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 | |
| 7 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 8 | from sgmllib import SGMLParser |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 9 | from formatter import AS_IS |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 10 | |
Skip Montanaro | 2dd4276 | 2001-01-23 15:35:05 +0000 | [diff] [blame] | 11 | __all__ = ["HTMLParser"] |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 12 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 13 | class HTMLParser(SGMLParser): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 14 | """This is the basic HTML parser class. |
| 15 | |
| 16 | It supports all entity names required by the HTML 2.0 specification |
| 17 | RFC 1866. It also defines handlers for all HTML 2.0 and many HTML 3.0 |
| 18 | and 3.2 elements. |
| 19 | |
| 20 | """ |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 21 | |
Guido van Rossum | 65126d5 | 1995-09-27 16:22:17 +0000 | [diff] [blame] | 22 | from htmlentitydefs import entitydefs |
| 23 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 24 | def __init__(self, formatter, verbose=0): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 25 | """Creates an instance of the HTMLParser class. |
| 26 | |
| 27 | The formatter parameter is the formatter instance associated with |
| 28 | the parser. |
| 29 | |
| 30 | """ |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 31 | SGMLParser.__init__(self, verbose) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 32 | self.formatter = formatter |
| 33 | self.savedata = None |
| 34 | self.isindex = 0 |
| 35 | self.title = None |
| 36 | self.base = None |
| 37 | self.anchor = None |
| 38 | self.anchorlist = [] |
| 39 | self.nofill = 0 |
| 40 | self.list_stack = [] |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 41 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 42 | # ------ Methods used internally; some may be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 43 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 44 | # --- Formatter interface, taking care of 'savedata' mode; |
| 45 | # shouldn't need to be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 46 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 47 | def handle_data(self, data): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 48 | if self.savedata is not None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 49 | self.savedata = self.savedata + data |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 50 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 51 | if self.nofill: |
| 52 | self.formatter.add_literal_data(data) |
| 53 | else: |
| 54 | self.formatter.add_flowing_data(data) |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 55 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 56 | # --- Hooks to save data; shouldn't need to be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 57 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 58 | def save_bgn(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 59 | """Begins saving character data in a buffer instead of sending it |
| 60 | to the formatter object. |
| 61 | |
| 62 | Retrieve the stored data via the save_end() method. Use of the |
| 63 | save_bgn() / save_end() pair may not be nested. |
| 64 | |
| 65 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 66 | self.savedata = '' |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 67 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 68 | def save_end(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 69 | """Ends buffering character data and returns all data saved since |
| 70 | the preceding call to the save_bgn() method. |
| 71 | |
| 72 | If the nofill flag is false, whitespace is collapsed to single |
| 73 | spaces. A call to this method without a preceding call to the |
| 74 | save_bgn() method will raise a TypeError exception. |
| 75 | |
| 76 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 77 | data = self.savedata |
| 78 | self.savedata = None |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 79 | if not self.nofill: |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 80 | data = ' '.join(data.split()) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 81 | return data |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 82 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 83 | # --- Hooks for anchors; should probably be overridden |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 84 | |
| 85 | def anchor_bgn(self, href, name, type): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 86 | """This method is called at the start of an anchor region. |
| 87 | |
| 88 | The arguments correspond to the attributes of the <A> tag with |
| 89 | the same names. The default implementation maintains a list of |
| 90 | hyperlinks (defined by the HREF attribute for <A> tags) within |
| 91 | the document. The list of hyperlinks is available as the data |
| 92 | attribute anchorlist. |
| 93 | |
| 94 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 95 | self.anchor = href |
| 96 | if self.anchor: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 97 | self.anchorlist.append(href) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 98 | |
| 99 | def anchor_end(self): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 100 | """This method is called at the end of an anchor region. |
| 101 | |
| 102 | The default implementation adds a textual footnote marker using an |
