source: pyyaml/trunk/lib/yaml/reader.py @ 136

Revision 136, 7.9 KB checked in by xi, 9 years ago (diff)

Major refactoring.

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):
123        if self.pointer+index+1 >= len(self.buffer):
124            self.update(index+1)
125        return self.buffer[self.pointer+index]
126
127    def prefix(self, length=1):
[45]128        if self.pointer+length >= len(self.buffer):
129            self.update(length)
130        return self.buffer[self.pointer:self.pointer+length]
[43]131
[45]132    def forward(self, length=1):
133        if self.pointer+length+1 >= len(self.buffer):
134            self.update(length+1)
135        for k in range(length):
136            ch = self.buffer[self.pointer]
137            self.pointer += 1
138            self.index += 1
139            if ch in u'\n\x85\u2028\u2029'  \
140                    or (ch == u'\r' and self.buffer[self.pointer+1] != u'\n'):
[43]141                self.line += 1
142                self.column = 0
[45]143            elif ch != u'\uFEFF':
[43]144                self.column += 1
145
[116]146    def get_mark(self):
[45]147        if self.stream is None:
[119]148            return Mark(self.name, self.index, self.line, self.column,
[45]149                    self.buffer, self.pointer)
150        else:
[119]151            return Mark(self.name, self.index, self.line, self.column,
152                    None, None)
[43]153
[45]154    def determine_encoding(self):
155        while not self.eof and len(self.raw_buffer) < 2:
156            self.update_raw()
[46]157        if not isinstance(self.raw_buffer, unicode):
158            if self.raw_buffer.startswith(codecs.BOM_UTF16_LE):
159                self.raw_decode = utf_16_le_decode
[130]160                self.encoding = 'utf-16-le'
[46]161            elif self.raw_buffer.startswith(codecs.BOM_UTF16_BE):
162                self.raw_decode = utf_16_be_decode
[130]163                self.encoding = 'utf-16-be'
[46]164            else:
165                self.raw_decode = utf_8_decode
[130]166                self.encoding = 'utf-8'
[45]167        self.update(1)
168
169    NON_PRINTABLE = re.compile(u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uD7FF\uE000-\uFFFD]')
170    def check_printable(self, data):
171        match = self.NON_PRINTABLE.search(data)
172        if match:
173            character = match.group()
174            position = self.index+(len(self.buffer)-self.pointer)+match.start()
[46]175            raise ReaderError(self.name, position, character,
176                    'unicode', "special characters are not allowed")
[45]177
178    def update(self, length):
179        if self.raw_buffer is None:
180            return
181        self.buffer = self.buffer[self.pointer:]
182        self.pointer = 0
183        while len(self.buffer) < length:
184            if not self.eof:
185                self.update_raw()
[46]186            if self.raw_decode is not None:
187                try:
188                    data, converted = self.raw_decode(self.raw_buffer,
189                            'strict', self.eof)
190                except UnicodeDecodeError, exc:
191                    character = exc.object[exc.start]
192                    if self.stream is not None:
193                        position = self.stream_pointer-len(self.raw_buffer)+exc.start
194                    else:
195                        position = exc.start
196                    raise ReaderError(self.name, position, character,
197                            exc.encoding, exc.reason)
198            else:
199                data = self.raw_buffer
200                converted = len(data)
[45]201            self.check_printable(data)
202            self.buffer += data
203            self.raw_buffer = self.raw_buffer[converted:]
204            if self.eof:
205                self.buffer += u'\0'
206                self.raw_buffer = None
207                break
208
209    def update_raw(self, size=1024):
210        data = self.stream.read(size)
211        if data:
212            self.raw_buffer += data
213            self.stream_pointer += len(data)
214        else:
215            self.eof = True
216
217#try:
218#    import psyco
[46]219#    psyco.bind(Reader)
[45]220#except ImportError:
221#    pass
222
Note: See TracBrowser for help on using the repository browser.