Migrate XML handling to ElementTree

11 files changed, 150 insertions, 230 deletions

diff --git a/layman/overlay.py b/layman/overlay.py
index 5b8c73f..5372ba7 100644
--- a/layman/overlay.py
+++ b/layman/overlay.py
@@ -15,6 +15,7 @@
 # Author(s):
 #             Gunnar Wrobel <wrobel@gentoo.org>
 #             Sebastian Pipping <sebastian@pipping.org>
+#             Christian Groschupp <christian@groschupp.org>
 #
 '''Main handler for overlays.'''
 
@@ -26,7 +27,8 @@ __version__ = "$Id: overlay.py 273 2006-12-30 15:54:50Z wrobel $"
 #
 #-------------------------------------------------------------------------------
 
-import sys, codecs, os, os.path, xml.dom.minidom
+import sys, codecs, os, os.path
+import elementtree.ElementTree as ET
 
 from   layman.overlays.bzr       import BzrOverlay
 from   layman.overlays.darcs     import DarcsOverlay
@@ -45,14 +47,16 @@ from   layman.debug              import OUT
 #
 #-------------------------------------------------------------------------------
 
-OVERLAY_TYPES = {'git'       : GitOverlay,
-                 'cvs'       : CvsOverlay,
-                 'svn'       : SvnOverlay,
-                 'rsync'     : RsyncOverlay,
-                 'tar'       : TarOverlay,
-                 'bzr'       : BzrOverlay,
-                 'mercurial' : MercurialOverlay,
-                 'darcs'     : DarcsOverlay}
+OVERLAY_TYPES = dict((e.type_key, e) for e in (
+    GitOverlay,
+    CvsOverlay,
+    SvnOverlay,
+    RsyncOverlay,
+    TarOverlay,
+    BzrOverlay,
+    MercurialOverlay,
+    DarcsOverlay
+))
 
 #===============================================================================
 #
@@ -81,17 +85,15 @@ class Overlays:
         '''Read the overlay definition file.'''
 
         try:
-
-            document = open(path).read()
+            document = open(path, 'r').read()
 
         except Exception, error:
             raise IOError('Failed to read the overlay list at ("'
                           + path + '")!\nError was:\n' + str(error))
 
-
         self.read(document)
 
-    def read(self, document):
+    def read(self, text):
         '''
         Read an xml list of overlays.
 
@@ -101,34 +103,28 @@ class Overlays:
         >>> a.overlays.keys()
         [u'wrobel', u'wrobel-stable']
 
-        >>> a.overlays['wrobel-stable'].data['&src']
+        >>> a.overlays['wrobel-stable'].src
         u'rsync://gunnarwrobel.de/wrobel-stable'
         '''
-        try:
-
-            document = xml.dom.minidom.parseString(document)
-
-        except Exception, error:
-            raise Exception('Failed to parse the overlay list!\nError was:\n'
-                            + str(error))
-
-        overlays = document.getElementsByTagName('overlay') + \
-                document.getElementsByTagName('repo')
+        document = ET.fromstring(text)
+        overlays = document.findall('overlay') + \
+                document.findall('repo')
 
         for overlay in overlays:
-
             OUT.debug('Parsing overlay entry', 8)
-            try:
-                element_to_scan = overlay.getElementsByTagName('source')[0]
-            except IndexError:
+            
+            _source = overlay.find('source')
+            if _source != None:
+                element_to_scan = _source
+            else:
                 element_to_scan = overlay
 
-            for index in range(0, element_to_scan.attributes.length):
-                attr = element_to_scan.attributes.item(index)
-                if attr.name == 'type':
-                    if attr.nodeValue in OVERLAY_TYPES.keys():
+            for attr_name in element_to_scan.attrib.keys():
+                if attr_name == 'type':
+                    overlay_type = element_to_scan.attrib['type']
+                    if overlay_type in OVERLAY_TYPES.keys():
                         try:
-                            ovl = OVERLAY_TYPES[attr.nodeValue](overlay,
+                            ovl = OVERLAY_TYPES[overlay_type](overlay,
                                                                 self.ignore,
                                                                 self.quiet)
                             self.overlays[ovl.name] = ovl
@@ -136,7 +132,8 @@ class Overlays:
                             OUT.warn(str(error), 3)
                     else:
                         raise Exception('Unknown overlay type "' +
-                                        attr.nodeValue + '"!')
+                                        overlay_type + '"!')
+                    break
 
     def write(self, path):
         '''
@@ -156,22 +153,16 @@ class Overlays:
         >>> os.unlink(write)
         '''
 
