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