jpayne@68: """A parser for HTML and XHTML."""
jpayne@68: 
jpayne@68: # This file is based on sgmllib.py, but the API is slightly different.
jpayne@68: 
jpayne@68: # XXX There should be a way to distinguish between PCDATA (parsed
jpayne@68: # character data -- the normal case), RCDATA (replaceable character
jpayne@68: # data -- only char and entity references and end tags are special)
jpayne@68: # and CDATA (character data -- only end tags are special).
jpayne@68: 
jpayne@68: 
jpayne@68: import re
jpayne@68: import warnings
jpayne@68: import _markupbase
jpayne@68: 
jpayne@68: from html import unescape
jpayne@68: 
jpayne@68: 
jpayne@68: __all__ = ['HTMLParser']
jpayne@68: 
jpayne@68: # Regular expressions used for parsing
jpayne@68: 
jpayne@68: interesting_normal = re.compile('[&<]')
jpayne@68: incomplete = re.compile('&[a-zA-Z#]')
jpayne@68: 
jpayne@68: entityref = re.compile('&([a-zA-Z][-.a-zA-Z0-9]*)[^a-zA-Z0-9]')
jpayne@68: charref = re.compile('&#(?:[0-9]+|[xX][0-9a-fA-F]+)[^0-9a-fA-F]')
jpayne@68: 
jpayne@68: starttagopen = re.compile('<[a-zA-Z]')
jpayne@68: piclose = re.compile('>')
jpayne@68: commentclose = re.compile(r'--\s*>')
jpayne@68: # Note:
jpayne@68: #  1) if you change tagfind/attrfind remember to update locatestarttagend too;
jpayne@68: #  2) if you change tagfind/attrfind and/or locatestarttagend the parser will
jpayne@68: #     explode, so don't do it.
jpayne@68: # see http://www.w3.org/TR/html5/tokenization.html#tag-open-state
jpayne@68: # and http://www.w3.org/TR/html5/tokenization.html#tag-name-state
jpayne@68: tagfind_tolerant = re.compile(r'([a-zA-Z][^\t\n\r\f />\x00]*)(?:\s|/(?!>))*')
jpayne@68: attrfind_tolerant = re.compile(
jpayne@68:     r'((?<=[\'"\s/])[^\s/>][^\s/=>]*)(\s*=+\s*'
jpayne@68:     r'(\'[^\']*\'|"[^"]*"|(?![\'"])[^>\s]*))?(?:\s|/(?!>))*')
jpayne@68: locatestarttagend_tolerant = re.compile(r"""
jpayne@68:   <[a-zA-Z][^\t\n\r\f />\x00]*       # tag name
jpayne@68:   (?:[\s/]*                          # optional whitespace before attribute name
jpayne@68:     (?:(?<=['"\s/])[^\s/>][^\s/=>]*  # attribute name
jpayne@68:       (?:\s*=+\s*                    # value indicator
jpayne@68:         (?:'[^']*'                   # LITA-enclosed value
jpayne@68:           |"[^"]*"                   # LIT-enclosed value
jpayne@68:           |(?!['"])[^>\s]*           # bare value
jpayne@68:          )
jpayne@68:          (?:\s*,)*                   # possibly followed by a comma
jpayne@68:        )?(?:\s|/(?!>))*
jpayne@68:      )*
jpayne@68:    )?
jpayne@68:   \s*                                # trailing whitespace
jpayne@68: """, re.VERBOSE)
jpayne@68: endendtag = re.compile('>')
jpayne@68: # the HTML 5 spec, section 8.1.2.2, doesn't allow spaces between
jpayne@68: # </ and the tag name, so maybe this should be fixed
jpayne@68: endtagfind = re.compile(r'</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>')
jpayne@68: 
jpayne@68: 
jpayne@68: 
jpayne@68: class HTMLParser(_markupbase.ParserBase):
jpayne@68:     """Find tags and other markup and call handler functions.