-        imp = xml.dom.minidom.getDOMImplementation()
-
-        doc = imp.createDocument('layman', 'repositories', None)
-
-        root =  doc.childNodes[0]
-
-        for name, overlay in self.overlays.items():
-
-            root.appendChild(overlay.to_minidom(doc))
-
+        xml = ET.Element('repositories', version="1.0")
+        xml[:] = [e.to_minidom() for e in self.overlays.values()]
+        tree = ET.ElementTree(xml)
         try:
-
-            out_file = codecs.open(path, 'w', 'utf-8')
-
-            doc.writexml(out_file, '', '  ', '\n')
-
+            f = open(path, 'w')
+            f.write("""\
+<?xml version="1.0" encoding="UTF-8"?>
+""")
+            tree.write(f, encoding='utf-8')
+            f.close()
         except Exception, error:
             raise Exception('Failed to write to local overlays file: '
                             + path + '\nError was:\n' + str(error))
@@ -182,7 +173,7 @@ class Overlays:
 
         >>> here = os.path.dirname(os.path.realpath(__file__))
         >>> a = Overlays([here + '/tests/testfiles/global-overlays.xml', ])
-        >>> a.select('wrobel-stable').data['&src']
+        >>> a.select('wrobel-stable').src
         u'rsync://gunnarwrobel.de/wrobel-stable'
         '''
         if overlay in self.overlays.keys():
diff --git a/layman/overlays/bzr.py b/layman/overlays/bzr.py
index 4c14278..64fbbdd 100644
--- a/layman/overlays/bzr.py
+++ b/layman/overlays/bzr.py
@@ -39,6 +39,7 @@ class BzrOverlay(Overlay):
     ''' Handles bzr overlays.'''
 
     type = 'Bzr'
+    type_key = 'bzr'
 
     binary_command  = '/usr/bin/bzr'
 
diff --git a/layman/overlays/cvs.py b/layman/overlays/cvs.py
index d9710de..a28889a 100644
--- a/layman/overlays/cvs.py
+++ b/layman/overlays/cvs.py
@@ -37,6 +37,7 @@ class CvsOverlay(Overlay):
     ''' Handles cvs overlays.'''
 
     type = 'cvs'
+    type_key = 'cvs'
 
     binary = '/usr/bin/cvs'
 
@@ -44,8 +45,8 @@ class CvsOverlay(Overlay):
 
         Overlay.__init__(self, xml, ignore, quiet)
 
-        if '&subpath' in self.data.keys():
-            self.subpath = self.data['&subpath']
+        if 'subpath' in xml.attrib:
+            self.subpath = xml.attrib['subpath']
         else:
             self.subpath = ''
 
diff --git a/layman/overlays/darcs.py b/layman/overlays/darcs.py
index 0d27d5e..e066eee 100644
--- a/layman/overlays/darcs.py
+++ b/layman/overlays/darcs.py
@@ -38,6 +38,7 @@ class DarcsOverlay(Overlay):
     ''' Handles darcs overlays.'''
 
     type = 'Darcs'
+    type_key = 'darcs'
 
     binary_command  = '/usr/bin/darcs'
 
diff --git a/layman/overlays/git.py b/layman/overlays/git.py
index bda2df0..3bcb3e3 100644
--- a/layman/overlays/git.py
+++ b/layman/overlays/git.py
@@ -37,6 +37,7 @@ class GitOverlay(Overlay):
     ''' Handles git overlays.'''
 
     type = 'Git'
+    type_key = 'git'
 
     binary_command  = '/usr/bin/git'
 
diff --git a/layman/overlays/mercurial.py b/layman/overlays/mercurial.py
index b010c8b..03af63b 100644
--- a/layman/overlays/mercurial.py
+++ b/layman/overlays/mercurial.py
@@ -38,6 +38,7 @@ class MercurialOverlay(Overlay):
     ''' Handles mercurial overlays.'''
 
     type = 'Mercurial'
+    type_key = 'mercurial'
 
     binary_command  = '/usr/bin/hg'
 
diff --git a/layman/overlays/overlay.py b/layman/overlays/overlay.py
index bee7529..f1e38a3 100644
--- a/layman/overlays/overlay.py
+++ b/layman/overlays/overlay.py
@@ -15,6 +15,7 @@
 # Author(s):
 #             Gunnar Wrobel <wrobel@gentoo.org>
 #             Sebastian Pipping <sebastian@pipping.org>
+#             Christian Groschupp <christian@groschupp.org>
 #
 ''' Basic overlay class.'''
 
@@ -27,8 +28,9 @@ __version__ = "$Id: overlay.py 273 2006-12-30 15:54:50Z wrobel $"
 #-------------------------------------------------------------------------------
 
 import sys, types, re, os, os.path, shutil, subprocess
+import elementtree.ElementTree as ET
 
-from   layman.utils             import node_to_dict, dict_to_node, path
+from   layman.utils             import path, ensure_unicode
 
 from   layman.debug             import OUT
 
@@ -46,10 +48,9 @@ class Overlay:
     def __init__(self, xml, ignore = 0, quiet = False):
         '''
         >>> here = os.path.dirname(os.path.realpath(__file__))
-        >>> document = open(here + '/../tests/testfiles/global-overlays.xml').read()
-        >>> import xml.dom.minidom
-        >>> document = xml.dom.minidom.parseString(document)
-        >>> overlays = document.getElementsByTagName('overlay')
+        >>> import elementtree.ElementTree as ET
+        >>> document = ET.parse(here + '/../tests/testfiles/global-overlays.xml')
+        >>> overlays = document.findall('overlay') + document.findall('repo')
         >>> a = Overlay(overlays[0])
         >>> a.name
         u'wrobel'
@@ -69,32 +70,36 @@ class Overlay:
         '''
         self.quiet = quiet
 
-        self.data = node_to_dict(xml)
-
-        if '<name>1' in self.data.keys():
-            self.name = self.data['<name>1']['@'].strip()
-        elif '&name' in self.data.keys():
-            self.name = self.data['&name']
+        _name = xml.find('name')
+        if _name != None:
+            self.name = ensure_unicode(_name.text)
+        elif 'name' in xml.attrib:
+            self.name = ensure_unicode(xml.attrib['name'])
         else:
             raise Exception('Overlay is missing a "name" entry!')
 
-        if '<source>1' in self.data.keys():
-            self.src = self.data['<source>1']['@'].strip()
-        elif '&src' in self.data.keys():
-            self.src = self.data['&src']
+        _source = xml.find('source')
+        if _source != None:
+            self.src = ensure_unicode(_source.text.strip())
+        elif 'src' in xml.attrib:
+            self.src = ensure_unicode(xml.attrib['src'])
+        else:
+            raise Exception('Overlay "' + self.name + '" is missing a "source" entry!')
+
+        _owner = xml.find('owner')
+        if _owner == None:
+            _email = None
         else:
-            raise Exception('Overlay "' + self.name + '" is missing a "source" '
-                            'entry!')
-
-        if '<owner>1' in self.data.keys() and \
-                '<email>1' in self.data['<owner>1']:
-            self.owner_email = self.data['<owner>1']['<email>1']['@'].strip()
-            if '<name>1' in self.data['<owner>1']:
-                self.owner_name = self.data['<owner>1']['<name>1']['@'].strip()
+            _email = _owner.find('email')
+        if _owner != None and _email != None:
+            self.owner_email = ensure_unicode(_email.text.strip())
+            _name = _owner.find('name')
+            if _name != None:
+                self.owner_name = ensure_unicode(_name.text.strip())
             else:
                 self.owner_name = None
-        elif '&contact' in self.data.keys():
-            self.owner_email = self.data['&contact']
+        elif 'contact' in xml.attrib:
+            self.owner_email = ensure_unicode(xml.attrib['contact'])
             self.owner_name = None
         else:
             self.owner_email = ''
@@ -106,37 +111,73 @@ class Overlay:
                 OUT.warn('Overlay "' + self.name + '" is missing a '
                          '"owner.email" entry!', 4)
 
-        if '<description>1' in self.data.keys():
-            self.description = self.data['<description>1']['@'].strip()
+
+        _desc = xml.find('description')
+        if _desc != None:
+            self.description = ensure_unicode(_desc.text.strip())
         else:
             self.description = ''
             if not ignore:
                 raise Exception('Overlay "' + self.name + '" is missing a '
-				'"description" entry!')
+	                                '"description" entry!')
             elif ignore == 1:
                 OUT.warn('Overlay "' + self.name + '" is missing a '
-			 '"description" entry!', 4)
+                         '"description" entry!', 4)
 
-        if '&status' in self.data.keys():
-            self.status = self.data['&status']
+        if 'status' in xml.attrib:
+            self.status = ensure_unicode(xml.attrib['status'])
         else:
-            self.status = ''
+            self.status = None
 
-        if '&priority' in self.data.keys():
-            self.priority = int(self.data['&priority'])
+        if 'priority' in xml.attrib:
+            self.priority = int(xml.attrib['priority'])
         else:
             self.priority = 50
 
+        h = xml.find('homepage')
+        l = xml.find('link')
+        if h != None:
+            self.homepage = ensure_unicode(h.text.strip())
+        elif l != None:
+            self.homepage = ensure_unicode(l.text.strip())
+        else:
+            self.homepage = None
+
     def set_priority(self, priority):
         '''Set the priority of this overlay.'''
 
-        self.data['&priority'] = str(priority)
         self.priority = int(priority)
 
-    def to_minidom(self, document):
+    def to_minidom(self):
         '''Convert to xml.'''
 
-        return dict_to_node(self.data, document, 'overlay')
+        repo = ET.Element('repo')
+        if self.status != None:
+            repo.attrib['status'] = self.status
+        repo.attrib['priority'] = str(self.priority)
+        name = ET.Element('name')
+        name.text = self.name
+        repo.append(name)
+        desc = ET.Element('description')
+        desc.text = self.description
+        repo.append(desc)
+        if self.homepage != None:
+            homepage = ET.Element('homepage')
+            homepage.text = self.homepage
+            repo.append(homepage)
+        owner = ET.Element('owner')
+        repo.append(owner)
+        owner_email = ET.Element('email')
+        owner_email.text = self.owner_email
+        owner.append(owner_email)
+        if self.owner_name != None:
+            owner_name = ET.Element('name')
+            owner_name.text = self.owner_name
+            owner.append(owner_name)
+        source = ET.Element('source', type=self.__class__.type_key)
+        source.text = self.src
+        repo.append(source)
+        return repo
 
     def add(self, base, quiet = False):
         '''Add the overlay.'''
@@ -186,10 +227,9 @@ class Overlay:
     def __str__(self):
         '''
         >>> here = os.path.dirname(os.path.realpath(__file__))
-        >>> document = open(here + '/../tests/testfiles/global-overlays.xml').read()
-        >>> import xml.dom.minidom
-        >>> document = xml.dom.minidom.parseString(document)
-        >>> overlays = document.getElementsByTagName('overlay')
+        >>> import elementtree.ElementTree as ET
+        >>> document = ET.parse(here + '/../tests/testfiles/global-overlays.xml')
+        >>> overlays = document.findall('overlay') + document.findall('repo')
         >>> a = Overlay(overlays[0])
         >>> print str(a)
         wrobel
@@ -222,24 +262,22 @@ class Overlay:
         result += u'\n  '.join((u'\n' + description).split(u'\n'))
         result += u'\n'
 
-        for key in (e for e in ('<homepage>1', '<link>1') if e in self.data.keys()):
-            link = self.data[key]['@'].strip()
+        if self.homepage != None:
+            link = self.homepage
             link = re.compile(u' +').sub(u' ', link)
             link = re.compile(u'\n ').sub(u'\n', link)
             result += u'\nLink:\n'
             result += u'\n  '.join((u'\n' + link).split(u'\n'))
             result += u'\n'
-            break
 
