Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 1 | """HTML 2.0 parser. |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 2 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 3 | See the HTML 2.0 specification: |
| 4 | http://www.w3.org/hypertext/WWW/MarkUp/html-spec/html-spec_toc.html |
| 5 | """ |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 6 | |
Georg Brandl | ac19d85 | 2008-06-01 21:19:14 +0000 | [diff] [blame] | 7 | from warnings import warnpy3k |
| 8 | warnpy3k("the htmllib module has been removed in Python 3.0", |
| 9 | stacklevel=2) |
| 10 | del warnpy3k |
| 11 | |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 12 | import sgmllib |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 13 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 14 | from formatter import AS_IS |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 15 | |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 16 | __all__ = ["HTMLParser", "HTMLParseError"] |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 17 | |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 18 | |
| 19 | class HTMLParseError(sgmllib.SGMLParseError): |
| 20 | """Error raised when an HTML document can't be parsed.""" |
| 21 | |
| 22 | |
| 23 | class HTMLParser(sgmllib.SGMLParser): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 24 | """This is the basic HTML parser class. |
| 25 | |
Andrew M. Kuchling | 0f10343 | 2003-10-27 15:47:48 +0000 | [diff] [blame] | 26 | It supports all entity names required by the XHTML 1.0 Recommendation. |
| 27 | It also defines handlers for all HTML 2.0 and many HTML 3.0 and 3.2 |
| 28 | elements. |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 29 | |
| 30 | """ |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 31 | |
Fred Drake | d995e11 | 2008-05-20 06:08:38 +0000 | [diff] [blame] | 32 | from htmlentitydefs import entitydefs |
Guido van Rossum | 65126d5 | 1995-09-27 16:22:17 +0000 | [diff] [blame] | 33 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 34 | def __init__(self, formatter, verbose=0): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 35 | """Creates an instance of the HTMLParser class. |
| 36 | |
| 37 | The formatter parameter is the formatter instance associated with |
| 38 | the parser. |
| 39 | |
| 40 | """ |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 41 | sgmllib.SGMLParser.__init__(self, verbose) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 42 | self.formatter = formatter |
Martin v. Löwis | cbe81f2 | 2003-09-12 16:38:00 +0000 | [diff] [blame] | 43 | |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 44 | def error(self, message): |
| 45 | raise HTMLParseError(message) |
| 46 | |
Martin v. Löwis | cbe81f2 | 2003-09-12 16:38:00 +0000 | [diff] [blame] | 47 | def reset(self): |
Fred Drake | 583359e | 2004-09-09 02:24:13 +0000 | [diff] [blame] | 48 | sgmllib.SGMLParser.reset(self) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 49 | self.savedata = None |
| 50 | self.isindex = 0 |
| 51 | self.title = None |
| 52 | self.base = None |
| 53 | self.anchor = None |
| 54 | self.anchorlist = [] |
| 55 | self.nofill = 0 |
| 56 | self.list_stack = [] |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 57 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 58 | # ------ Methods used internally; some may be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 59 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 60 | # --- Formatter interface, taking care of 'savedata' mode; |
| 61 | # shouldn't need to be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 62 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 63 | def handle_data(self, data): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 64 | if self.savedata is not None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 65 | self.savedata = self.savedata + data |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 66 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 67 | if self.nofill: |
| 68 | self.formatter.add_literal_data(data) |
| 69 | else: |
| 70 | self.formatter.add_flowing_data(data) |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 71 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 72 | # --- Hooks to save data; shouldn't need to be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 73 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 74 | def save_bgn(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 75 | """Begins saving character data in a buffer instead of sending it |
| 76 | to the formatter object. |
| 77 | |
| 78 | Retrieve the stored data via the save_end() method. Use of the |
| 79 | save_bgn() / save_end() pair may not be nested. |
| 80 | |
| 81 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 82 | self.savedata = '' |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 83 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 84 | def save_end(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 85 | """Ends buffering character data and returns all data saved since |
| 86 | the preceding call to the save_bgn() method. |
| 87 | |
| 88 | If the nofill flag is false, whitespace is collapsed to single |
| 89 | spaces. A call to this method without a preceding call to the |
| 90 | save_bgn() method will raise a TypeError exception. |
| 91 | |
| 92 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 93 | data = self.savedata |
| 94 | self.savedata = None |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 95 | if not self.nofill: |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 96 | data = ' '.join(data.split()) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 97 | return data |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 98 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 99 | # --- Hooks for anchors; should probably be overridden |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 100 | |
| 101 | def anchor_bgn(self, href, name, type): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 102 | """This method is called at the start of an anchor region. |