jpayne@68: 
jpayne@68:     Usage:
jpayne@68:         p = HTMLParser()
jpayne@68:         p.feed(data)
jpayne@68:         ...
jpayne@68:         p.close()
jpayne@68: 
jpayne@68:     Start tags are handled by calling self.handle_starttag() or
jpayne@68:     self.handle_startendtag(); end tags by self.handle_endtag().  The
jpayne@68:     data between tags is passed from the parser to the derived class
jpayne@68:     by calling self.handle_data() with the data as argument (the data
jpayne@68:     may be split up in arbitrary chunks).  If convert_charrefs is
jpayne@68:     True the character references are converted automatically to the
jpayne@68:     corresponding Unicode character (and self.handle_data() is no
jpayne@68:     longer split in chunks), otherwise they are passed by calling
jpayne@68:     self.handle_entityref() or self.handle_charref() with the string
jpayne@68:     containing respectively the named or numeric reference as the
jpayne@68:     argument.
jpayne@68:     """
jpayne@68: 
jpayne@68:     CDATA_CONTENT_ELEMENTS = ("script", "style")
jpayne@68: 
jpayne@68:     def __init__(self, *, convert_charrefs=True):
jpayne@68:         """Initialize and reset this instance.
jpayne@68: 
jpayne@68:         If convert_charrefs is True (the default), all character references
jpayne@68:         are automatically converted to the corresponding Unicode characters.
jpayne@68:         """
jpayne@68:         self.convert_charrefs = convert_charrefs
jpayne@68:         self.reset()
jpayne@68: 
jpayne@68:     def reset(self):
jpayne@68:         """Reset this instance.  Loses all unprocessed data."""
jpayne@68:         self.rawdata = ''
jpayne@68:         self.lasttag = '???'
jpayne@68:         self.interesting = interesting_normal
jpayne@68:         self.cdata_elem = None
jpayne@68:         _markupbase.ParserBase.reset(self)
jpayne@68: 
jpayne@68:     def feed(self, data):
jpayne@68:         r"""Feed data to the parser.
jpayne@68: 
jpayne@68:         Call this as often as you want, with as little or as much text
jpayne@68:         as you want (may include '\n').
jpayne@68:         """
jpayne@68:         self.rawdata = self.rawdata + data
jpayne@68:         self.goahead(0)
jpayne@68: 
jpayne@68:     def close(self):
jpayne@68:         """Handle any buffered data."""
jpayne@68:         self.goahead(1)
jpayne@68: 
jpayne@68:     __starttag_text = None
jpayne@68: 
jpayne@68:     def get_starttag_text(self):
jpayne@68:         """Return full source of start tag: '<...>'."""
jpayne@68:         return self.__starttag_text
jpayne@68: 
jpayne@68:     def set_cdata_mode(self, elem):
jpayne@68:         self.cdata_elem = elem.lower()
jpayne@68:         self.interesting = re.compile(r'</\s*%s\s*>' % self.cdata_elem, re.I)
jpayne@68: 
jpayne@68:     def clear_cdata_mode(self):
jpayne@68:         self.interesting = interesting_normal
jpayne@68:         self.cdata_elem = None
jpayne@68: 
jpayne@68:     # Internal -- handle data as far as reasonable.  May leave state
jpayne@68:     # and data to be processed by a subsequent call.  If 'end' is
jpayne@68:     # true, force handling all data as if followed by EOF marker.
jpayne@68:     def goahead(self, end):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         i = 0
jpayne@68:         n = len(rawdata)
jpayne@68:         while i < n:
jpayne@68:             if self.convert_charrefs and not self.cdata_elem:
jpayne@68:                 j = rawdata.find('<', i)
jpayne@68:                 if j < 0:
jpayne@68:                     # if we can't find the next <, either we are at the end
jpayne@68:                     # or there's more text incoming.  If the latter is True,
jpayne@68:                     # we can't pass the text to handle_data in case we have
jpayne@68:                     # a charref cut in half at end.  Try to determine if
jpayne@68:                     # this is the case before proceeding by looking for an
jpayne@68:                     # & near the end and see if it's followed by a space or ;.
