As I was looking at #138, I noticed in the codebase that the issue around decoding was already addressed for HTMLParser class (Modest impl), but not for LexborHMLParser (Lexbor impl).
Also it can be error prone, having to check two different places. So to make it easier to maintain, preferably there would be a single source of truth for the public API, while the implementation-specific C code would be abstracted away.
It could look maybe something like this:
class ModestParserImpl:
def get_body(self, myhtml_tree_t* html_tree):
cdef myhtml_tree_node_t* body
body = myhtml_tree_get_node_body(html_tree)
return body
class LexborParserImpl:
...
class HTMLParserABC:
self.impl = None
@property
def body(self):
cdef myhtml_tree_node_t* body
body = self.impl.get_body(self.html_tree)
if body != NULL:
node = Node()
node._init(body, self)
return node
return None
class HTMLParser(HTMLParserABC):
impl = ModestParserImpl()
class HTMLParser(HTMLParserABC):
self.impl = ParserImpl(self)
As I was looking at #138, I noticed in the codebase that the issue around decoding was already addressed for
HTMLParserclass (Modest impl), but not for LexborHMLParser (Lexbor impl).Also it can be error prone, having to check two different places. So to make it easier to maintain, preferably there would be a single source of truth for the public API, while the implementation-specific C code would be abstracted away.
It could look maybe something like this: