| 1 | #!/usr/bin/env python2
|
| 2 | """ul_table.py: Markdown Tables Without New Syntax."""
|
| 3 |
|
| 4 | try:
|
| 5 | from cStringIO import StringIO
|
| 6 | except ImportError:
|
| 7 | from io import StringIO
|
| 8 | import re
|
| 9 | import sys
|
| 10 |
|
| 11 | from doctools.util import log
|
| 12 | from lazylex import html
|
| 13 |
|
| 14 | _WHITESPACE_RE = re.compile(r'\s*')
|
| 15 |
|
| 16 |
|
| 17 | class UlTableParser(object):
|
| 18 |
|
| 19 | def __init__(self, lexer, tag_lexer):
|
| 20 | self.lexer = lexer
|
| 21 | self.tag_lexer = tag_lexer
|
| 22 |
|
| 23 | self.tok_id = html.Invalid
|
| 24 | self.start_pos = 0
|
| 25 | self.end_pos = 0
|
| 26 |
|
| 27 | def _CurrentString(self):
|
| 28 | part = self.lexer.s[self.start_pos:self.end_pos]
|
| 29 | return part
|
| 30 |
|
| 31 | def _Next(self):
|
| 32 | """
|
| 33 | Advance and set self.tok_id, self.start_pos, self.end_pos
|
| 34 | """
|
| 35 | self.start_pos = self.end_pos
|
| 36 | self.tok_id, self.end_pos = self.lexer.Read()
|
| 37 | if 0:
|
| 38 | part = self._CurrentString()
|
| 39 | log('[%3d - %3d] %r', self.start_pos, self.end_pos, part)
|
| 40 |
|
| 41 | #self.tok_id = html.EndOfStream
|
| 42 | # Don't change self.end_pos
|
| 43 |
|
| 44 | def _EatRawData(self, regex):
|
| 45 | # type: (str) -> None
|
| 46 | """
|
| 47 | Assert that we got text data matching a regex, and advance
|
| 48 | """
|
| 49 | if self.tok_id != html.RawData:
|
| 50 | raise html.ParseError('Expected RawData, got %s',
|
| 51 | html.TokenName(self.tok_id))
|
| 52 | actual = self._CurrentString()
|
| 53 | m = re.match(regex, actual) # could compile this
|
| 54 | if m is None:
|
| 55 | raise html.ParseError('Expected to match %r, got %r', regex,
|
| 56 | actual)
|
| 57 | self._Next()
|
| 58 |
|
| 59 | def _Eat(self, expected_id, expected_tag):
|
| 60 | """
|
| 61 | Assert that we got a start or end tag, with the given name, and advance
|
| 62 |
|
| 63 | Args:
|
| 64 | expected_id: html.StartTag or html.EndTag
|
| 65 | expected_tag: 'a', 'span', etc.
|
| 66 | """
|
| 67 | assert expected_id in (html.StartTag,
|
| 68 | html.EndTag), html.TokenName(expected_id)
|
| 69 |
|
| 70 | if self.tok_id != expected_id:
|
| 71 | raise html.ParseError('Expected token %s, got %s',
|
| 72 | html.TokenName(expected_id),
|
| 73 | html.TokenName(self.tok_id))
|
| 74 | self.tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 75 | tag_name = self.tag_lexer.TagName()
|
| 76 | if expected_tag != tag_name:
|
| 77 | raise html.ParseError('Expected tag %r, got %r', expected_tag,
|
| 78 | tag_name)
|
| 79 |
|
| 80 | self._Next()
|
| 81 |
|
| 82 | def _WhitespaceOk(self):
|
| 83 | """
|
| 84 | Optional whitespace
|
| 85 | """
|
| 86 | if (self.tok_id == html.RawData and
|
| 87 | _WHITESPACE_RE.match(self.lexer.s, self.start_pos)):
|
| 88 | self._Next()
|
| 89 |
|
| 90 | def FindUlTable(self):
|
| 91 | """Find <table ...> <ul>
|
| 92 |
|
| 93 | Return the START position of the <ul>
|
| 94 | Similar algorithm as html.ReadUntilStartTag()
|
| 95 | """
|
| 96 | tag_lexer = self.tag_lexer
|
| 97 |
|
| 98 | # Find first table
|
| 99 | while True:
|
| 100 | self._Next()
|
| 101 | if self.tok_id == html.EndOfStream:
|
| 102 | return -1
|
| 103 |
|
| 104 | tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 105 | if (self.tok_id == html.StartTag and
|
| 106 | tag_lexer.TagName() == 'table'):
|
| 107 | while True:
|
| 108 | self._Next()
|
| 109 | if self.tok_id != html.RawData:
|
| 110 | break
|
| 111 |
|
| 112 | tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 113 | if (self.tok_id == html.StartTag and
|
| 114 | tag_lexer.TagName() == 'ul'):
|
| 115 | return self.start_pos
|
| 116 | return -1
|
| 117 |
|
| 118 | def _ListItem(self):
|
| 119 | """Parse a list item nested below thead or tr.
