pass
#Getting detailed bug information and downloading an attachment body is not possible without logging in to Novell bugzilla
#get_novell_bug_via_xml function is a workaround for that situation
get_bug_function = get_novell_bug_via_xml if prefix == "novell" else get_from_bug_url_via_xml
def process(query, full, have=[]):
url = queryurl + '?' + '&'.join(['='.join(kv) for kv in query.items()])
print('url is ' + url)
d = feedparser.parse(url)
print(str(len(d['entries'])) + ' bugs to process')
entries = []
for entry in d['entries']:
bugid = entry['id'].split('=')[-1]
entries.append(entry)
if full:
available = set([str(entry['id'].split('=')[-1]) for entry in entries])
# we already have files from all available bugs
if available.difference(set(have)) == set():
print("assuming all downloaded files are up to date")
return
for entry in entries:
try:
get_bug_function(entry['id'], mimetype, prefix, suffix)
except KeyboardInterrupt:
raise # Ctrl+C should work
except:
print(entry['id'] + " failed: " + str(sys.exc_info()[0]))
pass
query = create_query(escape(mimetype.replace("+","%2B")))
query['ctype'] = 'rss'
files = get_downloaded_files(prefix, suffix)
if files != []:
print('looking for updated bugs having %s attachment(s)' % mimetype)
query_changed = query.copy()
query_changed['field0-1-0'] = 'delta_ts'
query_changed['type0-1-0'] = 'greaterthaneq'
query_changed['value0-1-0'] = get_changed_date(files).isoformat()
process(query_changed, False)
print('looking for all bugs having %s attachment(s)' % mimetype)
process(query, True, get_file_bz_ids(files, prefix))
#since searching bugs having attachments with specific mimetypes is not available in launchpad API
#we're iterating over all bugs of the most interesting source packages
launchpad_pkgs = (
"abiword",
"calibre",
"calligra",
"gnumeric",
"inkscape",
"koffice",
"libabw",
"libcdr",
"libe-book",
"libetonyek",
"libfreehand",
"libmspub",
"libmwaw",
"liborcus",
"libpagemaker",
"libreoffice",
"libvisio",
"libwpd",
"libwpg",
"libwps",
"openoffice.org",
"python-uniconvertor",
"scribus",
"sk1",
"unoconv",
)
def get_launchpad_bugs(prefix):
#launchpadlib python module is required to download launchpad attachments
from launchpadlib.launchpad import Launchpad
launchpad = Launchpad.login_anonymously("attachmentdownload", "production")
ubuntu = launchpad.distributions["ubuntu"]
for pkg in launchpad_pkgs:
srcpkg = ubuntu.getSourcePackage(name=pkg)
pkgbugs = srcpkg.searchTasks(status=["New", "Fix Committed", "Invalid", "Won't Fix", "Confirmed", "Triaged", "In Progress", "Incomplete", "Incomplete (with response)", "Incomplete (without response)", "Fix Released", "Opinion", "Expired"])
for bugtask in pkgbugs:
bug = bugtask.bug
id = str(bug.id)
print("parsing " + id + " status: " + bugtask.status + " title: " + bug.title[:50])
attachmentid = 0
for attachment in bug.attachments:
attachmentid += 1
handle = attachment.data.open()
if not handle.content_type in mimetypes:
#print "skipping"
continue
suffix = mimetypes[handle.content_type]
if not os.path.isdir(suffix):
try:
os.mkdir(suffix)
except:
pass
download = suffix + '/' + prefix + id + '-' + str(attachmentid) + '.' + suffix
if os.path.isfile(download):
print("assuming " + id + " is up to date")
break
print('mimetype is ' + handle.content_type + ' downloading as ' + download)
tmpfile = download + ".tmp"
f = open(tmpfile, "wb")
f.write(handle.read())
f.close()
os.rename(tmpfile, download)
rss_bugzillas = (
( 'abi', 'http://bugzilla.abisource.com/buglist.cgi' ), #added for abiword
( 'fdo', 'http://bugs.freedesktop.org/buglist.cgi' ),
( 'gentoo', 'http://bugs.gentoo.org/buglist.cgi' ),
( 'gnome', 'http://bugzilla.gnome.org/buglist.cgi' ), # added for gnumeric
( 'kde', 'http://bugs.kde.org/buglist.cgi' ), # added for koffice/calligra
( 'mandriva', 'https://qa.mandriva.com/buglist.cgi' ),
( 'moz', 'https://bugzilla.mozilla.org/buglist.cgi' ),
# It seems something has changed and it is no longer possible to
# download any files from there.
# NOTE: I am leaving it in the list, commented out, just so someone
# does not add it back immediately .-)
# 'novell': 'https://bugzilla.novell.com/buglist.cgi',
# note: running this script against bz.apache.org apparently causes one's IP
# to be banned or something; you won't get new files in any case...
# ( 'ooo', 'https://bz.apache.org/ooo/buglist.cgi' ),
( 'tdf', 'http://bugs.documentfoundation.org/buglist.cgi' ),
)
redhatrpc = 'https://bugzilla.redhat.com/xmlrpc.cgi'
redhatbug = 'https://bugzilla.redhat.com/show_bug.cgi?id='
#Novell Bugzilla requires users to log in, in order to get details of the bugs such as attachment bodies etc.
#As a dirty workaround, we parse comments containing "Created an attachment (id=xxxxxx)" and download attachments manually
#python-bugzilla claims that it supports Novell bugzilla login but it's not working right now and novell bugzilla login
#system is a nightmare
novellattach = 'https://bugzilla.novell.com/attachment.cgi?id='
class manage_threads(threading.Thread):
def run(self):
#print(threading.current_thread().get_ident())
while 1:
# Try to receive a job from queue
try:
# Get job from queue
# Use job parameters to call our query
# Then let the queue know we are done with this job
(uri, mimetype, prefix, extension) = jobs.get(True,6)
try:
get_through_rss_query(uri, mimetype, prefix, extension)
finally:
jobs.task_done()
except KeyboardInterrupt:
raise # Ctrl+C should work
except queue.Empty:
break
def generate_multi_threading():
# Initialize threads
for i in range(max_threads):
manage_threads().start()
for (prefix, uri) in rss_bugzillas:
# Create a job for every mimetype for a bugzilla
for (mimetype,extension) in mimetypes.items():
# It seems that bugzilla has problems returning that many results
# (10000 results is probably a limit set somewhere) so we always
# end processing the complete list.
if mimetype == 'text/html' and prefix == 'moz':
continue
jobs.put([uri, mimetype, prefix, extension], block=True)
print("successfully placed a job in the queue searching for " + mimetype + " in bugtracker " + prefix)
# Continue when all mimetypes are done for a bugzilla
print("STARTED all bugtracker " + prefix)
jobs.join()
max_threads = 20 # Number of threads to create, (1 = without multi-threading)
jobs = queue.Queue()
generate_multi_threading()
for (mimetype,extension) in mimetypes.items():
get_through_rpc_query(redhatrpc, redhatbug, mimetype, "rhbz", extension)
try:
get_launchpad_bugs("lp")
except ImportError:
print("launchpadlib unavailable, skipping Ubuntu tracker")
# vim:set shiftwidth=4 softtabstop=4 expandtab: