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 | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 8 | import sys |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 9 | import regsub |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 10 | import string |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 11 | from sgmllib import SGMLParser |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 12 | from formatter import AS_IS |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 13 | |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 14 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 15 | class HTMLParser(SGMLParser): |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 16 | |
Guido van Rossum | 65126d5 | 1995-09-27 16:22:17 +0000 | [diff] [blame] | 17 | from htmlentitydefs import entitydefs |
| 18 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 19 | def __init__(self, formatter, verbose=0): |
| 20 | SGMLParser.__init__(self, verbose) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 21 | self.formatter = formatter |
| 22 | self.savedata = None |
| 23 | self.isindex = 0 |
| 24 | self.title = None |
| 25 | self.base = None |
| 26 | self.anchor = None |
| 27 | self.anchorlist = [] |
| 28 | self.nofill = 0 |
| 29 | self.list_stack = [] |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 30 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 31 | # ------ Methods used internally; some may be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 32 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 33 | # --- Formatter interface, taking care of 'savedata' mode; |
| 34 | # shouldn't need to be overridden |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 35 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 36 | def handle_data(self, data): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 37 | if self.savedata is not None: |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 38 | self.savedata = self.savedata + data |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 39 | else: |
| 40 | if self.nofill: |
| 41 | self.formatter.add_literal_data(data) |
| 42 | else: |
| 43 | self.formatter.add_flowing_data(data) |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 44 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 45 | # --- Hooks to save data; 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 save_bgn(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 48 | self.savedata = '' |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 49 | |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 50 | def save_end(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 51 | data = self.savedata |
| 52 | self.savedata = None |
Guido van Rossum | fd504d9 | 1995-09-01 20:33:32 +0000 | [diff] [blame] | 53 | if not self.nofill: |
| 54 | data = string.join(string.split(data)) |
| 55 | return data |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 56 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 57 | # --- Hooks for anchors; should probably be overridden |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 58 | |
| 59 | def anchor_bgn(self, href, name, type): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 60 | self.anchor = href |
| 61 | if self.anchor: |
| 62 | self.anchorlist.append(href) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 63 | |
| 64 | def anchor_end(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 65 | if self.anchor: |
| 66 | self.handle_data("[%d]" % len(self.anchorlist)) |
| 67 | self.anchor = None |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 68 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 69 | # --- Hook for images; should probably be overridden |
| 70 | |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 71 | def handle_image(self, src, alt, *args): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 72 | self.handle_data(alt) |
| 73 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 74 | # --------- Top level elememts |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 75 | |
| 76 | def start_html(self, attrs): pass |
| 77 | def end_html(self): pass |
| 78 | |
| 79 | def start_head(self, attrs): pass |
| 80 | def end_head(self): pass |
| 81 | |
| 82 | def start_body(self, attrs): pass |
| 83 | def end_body(self): pass |
| 84 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 85 | # ------ Head elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 86 | |
| 87 | def start_title(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 88 | self.save_bgn() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 89 | |
| 90 | def end_title(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 91 | self.title = self.save_end() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 92 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 93 | def do_base(self, attrs): |
| 94 | for a, v in attrs: |
| 95 | if a == 'href': |
| 96 | self.base = v |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 97 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 98 | def do_isindex(self, attrs): |
| 99 | self.isindex = 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 100 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 101 | def do_link(self, attrs): |
| 102 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 103 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 104 | def do_meta(self, attrs): |
| 105 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 106 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 107 | def do_nextid(self, attrs): # Deprecated |
| 108 | pass |
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 | # ------ Body elements |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 111 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 112 | # --- Headings |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 113 | |