|
| 120 |
|
| 121 | Returns:
|
| 122 | A pair (td_attrs, inner_html)
|
| 123 |
|
| 124 | Grammar:
|
| 125 |
|
| 126 | LIST_ITEM =
|
| 127 | [RawData \s*]?
|
| 128 | [StartTag 'li']
|
| 129 | ANY* # NOT context-free:
|
| 130 | # - we MATCH <li> and </li> with a tack
|
| 131 | # - We search for [StartEndTag 'cell-attrs']?
|
| 132 | [EndTag 'li']
|
| 133 |
|
| 134 | Example of attribute borrowing:
|
| 135 |
|
| 136 | - hi there ==>
|
| 137 | <li>hi there</li> ==>
|
| 138 | <td>hi there</td>
|
| 139 |
|
| 140 | - <cell-attrs class=foo /> hi there ==>
|
| 141 | <li><cell-attrs class=foo /> hi there </li> ==>
|
| 142 | <td class=foo> hi there </td> ==>
|
| 143 | """
|
| 144 | self._WhitespaceOk()
|
| 145 |
|
| 146 | if self.tok_id != html.StartTag:
|
| 147 | return None, None
|
| 148 |
|
| 149 | inner_html = None
|
| 150 | td_attrs = None # Can we also have col-attrs?
|
| 151 | td_attrs_span = None
|
| 152 |
|
| 153 | self._Eat(html.StartTag, 'li')
|
| 154 |
|
| 155 | left = self.start_pos
|
| 156 |
|
| 157 | # Find the closing </li>, taking into accounted NESTED tags:
|
| 158 | # <li> <li>foo</li> </li>
|
| 159 | # because cells can have bulleted lists
|
| 160 | balance = 0
|
| 161 | while True:
|
| 162 | if self.tok_id == html.StartEndTag:
|
| 163 | self.tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 164 | tag_name = self.tag_lexer.TagName()
|
| 165 | # TODO: remove td-attrs backward compat
|
| 166 | if tag_name in ('td-attrs', 'cell-attrs'):
|
| 167 | td_attrs_span = self.start_pos, self.end_pos
|
| 168 | td_attrs = self.tag_lexer.AllAttrsRaw()
|
| 169 | #log('CELL ATTRS %r', self._CurrentString())
|
| 170 |
|
| 171 | elif self.tok_id == html.StartTag:
|
| 172 | self.tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 173 | if self.tag_lexer.TagName() == 'li':
|
| 174 | balance += 1
|
| 175 |
|
| 176 | elif self.tok_id == html.EndTag:
|
| 177 | self.tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 178 | if self.tag_lexer.TagName() == 'li':
|
| 179 | balance -= 1
|
| 180 | if balance < 0:
|
| 181 | break
|
| 182 | self._Next()
|
| 183 |
|
| 184 | right = self.start_pos # start of the end tag
|
| 185 |
|
| 186 | s = self.tag_lexer.s
|
| 187 | if td_attrs_span:
|
| 188 | # everything except the <cell-attrs />
|
| 189 | inner_html = s[left:td_attrs_span[0]] + s[td_attrs_span[1]:right]
|
| 190 | #log('LEFT %r', s[left:td_attrs_span[0]])
|
| 191 | #log('RIGHT %r', s[td_attrs_span[1]:right])
|
| 192 | else:
|
| 193 | inner_html = s[left:right]
|
| 194 | #log('RAW inner html %r', inner_html)
|
| 195 |
|
| 196 | #self._Eat(html.EndTag, 'li')
|
| 197 | self._Next()
|
| 198 |
|
| 199 | return td_attrs, inner_html
|
| 200 |
|
| 201 | def _ParseTHead(self):
|
| 202 | """
|
| 203 | Assume we're looking at the first <ul> tag. Now we want to find
|
| 204 | <li>thead and the nested <ul>
|
| 205 |
|
| 206 | Grammar:
|
| 207 |
|
| 208 | THEAD =
|
| 209 | [StartTag 'ul']
|
| 210 | [RawData \s*]?
