symbian-qemu-0.9.1-12/python-win32-2.6.1/lib/markupbase.py
author martin.trojer@nokia.com
Fri, 31 Jul 2009 15:01:17 +0100
changeset 1 2fb8b9db1c86
permissions -rw-r--r--
Initial QEMU (symbian-qemu-0.9.1-12) import
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     1
"""Shared support for scanning document type declarations in HTML and XHTML.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     3
This module is used as a foundation for the HTMLParser and sgmllib
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     4
modules (indirectly, for htmllib as well).  It has no documented
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     5
public API and should not be used directly.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     6
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     7
"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     8
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     9
import re
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    10
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    11
_declname_match = re.compile(r'[a-zA-Z][-_.a-zA-Z0-9]*\s*').match
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    12
_declstringlit_match = re.compile(r'(\'[^\']*\'|"[^"]*")\s*').match
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    13
_commentclose = re.compile(r'--\s*>')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    14
_markedsectionclose = re.compile(r']\s*]\s*>')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    15
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    16
# An analysis of the MS-Word extensions is available at
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    17
# http://www.planetpublish.com/xmlarena/xap/Thursday/WordtoXML.pdf
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    18
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    19
_msmarkedsectionclose = re.compile(r']\s*>')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    20
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    21
del re
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    22
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    23
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    24
class ParserBase:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    25
    """Parser base class which provides some common support methods used
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    26
    by the SGML/HTML and XHTML parsers."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    27
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    28
    def __init__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    29
        if self.__class__ is ParserBase:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    30
            raise RuntimeError(
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    31
                "markupbase.ParserBase must be subclassed")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    32
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    33
    def error(self, message):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    34
        raise NotImplementedError(
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    35
            "subclasses of ParserBase must override error()")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    36
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    37
    def reset(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    38
        self.lineno = 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    39
        self.offset = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    40
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    41
    def getpos(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    42
        """Return current line number and offset."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    43
        return self.lineno, self.offset
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    44
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    45
    # Internal -- update line number and offset.  This should be
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    46
    # called for each piece of data exactly once, in order -- in other
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    47
    # words the concatenation of all the input strings to this
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    48
    # function should be exactly the entire input.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    49
    def updatepos(self, i, j):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    50
        if i >= j:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    51
            return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    52
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    53
        nlines = rawdata.count("\n", i, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    54
        if nlines:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    55
            self.lineno = self.lineno + nlines
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    56
            pos = rawdata.rindex("\n", i, j) # Should not fail
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    57
            self.offset = j-(pos+1)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    58
        else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    59
            self.offset = self.offset + j-i
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    60
        return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    61
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    62
    _decl_otherchars = ''
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    63
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    64
    # Internal -- parse declaration (for use by subclasses).
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    65
    def parse_declaration(self, i):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    66
        # This is some sort of declaration; in "HTML as
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    67
        # deployed," this should only be the document type
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    68
        # declaration ("<!DOCTYPE html...>").
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    69
        # ISO 8879:1986, however, has more complex
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    70
        # declaration syntax for elements in <!...>, including:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    71
        # --comment--
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    72
        # [marked section]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    73
        # name in the following list: ENTITY, DOCTYPE, ELEMENT,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    74
        # ATTLIST, NOTATION, SHORTREF, USEMAP,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    75
        # LINKTYPE, LINK, IDLINK, USELINK, SYSTEM
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    76
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    77
        j = i + 2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    78
        assert rawdata[i:j] == "<!", "unexpected call to parse_declaration"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    79
        if rawdata[j:j+1] == ">":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    80
            # the empty comment <!>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    81
            return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    82
        if rawdata[j:j+1] in ("-", ""):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    83
            # Start of comment followed by buffer boundary,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    84
            # or just a buffer boundary.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    85
            return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    86
        # A simple, practical version could look like: ((name|stringlit) S*) + '>'
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    87
        n = len(rawdata)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    88
        if rawdata[j:j+2] == '--': #comment
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    89
            # Locate --.*-- as the body of the comment
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    90
            return self.parse_comment(i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    91
        elif rawdata[j] == '[': #marked section
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    92
            # Locate [statusWord [...arbitrary SGML...]] as the body of the marked section
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    93
            # Where statusWord is one of TEMP, CDATA, IGNORE, INCLUDE, RCDATA
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    94
            # Note that this is extended by Microsoft Office "Save as Web" function
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    95
            # to include [if...] and [endif].
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    96
            return self.parse_marked_section(i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    97
        else: #all other declaration elements
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    98
            decltype, j = self._scan_name(j, i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    99
        if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   100
            return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   101
        if decltype == "doctype":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   102
            self._decl_otherchars = ''
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   103
        while j < n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   104
            c = rawdata[j]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   105
            if c == ">":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   106
                # end of declaration syntax
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   107
                data = rawdata[i+2:j]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   108
                if decltype == "doctype":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   109
                    self.handle_decl(data)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   110
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   111
                    self.unknown_decl(data)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   112
                return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   113
            if c in "\"'":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   114
                m = _declstringlit_match(rawdata, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   115
                if not m:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   116
                    return -1 # incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   117
                j = m.end()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   118
            elif c in "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   119
                name, j = self._scan_name(j, i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   120
            elif c in self._decl_otherchars:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   121
                j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   122
            elif c == "[":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   123
                # this could be handled in a separate doctype parser
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   124
                if decltype == "doctype":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   125
                    j = self._parse_doctype_subset(j + 1, i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   126
                elif decltype in ("attlist", "linktype", "link", "element"):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   127
                    # must tolerate []'d groups in a content model in an element declaration
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   128
                    # also in data attribute specifications of attlist declaration
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   129
                    # also link type declaration subsets in linktype declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   130
                    # also link attribute specification lists in link declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   131
                    self.error("unsupported '[' char in %s declaration" % decltype)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   132
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   133
                    self.error("unexpected '[' char in declaration")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   134
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   135
                self.error(
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   136
                    "unexpected %r char in declaration" % rawdata[j])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   137
            if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   138
                return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   139
        return -1 # incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   140
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   141
    # Internal -- parse a marked section
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   142
    # Override this to handle MS-word extension syntax <![if word]>content<![endif]>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   143
    def parse_marked_section(self, i, report=1):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   144
        rawdata= self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   145
        assert rawdata[i:i+3] == '<![', "unexpected call to parse_marked_section()"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   146
        sectName, j = self._scan_name( i+3, i )
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   147
        if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   148
            return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   149
        if sectName in ("temp", "cdata", "ignore", "include", "rcdata"):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   150
            # look for standard ]]> ending
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   151
            match= _markedsectionclose.search(rawdata, i+3)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   152
        elif sectName in ("if", "else", "endif"):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   153
            # look for MS Office ]> ending
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   154
            match= _msmarkedsectionclose.search(rawdata, i+3)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   155
        else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   156
            self.error('unknown status keyword %r in marked section' % rawdata[i+3:j])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   157
        if not match:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   158
            return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   159
        if report:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   160
            j = match.start(0)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   161
            self.unknown_decl(rawdata[i+3: j])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   162
        return match.end(0)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   163
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   164
    # Internal -- parse comment, return length or -1 if not terminated
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   165
    def parse_comment(self, i, report=1):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   166
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   167
        if rawdata[i:i+4] != '<!--':
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   168
            self.error('unexpected call to parse_comment()')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   169
        match = _commentclose.search(rawdata, i+4)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   170
        if not match:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   171
            return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   172
        if report:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   173
            j = match.start(0)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   174
            self.handle_comment(rawdata[i+4: j])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   175
        return match.end(0)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   176
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   177
    # Internal -- scan past the internal subset in a <!DOCTYPE declaration,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   178
    # returning the index just past any whitespace following the trailing ']'.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   179
    def _parse_doctype_subset(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   180
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   181
        n = len(rawdata)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   182
        j = i
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   183
        while j < n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   184
            c = rawdata[j]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   185
            if c == "<":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   186
                s = rawdata[j:j+2]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   187
                if s == "<":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   188
                    # end of buffer; incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   189
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   190
                if s != "<!":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   191
                    self.updatepos(declstartpos, j + 1)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   192
                    self.error("unexpected char in internal subset (in %r)" % s)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   193
                if (j + 2) == n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   194
                    # end of buffer; incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   195
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   196
                if (j + 4) > n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   197
                    # end of buffer; incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   198
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   199
                if rawdata[j:j+4] == "<!--":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   200
                    j = self.parse_comment(j, report=0)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   201
                    if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   202
                        return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   203
                    continue
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   204
                name, j = self._scan_name(j + 2, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   205
                if j == -1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   206
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   207
                if name not in ("attlist", "element", "entity", "notation"):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   208
                    self.updatepos(declstartpos, j + 2)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   209
                    self.error(
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   210
                        "unknown declaration %r in internal subset" % name)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   211
                # handle the individual names
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   212
                meth = getattr(self, "_parse_doctype_" + name)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   213
                j = meth(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   214
                if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   215
                    return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   216
            elif c == "%":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   217
                # parameter entity reference
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   218
                if (j + 1) == n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   219
                    # end of buffer; incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   220
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   221
                s, j = self._scan_name(j + 1, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   222
                if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   223
                    return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   224
                if rawdata[j] == ";":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   225
                    j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   226
            elif c == "]":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   227
                j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   228
                while j < n and rawdata[j].isspace():
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   229
                    j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   230
                if j < n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   231
                    if rawdata[j] == ">":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   232
