source: pyyaml/trunk/lib/yaml/reader.py @ 188

Revision 188, 7.9 KB checked in by xi, 8 years ago (diff)

Fix several problems caused by ill-formed documents.

The line number is not calculated correctly for DOS-style line breaks.

Fix error reporting in remove_possible_simple_key. The problem is caused by the document:

+foo: &A bar
+*A ]

Raise an error for a complex key which is not indented correctly, for instance:

? "foo"
 : "bar"
RevLine 
[45]1# This module contains abstractions for the input stream. You don't have to
2# looks further, there are no pretty code.
3#
4# We define two classes here.
5#
[116]6#   Mark(source, line, column)
[45]7# It's just a record and its only use is producing nice error messages.
8# Parser does not use it for any other purposes.
9#
[46]10#   Reader(source, data)
11# Reader determines the encoding of `data` and converts it to unicode.
12# Reader provides the following methods and attributes:
13#   reader.peek(length=1) - return the next `length` characters
14#   reader.forward(length=1) - move the current position to `length` characters.
15#   reader.index - the number of the current character.
16#   reader.line, stream.column - the line and the column of the current character.
[43]17
[52]18__all__ = ['Reader', 'ReaderError']
[43]19
[116]20from error import YAMLError, Mark
[45]21
22import codecs, re
23
24# Unfortunately, codec functions in Python 2.3 does not support the `finish`
25# arguments, so we have to write our own wrappers.
26
27try:
28    codecs.utf_8_decode('', 'strict', False)
29    from codecs import utf_8_decode, utf_16_le_decode, utf_16_be_decode
30
31except TypeError:
32
33    def utf_16_le_decode(data, errors, finish=False):
34        if not finish and len(data) % 2 == 1:
35            data = data[:-1]
36        return codecs.utf_16_le_decode(data, errors)
37
38    def utf_16_be_decode(data, errors, finish=False):
39        if not finish and len(data) % 2 == 1:
40            data = data[:-1]
41        return codecs.utf_16_be_decode(data, errors)
42
43    def utf_8_decode(data, errors, finish=False):
44        if not finish:
45            # We are trying to remove a possible incomplete multibyte character
46            # from the suffix of the data.
47            # The first byte of a multi-byte sequence is in the range 0xc0 to 0xfd.
48            # All further bytes are in the range 0x80 to 0xbf.
49            # UTF-8 encoded UCS characters may be up to six bytes long.
50            count = 0
51            while count < 5 and count < len(data)   \
52                    and '\x80' <= data[-count-1] <= '\xBF':
53                count -= 1
54            if count < 5 and count < len(data)  \
55                    and '\xC0' <= data[-count-1] <= '\xFD':
56                data = data[:-count-1]
57        return codecs.utf_8_decode(data, errors)
58
[46]59class ReaderError(YAMLError):
[45]60
[46]61    def __init__(self, name, position, character, encoding, reason):
62        self.name = name
[45]63        self.character = character
64        self.position = position
[46]65        self.encoding = encoding
[45]66        self.reason = reason
67
68    def __str__(self):
69        if isinstance(self.character, str):
70            return "'%s' codec can't decode byte #x%02x: %s\n"  \
[47]71                    "  in \"%s\", position %d"    \
[45]72                    % (self.encoding, ord(self.character), self.reason,
[46]73                            self.name, self.position)
[45]74        else:
75            return "unacceptable character #x%04x: %s\n"    \
[47]76                    "  in \"%s\", position %d"    \
[45]77                    % (ord(self.character), self.reason,
[46]78                            self.name, self.position)
[45]79
[46]80class Reader:
81    # Reader:
[45]82    # - determines the data encoding and converts it to unicode,
83    # - checks if characters are in allowed range,
84    # - adds '\0' to the end.
[43]85
[46]86    # Reader accepts
87    #  - a `str` object,
88    #  - a `unicode` object,
89    #  - a file-like object with its `read` method returning `str`,
90    #  - a file-like object with its `read` method returning `unicode`.
91
[45]92    # Yeah, it's ugly and slow.
93
[136]94    def __init__(self, stream):
[46]95        self.name = None
[45]96        self.stream = None
97        self.stream_pointer = 0
98        self.eof = True
99        self.buffer = u''
100        self.pointer = 0
101        self.raw_buffer = None
[46]102        self.raw_decode = None
[130]103        self.encoding = None
[43]104        self.index = 0
105        self.line = 0
106        self.column = 0
[136]107        if isinstance(stream, unicode):
[46]108            self.name = "<unicode string>"
[136]109            self.check_printable(stream)