|
| 211 | [StartTag 'li']
|
| 212 | [RawData thead\s*]
|
| 213 | [StartTag 'ul'] # Indented bullet that starts -
|
| 214 | LIST_ITEM+
|
| 215 | [RawData \s*]?
|
| 216 | [EndTag 'ul']
|
| 217 | [RawData thead\s+]
|
| 218 | [End 'li']
|
| 219 |
|
| 220 | Two Algorithms:
|
| 221 |
|
| 222 | 1. Replacement:
|
| 223 | - skip over the first ul 'thead' li, and ul 'tr' li
|
| 224 | - then replace the next ul -> tr, and li -> td
|
| 225 | 2. Parsing and Rendering:
|
| 226 | - parse them into a structure
|
| 227 | - skip all the text
|
| 228 | - print your own HTML
|
| 229 |
|
| 230 | I think the second one is better, because it allows attribute extensions
|
| 231 | to thead
|
| 232 |
|
| 233 | - thead
|
| 234 | - name [link][]
|
| 235 | - colgroup=foo align=left
|
| 236 | - age
|
| 237 | - colgroup=foo align=right
|
| 238 | """
|
| 239 | #log('*** _ParseTHead')
|
| 240 | cells = []
|
| 241 |
|
| 242 | self._WhitespaceOk()
|
| 243 | self._Eat(html.StartTag, 'li')
|
| 244 |
|
| 245 | # In CommonMark, r'thead\n' is enough, because it strips trailing
|
| 246 | # whitespace. I'm not sure if other Markdown processors do that, so
|
| 247 | # use r'thead\s+'.
|
| 248 | self._EatRawData(r'thead\s+')
|
| 249 |
|
| 250 | # This is the row data
|
| 251 | self._Eat(html.StartTag, 'ul')
|
| 252 |
|
| 253 | while True:
|
| 254 | td_attrs, inner_html = self._ListItem()
|
| 255 | if inner_html is None:
|
| 256 | break
|
| 257 | cells.append((td_attrs, inner_html))
|
| 258 | self._WhitespaceOk()
|
| 259 |
|
| 260 | self._Eat(html.EndTag, 'ul')
|
| 261 |
|
| 262 | self._WhitespaceOk()
|
| 263 | self._Eat(html.EndTag, 'li')
|
| 264 |
|
| 265 | #log('_ParseTHead %s ', html.TOKEN_NAMES[self.tok_id])
|
| 266 | return cells
|
| 267 |
|
| 268 | def _ParseTr(self):
|
| 269 | """
|
| 270 | Assume we're looking at the first <ul> tag. Now we want to find
|
| 271 | <li>tr and the nested <ul>
|
| 272 |
|
| 273 | Grammar:
|
| 274 |
|
| 275 | TR =
|
| 276 | [RawData \s*]?
|
| 277 | [StartTag 'li']
|
| 278 | [RawData thead\s*]
|
| 279 | [StartTag 'ul'] # Indented bullet that starts -
|
| 280 | ( [StartEndTag row-attrs] [RawData \s*] )?
|
| 281 | LIST_ITEM+ # Defined above
|
| 282 | [RawData \s*]?
