Source Code for Module lxml.html.defs

  1  # FIXME: this should all be confirmed against what a DTD says 
  2  # (probably in a test; this may not match the DTD exactly, but we 
  3  # should document just how it differs). 
  4   
  5  # Data taken from http://www.w3.org/TR/html401/index/elements.html 
  6   
  7  try: 
  8      frozenset 
  9  except NameError: 
 10      from sets import Set as frozenset 
 11   
 12   
 13  empty_tags = frozenset([ 
 14      'area', 'base', 'basefont', 'br', 'col', 'frame', 'hr', 
 15      'img', 'input', 'isindex', 'link', 'meta', 'param']) 
 16   
 17  deprecated_tags = frozenset([ 
 18      'applet', 'basefont', 'center', 'dir', 'font', 'isindex', 
 19      'menu', 's', 'strike', 'u']) 
 20   
 21  # archive actually takes a space-separated list of URIs 
 22  link_attrs = frozenset([ 
 23      'action', 'archive', 'background', 'cite', 'classid', 
 24      'codebase', 'data', 'href', 'longdesc', 'profile', 'src', 
 25      'usemap', 
 26      # Not standard: 
 27      'dynsrc', 'lowsrc', 
 28      ]) 
 29   
 30  # Not in the HTML 4 spec: 
 31  # onerror, onresize 
 32  event_attrs = frozenset([ 
 33      'onblur', 'onchange', 'onclick', 'ondblclick', 'onerror', 
 34      'onfocus', 'onkeydown', 'onkeypress', 'onkeyup', 'onload', 
 35      'onmousedown', 'onmousemove', 'onmouseout', 'onmouseover', 
 36      'onmouseup', 'onreset', 'onresize', 'onselect', 'onsubmit', 
 37      'onunload', 
 38      ]) 
 39   
 40  safe_attrs = frozenset([ 
 41      'abbr', 'accept', 'accept-charset', 'accesskey', 'action', 'align', 
 42      'alt', 'axis', 'border', 'cellpadding', 'cellspacing', 'char', 'charoff', 
 43      'charset', 'checked', 'cite', 'class', 'clear', 'cols', 'colspan', 
 44      'color', 'compact', 'coords', 'datetime', 'dir', 'disabled', 'enctype', 
 45      'for', 'frame', 'headers', 'height', 'href', 'hreflang', 'hspace', 'id', 
 46      'ismap', 'label', 'lang', 'longdesc', 'maxlength', 'media', 'method', 
 47      'multiple', 'name', 'nohref', 'noshade', 'nowrap', 'prompt', 'readonly', 
 48      'rel', 'rev', 'rows', 'rowspan', 'rules', 'scope', 'selected', 'shape', 
 49      'size', 'span', 'src', 'start', 'summary', 'tabindex', 'target', 'title', 
 50      'type', 'usemap', 'valign', 'value', 'vspace', 'width']) 
 51   
 52  # From http://htmlhelp.com/reference/html40/olist.html 
 53  top_level_tags = frozenset([ 
 54      'html', 'head', 'body', 'frameset', 
 55      ]) 
 56   
 57  head_tags = frozenset([ 
 58      'base', 'isindex', 'link', 'meta', 'script', 'style', 'title', 
 59      ]) 
 60   
 61  general_block_tags = frozenset([ 
 62      'address', 
 63      'blockquote', 
 64      'center', 
 65      'del', 
 66      'div', 
 67      'h1', 
 68      'h2', 
 69      'h3', 
 70      'h4', 
 71      'h5', 
 72      'h6', 
 73      'hr', 
 74      'ins', 
 75      'isindex', 
 76      'noscript', 
 77      'p', 
 78      'pre', 
 79      ]) 
 80   
 81  list_tags = frozenset([ 
 82      'dir', 'dl', 'dt', 'dd', 'li', 'menu', 'ol', 'ul', 
 83      ]) 
 84   
 85  table_tags = frozenset([ 
 86      'table', 'caption', 'colgroup', 'col', 
 87      'thead', 'tfoot', 'tbody', 'tr', 'td', 'th', 
 88      ]) 
 89   
 90  # just this one from 
 91  # http://www.georgehernandez.com/h/XComputers/HTML/2BlockLevel.htm 
 92  block_tags = general_block_tags | list_tags | table_tags | frozenset([ 
 93      # Partial form tags 
 94      'fieldset', 'form', 'legend', 'optgroup', 'option', 
 95      ]) 
 96   
 97  form_tags = frozenset([ 
 98      'form', 'button', 'fieldset', 'legend', 'input', 'label', 
 99      'select', 'optgroup', 'option', 'textarea', 
100      ]) 
101   
102  special_inline_tags = frozenset([ 
103      'a', 'applet', 'basefont', 'bdo', 'br', 'embed', 'font', 'iframe', 
104      'img', 'map', 'area', 'object', 'param', 'q', 'script', 
105      'span', 'sub', 'sup', 
106      ]) 
107   
108  phrase_tags = frozenset([ 
109      'abbr', 'acronym', 'cite', 'code', 'del', 'dfn', 'em', 
110      'ins', 'kbd', 'samp', 'strong', 'var', 
111      ]) 
112   
113  font_style_tags = frozenset([ 
114      'b', 'big', 'i', 's', 'small', 'strike', 'tt', 'u', 
115      ]) 
116   
117  frame_tags = frozenset([ 
118      'frameset', 'frame', 'noframes', 
119      ]) 
120   
121  # These tags aren't standard 
122  nonstandard_tags = frozenset(['blink', 'marque']) 
123   
124  tags = (top_level_tags | head_tags | general_block_tags | list_tags 
125          | table_tags | form_tags | special_inline_tags | phrase_tags 
126          | font_style_tags | nonstandard_tags) 
127