| 修订版 | 3a17d1c05da239c63cec678f8e139ecfa60d8a25 (tree) |
|---|---|
| 时间 | 2011-07-21 17:29:33 |
| 作者 | Hiromichi MATSUSHIMA <hirom@offi...> |
| Commiter | Hiromichi MATSUSHIMA |
add SGML declare handler
htmltree.py (diff)| @@ -60,13 +60,15 @@ class HTMLRenderer(Renderer): | ||
| 60 | 60 | elif elem.is_root(): |
| 61 | 61 | for child in elem: |
| 62 | 62 | self._recursive(child, texts) |
| 63 | - | |
| 63 | + elif elem.is_decl(): | |
| 64 | + texts.append("<!" + elem.name + ">") | |
| 64 | 65 | |
| 65 | 66 | class HTMLElement(list): |
| 66 | 67 | """HTML element object to use as tree nodes.""" |
| 67 | 68 | ROOT = 0 |
| 68 | 69 | TAG = 100 |
| 69 | 70 | TEXT = 200 |
| 71 | + DECL = 300 | |
| 70 | 72 | |
| 71 | 73 | def __init__(self, type, name="", attrs={}): |
| 72 | 74 | """ |
| @@ -94,6 +96,8 @@ class HTMLElement(list): | ||
| 94 | 96 | def __repr__(self): |
| 95 | 97 | if self.type == HTMLElement.TAG: |
| 96 | 98 | return "<TAG:%s %s>" % (self.name, self._attrs2str()) |
| 99 | + elif self.type == HTMLElement.DECL: | |
| 100 | + return "<DECL:'%s'>" % self.name | |
| 97 | 101 | elif self.type == HTMLElement.TEXT: |
| 98 | 102 | return "<TEXT:'%s'>" % self._text |
| 99 | 103 | else: |
| @@ -244,6 +248,9 @@ class HTMLElement(list): | ||
| 244 | 248 | def is_root(self): |
| 245 | 249 | return self.type == HTMLElement.ROOT |
| 246 | 250 | |
| 251 | + def is_decl(self): | |
| 252 | + return self.type == HTMLElement.DECL | |
| 253 | + | |
| 247 | 254 | def is_descendant(self, tagname): |
| 248 | 255 | p = self.parent() |
| 249 | 256 | while p != None: |
| @@ -408,6 +415,11 @@ class HTMLTree(HTMLParser.HTMLParser): | ||
| 408 | 415 | elem._text = data |
| 409 | 416 | self._cursor.append(elem) |
| 410 | 417 | |
| 418 | + def handle_decl(self, decl): | |
| 419 | + elem = HTMLElement(HTMLElement.DECL, decl) | |
| 420 | + elem._parent = self._cursor | |
| 421 | + self._cursor.append(elem) | |
| 422 | + | |
| 411 | 423 | # Accessor |
| 412 | 424 | def root(self): |
| 413 | 425 | return self._htmlroot |
Fork