         return result
 
     def short_list(self, width = 0):
         '''
         >>> here = os.path.dirname(os.path.realpath(__file__))
-        >>> document = open(here + '/../tests/testfiles/global-overlays.xml').read()
-        >>> import xml.dom.minidom
-        >>> document = xml.dom.minidom.parseString(document)
-        >>> overlays = document.getElementsByTagName('overlay')
+        >>> import elementtree.ElementTree as ET
+        >>> document = ET.parse(here + '/../tests/testfiles/global-overlays.xml')
+        >>> overlays = document.findall('repo') + document.findall('overlay')
         >>> a = Overlay(overlays[0])
         >>> print a.short_list(80)
         wrobel                    [None      ] (https://o.g.o/svn/dev/wrobel         )
diff --git a/layman/overlays/rsync.py b/layman/overlays/rsync.py
index 6052bd0..1fc767f 100644
--- a/layman/overlays/rsync.py
+++ b/layman/overlays/rsync.py
@@ -37,6 +37,7 @@ class RsyncOverlay(Overlay):
     ''' Handles rsync overlays.'''
 
     type = 'Rsync'
+    type_key = 'rsync'
 
     binary = '/usr/bin/rsync'
 
diff --git a/layman/overlays/svn.py b/layman/overlays/svn.py
index 4ff3da4..8fef998 100644
--- a/layman/overlays/svn.py
+++ b/layman/overlays/svn.py
@@ -37,6 +37,7 @@ class SvnOverlay(Overlay):
     ''' Handles subversion overlays.'''
 
     type = 'Subversion'
+    type_key = 'svn'
 
     binary = '/usr/bin/svn'
 
diff --git a/layman/overlays/tar.py b/layman/overlays/tar.py
index 5786931..4990b2b 100644
--- a/layman/overlays/tar.py
+++ b/layman/overlays/tar.py
@@ -62,6 +62,7 @@ class TarOverlay(Overlay):
     '''
 
     type = 'Tar'
+    type_key = 'tar'
 
     binary = u'/bin/tar'
 
@@ -69,22 +70,22 @@ class TarOverlay(Overlay):
 
         Overlay.__init__(self, xml, ignore)
 
-        if '&format' in self.data.keys():
-            self.format = self.data['&format']
+        if 'format' in xml.attrib:
+            self.format = xml.attrib['format']
         else:
             self.format = ''
 
-        if '&subpath' in self.data.keys():
-            self.subpath = self.data['&subpath']
+        if 'subpath' in xml.attrib:
+            self.subpath = xml.attrib['subpath']
         else:
             self.subpath = ''
 
-        if '&category' in self.data.keys():
+        if 'category' in xml.attrib:
             if self.subpath:
                 raise Exception('Cannot use "category" and "subpath" at the same'
                                 ' time!')
 
-            self.category = self.data['&category']
+            self.category = xml.attrib['category']
         else:
             self.category = ''
 
diff --git a/layman/utils.py b/layman/utils.py
index 695b996..55df876 100644
--- a/layman/utils.py
+++ b/layman/utils.py
@@ -13,6 +13,8 @@
 #
 # Author(s):
 #             Gunnar Wrobel <wrobel@gentoo.org>
+#             Sebastian Pipping <sebastian@pipping.org>
+#             Christian Groschupp <christian@groschupp.org>
 #
 
 '''Utility functions to deal with xml nodes. '''
@@ -33,130 +35,11 @@ import types, re
 #
 #-------------------------------------------------------------------------------
 
