X-Git-Url: https://git.openstreetmap.org./osqa.git/blobdiff_plain/23b2a2bfa95f12d0333a82c6e3d34456698522b7..626876f896404c5b8ed10e200d95cbedc5158f5b:/forum_modules/exporter/exporter.py diff --git a/forum_modules/exporter/exporter.py b/forum_modules/exporter/exporter.py index 5559e44..d652211 100644 --- a/forum_modules/exporter/exporter.py +++ b/forum_modules/exporter/exporter.py @@ -1,4 +1,6 @@ -import os, tarfile, datetime, logging, re, ConfigParser, shutil +from __future__ import with_statement + +import os, tarfile, datetime, logging, re, ConfigParser, shutil, zipfile from django.core.cache import cache from django.utils.translation import ugettext as _ @@ -7,19 +9,43 @@ from forum.settings import APP_URL from forum.templatetags.extra_tags import diff_date import xml.etree.ElementTree from xml.etree import ElementTree as ET -from xml.etree.ElementTree import Comment, _encode, ProcessingInstruction, QName, fixtag, _escape_attrib, _escape_cdata +from xml.etree.ElementTree import Comment, _encode, ProcessingInstruction, QName, _escape_attrib, _escape_cdata, _namespace_map from forum import settings from django.conf import settings as djsettings import settings as selfsettings +import string + +try: + from xml.etree.ElementTree import fixtag +except ImportError: + def fixtag(tag, namespaces): + # given a decorated tag (of the form {uri}tag), return prefixed + # tag and namespace declaration, if any + if isinstance(tag, QName): + tag = tag.text + namespace_uri, tag = string.split(tag[1:], "}", 1) + prefix = namespaces.get(namespace_uri) + if prefix is None: + prefix = _namespace_map.get(namespace_uri) + if prefix is None: + prefix = "ns%d" % len(namespaces) + namespaces[namespace_uri] = prefix + if prefix == "xml": + xmlns = None + else: + xmlns = ("xmlns:%s" % prefix, namespace_uri) + else: + xmlns = None + return "%s:%s" % (prefix, tag), xmlns CACHE_KEY = "%s_exporter_state" % APP_URL EXPORT_STEPS = [] TMP_FOLDER = os.path.join(os.path.dirname(__file__), 'tmp') -LAST_BACKUP = os.path.join(TMP_FOLDER, 'backup.tar.gz') DATE_AND_AUTHOR_INF_SECTION = 'DateAndAuthor' OPTIONS_INF_SECTION = 'Options' +META_INF_SECTION = 'Meta' DATETIME_FORMAT = "%Y-%m-%d %H:%M:%S" DATE_FORMAT = "%Y-%m-%d" @@ -56,15 +82,15 @@ def Etree_pretty__write(self, file, node, encoding, namespaces, if xmlns: xmlns_items.append(xmlns) except TypeError: raise #_raise_serialization_error(v) - file.write(" %s=\"%s\"" % (_encode(k, encoding), + file.write(u" %s=\"%s\"" % (_encode(k, encoding), _escape_attrib(v, encoding))) for k, v in xmlns_items: - file.write(" %s=\"%s\"" % (_encode(k, encoding), + file.write(u" %s=\"%s\"" % (_encode(k, encoding), _escape_attrib(v, encoding))) if node.text or len(node): file.write(">") if node.text: - file.write(_escape_cdata(node.text.replace("\n", (level + 1) * identator + "\n"), encoding)) + file.write(_escape_cdata(node.text, encoding)) for n in node: self._write(file, n, encoding, namespaces, level + 1, identator) if node.text and len(node.text) < 125: @@ -76,19 +102,28 @@ def Etree_pretty__write(self, file, node, encoding, namespaces, for k, v in xmlns_items: del namespaces[v] if node.tail: - file.write(_escape_cdata(node.tail.replace("\n", level * identator + "\n"), encoding)) + file.write(_escape_cdata(node.tail.replace("\n", (level * identator )+ "\n"), encoding)) + +def make_date(date, with_time=True): + try: + return date.strftime(with_time and DATETIME_FORMAT or DATE_FORMAT) + except ValueError, e: + return