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# Mark(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__ = ['Reader', 'ReaderError'] 19 20from error import YAMLError, Mark 21 22import codecs, re, sys 23 24has_ucs4 = sys.maxunicode > 0xffff 25 26class ReaderError(YAMLError): 27 28 def __init__(self, name, position, character, encoding, reason): 29 self.name = name 30 self.character = character 31 self.position = position 32 self.encoding = encoding 33 self.reason = reason 34 35 def __str__(self): 36 if isinstance(self.character, str): 37 return "'%s' codec can't decode byte #x%02x: %s\n" \ 38 " in \"%s\", position %d" \ 39 % (self.encoding, ord(self.character), self.reason, 40 self.name, self.position) 41 else: 42 return "unacceptable character #x%04x: %s\n" \ 43 " in \"%s\", position %d" \ 44 % (self.character, self.reason, 45 self.name, self.position) 46 47class Reader(object): 48 # Reader: 49 # - determines the data encoding and converts it to unicode, 50 # - checks if characters are in allowed range, 51 # - adds '\0' to the end. 52 53 # Reader accepts 54 # - a `str` object, 55 # - a `unicode` object, 56 # - a file-like object with its `read` method returning `str`, 57 # - a file-like object with its `read` method returning `unicode`. 58 59 # Yeah, it's ugly and slow. 60 61 def __init__(self, stream): 62 self.name = None 63 self.stream = None 64 self.stream_pointer = 0 65 self.eof = True 66 self.buffer = u'' 67 self.pointer = 0 68 self.raw_buffer = None 69 self.raw_decode = None 70 self.encoding = None 71 self.index = 0 72 self.line = 0 73 self.column = 0 74 if isinstance(stream, unicode): 75 self.name = "<unicode string>" 76 self.check_printable(stream) 77 self.buffer = stream+u'\0' 78 elif isinstance(stream, str): 79 self.name = "<string>" 80 self.raw_buffer = stream 81 self.determine_encoding() 82 else: 83 self.stream = stream 84 self.name = getattr(stream, 'name', "<file>") 85 self.eof = False 86 self.raw_buffer = '' 87 self.determine_encoding() 88 89 def peek(self, index=0): 90 try: 91 return self.buffer[self.pointer+index] 92 except IndexError: 93 self.update(index+1) 94 return self.buffer[self.pointer+index] 95 96 def prefix(self, length=1): 97 if self.pointer+length >= len(self.buffer): 98 self.update(length) 99 return self.buffer[self.pointer:self.pointer+length] 100 101 def forward(self, length=1): 102 if self.pointer+length+1 >= len(self.buffer): 103 self.update(length+1) 104 while length: 105 ch = self.buffer[self.pointer] 106 self.pointer += 1 107 self.index += 1 108 if ch in u'\n\x85\u2028\u2029' \ 109 or (ch == u'\r' and self.buffer[self.pointer] != u'\n'): 110 self.line += 1 111 self.column = 0 112 elif ch != u'\uFEFF': 113 self.column += 1 114 length -= 1 115 116 def get_mark(self): 117 if self.stream is None: 118 return Mark(self.name, self.index, self.line, self.column, 119 self.buffer, self.pointer) 120 else: 121 return Mark(self.name, self.index, self.line, self.column, 122 None, None) 123 124 def determine_encoding(self): 125 while not self.eof and len(self.raw_buffer) < 2: 126 self.update_raw() 127 if not isinstance(self.raw_buffer, unicode): 128 if self.raw_buffer.startswith(codecs.BOM_UTF16_LE): 129 self.raw_decode = codecs.utf_16_le_decode 130 self.encoding = 'utf-16-le' 131 elif self.raw_buffer.startswith(codecs.BOM_UTF16_BE): 132 self.raw_decode = codecs.utf_16_be_decode 133 self.encoding = 'utf-16-be' 134 else: 135 self.raw_decode = codecs.utf_8_decode 136 self.encoding = 'utf-8' 137 self.update(1) 138 139 if has_ucs4: 140 NON_PRINTABLE = u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uD7FF\uE000-\uFFFD\U00010000-\U0010ffff]' 141 elif sys.platform.startswith('java'): 142 # Jython doesn't support lone surrogates https://bugs.jython.org/issue2048 143 NON_PRINTABLE = u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uD7FF\uE000-\uFFFD]' 144 else: 145 # Need to use eval here due to the above Jython issue 146 NON_PRINTABLE = eval(r"u'[^\x09\x0A\x0D\x20-\x7E\x85\xA0-\uFFFD]|(?:^|[^\uD800-\uDBFF])[\uDC00-\uDFFF]|[\uD800-\uDBFF](?:[^\uDC00-\uDFFF]|$)'") 147 NON_PRINTABLE = re.compile(NON_PRINTABLE) 148 def check_printable(self, data): 149 match = self.NON_PRINTABLE.search(data) 150 if match: 151 character = match.group() 152 position = self.index+(len(self.buffer)-self.pointer)+match.start() 153 raise ReaderError(self.name, position, ord(character), 154 'unicode', "special characters are not allowed") 155 156 def update(self, length): 157 if self.raw_buffer is None: 158 return 159 self.buffer = self.buffer[self.pointer:] 160 self.pointer = 0 161 while len(self.buffer) < length: 162 if not self.eof: 163 self.update_raw() 164 if self.raw_decode is not None: 165 try: 166 data, converted = self.raw_decode(self.raw_buffer, 167 'strict', self.eof) 168 except UnicodeDecodeError, exc: 169 character = exc.object[exc.start] 170 if self.stream is not None: 171 position = self.stream_pointer-len(self.raw_buffer)+exc.start 172 else: 173 position = exc.start 174 raise ReaderError(self.name, position, character, 175 exc.encoding, exc.reason) 176 else: 177 data = self.raw_buffer 178 converted = len(data) 179 self.check_printable(data) 180 self.buffer += data 181 self.raw_buffer = self.raw_buffer[converted:] 182 if self.eof: 183 self.buffer += u'\0' 184 self.raw_buffer = None 185 break 186 187 def update_raw(self, size=1024): 188 data = self.stream.read(size) 189 if data: 190 self.raw_buffer += data 191 self.stream_pointer += len(data) 192 else: 193 self.eof = True 194