merge.py

   1 from xml.dom import minidom, XMLNS_NAMESPACE, Node
   2 from zeroinstall.injector.namespaces import XMLNS_IFACE
   3 import xmltools
   4
   5 def childNodes(parent, namespaceURI = None, localName = None):
   6         for x in parent.childNodes:
   7                 if x.nodeType != Node.ELEMENT_NODE: continue
   8                 if namespaceURI is not None and x.namespaceURI != namespaceURI: continue
   9
  10                 if localName is None or x.localName == localName:
  11                         yield x
  12
  13 class Context:
  14         def __init__(self, impl):
  15                 doc = impl.ownerDocument
  16                 self.attribs = {}
  17                 self.requires = []
  18                 self.commands = {}
  19
  20                 node = impl
  21                 while True:
  22                         for name, value in node.attributes.itemsNS():
  23                                 if name[0] == XMLNS_NAMESPACE:
  24                                         xmltools.register_namespace(value, name[1])
  25                                 elif name not in self.attribs:
  26                                         self.attribs[name] = value
  27                         if node.nodeName == 'group':
  28                                 # We don't care about <requires> or <command> inside <implementation>;
  29                                 # they'll get copied over anyway
  30                                 for x in childNodes(node, XMLNS_IFACE, 'requires'):
  31                                         self.requires.append(x)
  32                                 for x in childNodes(node, XMLNS_IFACE, 'restricts'):
  33                                         self.requires.append(x)
  34                                 for x in childNodes(node, XMLNS_IFACE, 'command'):
  35                                         command_name = x.getAttribute('name')
  36                                         if command_name not in self.commands:
  37                                                 self.commands[command_name] = x
  38                                         # (else the existing definition on the child should be used)
  39                         node = node.parentNode
  40                         if node.nodeName != 'group':
  41                                 break
  42
  43 def find_impls(parent):
  44         """Return all <implementation> children, including those inside groups."""
  45         for x in childNodes(parent, XMLNS_IFACE):
  46                 if x.localName == 'implementation':
  47                         yield x
  48                 elif x.localName == 'group':
  49                         for y in find_impls(x):
  50                                 yield y
  51
  52 def find_groups(parent):
  53         """Return all <group> children, including those inside other groups."""
  54         for x in childNodes(parent, XMLNS_IFACE, 'group'):
  55                 yield x
  56                 for y in find_groups(x):
  57                         yield y
  58
  59 def nodesEqual(a, b):
  60         assert a.nodeType == Node.ELEMENT_NODE
  61         assert b.nodeType == Node.ELEMENT_NODE
  62
  63         if a.namespaceURI != b.namespaceURI:
  64                 return False
  65
  66         if a.nodeName != b.nodeName:
  67                 return False
  68
  69         a_attrs = set(["%s %s" % (name, value) for name, value in a.attributes.itemsNS()])
  70         b_attrs = set(["%s %s" % (name, value) for name, value in b.attributes.itemsNS()])
  71
  72         if a_attrs != b_attrs:
  73                 #print "%s != %s" % (a_attrs, b_attrs)
  74                 return False
  75
  76         a_children = list(childNodes(a))
  77         b_children = list(childNodes(b))
  78
  79         if len(a_children) != len(b_children):
  80                 return False
  81
  82         for a_child, b_child in zip(a_children, b_children):
  83                 if not nodesEqual(a_child, b_child):
  84                         return False
  85
  86         return True
  87
  88 def score_subset(group, impl):
  89         """Returns (is_subset, goodness)"""
  90         for key in group.attribs:
  91                 if key not in impl.attribs.keys():
  92                         #print "BAD", key
  93                         return (0,)             # Group sets an attribute the impl doesn't want
  94         matching_commands = 0
  95         for name, g_command in group.commands.iteritems():
  96                 if name not in impl.commands:
  97                         return (0,)             # Group sets a command the impl doesn't want
  98                 if nodesEqual(g_command, impl.commands[name]):
  99                         # Prefer matching commands to overriding them
 100                         matching_commands += 1
 101         for g_req in group.requires:
 102                 for i_req in impl.requires:
 103                         if nodesEqual(g_req, i_req): break
 104                 else:
 105                         return (0,)             # Group adds a requires that the impl doesn't want
 106         # Score result so we get groups that have all the same requires/commands first, then ones with all the same attribs
 107         return (1, len(group.requires) + len(group.commands), len(group.attribs) + matching_commands)
 108
 109 # Note: the namespace stuff isn't quite right yet.
 110 # Might get conflicts if both documents use the same prefix for different things.
 111 def merge(data, local):
 112         local_doc = minidom.parse(local)
 113         master_doc = minidom.parseString(data)
 114
 115         # Merge each implementation in the local feed in turn (normally there will only be one)
 116         for impl in find_impls(local_doc.documentElement):
 117                 # 1. Get the context of the implementation to add. This is:
 118                 #    - The set of its requirements
 119                 #    - The set of its commands
 120                 #    - Its attributes
 121                 new_impl_context = Context(impl)
 122
 123                 # 2. For each <group> in the master feed, see if it provides a compatible context:
 124                 #    - A subset of the new implementation's requirements
 125                 #    - A subset of the new implementation's command names
 126                 #    - A subset of the new implementation's attributes (names, not values)
 127                 #    Choose the most compatible <group> (the root counts as a minimally compatible group)
 128
 129                 best_group = ((1, 0, 0), master_doc.documentElement)    # (score, element)
 130
 131                 for group in find_groups(master_doc.documentElement):
 132                         group_context = Context(group)
 133                         score = score_subset(group_context, new_impl_context)
 134                         if score > best_group[0]:
 135                                 best_group = (score, group)
 136
 137                 group = best_group[1]
 138                 group_context = Context(group)
 139
 140                 new_commands = []
 141                 for name, new_command in new_impl_context.commands.iteritems():
 142                         old_command = group_context.commands.get(name, None)
 143                         if not (old_command and nodesEqual(old_command, new_command)):
 144                                 new_commands.append(master_doc.importNode(new_command, True))
 145
 146                 # If we have additional requirements, we'll need to create a subgroup and add them
 147                 if len(new_impl_context.requires) > len(group_context.requires) or new_commands:
 148                         subgroup = xmltools.create_element(group, 'group')
 149                         group = subgroup
 150                         #group_context = Context(group)
 151                         for x in new_impl_context.requires:
 152                                 for y in group_context.requires:
 153                                         if nodesEqual(x, y): break
 154                                 else:
 155                                         req = master_doc.importNode(x, True)
 156                                         #print "Add", req
 157                                         xmltools.insert_element(req, group)
 158                         for c in new_commands:
 159                                 xmltools.insert_element(c, group)
 160
 161                 new_impl = master_doc.importNode(impl, True)
 162
 163                 # Attributes might have been set on a parent group; move to the impl
 164                 for name in new_impl_context.attribs:
 165                         #print "Set", name, value
 166                         xmltools.add_attribute_ns(new_impl, name[0], name[1], new_impl_context.attribs[name])
 167
 168                 for name, value in new_impl.attributes.itemsNS():
 169                         if name[0] == XMLNS_NAMESPACE or \
 170                            (name in group_context.attribs and group_context.attribs[name] == value):
 171                                 #print "Deleting duplicate attribute", name, value
 172                                 new_impl.removeAttributeNS(name[0], name[1])
 173
 174                 xmltools.insert_element(new_impl, group)
 175
 176         return master_doc.toxml('utf-8')