| 103 | |
| 104 | The arguments correspond to the attributes of the <A> tag with |
| 105 | the same names. The default implementation maintains a list of |
| 106 | hyperlinks (defined by the HREF attribute for <A> tags) within |
| 107 | the document. The list of hyperlinks is available as the data |
| 108 | attribute anchorlist. |
| 109 | |
| 110 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 111 | self.anchor = href |
| 112 | if self.anchor: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 113 | self.anchorlist.append(href) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 114 | |
| 115 | def anchor_end(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 116 | """This method is called at the end of an anchor region. |
| 117 | |
| 118 | The default implementation adds a textual footnote marker using an |
| 119 | index into the list of hyperlinks created by the anchor_bgn()method. |
| 120 | |
| 121 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 122 | if self.anchor: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 123 | self.handle_data("[%d]" % len(self.anchorlist)) |
| 124 | self.anchor = None |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 125 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 126 | # --- Hook for images; should probably be overridden |
| 127 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 128 | def handle_image(self, src, alt, *args): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 129 | """This method is called to handle images. |
| 130 | |
| 131 | The default implementation simply passes the alt value to the |
| 132 | handle_data() method. |
| 133 | |
| 134 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 135 | self.handle_data(alt) |
| 136 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 137 | # --------- Top level elememts |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 138 | |
| 139 | def start_html(self, attrs): pass |
| 140 | def end_html(self): pass |
| 141 | |
| 142 | def start_head(self, attrs): pass |
| 143 | def end_head(self): pass |
| 144 | |
| 145 | def start_body(self, attrs): pass |
| 146 | def end_body(self): pass |
| 147 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 148 | # ------ Head elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 149 | |
| 150 | def start_title(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 151 | self.save_bgn() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 152 | |
| 153 | def end_title(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 154 | self.title = self.save_end() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 155 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 156 | def do_base(self, attrs): |
| 157 | for a, v in attrs: |
| 158 | if a == 'href': |
| 159 | self.base = v |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 160 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 161 | def do_isindex(self, attrs): |
| 162 | self.isindex = 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 163 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 164 | def do_link(self, attrs): |
| 165 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 166 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 167 | def do_meta(self, attrs): |
| 168 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 169 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 170 | def do_nextid(self, attrs): # Deprecated |
| 171 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 172 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 173 | # ------ Body elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 174 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 175 | # --- Headings |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 176 | |
| 177 | def start_h1(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 178 | self.formatter.end_paragraph(1) |
| 179 | self.formatter.push_font(('h1', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 180 | |
| 181 | def end_h1(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 182 | self.formatter.end_paragraph(1) |
| 183 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 184 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 185 | def start_h2(self, attrs): |
| 186 | self.formatter.end_paragraph(1) |
| 187 | self.formatter.push_font(('h2', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 188 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 189 | def end_h2(self): |
| 190 | self.formatter.end_paragraph(1) |
| 191 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 192 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 193 | def start_h3(self, attrs): |
| 194 | self.formatter.end_paragraph(1) |
| 195 | self.formatter.push_font(('h3', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 196 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 197 | def end_h3(self): |
| 198 | self.formatter.end_paragraph(1) |
| 199 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 200 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 201 | def start_h4(self, attrs): |
| 202 | self.formatter.end_paragraph(1) |
| 203 | self.formatter.push_font(('h4', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 204 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 205 | def end_h4(self): |
| 206 | self.formatter.end_paragraph(1) |
| 207 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 208 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 209 | def start_h5(self, attrs): |
| 210 | self.formatter.end_paragraph(1) |
| 211 | self.formatter.push_font(('h5', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 212 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 213 | def end_h5(self): |
| 214 | self.formatter.end_paragraph(1) |
| 215 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 216 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 217 | def start_h6(self, attrs): |
| 218 | self.formatter.end_paragraph(1) |