| 103 | index into the list of hyperlinks created by the anchor_bgn()method. |
| 104 | |
| 105 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 106 | if self.anchor: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 107 | self.handle_data("[%d]" % len(self.anchorlist)) |
| 108 | self.anchor = None |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 109 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 110 | # --- Hook for images; should probably be overridden |
| 111 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 112 | def handle_image(self, src, alt, *args): |
Raymond Hettinger | aef22fb | 2002-05-29 16:18:42 +0000 | [diff] [blame] | 113 | """This method is called to handle images. |
| 114 | |
| 115 | The default implementation simply passes the alt value to the |
| 116 | handle_data() method. |
| 117 | |
| 118 | """ |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 119 | self.handle_data(alt) |
| 120 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 121 | # --------- Top level elememts |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 122 | |
| 123 | def start_html(self, attrs): pass |
| 124 | def end_html(self): pass |
| 125 | |
| 126 | def start_head(self, attrs): pass |
| 127 | def end_head(self): pass |
| 128 | |
| 129 | def start_body(self, attrs): pass |
| 130 | def end_body(self): pass |
| 131 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 132 | # ------ Head elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 133 | |
| 134 | def start_title(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 135 | self.save_bgn() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 136 | |
| 137 | def end_title(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 138 | self.title = self.save_end() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 139 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 140 | def do_base(self, attrs): |
| 141 | for a, v in attrs: |
| 142 | if a == 'href': |
| 143 | self.base = v |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 144 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 145 | def do_isindex(self, attrs): |
| 146 | self.isindex = 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 147 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 148 | def do_link(self, attrs): |
| 149 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 150 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 151 | def do_meta(self, attrs): |
| 152 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 153 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 154 | def do_nextid(self, attrs): # Deprecated |
| 155 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 156 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 157 | # ------ Body elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 158 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 159 | # --- Headings |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 160 | |
| 161 | def start_h1(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 162 | self.formatter.end_paragraph(1) |
| 163 | self.formatter.push_font(('h1', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 164 | |
| 165 | def end_h1(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 166 | self.formatter.end_paragraph(1) |
| 167 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 168 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 169 | def start_h2(self, attrs): |
| 170 | self.formatter.end_paragraph(1) |
| 171 | self.formatter.push_font(('h2', 0, 1, 0)) |
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 | def end_h2(self): |
| 174 | self.formatter.end_paragraph(1) |
| 175 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 176 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 177 | def start_h3(self, attrs): |
| 178 | self.formatter.end_paragraph(1) |
| 179 | self.formatter.push_font(('h3', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 180 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 181 | def end_h3(self): |
| 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_h4(self, attrs): |
| 186 | self.formatter.end_paragraph(1) |
| 187 | self.formatter.push_font(('h4', 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_h4(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_h5(self, attrs): |
| 194 | self.formatter.end_paragraph(1) |
| 195 | self.formatter.push_font(('h5', 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_h5(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_h6(self, attrs): |
| 202 | self.formatter.end_paragraph(1) |
| 203 | self.formatter.push_font(('h6', 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_h6(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 | # --- Block Structuring Elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 210 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 211 | def do_p(self, attrs): |
| 212 | self.formatter.end_paragraph(1) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 213 | |
| 214 | def start_pre(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 215 | self.formatter.end_paragraph(1) |
| 216 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
| 217 | self.nofill = self.nofill + 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 218 | |
| 219 | def end_pre(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 220 | self.formatter.end_paragraph(1) |
| 221 | self.formatter.pop_font() |
| 222 | self.nofill = max(0, self.nofill - 1) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 223 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 224 | def start_xmp(self, attrs): |