110            self.buffer = stream+u'\0'
111        elif isinstance(stream, str):
[46]112            self.name = "<string>"
[136]113            self.raw_buffer = stream
[45]114            self.determine_encoding()
115        else:
[136]116            self.stream = stream
117            self.name = getattr(stream, 'name', "<file>")
[45]118            self.eof = False
119            self.raw_buffer = ''
120            self.determine_encoding()
[43]121
[48]122    def peek(self, index=0):
[173]123        try:
124            return self.buffer[self.pointer+index]
125        except IndexError:
[48]126            self.update(index+1)
[173]127            return self.buffer[self.pointer+index]
[48]128
129    def prefix(self, length=1):
[45]130        if self.pointer+length >= len(self.buffer):
131            self.update(length)
132        return self.buffer[self.pointer:self.pointer+length]
[43]133
[45]134    def forward(self, length=1):
135        if self.pointer+length+1 >= len(self.buffer):
136            self.update(length+1)
[173]137        while length:
[45]138            ch = self.buffer[self.pointer]
139            self.pointer += 1
140            self.index += 1
141            if ch in u'\n\x85\u2028\u2029'  \
[188]142                    or (ch == u'\r' and self.buffer[self.pointer] != u'\n'):
[43]143                self.line += 1
144                self.column = 0
[45]145            elif ch != u'\uFEFF':
[43]146                self.column += 1
[173]147            length -= 1
[43]148
[116]149    def get_mark(self):
[45]150        if self.stream is None:
[119]151            return Mark(self.name, self.index, self.line, self.column,
[45]152                    self.buffer, self.pointer)
153        else:
[119]154            return Mark(self.name, self.index, self.line, self.column,
155                    None, None)
[43]156
[45]157    def determine_encoding(self):
158        while not self.eof and len(self.raw_buffer) < 2:
159            self.update_raw()
[46]160        if not isinstance(self.raw_buffer, unicode):
161            if self.raw_buffer.startswith(codecs.BOM_UTF16_LE):
162                self.raw_decode = utf_16_le_decode
[130]163                self.encoding = 'utf-16-le'
[46]164            elif self.raw_buffer.startswith(codecs.BOM_UTF16_BE):
165                self.raw_decode = utf_16_be_decode
[130]166                self.encoding = 'utf-16-be'
[46]167            else:
168                self.raw_decode = utf_8_decode
[130]169                self.encoding = 'utf-8'
[45]170        self.update(1)
171
172    NON_PRINTABLE = re.compile(u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uD7FF\uE000-\uFFFD]')
173    def check_printable(self, data):
174        match = self.NON_PRINTABLE.search(data)
175        if match:
176            character = match.group()
177            position = self.index+(len(self.buffer)-self.pointer)+match.start()
[46]178            raise ReaderError(self.name, position, character,
179                    'unicode', "special characters are not allowed")
[45]180
181    def update(self, length):
182        if self.raw_buffer is None:
183            return
184        self.buffer = self.buffer[self.pointer:]
185        self.pointer = 0
186        while len(self.buffer) < length:
187            if not self.eof:
188                self.update_raw()
[46]189            if self.raw_decode is not None:
190                try:
191                    data, converted = self.raw_decode(self.raw_buffer,
192                            'strict', self.eof)
193                except UnicodeDecodeError, exc:
194                    character = exc.object[exc.start]
195                    if self.stream is not None:
196                        position = self.stream_pointer-len(self.raw_buffer)+exc.start
197                    else:
198                        position = exc.start
199                    raise ReaderError(self.name, position, character,
200                            exc.encoding, exc.reason)
201            else:
202                data = self.raw_buffer
203                converted = len(data)
[45]204            self.check_printable(data)
205            self.buffer += data
206            self.raw_buffer = self.raw_buffer[converted:]
207            if self.eof:
208                self.buffer += u'\0'
209                self.raw_buffer = None
210                break
211
212    def update_raw(self, size=1024):
213        data = self.stream.read(size)
214        if data:
215            self.raw_buffer += data
216            self.stream_pointer += len(data)
217        else:
218            self.eof = True
219
220#try:
221#    import psyco
[46]222#    psyco.bind(Reader)
[45]223#except ImportError:
224#    pass
225
Note: See TracBrowser for help on using the repository browser.