| 219 | self.formatter.push_font(('h6', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 220 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 221 | def end_h6(self): |
| 222 | self.formatter.end_paragraph(1) |
| 223 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 224 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 225 | # --- Block Structuring Elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 226 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 227 | def do_p(self, attrs): |
| 228 | self.formatter.end_paragraph(1) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 229 | |
| 230 | def start_pre(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 231 | self.formatter.end_paragraph(1) |
| 232 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
| 233 | self.nofill = self.nofill + 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 234 | |
| 235 | def end_pre(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 236 | self.formatter.end_paragraph(1) |
| 237 | self.formatter.pop_font() |
| 238 | self.nofill = max(0, self.nofill - 1) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 239 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 240 | def start_xmp(self, attrs): |
| 241 | self.start_pre(attrs) |
| 242 | self.setliteral('xmp') # Tell SGML parser |
| 243 | |
| 244 | def end_xmp(self): |
| 245 | self.end_pre() |
| 246 | |
| 247 | def start_listing(self, attrs): |
| 248 | self.start_pre(attrs) |
| 249 | self.setliteral('listing') # Tell SGML parser |
| 250 | |
| 251 | def end_listing(self): |
| 252 | self.end_pre() |
| 253 | |
| 254 | def start_address(self, attrs): |
| 255 | self.formatter.end_paragraph(0) |
| 256 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
| 257 | |
| 258 | def end_address(self): |
| 259 | self.formatter.end_paragraph(0) |
| 260 | self.formatter.pop_font() |
| 261 | |
| 262 | def start_blockquote(self, attrs): |
| 263 | self.formatter.end_paragraph(1) |
| 264 | self.formatter.push_margin('blockquote') |
| 265 | |
| 266 | def end_blockquote(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 267 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 268 | self.formatter.pop_margin() |
| 269 | |
| 270 | # --- List Elements |
| 271 | |
| 272 | def start_ul(self, attrs): |
| 273 | self.formatter.end_paragraph(not self.list_stack) |
| 274 | self.formatter.push_margin('ul') |
| 275 | self.list_stack.append(['ul', '*', 0]) |
| 276 | |
| 277 | def end_ul(self): |
| 278 | if self.list_stack: del self.list_stack[-1] |
| 279 | self.formatter.end_paragraph(not self.list_stack) |
| 280 | self.formatter.pop_margin() |
| 281 | |
| 282 | def do_li(self, attrs): |
| 283 | self.formatter.end_paragraph(0) |
| 284 | if self.list_stack: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 285 | [dummy, label, counter] = top = self.list_stack[-1] |
| 286 | top[2] = counter = counter+1 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 287 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 288 | label, counter = '*', 0 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 289 | self.formatter.add_label_data(label, counter) |
| 290 | |
| 291 | def start_ol(self, attrs): |
| 292 | self.formatter.end_paragraph(not self.list_stack) |
| 293 | self.formatter.push_margin('ol') |
| 294 | label = '1.' |
| 295 | for a, v in attrs: |
| 296 | if a == 'type': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 297 | if len(v) == 1: v = v + '.' |
| 298 | label = v |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 299 | self.list_stack.append(['ol', label, 0]) |
| 300 | |
| 301 | def end_ol(self): |
| 302 | if self.list_stack: del self.list_stack[-1] |
| 303 | self.formatter.end_paragraph(not self.list_stack) |
| 304 | self.formatter.pop_margin() |
| 305 | |
| 306 | def start_menu(self, attrs): |
| 307 | self.start_ul(attrs) |
| 308 | |
| 309 | def end_menu(self): |
| 310 | self.end_ul() |
| 311 | |
| 312 | def start_dir(self, attrs): |
| 313 | self.start_ul(attrs) |
| 314 | |
| 315 | def end_dir(self): |
| 316 | self.end_ul() |
| 317 | |
| 318 | def start_dl(self, attrs): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 319 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 320 | self.list_stack.append(['dl', '', 0]) |
| 321 | |
| 322 | def end_dl(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 323 | self.ddpop(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 324 | if self.list_stack: del self.list_stack[-1] |
| 325 | |
| 326 | def do_dt(self, attrs): |
| 327 | self.ddpop() |
| 328 | |
| 329 | def do_dd(self, attrs): |
| 330 | self.ddpop() |
| 331 | self.formatter.push_margin('dd') |
| 332 | self.list_stack.append(['dd', '', 0]) |
| 333 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 334 | def ddpop(self, bl=0): |
| 335 | self.formatter.end_paragraph(bl) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 336 | if self.list_stack: |
| 337 | if self.list_stack[-1][0] == 'dd': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 338 | del self.list_stack[-1] |
| 339 | self.formatter.pop_margin() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 340 | |
| 341 | # --- Phrase Markup |
| 342 | |
| 343 | # Idiomatic Elements |
| 344 | |
| 345 | def start_cite(self, attrs): self.start_i(attrs) |
| 346 | def end_cite(self): self.end_i() |
| 347 | |
| 348 | def start_code(self, attrs): self.start_tt(attrs) |
| 349 | def end_code(self): self.end_tt() |
| 350 | |
| 351 | def start_em(self, attrs): self.start_i(attrs) |
| 352 | def end_em(self): self.end_i() |
| 353 | |
| 354 | def start_kbd(self, attrs): self.start_tt(attrs) |
| 355 | def end_kbd(self): self.end_tt() |
| 356 | |
| 357 | def start_samp(self, attrs): self.start_tt(attrs) |
| 358 | def end_samp(self): self.end_tt() |
| 359 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 360 | def start_strong(self, attrs): self.start_b(attrs) |
| 361 | def end_strong(self): self.end_b() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 362 | |
| 363 | def start_var(self, attrs): self.start_i(attrs) |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 364 | def end_var(self): self.end_i() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 365 | |
