25 |
return [default, spec] |
return [default, spec] |
26 |
|
|
27 |
class _parse_in_progress(HTMLParser): |
class _parse_in_progress(HTMLParser): |
28 |
def __init__(self): |
def __init__(self, module): |
29 |
self.result = Document() |
self._module = module |
30 |
|
try: |
31 |
|
self.result = module.Document() |
32 |
|
except AttributeError: |
33 |
|
self.result = Document() |
34 |
self.stack = [self.result] |
self.stack = [self.result] |
35 |
self.namespaces = [] |
self.namespaces = [] |
36 |
self.lame_namespaces = {} |
self.lame_namespaces = {} |
66 |
else: |
else: |
67 |
nsdecls = {} |
nsdecls = {} |
68 |
ns = self.ns(ns) |
ns = self.ns(ns) |
69 |
element = core.Element(ns, name, attributes, nsdecls, self.__raw()) |
element = self._module.Element(ns, name, attributes, nsdecls, self.__raw()) |
70 |
element.rawname = rawname |
element.rawname = rawname |
71 |
self.stack.append(element) |
self.stack.append(element) |
72 |
self.ns_pushed = False |
self.ns_pushed = False |
87 |
self.stack.pop() |
self.stack.pop() |
88 |
HTMLParser.close(self) |
HTMLParser.close(self) |
89 |
for element in self.result: |
for element in self.result: |
90 |
if isinstance(element, core.Element): |
if isinstance(element, self._module.Element): |
91 |
element.lame_namespaces = self.lame_namespaces |
element.lame_namespaces = self.lame_namespaces |
92 |
|
|
93 |
def handle_decl(self, data): |
def handle_decl(self, data): |
94 |
self.stack[-1].append(core.Declaration(data)) |
self.stack[-1].append(self._module.Declaration(data)) |
95 |
|
|
96 |
def handle_data(self, data): |
def handle_data(self, data): |
97 |
self.stack[-1].append(data) |
self.stack[-1].append(data) |
98 |
|
|
99 |
def handle_charref(self, data): |
def handle_charref(self, data): |
100 |
self.stack[-1].append(core.CharRef(int(data))) |
self.stack[-1].append(self._module.CharRef(int(data))) |
101 |
|
|
102 |
def handle_entityref(self, data): |
def handle_entityref(self, data): |
103 |
self.stack[-1].append(core.EntityRef(data)) |
self.stack[-1].append(self._module.EntityRef(data)) |
104 |
|
|
105 |
def add_namespace(self, prefix, url): |
def add_namespace(self, prefix, url): |
106 |
if self.ns_pushed: |
if self.ns_pushed: |
117 |
self.ns_popped = True |
self.ns_popped = True |
118 |
|
|
119 |
def handle_comment(self, data): |
def handle_comment(self, data): |
120 |
self.stack[-1].append(core.Comment(data)) |
self.stack[-1].append(self._module.Comment(data)) |
121 |
|
|
122 |
def parse(data, builtin_namespaces=None, encoding=None): |
def parse(data, builtin_namespaces=None, encoding=None, module=core): |
123 |
if encoding is None: |
if encoding is None: |
124 |
# FIXME: can we get the encoding from the HTML? |
# FIXME: can we get the encoding from the HTML? |
125 |
encoding = 'latin1' |
encoding = 'latin1' |
126 |
if type(data) is type(''): |
if type(data) is type(''): |
127 |
data = unicode(data, encoding) |
data = unicode(data, encoding) |
128 |
parser = _parse_in_progress() |
parser = _parse_in_progress(module) |
129 |
if builtin_namespaces is not None: |
if builtin_namespaces is not None: |
130 |
parser.namespaces.append(builtin_namespaces) |
parser.namespaces.append(builtin_namespaces) |
131 |
parser.feed(data) |
parser.feed(data) |