                        return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   233
                    self.updatepos(declstartpos, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   234
                    self.error("unexpected char after internal subset")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   235
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   236
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   237
            elif c.isspace():
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   238
                j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   239
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   240
                self.updatepos(declstartpos, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   241
                self.error("unexpected char %r in internal subset" % c)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   242
        # end of buffer reached
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   243
        return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   244
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   245
    # Internal -- scan past <!ELEMENT declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   246
    def _parse_doctype_element(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   247
        name, j = self._scan_name(i, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   248
        if j == -1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   249
            return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   250
        # style content model; just skip until '>'
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   251
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   252
        if '>' in rawdata[j:]:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   253
            return rawdata.find(">", j) + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   254
        return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   255
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   256
    # Internal -- scan past <!ATTLIST declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   257
    def _parse_doctype_attlist(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   258
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   259
        name, j = self._scan_name(i, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   260
        c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   261
        if c == "":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   262
            return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   263
        if c == ">":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   264
            return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   265
        while 1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   266
            # scan a series of attribute descriptions; simplified:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   267
            #   name type [value] [#constraint]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   268
            name, j = self._scan_name(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   269
            if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   270
                return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   271
            c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   272
            if c == "":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   273
                return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   274
            if c == "(":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   275
                # an enumerated type; look for ')'
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   276
                if ")" in rawdata[j:]:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   277
                    j = rawdata.find(")", j) + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   278
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   279
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   280
                while rawdata[j:j+1].isspace():
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   281
                    j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   282
                if not rawdata[j:]:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   283
                    # end of buffer, incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   284
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   285
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   286
                name, j = self._scan_name(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   287
            c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   288
            if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   289
                return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   290
            if c in "'\"":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   291
                m = _declstringlit_match(rawdata, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   292
                if m:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   293
                    j = m.end()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   294
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   295
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   296
                c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   297
                if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   298
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   299
            if c == "#":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   300
                if rawdata[j:] == "#":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   301
                    # end of buffer
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   302
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   303
                name, j = self._scan_name(j + 1, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   304
                if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   305
                    return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   306
                c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   307
                if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   308
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   309
            if c == '>':
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   310
                # all done
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   311
                return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   312
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   313
    # Internal -- scan past <!NOTATION declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   314
    def _parse_doctype_notation(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   315
        name, j = self._scan_name(i, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   316
        if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   317
            return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   318
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   319
        while 1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   320
            c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   321
            if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   322
                # end of buffer; incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   323
                return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   324
            if c == '>':
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   325
                return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   326
            if c in "'\"":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   327
                m = _declstringlit_match(rawdata, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   328
                if not m:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   329
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   330
                j = m.end()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   331
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   332
                name, j = self._scan_name(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   333
                if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   334
                    return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   335
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   336
    # Internal -- scan past <!ENTITY declarations
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   337
    def _parse_doctype_entity(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   338
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   339
        if rawdata[i:i+1] == "%":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   340
            j = i + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   341
            while 1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   342
                c = rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   343
                if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   344
                    return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   345
                if c.isspace():
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   346
                    j = j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   347
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   348
                    break
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   349
        else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   350
            j = i
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   351
        name, j = self._scan_name(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   352
        if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   353
            return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   354
        while 1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   355
            c = self.rawdata[j:j+1]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   356
            if not c:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   357
                return -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   358
            if c in "'\"":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   359
                m = _declstringlit_match(rawdata, j)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   360
                if m:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   361
                    j = m.end()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   362
                else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   363
                    return -1    # incomplete
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   364
            elif c == ">":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   365
                return j + 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   366
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   367
                name, j = self._scan_name(j, declstartpos)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   368
                if j < 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   369
                    return j
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   370
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   371
    # Internal -- scan a name token and the new position and the token, or
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   372
    # return -1 if we've reached the end of the buffer.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   373
    def _scan_name(self, i, declstartpos):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   374
        rawdata = self.rawdata
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   375
        n = len(rawdata)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   376
        if i == n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   377
            return None, -1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   378
        m = _declname_match(rawdata, i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   379
        if m:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   380
            s = m.group()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   381
            name = s.strip()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   382
            if (i + len(s)) == n:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   383
                return None, -1  # end of buffer
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   384
            return name.lower(), m.end()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   385
        else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   386
            self.updatepos(declstartpos, i)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   387
            self.error("expected name token at %r"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   388
                       % rawdata[declstartpos:declstartpos+20])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   389
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   390
    # To be overridden -- handlers for unknown objects
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   391
    def unknown_decl(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   392
        pass