-def node_to_text(node):
-    '''
-    Reduces an xml node to its text elements. The function does not
-    collect the text nodes recursively.
-
-    >>> import xml.dom.minidom
-    >>> imp = xml.dom.minidom.getDOMImplementation()
-    >>> doc = imp.createDocument('test', 'root', None)
-    >>> root =  doc.childNodes[0]
-    >>> node = doc.createTextNode('text')
-    >>> a = root.appendChild(node)
-    >>> node = doc.createElement('text')
-    >>> node2 = doc.createTextNode('text')
-    >>> a = node.appendChild(node2)
-    >>> a = root.appendChild(node)
-    >>> node = doc.createTextNode('text')
-    >>> a = root.appendChild(node)
-    >>> doc.toprettyxml('', '') #doctest: +ELLIPSIS
-    '<?xml version="1.0" ?>...<root>text<text>text</text>text</root>'
-
-    >>> node_to_text(root)
-    'texttext'
-
-    '''
-    text = ''
-
-    for child in node.childNodes:
-        if child.nodeType == child.TEXT_NODE:
-            text = text + child.data
-
-    return text
-
-def node_to_dict(node):
-    ''' Converts a xml node to a dictionary. The function collects the
-    nodes recursively. Attributes will be prepended with '&', child
-    nodes will be surrounded with tags. An index will be appended
-    since several child nodes with the same tag may exist. Text
-    elements will be collapsed and stored in a n entry prepended with
-    '@'. Comments will be ignored.
-
-    >>> import xml.dom.minidom
-    >>> imp = xml.dom.minidom.getDOMImplementation()
-    >>> doc = imp.createDocument('test', 'root', None)
-    >>> root =  doc.childNodes[0]
-    >>> node = doc.createTextNode('text')
-    >>> a = root.appendChild(node)
-    >>> node = doc.createElement('text')
-    >>> node2 = doc.createTextNode('text')
-    >>> comm = doc.createComment('comment')
-    >>> attr = doc.createAttribute('&attr')
-    >>> a = node.appendChild(node2)
-    >>> a = root.appendChild(comm)
-    >>> node.setAttributeNode(attr)
-    >>> node.setAttribute('&attr','test')
-    >>> a = root.appendChild(node)
-    >>> node3 = doc.createElement('text')
-    >>> a = root.appendChild(node3)
-    >>> node = doc.createTextNode('text')
-    >>> a = root.appendChild(node)
-    >>> doc.toprettyxml('', '') #doctest: +ELLIPSIS
-    '<?xml version="1.0" ?>...<root>text<!--comment--><text &attr="test">text</text><text/>text</root>'
-
-    >>> node_to_dict(root)
-    {'<text>1': {'@': 'text', '&&attr': 'test'}, '<text>2': {'@': ''}, '@': 'texttext'}
-
-    '''
-    result = {}
-
-    # Map the attributes
-    for index in range(0, node.attributes.length):
-        attr = node.attributes.item(index)
-        result['&' +  attr.name] = attr.nodeValue
-
-    text = ''
-
-    # Map the nodes
-    for child in node.childNodes:
-        if child.nodeType == child.TEXT_NODE:
-            text = text + child.data
-        elif child.nodeType == child.CDATA_SECTION_NODE:
-            text = text + child.data
-        elif child.nodeType == child.ELEMENT_NODE:
-            index = 1
-            while ('<' + child.tagName + '>' + str(index)) in result.keys():
-                index += 1
-            result['<' + child.tagName + '>' + str(index)] = node_to_dict(child)
-
-    result['@'] = text
-
-    return result
-
-def dict_to_node(data, document, root_name):
-    ''' Reverts the node_to_dict operation.
-
-    >>> import xml.dom.minidom
-    >>> imp = xml.dom.minidom.getDOMImplementation()
-    >>> doc = imp.createDocument('test', 'root', None)
-    >>> a = {'<text>1': {'@': 'text', '&&attr': 'test'}, '<text>2': {'@': ''}, '@': 'texttext'}
-    >>> doc.childNodes[0] = dict_to_node(a, doc, 'root')
-    >>> doc.toprettyxml('', '') #doctest: +ELLIPSIS
-    '<?xml version="1.0" ?>...<root><text &attr="test">text</text><text></text>texttext</root>'
-
-    '''
-    node = document.createElement(root_name)
-
-    for i, j in data.items():
-
-        if i[0] == '&':
-            attr = document.createAttribute(i[1:])
-            node.setAttributeNode(attr)
-            node.setAttribute(i[1:], j)
-        if i[0] == '<':
-            k = i[1:]
-            while k[-1] in '0123456789':
-                k = k[:-1]
-            child = dict_to_node(data[i],
-                                 document,
-                                 k[:-1])
-            node.appendChild(child)
-        if i[0] == '@':
-            child = document.createTextNode(j)
-            node.appendChild(child)
-
-    return node
+def ensure_unicode(obj, encoding='utf-8'):
+    if isinstance(obj, basestring):
+        if not isinstance(obj, unicode):
+            obj = unicode(obj, encoding)
+    return obj
 
 def path(path_elements):
     '''