| 225 | self.start_pre(attrs) |
| 226 | self.setliteral('xmp') # Tell SGML parser |
| 227 | |
| 228 | def end_xmp(self): |
| 229 | self.end_pre() |
| 230 | |
| 231 | def start_listing(self, attrs): |
| 232 | self.start_pre(attrs) |
| 233 | self.setliteral('listing') # Tell SGML parser |
| 234 | |
| 235 | def end_listing(self): |
| 236 | self.end_pre() |
| 237 | |
| 238 | def start_address(self, attrs): |
| 239 | self.formatter.end_paragraph(0) |
| 240 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
| 241 | |
| 242 | def end_address(self): |
| 243 | self.formatter.end_paragraph(0) |
| 244 | self.formatter.pop_font() |
| 245 | |
| 246 | def start_blockquote(self, attrs): |
| 247 | self.formatter.end_paragraph(1) |
| 248 | self.formatter.push_margin('blockquote') |
| 249 | |
| 250 | def end_blockquote(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 251 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 252 | self.formatter.pop_margin() |
| 253 | |
| 254 | # --- List Elements |
| 255 | |
| 256 | def start_ul(self, attrs): |
| 257 | self.formatter.end_paragraph(not self.list_stack) |
| 258 | self.formatter.push_margin('ul') |
| 259 | self.list_stack.append(['ul', '*', 0]) |
| 260 | |
| 261 | def end_ul(self): |
| 262 | if self.list_stack: del self.list_stack[-1] |
| 263 | self.formatter.end_paragraph(not self.list_stack) |
| 264 | self.formatter.pop_margin() |
| 265 | |
| 266 | def do_li(self, attrs): |
| 267 | self.formatter.end_paragraph(0) |
| 268 | if self.list_stack: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 269 | [dummy, label, counter] = top = self.list_stack[-1] |
| 270 | top[2] = counter = counter+1 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 271 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 272 | label, counter = '*', 0 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 273 | self.formatter.add_label_data(label, counter) |
| 274 | |
| 275 | def start_ol(self, attrs): |
| 276 | self.formatter.end_paragraph(not self.list_stack) |
| 277 | self.formatter.push_margin('ol') |
| 278 | label = '1.' |
| 279 | for a, v in attrs: |
| 280 | if a == 'type': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 281 | if len(v) == 1: v = v + '.' |
| 282 | label = v |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 283 | self.list_stack.append(['ol', label, 0]) |
| 284 | |
| 285 | def end_ol(self): |
| 286 | if self.list_stack: del self.list_stack[-1] |
| 287 | self.formatter.end_paragraph(not self.list_stack) |
| 288 | self.formatter.pop_margin() |
| 289 | |
| 290 | def start_menu(self, attrs): |
| 291 | self.start_ul(attrs) |
| 292 | |
| 293 | def end_menu(self): |
| 294 | self.end_ul() |
| 295 | |
| 296 | def start_dir(self, attrs): |
| 297 | self.start_ul(attrs) |
| 298 | |
| 299 | def end_dir(self): |
| 300 | self.end_ul() |
| 301 | |
| 302 | def start_dl(self, attrs): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 303 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 304 | self.list_stack.append(['dl', '', 0]) |
| 305 | |
| 306 | def end_dl(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 307 | self.ddpop(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 308 | if self.list_stack: del self.list_stack[-1] |
| 309 | |
| 310 | def do_dt(self, attrs): |
| 311 | self.ddpop() |
| 312 | |
| 313 | def do_dd(self, attrs): |
| 314 | self.ddpop() |
| 315 | self.formatter.push_margin('dd') |
| 316 | self.list_stack.append(['dd', '', 0]) |
| 317 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 318 | def ddpop(self, bl=0): |
| 319 | self.formatter.end_paragraph(bl) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 320 | if self.list_stack: |
| 321 | if self.list_stack[-1][0] == 'dd': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 322 | del self.list_stack[-1] |
| 323 | self.formatter.pop_margin() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 324 | |
| 325 | # --- Phrase Markup |
| 326 | |
| 327 | # Idiomatic Elements |
| 328 | |
| 329 | def start_cite(self, attrs): self.start_i(attrs) |
| 330 | def end_cite(self): self.end_i() |
| 331 | |
| 332 | def start_code(self, attrs): self.start_tt(attrs) |
| 333 | def end_code(self): self.end_tt() |
| 334 | |
| 335 | def start_em(self, attrs): self.start_i(attrs) |
| 336 | def end_em(self): self.end_i() |
| 337 | |
| 338 | def start_kbd(self, attrs): self.start_tt(attrs) |
| 339 | def end_kbd(self): self.end_tt() |
| 340 | |
| 341 | def start_samp(self, attrs): self.start_tt(attrs) |
| 342 | def end_samp(self): self.end_tt() |
| 343 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 344 | def start_strong(self, attrs): self.start_b(attrs) |
| 345 | def end_strong(self): self.end_b() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 346 | |
| 347 | def start_var(self, attrs): self.start_i(attrs) |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 348 | def end_var(self): self.end_i() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 349 | |
| 350 | # Typographic Elements |
| 351 | |
| 352 | def start_i(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 353 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 354 | def end_i(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 355 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 356 | |
| 357 | def start_b(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 358 | self.formatter.push_font((AS_IS, AS_IS, 1, AS_IS)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 359 | def end_b(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 360 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 361 | |