jpayne@68:                     amppos = rawdata.rfind('&', max(i, n-34))
jpayne@68:                     if (amppos >= 0 and
jpayne@68:                         not re.compile(r'[\s;]').search(rawdata, amppos)):
jpayne@68:                         break  # wait till we get all the text
jpayne@68:                     j = n
jpayne@68:             else:
jpayne@68:                 match = self.interesting.search(rawdata, i)  # < or &
jpayne@68:                 if match:
jpayne@68:                     j = match.start()
jpayne@68:                 else:
jpayne@68:                     if self.cdata_elem:
jpayne@68:                         break
jpayne@68:                     j = n
jpayne@68:             if i < j:
jpayne@68:                 if self.convert_charrefs and not self.cdata_elem:
jpayne@68:                     self.handle_data(unescape(rawdata[i:j]))
jpayne@68:                 else:
jpayne@68:                     self.handle_data(rawdata[i:j])
jpayne@68:             i = self.updatepos(i, j)
jpayne@68:             if i == n: break
jpayne@68:             startswith = rawdata.startswith
jpayne@68:             if startswith('<', i):
jpayne@68:                 if starttagopen.match(rawdata, i): # < + letter
jpayne@68:                     k = self.parse_starttag(i)
jpayne@68:                 elif startswith("</", i):
jpayne@68:                     k = self.parse_endtag(i)
jpayne@68:                 elif startswith("<!--", i):
jpayne@68:                     k = self.parse_comment(i)
jpayne@68:                 elif startswith("<?", i):
jpayne@68:                     k = self.parse_pi(i)
jpayne@68:                 elif startswith("<!", i):
jpayne@68:                     k = self.parse_html_declaration(i)
jpayne@68:                 elif (i + 1) < n:
jpayne@68:                     self.handle_data("<")
jpayne@68:                     k = i + 1
jpayne@68:                 else:
jpayne@68:                     break
jpayne@68:                 if k < 0:
jpayne@68:                     if not end:
jpayne@68:                         break
jpayne@68:                     k = rawdata.find('>', i + 1)
jpayne@68:                     if k < 0:
jpayne@68:                         k = rawdata.find('<', i + 1)
jpayne@68:                         if k < 0:
jpayne@68:                             k = i + 1
jpayne@68:                     else:
jpayne@68:                         k += 1
jpayne@68:                     if self.convert_charrefs and not self.cdata_elem:
jpayne@68:                         self.handle_data(unescape(rawdata[i:k]))
jpayne@68:                     else:
jpayne@68:                         self.handle_data(rawdata[i:k])
jpayne@68:                 i = self.updatepos(i, k)
jpayne@68:             elif startswith("&#", i):
jpayne@68:                 match = charref.match(rawdata, i)
jpayne@68:                 if match:
jpayne@68:                     name = match.group()[2:-1]
jpayne@68:                     self.handle_charref(name)
jpayne@68:                     k = match.end()
jpayne@68:                     if not startswith(';', k-1):
jpayne@68:                         k = k - 1
jpayne@68:                     i = self.updatepos(i, k)
jpayne@68:                     continue
jpayne@68:                 else:
jpayne@68:                     if ";" in rawdata[i:]:  # bail by consuming &#
jpayne@68:                         self.handle_data(rawdata[i:i+2])
jpayne@68:                         i = self.updatepos(i, i+2)
jpayne@68:                     break
jpayne@68:             elif startswith('&', i):
jpayne@68:                 match = entityref.match(rawdata, i)
jpayne@68:                 if match:
jpayne@68:                     name = match.group(1)
jpayne@68:                     self.handle_entityref(name)
jpayne@68:                     k = match.end()
jpayne@68:                     if not startswith(';', k-1):
jpayne@68:                         k = k - 1
jpayne@68:                     i = self.updatepos(i, k)
jpayne@68:                     continue