| 114 | def start_h1(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 115 | self.formatter.end_paragraph(1) |
| 116 | self.formatter.push_font(('h1', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 117 | |
| 118 | def end_h1(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 119 | self.formatter.end_paragraph(1) |
| 120 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 121 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 122 | def start_h2(self, attrs): |
| 123 | self.formatter.end_paragraph(1) |
| 124 | self.formatter.push_font(('h2', 0, 1, 0)) |
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 | def end_h2(self): |
| 127 | self.formatter.end_paragraph(1) |
| 128 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 129 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 130 | def start_h3(self, attrs): |
| 131 | self.formatter.end_paragraph(1) |
| 132 | self.formatter.push_font(('h3', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 133 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 134 | def end_h3(self): |
| 135 | self.formatter.end_paragraph(1) |
| 136 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 137 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 138 | def start_h4(self, attrs): |
| 139 | self.formatter.end_paragraph(1) |
| 140 | self.formatter.push_font(('h4', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 141 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 142 | def end_h4(self): |
| 143 | self.formatter.end_paragraph(1) |
| 144 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 145 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 146 | def start_h5(self, attrs): |
| 147 | self.formatter.end_paragraph(1) |
| 148 | self.formatter.push_font(('h5', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 149 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 150 | def end_h5(self): |
| 151 | self.formatter.end_paragraph(1) |
| 152 | self.formatter.pop_font() |
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 start_h6(self, attrs): |
| 155 | self.formatter.end_paragraph(1) |
| 156 | self.formatter.push_font(('h6', 0, 1, 0)) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 157 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 158 | def end_h6(self): |
| 159 | self.formatter.end_paragraph(1) |
| 160 | self.formatter.pop_font() |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 161 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 162 | # --- Block Structuring Elements |
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_p(self, attrs): |
| 165 | self.formatter.end_paragraph(1) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 166 | |
| 167 | def start_pre(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 168 | self.formatter.end_paragraph(1) |
| 169 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
| 170 | self.nofill = self.nofill + 1 |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 171 | |
| 172 | def end_pre(self): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 173 | self.formatter.end_paragraph(1) |
| 174 | self.formatter.pop_font() |
| 175 | self.nofill = max(0, self.nofill - 1) |
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_xmp(self, attrs): |
| 178 | self.start_pre(attrs) |
| 179 | self.setliteral('xmp') # Tell SGML parser |
| 180 | |
| 181 | def end_xmp(self): |
| 182 | self.end_pre() |
| 183 | |
| 184 | def start_listing(self, attrs): |
| 185 | self.start_pre(attrs) |
| 186 | self.setliteral('listing') # Tell SGML parser |
| 187 | |
| 188 | def end_listing(self): |
| 189 | self.end_pre() |
| 190 | |
| 191 | def start_address(self, attrs): |
| 192 | self.formatter.end_paragraph(0) |
| 193 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
| 194 | |
| 195 | def end_address(self): |
| 196 | self.formatter.end_paragraph(0) |
| 197 | self.formatter.pop_font() |
| 198 | |
| 199 | def start_blockquote(self, attrs): |
| 200 | self.formatter.end_paragraph(1) |
| 201 | self.formatter.push_margin('blockquote') |
| 202 | |
| 203 | def end_blockquote(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 204 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 205 | self.formatter.pop_margin() |
| 206 | |
| 207 | # --- List Elements |
| 208 | |
| 209 | def start_ul(self, attrs): |
| 210 | self.formatter.end_paragraph(not self.list_stack) |
| 211 | self.formatter.push_margin('ul') |
| 212 | self.list_stack.append(['ul', '*', 0]) |
| 213 | |
| 214 | def end_ul(self): |
| 215 | if self.list_stack: del self.list_stack[-1] |
| 216 | self.formatter.end_paragraph(not self.list_stack) |
| 217 | self.formatter.pop_margin() |
| 218 | |
| 219 | def do_li(self, attrs): |
| 220 | self.formatter.end_paragraph(0) |
| 221 | if self.list_stack: |
| 222 | [dummy, label, counter] = top = self.list_stack[-1] |
| 223 | top[2] = counter = counter+1 |
| 224 | else: |
| 225 | label, counter = '*', 0 |
| 226 | self.formatter.add_label_data(label, counter) |
| 227 | |
| 228 | def start_ol(self, attrs): |
| 229 | self.formatter.end_paragraph(not self.list_stack) |
| 230 | self.formatter.push_margin('ol') |
| 231 | label = '1.' |
| 232 | for a, v in attrs: |
| 233 | if a == 'type': |
| 234 | if len(v) == 1: v = v + '.' |
| 235 | label = v |
| 236 | self.list_stack.append(['ol', label, 0]) |
| 237 | |
| 238 | def end_ol(self): |
| 239 | if self.list_stack: del self.list_stack[-1] |
| 240 | self.formatter.end_paragraph(not self.list_stack) |
| 241 | self.formatter.pop_margin() |
| 242 | |
| 243 | def start_menu(self, attrs): |
| 244 | self.start_ul(attrs) |
| 245 | |
| 246 | def end_menu(self): |
| 247 | self.end_ul() |
| 248 | |
| 249 | def start_dir(self, attrs): |
| 250 | self.start_ul(attrs) |
| 251 | |
| 252 | def end_dir(self): |
| 253 | self.end_ul() |
| 254 | |
| 255 | def start_dl(self, attrs): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 256 | self.formatter.end_paragraph(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 257 | self.list_stack.append(['dl', '', 0]) |