| 362 | def start_tt(self, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 363 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 364 | def end_tt(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 365 | self.formatter.pop_font() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 366 | |
| 367 | def start_a(self, attrs): |
| 368 | href = '' |
| 369 | name = '' |
| 370 | type = '' |
| 371 | for attrname, value in attrs: |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 372 | value = value.strip() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 373 | if attrname == 'href': |
| 374 | href = value |
| 375 | if attrname == 'name': |
| 376 | name = value |
| 377 | if attrname == 'type': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 378 | type = value.lower() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 379 | self.anchor_bgn(href, name, type) |
| 380 | |
| 381 | def end_a(self): |
| 382 | self.anchor_end() |
| 383 | |
| 384 | # --- Line Break |
| 385 | |
| 386 | def do_br(self, attrs): |
| 387 | self.formatter.add_line_break() |
| 388 | |
| 389 | # --- Horizontal Rule |
| 390 | |
| 391 | def do_hr(self, attrs): |
| 392 | self.formatter.add_hor_rule() |
| 393 | |
| 394 | # --- Image |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 395 | |
| 396 | def do_img(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 397 | align = '' |
| 398 | alt = '(image)' |
| 399 | ismap = '' |
| 400 | src = '' |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 401 | width = 0 |
| 402 | height = 0 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 403 | for attrname, value in attrs: |
| 404 | if attrname == 'align': |
| 405 | align = value |
| 406 | if attrname == 'alt': |
| 407 | alt = value |
| 408 | if attrname == 'ismap': |
| 409 | ismap = value |
| 410 | if attrname == 'src': |
| 411 | src = value |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 412 | if attrname == 'width': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 413 | try: width = int(value) |
Fred Drake | 1b7e079 | 2001-05-11 18:45:52 +0000 | [diff] [blame] | 414 | except ValueError: pass |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 415 | if attrname == 'height': |
Eric S. Raymond | 373c55e | 2001-02-09 08:25:29 +0000 | [diff] [blame] | 416 | try: height = int(value) |
Fred Drake | 1b7e079 | 2001-05-11 18:45:52 +0000 | [diff] [blame] | 417 | except ValueError: pass |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 418 | self.handle_image(src, alt, ismap, align, width, height) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 419 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 420 | # --- Really Old Unofficial Deprecated Stuff |
| 421 | |
| 422 | def do_plaintext(self, attrs): |
| 423 | self.start_pre(attrs) |
| 424 | self.setnomoretags() # Tell SGML parser |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 425 | |
| 426 | # --- Unhandled tags |
| 427 | |
| 428 | def unknown_starttag(self, tag, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 429 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 430 | |
| 431 | def unknown_endtag(self, tag): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 432 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 433 | |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 434 | |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 435 | def test(args = None): |
| 436 | import sys, formatter |
| 437 | |
| 438 | if not args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 439 | args = sys.argv[1:] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 440 | |
| 441 | silent = args and args[0] == '-s' |
| 442 | if silent: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 443 | del args[0] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 444 | |
| 445 | if args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 446 | file = args[0] |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 447 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 448 | file = 'test.html' |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 449 | |
| 450 | if file == '-': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 451 | f = sys.stdin |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 452 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 453 | try: |
| 454 | f = open(file, 'r') |
| 455 | except IOError, msg: |
| 456 | print file, ":", msg |
| 457 | sys.exit(1) |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 458 | |
| 459 | data = f.read() |
| 460 | |
| 461 | if f is not sys.stdin: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 462 | f.close() |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 463 | |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 464 | if silent: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 465 | f = formatter.NullFormatter() |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 466 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 467 | f = formatter.AbstractFormatter(formatter.DumbWriter()) |
Guido van Rossum | a98b0a1 | 1996-02-13 00:02:10 +0000 | [diff] [blame] | 468 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 469 | p = HTMLParser(f) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 470 | p.feed(data) |
| 471 | p.close() |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 472 | |
| 473 | |
| 474 | if __name__ == '__main__': |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 475 | test() |