| 366 | # Typographic Elements |
| 367 | |
| 368 | def start_i(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 369 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 370 | def end_i(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 371 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 372 | |
| 373 | def start_b(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 374 | self.formatter.push_font((AS_IS, AS_IS, 1, AS_IS)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 375 | def end_b(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 376 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 377 | |
| 378 | def start_tt(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 379 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 380 | def end_tt(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 381 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 382 | |
| 383 | def start_a(self, attrs): |
| 384 | href = '' |
| 385 | name = '' |
| 386 | type = '' |
| 387 | for attrname, value in attrs: |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 388 | value = value.strip() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 389 | if attrname == 'href': |
| 390 | href = value |
| 391 | if attrname == 'name': |
| 392 | name = value |
| 393 | if attrname == 'type': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 394 | type = value.lower() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 395 | self.anchor_bgn(href, name, type) |
| 396 | |
| 397 | def end_a(self): |
| 398 | self.anchor_end() |
| 399 | |
| 400 | # --- Line Break |
| 401 | |
| 402 | def do_br(self, attrs): |
| 403 | self.formatter.add_line_break() |
| 404 | |
| 405 | # --- Horizontal Rule |
| 406 | |
| 407 | def do_hr(self, attrs): |
| 408 | self.formatter.add_hor_rule() |
| 409 | |
| 410 | # --- Image |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 411 | |
| 412 | def do_img(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 413 | align = '' |
| 414 | alt = '(image)' |
| 415 | ismap = '' |
| 416 | src = '' |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 417 | width = 0 |
| 418 | height = 0 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 419 | for attrname, value in attrs: |
| 420 | if attrname == 'align': |
| 421 | align = value |
| 422 | if attrname == 'alt': |
| 423 | alt = value |
| 424 | if attrname == 'ismap': |
| 425 | ismap = value |
| 426 | if attrname == 'src': |
| 427 | src = value |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 428 | if attrname == 'width': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 429 | try: width = int(value) |
Fred Drake | 1b7e079 | 2001-05-11 18:45:52 +0000 | [diff] [blame] | 430 | except ValueError: pass |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 431 | if attrname == 'height': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 432 | try: height = int(value) |
Fred Drake | 1b7e079 | 2001-05-11 18:45:52 +0000 | [diff] [blame] | 433 | except ValueError: pass |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 434 | self.handle_image(src, alt, ismap, align, width, height) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 435 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 436 | # --- Really Old Unofficial Deprecated Stuff |
| 437 | |
| 438 | def do_plaintext(self, attrs): |
| 439 | self.start_pre(attrs) |
| 440 | self.setnomoretags() # Tell SGML parser |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 441 | |
| 442 | # --- Unhandled tags |
| 443 | |
| 444 | def unknown_starttag(self, tag, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 445 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 446 | |
| 447 | def unknown_endtag(self, tag): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 448 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 449 | |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 450 | |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 451 | def test(args = None): |
| 452 | import sys, formatter |
| 453 | |
| 454 | if not args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 455 | args = sys.argv[1:] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 456 | |
| 457 | silent = args and args[0] == '-s' |
| 458 | if silent: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 459 | del args[0] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 460 | |
| 461 | if args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 462 | file = args[0] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 463 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 464 | file = 'test.html' |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 465 | |
| 466 | if file == '-': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 467 | f = sys.stdin |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 468 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 469 | try: |
| 470 | f = open(file, 'r') |
| 471 | except IOError, msg: |
| 472 | print file, ":", msg |
| 473 | sys.exit(1) |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 474 | |
| 475 | data = f.read() |
| 476 | |
| 477 | if f is not sys.stdin: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 478 | f.close() |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 479 | |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 480 | if silent: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 481 | f = formatter.NullFormatter() |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 482 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 483 | f = formatter.AbstractFormatter(formatter.DumbWriter()) |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 484 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 485 | p = HTMLParser(f) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 486 | p.feed(data) |
| 487 | p.close() |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 488 | |
| 489 | |
| 490 | if __name__ == '__main__': |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 491 | test() |