|
| 283 | [EndTag 'ul']
|
| 284 | """
|
| 285 | #log('*** _ParseTr')
|
| 286 |
|
| 287 | cells = []
|
| 288 |
|
| 289 | self._WhitespaceOk()
|
| 290 |
|
| 291 | # Could be a </ul>
|
| 292 | if self.tok_id != html.StartTag:
|
| 293 | return None, None
|
| 294 |
|
| 295 | self._Eat(html.StartTag, 'li')
|
| 296 |
|
| 297 | self._EatRawData(r'tr\s*')
|
| 298 |
|
| 299 | tr_attrs = None
|
| 300 | if self.tok_id == html.StartEndTag:
|
| 301 | self.tag_lexer.Reset(self.start_pos, self.end_pos)
|
| 302 | tag_name = self.tag_lexer.TagName()
|
| 303 | if tag_name != 'row-attrs':
|
| 304 | raise html.ParseError('Expected row-attrs, got %r' % tag_name)
|
| 305 | tr_attrs = self.tag_lexer.AllAttrsRaw()
|
| 306 | self._Next()
|
| 307 | self._WhitespaceOk()
|
| 308 |
|
| 309 | # This is the row data
|
| 310 | self._Eat(html.StartTag, 'ul')
|
| 311 |
|
| 312 | while True:
|
| 313 | td_attrs, inner_html = self._ListItem()
|
| 314 | if inner_html is None:
|
| 315 | break
|
| 316 | cells.append((td_attrs, inner_html))
|
| 317 | # TODO: assert
|
| 318 |
|
| 319 | self._WhitespaceOk()
|
| 320 |
|
| 321 | self._Eat(html.EndTag, 'ul')
|
| 322 |
|
| 323 | self._WhitespaceOk()
|
| 324 | self._Eat(html.EndTag, 'li')
|
| 325 |
|
| 326 | #log('_ParseTHead %s ', html.TOKEN_NAMES[self.tok_id])
|
| 327 | return tr_attrs, cells
|
| 328 |
|
| 329 | def ParseTable(self):
|
| 330 | """
|
| 331 | Returns a structure like this
|
| 332 | { 'thead': [ 'col1', 'col2' ], # TODO: columns can have CSS attributes
|
| 333 | 'tr': [ # raw HTML that you surround with <td>
|
| 334 | [ 'cell1 html', 'cell2 html' ],
|
| 335 | [ 'cell1 html', 'cell2 html' ],
|
| 336 | ]
|
| 337 | }
|
| 338 |
|
| 339 | Grammar:
|
| 340 |
|
| 341 | UL_TABLE =
|
| 342 | [StartTag 'ul']
|
| 343 | THEAD # this this returns the number of cells, so it's NOT context
|
| 344 | # free
|
| 345 | TR*
|
| 346 | [EndTag 'ul']
|
| 347 | """
|
| 348 | table = {'tr': []}
|
| 349 |
|
| 350 | ul_start = self.start_pos
|
| 351 | self._Eat(html.StartTag, 'ul')
|
| 352 |
|
| 353 | # Look ahead 2 or 3 tokens:
|
| 354 | if self.lexer.LookAhead(r'\s*<li>thead\s+'):
|
| 355 | thead = self._ParseTHead()
|
| 356 | else:
|
| 357 | thead = None
|
| 358 | #log('___ THEAD %s', thead)
|
| 359 |
|
| 360 | while True:
|
| 361 | tr_attrs, tr = self._ParseTr()
|
| 362 | if tr is None:
|
| 363 | break
|
| 364 | # Not validating because of colspan
|
| 365 | if 0:
|
| 366 | if thead and len(tr) != len(thead):
|
| 367 | raise html.ParseError('Expected %d cells, got %d: %s',
|
| 368 | len(thead), len(tr), tr)
|
| 369 |
|
| 370 | #log('___ TR %s', tr)
|
| 371 | table['tr'].append((tr_attrs, tr))
|
| 372 |
|
| 373 | self._Eat(html.EndTag, 'ul')
|
| 374 |
|
| 375 | self._WhitespaceOk()
|
| 376 |
|
| 377 | ul_end = self.start_pos
|
| 378 |
|
| 379 | table['thead'] = thead
|
| 380 | table['ul_start'] = ul_start
|
| 381 | table['ul_end'] = ul_end
|
| 382 |
|
| 383 | if 0:
|
| 384 | log('table %s', table)
|
| 385 | from pprint import pprint
|
| 386 | pprint(table)
|
| 387 |
|
| 388 | return table
|
| 389 |
|
| 390 |
|
| 391 | def MergeAttrs(thead_td_attrs, row_td_attrs):
|
| 392 | merged_attrs = []
|
| 393 |
|
| 394 | if row_td_attrs is None:
|
| 395 | row_lookup = {}
|
| 396 | else:
|
| 397 | row_lookup = {n: v for n, v in row_td_attrs}