date.replace(year=1900).strftime(with_time and DATETIME_FORMAT or DATE_FORMAT) -def _add_tag(el, name, content = None): - tag = ET.SubElement(el, name) - if content: - tag.text = content - return tag def ET_Element_add_tag(el, tag_name, content = None, **attrs): tag = ET.SubElement(el, tag_name) if content: - tag.text = unicode(content).encode('utf-8') + try: + tag.text = unicode(content) + except Exception, e: + #logging.error('error converting unicode characters') + #import traceback + #logging.error(traceback.print_exc()) + + import string + tag.text = unicode("".join([c for c in content if c in string.printable])) for k, v in attrs.items(): tag.set(k, unicode(v)) @@ -128,17 +163,36 @@ def write_to_file(root, tmp, filename): tree = ET.ElementTree(root) tree.write(os.path.join(tmp, filename), encoding='UTF-8') -def create_targz(tmp, files, start_time, options, user, state, set_state): - if os.path.exists(LAST_BACKUP): - os.remove(LAST_BACKUP) - - t = tarfile.open(name=LAST_BACKUP, mode = 'w:gz') +def create_targz(tmp, files, start_time, options, user, state, set_state, file_format): + now = datetime.datetime.now() + domain = re.match('[\w-]+\.[\w-]+(\.[\w-]+)*', djsettings.APP_URL) + if domain: + domain = '_'.join(domain.get(0).split('.')) + else: + domain = 'localhost' + + fname = "%s-%s" % (domain, now.strftime('%Y%m%d%H%M')) + if file_format == 'zip': + full_fname = "%s.zip" % fname + else: + full_fname = "%s.tar.gz" % fname + + if file_format == 'zip': + t = zipfile.ZipFile(os.path.join(selfsettings.EXPORTER_BACKUP_STORAGE, full_fname), 'w') + + def add_to_file(f, a): + t.write(f, a) + else: + t = tarfile.open(os.path.join(selfsettings.EXPORTER_BACKUP_STORAGE, full_fname), mode = 'w:gz') + + def add_to_file(f, a): + t.add(f, a) state['overall']['status'] = _('Compressing xml files') set_state() for f in files: - t.add(os.path.join(tmp, f), arcname=f) + add_to_file(os.path.join(tmp, f), "/%s" % f) if options.get('uplodaded_files', False): state['overall']['status'] = _('Importing uploaded files') @@ -153,20 +207,11 @@ def create_targz(tmp, files, start_time, options, user, state, set_state): state['overall']['status'] = _('Writing inf file.') set_state() - now = datetime.datetime.now() - domain = re.match('[\w-]+\.[\w-]+(\.[\w-]+)*', djsettings.APP_URL) - if domain: - domain = '_'.join(domain.get(0).split('.')) - else: - domain = 'localhost' - - fname = "%s-%s" % (domain, now.strftime('%Y%m%d%H%M')) - inf = ConfigParser.SafeConfigParser() inf.add_section(DATE_AND_AUTHOR_INF_SECTION) - inf.set(DATE_AND_AUTHOR_INF_SECTION, 'file-name', "%s.tar.gz" % fname) + inf.set(DATE_AND_AUTHOR_INF_SECTION, 'file-name', full_fname) inf.set(DATE_AND_AUTHOR_INF_SECTION, 'author', unicode(user.id)) inf.set(DATE_AND_AUTHOR_INF_SECTION, 'site', djsettings.APP_URL) inf.set(DATE_AND_AUTHOR_INF_SECTION, 'started', start_time.strftime(DATETIME_FORMAT)) @@ -177,34 +222,43 @@ def create_targz(tmp, files, start_time, options, user, state, set_state): inf.set(OPTIONS_INF_SECTION, 'with-upfiles', str(options.get('uplodaded_files', False))) inf.set(OPTIONS_INF_SECTION, 'with-skins', str(options.get('import_skins_folder', False))) - with open(os.path.join(tmp, 'backup.inf'), 'wb') as inffile: + inf.add_section(META_INF_SECTION) + + for id, s in state.items(): + inf.set(META_INF_SECTION, id, str(s['count'])) + + with open(os.path.join(tmp, '%s.backup.inf' % fname), 'wb') as inffile: inf.write(inffile) - t.add(os.path.join(tmp, 