| 59 |
class SaxEventLister(xml.sax.handler.ContentHandler): |
class SaxEventLister(xml.sax.handler.ContentHandler): |
| 60 |
|
|
| 61 |
"""Create a normalized list representation containing the SAX events |
"""Create a normalized list representation containing the SAX events |
| 62 |
|
|
| 63 |
|
The normalization includes the following |
| 64 |
|
|
| 65 |
|
- The attribute dictionary of a staertElement event is converted to |
| 66 |
|
a sorted list of (key, value) pairs |
| 67 |
|
|
| 68 |
|
- ID and IDREF attribute values are normalized in such a way that |
| 69 |
|
two documents that only use different values for IDs can still |
| 70 |
|
lead to the same normalized representation. |
| 71 |
|
|
| 72 |
|
The implementation of this feature assumes that all IDs are |
| 73 |
|
defined before they are used. The normalized ID values are of the |
| 74 |
|
form 'D<NUM>' where <NUM> is a counter starting with 0, so the |
| 75 |
|
first ID value will become 'D0', the second 'D1', etc. |
| 76 |
|
|
| 77 |
|
Which attributes are IDs or IDREFS is defined with the |
| 78 |
|
correspoding constructor arguments. |
| 79 |
""" |
""" |
| 80 |
|
|
| 81 |
def __init__(self): |
def __init__(self, ids = (), idrefs = ()): |
| 82 |
|
"""Initialize the SaxEventLister |
| 83 |
|
|
| 84 |
|
The ids and idrefs parameters should be lists of (element, attr) |
| 85 |
|
pairs where name is the name of an attribute as passed to the |
| 86 |
|
startElementNS method and attr is the name of an attribute as |
| 87 |
|
used in the mapping passed to startElementNS, so both name and |
| 88 |
|
attr usually must include the namespace. |
| 89 |
|
""" |
| 90 |
self.eventlist = [] |
self.eventlist = [] |
| 91 |
|
self.ids = ids |
| 92 |
|
self.idrefs = idrefs |
| 93 |
|
self.idremap = {} |
| 94 |
|
|
| 95 |
def startElementNS(self, name, qname, attrs): |
def startElementNS(self, name, qname, attrs): |
| 96 |
items = attrs.items() |
items = attrs.items() |
| 97 |
items.sort() |
items.sort() |
| 98 |
|
for i, (attr, value) in zip(range(len(items)), items): |
| 99 |
|
#print '++++' |
| 100 |
|
#print self.idremap |
| 101 |
|
#print name, attr, value |
| 102 |
|
if (name, attr) in self.ids: |
| 103 |
|
newid = len(self.idremap) |
| 104 |
|
self.idremap[value] = "D" + str(newid) |
| 105 |
|
value = self.idremap[value] |
| 106 |
|
elif (name, attr) in self.idrefs: |
| 107 |
|
value = self.idremap[value] |
| 108 |
|
items[i] = (attr, value) |
| 109 |
|
#print name, attr, value |
| 110 |
self.eventlist.append(("start", name, qname, items)) |
self.eventlist.append(("start", name, qname, items)) |
| 111 |
|
|
| 112 |
def endElementNS(self, name, qname): |
def endElementNS(self, name, qname): |
| 113 |
self.eventlist.append(("end", name, qname)) |
self.eventlist.append(("end", name, qname)) |
| 114 |
|
|
| 115 |
|
|
| 116 |
def sax_eventlist(data = None, filename = None): |
def sax_eventlist(data = None, filename = None, |
| 117 |
|
ids = (), idrefs = ()): |
| 118 |
"""Return a list of SAX event generated for a given XML source |
"""Return a list of SAX event generated for a given XML source |
| 119 |
|
|
| 120 |
The xml source may either be a string with the actual XML, in which |
The xml source may either be a string with the actual XML, in which |
| 123 |
""" |
""" |
| 124 |
if filename is not None: |
if filename is not None: |
| 125 |
data = open(filename).read() |
data = open(filename).read() |
| 126 |
handler = SaxEventLister() |
handler = SaxEventLister(ids = ids, idrefs = idrefs) |
| 127 |
parser = make_parser() |
parser = make_parser() |
| 128 |
parser.setContentHandler(handler) |
parser.setContentHandler(handler) |
| 129 |
parser.setErrorHandler(ErrorHandler()) |
parser.setErrorHandler(ErrorHandler()) |
Parent Directory
| 
Revision Log
| 
Patch