| 258 | |
| 259 | def end_dl(self): |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 260 | self.ddpop(1) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 261 | if self.list_stack: del self.list_stack[-1] |
| 262 | |
| 263 | def do_dt(self, attrs): |
| 264 | self.ddpop() |
| 265 | |
| 266 | def do_dd(self, attrs): |
| 267 | self.ddpop() |
| 268 | self.formatter.push_margin('dd') |
| 269 | self.list_stack.append(['dd', '', 0]) |
| 270 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 271 | def ddpop(self, bl=0): |
| 272 | self.formatter.end_paragraph(bl) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 273 | if self.list_stack: |
| 274 | if self.list_stack[-1][0] == 'dd': |
| 275 | del self.list_stack[-1] |
| 276 | self.formatter.pop_margin() |
| 277 | |
| 278 | # --- Phrase Markup |
| 279 | |
| 280 | # Idiomatic Elements |
| 281 | |
| 282 | def start_cite(self, attrs): self.start_i(attrs) |
| 283 | def end_cite(self): self.end_i() |
| 284 | |
| 285 | def start_code(self, attrs): self.start_tt(attrs) |
| 286 | def end_code(self): self.end_tt() |
| 287 | |
| 288 | def start_em(self, attrs): self.start_i(attrs) |
| 289 | def end_em(self): self.end_i() |
| 290 | |
| 291 | def start_kbd(self, attrs): self.start_tt(attrs) |
| 292 | def end_kbd(self): self.end_tt() |
| 293 | |
| 294 | def start_samp(self, attrs): self.start_tt(attrs) |
| 295 | def end_samp(self): self.end_tt() |
| 296 | |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 297 | def start_strong(self, attrs): self.start_b(attrs) |
| 298 | def end_strong(self): self.end_b() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 299 | |
| 300 | def start_var(self, attrs): self.start_i(attrs) |
Guido van Rossum | aa76344 | 1995-08-09 02:31:00 +0000 | [diff] [blame] | 301 | def end_var(self): self.end_i() |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 302 | |
| 303 | # Typographic Elements |
| 304 | |
| 305 | def start_i(self, attrs): |
| 306 | self.formatter.push_font((AS_IS, 1, AS_IS, AS_IS)) |
| 307 | def end_i(self): |
| 308 | self.formatter.pop_font() |
| 309 | |
| 310 | def start_b(self, attrs): |
| 311 | self.formatter.push_font((AS_IS, AS_IS, 1, AS_IS)) |
| 312 | def end_b(self): |
| 313 | self.formatter.pop_font() |
| 314 | |
| 315 | def start_tt(self, attrs): |
| 316 | self.formatter.push_font((AS_IS, AS_IS, AS_IS, 1)) |
| 317 | def end_tt(self): |
| 318 | self.formatter.pop_font() |
| 319 | |
| 320 | def start_a(self, attrs): |
| 321 | href = '' |
| 322 | name = '' |
| 323 | type = '' |
| 324 | for attrname, value in attrs: |
Guido van Rossum | 84175ba | 1995-10-06 15:30:57 +0000 | [diff] [blame] | 325 | value = string.strip(value) |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 326 | if attrname == 'href': |
| 327 | href = value |
| 328 | if attrname == 'name': |
| 329 | name = value |
| 330 | if attrname == 'type': |
| 331 | type = string.lower(value) |
| 332 | self.anchor_bgn(href, name, type) |
| 333 | |
| 334 | def end_a(self): |
| 335 | self.anchor_end() |
| 336 | |
| 337 | # --- Line Break |
| 338 | |
| 339 | def do_br(self, attrs): |
| 340 | self.formatter.add_line_break() |
| 341 | |
| 342 | # --- Horizontal Rule |
| 343 | |
| 344 | def do_hr(self, attrs): |
| 345 | self.formatter.add_hor_rule() |
| 346 | |
| 347 | # --- Image |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 348 | |
| 349 | def do_img(self, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 350 | align = '' |
| 351 | alt = '(image)' |
| 352 | ismap = '' |
| 353 | src = '' |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 354 | width = 0 |
| 355 | height = 0 |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 356 | for attrname, value in attrs: |
| 357 | if attrname == 'align': |
| 358 | align = value |
| 359 | if attrname == 'alt': |
| 360 | alt = value |
| 361 | if attrname == 'ismap': |
| 362 | ismap = value |
| 363 | if attrname == 'src': |
| 364 | src = value |
Guido van Rossum | 453534a | 1995-09-22 00:55:50 +0000 | [diff] [blame] | 365 | if attrname == 'width': |
| 366 | try: width = string.atoi(value) |
| 367 | except: pass |
| 368 | if attrname == 'height': |
| 369 | try: height = string.atoi(value) |
| 370 | except: pass |
| 371 | self.handle_image(src, alt, ismap, align, width, height) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 372 | |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 373 | # --- Really Old Unofficial Deprecated Stuff |
| 374 | |
| 375 | def do_plaintext(self, attrs): |
| 376 | self.start_pre(attrs) |
| 377 | self.setnomoretags() # Tell SGML parser |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 378 | |
| 379 | # --- Unhandled tags |
| 380 | |
| 381 | def unknown_starttag(self, tag, attrs): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 382 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 383 | |
| 384 | def unknown_endtag(self, tag): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 385 | pass |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 386 | |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 387 | |
| 388 | def test(): |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 389 | import sys |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 390 | file = 'test.html' |
Guido van Rossum | f54d967 | 1995-08-07 20:07:44 +0000 | [diff] [blame] | 391 | if sys.argv[1:]: file = sys.argv[1] |
| 392 | fp = open(file, 'r') |
| 393 | data = fp.read() |
| 394 | fp.close() |
| 395 | from formatter import DumbWriter, AbstractFormatter |
| 396 | w = DumbWriter() |
| 397 | f = AbstractFormatter(w) |
| 398 | p = HTMLParser(f) |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 399 | p.feed(data) |
| 400 | p.close() |
Guido van Rossum | 7c750e1 | 1995-02-27 13:16:55 +0000 | [diff] [blame] | 401 | |
| 402 | |
| 403 | if __name__ == '__main__': |
Guido van Rossum | 7ff5d7f | 1995-08-04 04:23:30 +0000 | [diff] [blame] | 404 | test() |