]> git.openstreetmap.org Git - osqa.git/blobdiff - forum_modules/exporter/exporter.py
the python XML module Element Tree name-space has been changed and didn't include...
[osqa.git] / forum_modules / exporter / exporter.py
index 9f1aeb1690104f85d52f254589414dceb793b28f..c6734298eedc15c932677dbc433f6f6dabbc3abe 100644 (file)
@@ -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 +9,44 @@ 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 +83,15 @@ 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:
@@ -76,13 +103,7 @@ 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))
-
-def _add_tag(el, name, content = None):
-    tag = ET.SubElement(el, name)
-    if content:
-        tag.text = content
-    return tag
+        file.write(_escape_cdata(node.tail.replace("\n", (level * identator )+ "\n"), encoding))
 
 def make_date(date, with_time=True):
     try:
@@ -95,7 +116,15 @@ 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))
@@ -135,17 +164,36 @@ 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')
@@ -160,20 +208,11 @@ 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))
@@ -184,30 +223,33 @@ 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')
+        tf.add(folder, arcname='/upfiles')
 
 
 def export_skinsfolder(tf):
     folder = djsettings.TEMPLATE_DIRS[0]
 
     if os.path.exists(folder):
-        tf.add(folder, arcname='skins')
+        tf.add(folder, arcname='/skins')
 
 
 def export(options, user):
@@ -272,10 +314,11 @@ 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:
@@ -397,7 +440,7 @@ def export_nodes(n, el, anon_data):
         el.add('author', n.author.id)
     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 "")
@@ -406,15 +449,22 @@ def export_nodes(n, el, anon_data):
     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:
@@ -426,6 +476,7 @@ def export_nodes(n, el, anon_data):
         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 "")
@@ -435,7 +486,7 @@ 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)
@@ -452,7 +503,7 @@ 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')
@@ -489,20 +540,3 @@ 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)
-
-
-
-
-
-
-
-
-        
-
-
-
-
-
-
-
-