'backup.inf'), arcname='backup.inf') + add_to_file(os.path.join(tmp, '%s.backup.inf' % fname), '/backup.inf') state['overall']['status'] = _('Saving backup file') set_state() t.close() - shutil.copyfile(LAST_BACKUP, os.path.join(selfsettings.EXPORTER_BACKUP_STORAGE, "%s.tar.gz" % fname)) - shutil.copyfile(os.path.join(tmp, 'backup.inf'), os.path.join(selfsettings.EXPORTER_BACKUP_STORAGE, "%s.backup.inf" % fname)) + return full_fname - def export_upfiles(tf): folder = str(settings.UPFILES_FOLDER) if os.path.exists(folder): - tf.add(folder, arcname='upfiles') + if isinstance(tf, zipfile.ZipFile): + tf.write(folder, arcname='/upfiles') + else: + tf.add(folder, arcname='/upfiles') def export_skinsfolder(tf): folder = djsettings.TEMPLATE_DIRS[0] if os.path.exists(folder): - tf.add(folder, arcname='skins') + if isinstance(tf, zipfile.ZipFile): + tf.write(folder, arcname='/skins') + else: + tf.add(folder, arcname='/skins') def export(options, user): - original__write = xml.etree.ElementTree.ElementTree._write + original__write = xml.etree.ElementTree.ElementTree.write xml.etree.ElementTree.ElementTree._write = Etree_pretty__write xml.etree.ElementTree._ElementInterface.add = ET_Element_add_tag @@ -265,10 +319,11 @@ def export(options, user): state['overall']['status'] = _('Compressing files') set_state() - create_targz(tmp, dump_files, start_time, options, user, state, set_state) + fname = create_targz(tmp, dump_files, start_time, options, user, state, set_state, options['file_format']) full_state['running'] = False full_state['errors'] = False state['overall']['status'] = _('Done') + state['overall']['fname'] = fname set_state() except Exception, e: @@ -332,14 +387,14 @@ def export_users(u, el, anon_data): el.add('email', u.email, validated=u.email_isvalid and 'true' or 'false') el.add('reputation', u.reputation) el.add('badges', bronze=u.bronze, silver=u.silver, gold=u.gold) - el.add('joindate', u.date_joined.strftime(DATETIME_FORMAT)) + el.add('joindate', make_date(u.date_joined)) el.add('active', u.is_active and 'true' or 'false') el.add('realname', u.real_name) el.add('bio', u.about) el.add('location', u.location) el.add('website', u.website) - el.add('birthdate', u.date_of_birth and u.date_of_birth.strftime(DATE_FORMAT) or "") + el.add('birthdate', u.date_of_birth and make_date(u.date_of_birth, with_time=False) or "") roles = el.add('roles') @@ -355,29 +410,34 @@ def export_users(u, el, anon_data): key.add('provider', a.provider) key.add('key', a.key) + try: + ss = u.subscription_settings - ss = u.subscription_settings - - notify = el.add('notifications', enabled=ss.enable_notifications and 'true' or 'false') + notify = el.add('notifications', enabled=ss.enable_notifications and 'true' or 'false') - notify.add('notify', **dict([(t, ss.__dict__.get(t, 'n') == 'i' and 'true' or 'false') for t in ['member_joins', 'new_question', 'new_question_watched_tags', 'subscribed_questions']])) + notify.add('notify', **dict([(t, ss.__dict__.get(t, 'n') == 'i' and 'true' or 'false') for t in ['member_joins', 'new_question', 'new_question_watched_tags', 'subscribed_questions']])) - notify.add('autoSubscribe', **dict([(t, ss.__dict__.get(t, False) and 'true' or 'false') for t in [ - 'all_questions', 'all_questions_watched_tags', 'questions_asked', 'questions_answered', 'questions_commented', 'questions_viewed']])) + notify.add('autoSubscribe', **dict([(t, ss.