jpayne@68:                 match = incomplete.match(rawdata, i)
jpayne@68:                 if match:
jpayne@68:                     # match.group() will contain at least 2 chars
jpayne@68:                     if end and match.group() == rawdata[i:]:
jpayne@68:                         k = match.end()
jpayne@68:                         if k <= i:
jpayne@68:                             k = n
jpayne@68:                         i = self.updatepos(i, i + 1)
jpayne@68:                     # incomplete
jpayne@68:                     break
jpayne@68:                 elif (i + 1) < n:
jpayne@68:                     # not the end of the buffer, and can't be confused
jpayne@68:                     # with some other construct
jpayne@68:                     self.handle_data("&")
jpayne@68:                     i = self.updatepos(i, i + 1)
jpayne@68:                 else:
jpayne@68:                     break
jpayne@68:             else:
jpayne@68:                 assert 0, "interesting.search() lied"
jpayne@68:         # end while
jpayne@68:         if end and i < n and not self.cdata_elem:
jpayne@68:             if self.convert_charrefs and not self.cdata_elem:
jpayne@68:                 self.handle_data(unescape(rawdata[i:n]))
jpayne@68:             else:
jpayne@68:                 self.handle_data(rawdata[i:n])
jpayne@68:             i = self.updatepos(i, n)
jpayne@68:         self.rawdata = rawdata[i:]
jpayne@68: 
jpayne@68:     # Internal -- parse html declarations, return length or -1 if not terminated
jpayne@68:     # See w3.org/TR/html5/tokenization.html#markup-declaration-open-state
jpayne@68:     # See also parse_declaration in _markupbase
jpayne@68:     def parse_html_declaration(self, i):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         assert rawdata[i:i+2] == '<!', ('unexpected call to '
jpayne@68:                                         'parse_html_declaration()')
jpayne@68:         if rawdata[i:i+4] == '<!--':
jpayne@68:             # this case is actually already handled in goahead()
jpayne@68:             return self.parse_comment(i)
jpayne@68:         elif rawdata[i:i+3] == '<![':
jpayne@68:             return self.parse_marked_section(i)
jpayne@68:         elif rawdata[i:i+9].lower() == '<!doctype':
jpayne@68:             # find the closing >
jpayne@68:             gtpos = rawdata.find('>', i+9)
jpayne@68:             if gtpos == -1:
jpayne@68:                 return -1
jpayne@68:             self.handle_decl(rawdata[i+2:gtpos])
jpayne@68:             return gtpos+1
jpayne@68:         else:
jpayne@68:             return self.parse_bogus_comment(i)
jpayne@68: 
jpayne@68:     # Internal -- parse bogus comment, return length or -1 if not terminated
jpayne@68:     # see http://www.w3.org/TR/html5/tokenization.html#bogus-comment-state
jpayne@68:     def parse_bogus_comment(self, i, report=1):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         assert rawdata[i:i+2] in ('<!', '</'), ('unexpected call to '
jpayne@68:                                                 'parse_comment()')
jpayne@68:         pos = rawdata.find('>', i+2)
jpayne@68:         if pos == -1:
jpayne@68:             return -1
jpayne@68:         if report:
jpayne@68:             self.handle_comment(rawdata[i+2:pos])
jpayne@68:         return pos + 1
jpayne@68: 
jpayne@68:     # Internal -- parse processing instr, return end or -1 if not terminated
jpayne@68:     def parse_pi(self, i):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         assert rawdata[i:i+2] == '<?', 'unexpected call to parse_pi()'
jpayne@68:         match = piclose.search(rawdata, i+2) # >
jpayne@68:         if not match:
jpayne@68:             return -1
jpayne@68:         j = match.start()
jpayne@68:         self.handle_pi(rawdata[i+2: j])
jpayne@68:         j = match.end()
jpayne@68:         return j
jpayne@68: 
jpayne@68:     # Internal -- handle starttag, return end or -1 if not terminated
jpayne@68:     def parse_starttag(self, i):
jpayne@68:         self.__starttag_text = None
jpayne@68:         endpos = self.check_for_whole_start_tag(i)
jpayne@68:         if endpos < 0:
jpayne@68:             return endpos
jpayne@68:         rawdata = self.rawdata
jpayne@68:         self.__starttag_text = rawdata[i:endpos]
jpayne@68: 
jpayne@68:         # Now parse the data between i+1 and j into a tag and attrs
jpayne@68:         attrs = []
jpayne@68:         match = tagfind_tolerant.match(rawdata, i+1)
jpayne@68:         assert match, 'unexpected call to parse_starttag()'
jpayne@68:         k = match.end()
jpayne@68:         self.lasttag = tag = match.group(1).lower()
jpayne@68:         while k < endpos:
jpayne@68:             m = attrfind_tolerant.match(rawdata, k)
jpayne@68:             if not m:
jpayne@68:                 break
jpayne@68:             attrname, rest, attrvalue = m.group(1, 2, 3)
jpayne@68:             if not rest:
jpayne@68:                 attrvalue = None
jpayne@68:             elif attrvalue[:1] == '\'' == attrvalue[-1:] or \
jpayne@68:                  attrvalue[:1] == '"' == attrvalue[-1:]:
jpayne@68:                 attrvalue = attrvalue[1:-1]
jpayne@68:             if attrvalue:
jpayne@68:                 attrvalue = unescape(attrvalue)
jpayne@68:             attrs.append((attrname.lower(), attrvalue))
jpayne@68:             k = m.end()
jpayne@68: 
jpayne@68:         end = rawdata[k:endpos].strip()
jpayne@68:         if end not in (">", "/>"):
jpayne@68:             lineno, offset = self.getpos()
jpayne@68:             if "\n" in self.__starttag_text:
jpayne@68:                 lineno = lineno + self.__starttag_text.count("\n")
jpayne@68:                 offset = len(self.__starttag_text) \
jpayne@68:                          - self.__starttag_text.rfind("\n")
jpayne@68:             else:
jpayne@68:                 offset = offset + len(self.__starttag_text)
jpayne@68:             self.handle_data(rawdata[i:endpos])
jpayne@68:             return endpos
jpayne@68:         if end.endswith('/>'):
jpayne@68:             # XHTML-style empty tag: <span attr="value" />
jpayne@68:             self.handle_startendtag(tag, attrs)
jpayne@68:         else:
jpayne@68:             self.handle_starttag(tag, attrs)
jpayne@68:             if tag in self.CDATA_CONTENT_ELEMENTS:
jpayne@68:                 self.set_cdata_mode(tag)
jpayne@68:         return endpos
jpayne@68: 
jpayne@68:     # Internal -- check to see if we have a complete starttag; return end
jpayne@68:     # or -1 if incomplete.
jpayne@68:     def check_for_whole_start_tag(self, i):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         m = locatestarttagend_tolerant.match(rawdata, i)
jpayne@68:         if m:
jpayne@68:             j = m.end()
jpayne@68:             next = rawdata[j:j+1]
jpayne@68:             if next == ">":
jpayne@68:                 return j + 1
jpayne@68:             if next == "/":
jpayne@68:                 if rawdata.startswith("/>", j):
jpayne@68:                     return j + 2
jpayne@68:                 if rawdata.startswith("/", j):
jpayne@68:                     # buffer boundary
jpayne@68:                     return -1
jpayne@68:                 # else bogus input
jpayne@68:                 if j > i:
jpayne@68:                     return j
jpayne@68:                 else:
jpayne@68:                     return i + 1
jpayne@68:             if next == "":
jpayne@68:                 # end of input
jpayne@68:                 return -1
jpayne@68:             if next in ("abcdefghijklmnopqrstuvwxyz=/"
jpayne@68:                         "ABCDEFGHIJKLMNOPQRSTUVWXYZ"):
jpayne@68:                 # end of input in or before attribute value, or we have the
jpayne@68:                 # '/' from a '/>' ending
jpayne@68:                 return -1
jpayne@68:             if j > i:
jpayne@68:                 return j
jpayne@68:             else:
jpayne@68:                 return i + 1
jpayne@68:         raise AssertionError("we should not get here!")
jpayne@68: 
jpayne@68:     # Internal -- parse endtag, return end or -1 if incomplete
jpayne@68:     def parse_endtag(self, i):
jpayne@68:         rawdata = self.rawdata
jpayne@68:         assert rawdata[i:i+2] == "</", "unexpected call to parse_endtag"
jpayne@68:         match = endendtag.search(rawdata, i+1) # >
jpayne@68:         if not match:
jpayne@68:             return -1
jpayne@68:         gtpos = match.end()
jpayne@68:         match = endtagfind.match(rawdata, i) # </ + tag + >
jpayne@68:         if not match:
jpayne@68:             if self.cdata_elem is not None:
jpayne@68:                 self.handle_data(rawdata[i:gtpos])
jpayne@68:                 return gtpos
jpayne@68:             # find the name: w3.org/TR/html5/tokenization.html#tag-name-state
jpayne@68:             namematch = tagfind_tolerant.match(rawdata, i+2)
jpayne@68:             if not namematch:
jpayne@68:                 # w3.org/TR/html5/tokenization.html#end-tag-open-state
jpayne@68:                 if rawdata[i:i+3] == '</>':
jpayne@68:                     return i+3
jpayne@68:                 else:
jpayne@68:                     return self.parse_bogus_comment(i)
jpayne@68:             tagname = namematch.group(1).lower()
jpayne@68:             # consume and ignore other stuff between the name and the >
jpayne@68:             # Note: this is not 100% correct, since we might have things like
jpayne@68:             # </tag attr=">">, but looking for > after tha name should cover
jpayne@68:             # most of the cases and is much simpler
jpayne@68:             gtpos = rawdata.find('>', namematch.end())
jpayne@68:             self.handle_endtag(tagname)
jpayne@68:             return gtpos+1
jpayne@68: 
jpayne@68:         elem = match.group(1).lower() # script or style
jpayne@68:         if self.cdata_elem is not None:
jpayne@68:             if elem != self.cdata_elem:
jpayne@68:                 self.handle_data(rawdata[i:gtpos])
jpayne@68:                 return gtpos
jpayne@68: 
jpayne@68:         self.handle_endtag(elem)
jpayne@68:         self.clear_cdata_mode()
jpayne@68:         return gtpos
jpayne@68: 
jpayne@68:     # Overridable -- finish processing of start+end tag: <tag.../>
jpayne@68:     def handle_startendtag(self, tag, attrs):
jpayne@68:         self.handle_starttag(tag, attrs)
jpayne@68:         self.handle_endtag(tag)
jpayne@68: 
jpayne@68:     # Overridable -- handle start tag
jpayne@68:     def handle_starttag(self, tag, attrs):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle end tag
jpayne@68:     def handle_endtag(self, tag):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle character reference
jpayne@68:     def handle_charref(self, name):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle entity reference
jpayne@68:     def handle_entityref(self, name):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle data
jpayne@68:     def handle_data(self, data):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle comment
jpayne@68:     def handle_comment(self, data):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle declaration
jpayne@68:     def handle_decl(self, decl):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Overridable -- handle processing instruction
jpayne@68:     def handle_pi(self, data):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     def unknown_decl(self, data):
jpayne@68:         pass
jpayne@68: 
jpayne@68:     # Internal -- helper to remove special character quoting
jpayne@68:     def unescape(self, s):
jpayne@68:         warnings.warn('The unescape method is deprecated and will be removed '
jpayne@68:                       'in 3.5, use html.unescape() instead.',
jpayne@68:                       DeprecationWarning, stacklevel=2)
jpayne@68:         return unescape(s)