lib/yaml/reader.py

   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 #
   6 #   Marker(source, line, column)
   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 #
  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.
  17
  18 __all__ = ['Marker', 'Reader', 'ReaderError']
  19
  20 from error import YAMLError
  21
  22 import 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
  27 try:
  28     codecs.utf_8_decode('', 'strict', False)
  29     from codecs import utf_8_decode, utf_16_le_decode, utf_16_be_decode
  30
  31 except 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
  59 class Marker:
  60
  61     def __init__(self, name, line, column, buffer, pointer):
  62         self.name = name
  63         self.line = line
  64         self.column = column
  65         self.buffer = buffer
  66         self.pointer = pointer
  67
  68     def get_snippet(self, max_length=79):
  69         if self.buffer is None:
  70             return None
  71         head = ''
  72         start = self.pointer
  73         while start > 0 and self.buffer[start-1] not in u'\0\r\n\x85\u2028\u2029':
  74             start -= 1
  75             if self.pointer-start > max_length/2-1:
  76                 head = ' ... '
  77                 start += 5
  78                 break
  79         tail = ''
  80         end = self.pointer
  81         while end < len(self.buffer) and self.buffer[end] not in u'\0\r\n\x85\u2028\u2029':
  82             end += 1
  83             if end-self.pointer > max_length/2-1:
  84                 tail = ' ... '
  85                 end -= 5
  86                 break
  87         snippet = self.buffer[start:end].encode('utf-8')
  88         return head + snippet + tail + '\n'  \
  89                 + ' '*(self.pointer-start+len(head)) + '^' + '\n'
  90
  91 class ReaderError(YAMLError):
  92
  93     def __init__(self, name, position, character, encoding, reason):
  94         self.name = name
  95         self.character = character
  96         self.position = position
  97         self.encoding = encoding
  98         self.reason = reason
  99
 100     def __str__(self):
 101         if isinstance(self.character, str):
 102             return "'%s' codec can't decode byte #x%02x: %s\n"  \
 103                     "\tin '%s', position %d."   \
 104                     % (self.encoding, ord(self.character), self.reason,
 105                             self.name, self.position)
 106         else:
 107             return "unacceptable character #x%04x: %s\n"    \
 108                     "\tin '%s', position %d."   \
 109                     % (ord(self.character), self.reason,
 110                             self.name, self.position)
 111
 112 class Reader:
 113     # Reader:
 114     # - determines the data encoding and converts it to unicode,
 115     # - checks if characters are in allowed range,
 116     # - adds '\0' to the end.
 117
 118     # Reader accepts
 119     #  - a `str` object,
 120     #  - a `unicode` object,
 121     #  - a file-like object with its `read` method returning `str`,
 122     #  - a file-like object with its `read` method returning `unicode`.
 123
 124     # Yeah, it's ugly and slow.
 125
 126     def __init__(self, data):
 127         self.name = None
 128         self.stream = None
 129         self.stream_pointer = 0
 130         self.eof = True
 131         self.buffer = u''
 132         self.pointer = 0
 133         self.raw_buffer = None
 134         self.raw_decode = None
 135         self.index = 0
 136         self.line = 0
 137         self.column = 0
 138         if isinstance(data, unicode):
 139             self.name = "<unicode string>"
 140             self.check_printable(data)
 141             self.buffer = data+u'\0'
 142         elif isinstance(data, str):
 143             self.name = "<string>"
 144             self.raw_buffer = data
 145             self.determine_encoding()
 146         else:
 147             self.stream = data
 148             self.name = getattr(data, 'name', "<file>")
 149             self.eof = False
 150             self.raw_buffer = ''
 151             self.determine_encoding()
 152
 153     def peek(self, length=1):
 154         if self.pointer+length >= len(self.buffer):
 155             self.update(length)
 156         return self.buffer[self.pointer:self.pointer+length]
 157
 158     def forward(self, length=1):
 159         if self.pointer+length+1 >= len(self.buffer):
 160             self.update(length+1)
 161         for k in range(length):
 162             ch = self.buffer[self.pointer]
 163             self.pointer += 1
 164             self.index += 1
 165             if ch in u'\n\x85\u2028\u2029'  \
 166                     or (ch == u'\r' and self.buffer[self.pointer+1] != u'\n'):
 167                 self.line += 1
 168                 self.column = 0
 169             elif ch != u'\uFEFF':
 170                 self.column += 1
 171
 172     def get_marker(self):
 173         if self.stream is None:
 174             return Marker(self.name, self.line, self.column,
 175                     self.buffer, self.pointer)
 176         else:
 177             return Marker(self.name, self.line, self.column, None, None)
 178
 179     def determine_encoding(self):
 180         while not self.eof and len(self.raw_buffer) < 2:
 181             self.update_raw()
 182         if not isinstance(self.raw_buffer, unicode):
 183             if self.raw_buffer.startswith(codecs.BOM_UTF16_LE):
 184                 self.raw_decode = utf_16_le_decode
 185             elif self.raw_buffer.startswith(codecs.BOM_UTF16_BE):
 186                 self.raw_decode = utf_16_be_decode
 187             else:
 188                 self.raw_decode = utf_8_decode
 189         self.update(1)
 190
 191     NON_PRINTABLE = re.compile(u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uD7FF\uE000-\uFFFD]')
 192     def check_printable(self, data):
 193         match = self.NON_PRINTABLE.search(data)
 194         if match:
 195             character = match.group()
 196             position = self.index+(len(self.buffer)-self.pointer)+match.start()
 197             raise ReaderError(self.name, position, character,
 198                     'unicode', "special characters are not allowed")
 199
 200     def update(self, length):
 201         if self.raw_buffer is None:
 202             return
 203         self.buffer = self.buffer[self.pointer:]
 204         self.pointer = 0
 205         while len(self.buffer) < length:
 206             if not self.eof:
 207                 self.update_raw()
 208             if self.raw_decode is not None:
 209                 try:
 210                     data, converted = self.raw_decode(self.raw_buffer,
 211                             'strict', self.eof)
 212                 except UnicodeDecodeError, exc:
 213                     character = exc.object[exc.start]
 214                     if self.stream is not None:
 215                         position = self.stream_pointer-len(self.raw_buffer)+exc.start
 216                     else:
 217                         position = exc.start
 218                     raise ReaderError(self.name, position, character,
 219                             exc.encoding, exc.reason)
 220             else:
 221                 data = self.raw_buffer
 222                 converted = len(data)
 223             self.check_printable(data)
 224             self.buffer += data
 225             self.raw_buffer = self.raw_buffer[converted:]
 226             if self.eof:
 227                 self.buffer += u'\0'
 228                 self.raw_buffer = None
 229                 break
 230
 231     def update_raw(self, size=1024):
 232         data = self.stream.read(size)
 233         if data:
 234             self.raw_buffer += data
 235             self.stream_pointer += len(data)
 236         else:
 237             self.eof = True
 238
 239 #try:
 240 #    import psyco
 241 #    psyco.bind(Reader)
 242 #except ImportError:
 243 #    pass
 244