__dict__.get(t, False) and 'true' or 'false') for t in [ + 'all_questions', 'all_questions_watched_tags', 'questions_asked', 'questions_answered', 'questions_commented', 'questions_viewed']])) - notify.add('notifyOnSubscribed', **dict([(t, ss.__dict__.get("notify_%s" % t, False) and 'true' or 'false') for t in [ - 'answers', 'reply_to_comments', 'comments_own_post', 'comments', 'accepted']])) + notify.add('notifyOnSubscribed', **dict([(t, ss.__dict__.get("notify_%s" % t, False) and 'true' or 'false') for t in [ + 'answers', 'reply_to_comments', 'comments_own_post', 'comments', 'accepted']])) - notify.add('digest', ss.send_digest and 'on' or 'off') + notify.add('digest', ss.send_digest and 'on' or 'off') + except SubscriptionSettings.DoesNotExist: + pass watched = el.add('watchedTags') rejected = el.add('rejectedTags') for m in u.tag_selections.all(): - if m.reason == 'good': - watched.add('tag', m.tag.name) - else: - rejected.add('tag', m.tag.name) + try: + if m.reason == 'good': + watched.add('tag', m.tag.name) + else: + rejected.add('tag', m.tag.name) + except Tag.DoesNotExist: + pass @@ -388,37 +448,45 @@ def export_nodes(n, el, anon_data): if not anon_data: el.add('author', n.author.id) - el.add('date', n.added_at.strftime(DATETIME_FORMAT)) + el.add('date', make_date(n.added_at)) el.add('parent', n.parent and n.parent.id or "") - el.add('absparent', n.abs_parent and n.abs_parent or "") + el.add('absparent', n.abs_parent and n.abs_parent.id or "") act = el.add('lastactivity') act.add('by', n.last_activity_by and n.last_activity_by.id or "") - act.add('at', n.last_activity_at and n.last_activity_at.strftime(DATETIME_FORMAT) or "") + act.add('at', n.last_activity_at and make_date(n.last_activity_at) or "") el.add('title', n.title) el.add('body', n.body) + el.add('score', n.score) + tags = el.add('tags') for t in n.tagname_list(): tags.add('tag', t) - revs = el.add('revisions', active=n.active_revision and n.active_revision.revision or n.revisions.order_by('revision')[0].revision) + try: + active = n.active_revision and n.active_revision.revision or n.revisions.order_by('revision')[0].revision + except IndexError: + active = 0 + + revs = el.add('revisions', active=active) for r in n.revisions.order_by('revision'): - rev = _add_tag(revs, 'revision') + rev = revs.add('revision') rev.add('number', r.revision) rev.add('summary', r.summary) if not anon_data: rev.add('author', r.author.id) - rev.add('date', r.revised_at.strftime(DATETIME_FORMAT)) + rev.add('date', make_date(r.revised_at)) rev.add('title', r.title) rev.add('body', r.body) rev.add('tags', ", ".join(r.tagname_list())) el.add('marked', n.marked and 'true' or 'false') + el.add('wiki', n.nis.wiki and 'true' or 'false') el.add('extraRef', n.extra_ref and n.extra_ref.id or "") make_extra(el.add('extraData'), n.extra) el.add('extraCount', n.extra_count and n.extra_count or "") @@ -428,7 +496,7 @@ def export_nodes(n, el, anon_data): def export_actions(a, el, anon_data): el.add('id', a.id) el.add('type', a.action_type) - el.add('date', a.action_date) + el.add('date', make_date(a.action_date)) if not anon_data: el.add('user', a.user.id) @@ -445,7 +513,7 @@ def export_actions(a, el, anon_data): canceled.add('user', a.canceled_by.id) canceled.add('ip', a.canceled_ip) - canceled.add('date', a.canceled_at) + canceled.add('date', make_date(a.canceled_at)) if not anon_data: reputes = el.add('reputes') @@ -482,20 +550,3 @@ def export_awards(a, el, anon_data): def export_settings(s, el, anon_data): el.add('key', s.key) make_extra(el.add('value'), s.value) - - - - - - - - - - - - - - - - -