|
| 398 |
|
| 399 | done_for_row = set()
|
| 400 |
|
| 401 | if thead_td_attrs:
|
| 402 | for name, raw_value in thead_td_attrs:
|
| 403 | more_values = row_lookup.get(name)
|
| 404 | if more_values is not None:
|
| 405 | raw_value += ' %s' % more_values
|
| 406 | done_for_row.add(name)
|
| 407 | merged_attrs.append((name, raw_value))
|
| 408 |
|
| 409 | if row_td_attrs:
|
| 410 | for name, raw_value in row_td_attrs:
|
| 411 | if name in done_for_row:
|
| 412 | continue
|
| 413 | merged_attrs.append((name, raw_value))
|
| 414 |
|
| 415 | return merged_attrs
|
| 416 |
|
| 417 |
|
| 418 | def ReplaceTables(s, debug_out=None):
|
| 419 | """
|
| 420 | ul-table: Write tables using bulleted list
|
| 421 | """
|
| 422 | if debug_out is None:
|
| 423 | debug_out = []
|
| 424 |
|
| 425 | f = StringIO()
|
| 426 | out = html.Output(s, f)
|
| 427 |
|
| 428 | tag_lexer = html.TagLexer(s)
|
| 429 | lexer = html.Lexer(s)
|
| 430 |
|
| 431 | p = UlTableParser(lexer, tag_lexer)
|
| 432 |
|
| 433 | while True:
|
| 434 | ul_start = p.FindUlTable()
|
| 435 | if ul_start == -1:
|
| 436 | break
|
| 437 |
|
| 438 | #log('UL START %d', ul_start)
|
| 439 | out.PrintUntil(ul_start)
|
| 440 |
|
| 441 | table = p.ParseTable()
|
| 442 | #log('UL END %d', ul_end)
|
| 443 |
|
| 444 | # Don't write the matching </u> of the LAST row, but write everything
|
| 445 | # after that
|
| 446 | out.SkipTo(table['ul_end'])
|
| 447 |
|
| 448 | # Write the header
|
| 449 | thead = table['thead']
|
| 450 |
|
| 451 | col_attrs = {} # integer -> td_attrs
|
| 452 | if thead:
|
| 453 | out.Print('<thead>\n')
|
| 454 | out.Print('<tr>\n')
|
| 455 |
|
| 456 | i = 0
|
| 457 | for td_attrs, raw_html in thead:
|
| 458 | if td_attrs:
|
| 459 | col_attrs[i] = td_attrs
|
| 460 | # <th> tag is more semantic, and styled bold by default
|
| 461 | out.Print(' <th>')
|
| 462 | out.Print(raw_html)
|
| 463 | out.Print('</th>\n')
|
| 464 | i += 1
|
| 465 |
|
| 466 | out.Print('</tr>\n')
|
| 467 | out.Print('</thead>\n')
|
| 468 |
|
| 469 | # Write each row
|
| 470 | for tr_attrs, row in table['tr']:
|
| 471 |
|
| 472 | # Print tr tag and attrs
|
| 473 | out.Print('<tr')
|
| 474 | if tr_attrs:
|
| 475 | for name, raw_value in tr_attrs:
|
| 476 | out.Print(' ')
|
| 477 | out.Print(name)
|
| 478 | # No escaping because it's raw. It can't contain quotes.
|
| 479 | out.Print('="%s"' % raw_value)
|
| 480 | out.Print('>\n')
|
| 481 |
|
| 482 | # Print cells
|
| 483 | i = 0
|
| 484 | for row_td_attrs, raw_html in row:
|
| 485 | # Inherited from header
|
| 486 | thead_td_attrs = col_attrs.get(i)
|
| 487 | merged_attrs = MergeAttrs(thead_td_attrs, row_td_attrs)
|
| 488 |
|
| 489 | out.Print(' <td')
|
| 490 | for name, raw_value in merged_attrs:
|
| 491 | out.Print(' ')
|
| 492 | out.Print(name)
|
| 493 | # No escaping because it's raw. It can't contain quotes.
|
| 494 | out.Print('="%s"' % raw_value)
|
| 495 | out.Print('>')
|
| 496 |
|
| 497 | out.Print(raw_html)
|
| 498 | out.Print('</td>\n')
|
| 499 | i += 1
|
| 500 | out.Print('</tr>\n')
|
| 501 |
|
| 502 | out.PrintTheRest()
|
| 503 |
|
| 504 | return f.getvalue()
|
| 505 |
|
| 506 |
|
| 507 | if __name__ == '__main__':
|
| 508 | # Simple CLI filter
|
| 509 | sys.stdout.write(ReplaceTables(sys.stdin.read()))
|