Mercurial > cgi-bin > hgweb.cgi > curlyq
annotate runes.py @ 15:0be0586104b7
Plug the leak properly (I hope).
| author | David Barts <n5jrn@me.com> |
|---|---|
| date | Fri, 27 Dec 2019 13:35:51 -0800 |
| parents | 152f6aa87d62 |
| children | 61772bf1f77c |
| rev | line source |
|---|---|
| 10 | 1 #!/usr/bin/env python3 |
| 2 # -*- coding: utf-8 -*- | |
| 3 | |
| 4 # I m p o r t s | |
| 5 | |
| 6 import array | |
| 7 import codecs | |
| 8 import collections | |
| 9 import struct | |
| 10 import sys | |
| 11 | |
| 12 # C l a s s e s | |
| 13 | |
| 14 class Runes(object): | |
| 15 """ | |
| 16 A mutable, fixed-length sequence of UTF-16 runes. The attributes | |
| 17 encoding and codec contain the name of the encoding and the codec | |
| 18 used to generate the UTF-16. The attribute buffer contains the | |
| 19 buffer (an array of 16-bit unsigned integers) used to back this | |
| 20 object; modifications to that array will be reflected in this | |
| 21 object. | |
| 22 """ | |
| 23 # The most efficient 16-bit one on this platform | |
| 24 encoding = "UTF-16" + sys.byteorder[0].upper() + "E" | |
| 25 codec = codecs.lookup(encoding) | |
| 26 | |
| 27 def __init__(self, based_on=None): | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
28 if isinstance(based_on, (array.array, memoryview)): |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
29 format = based_on.typecode if isinstance(based_on, array.array) else based_on.format |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
30 if format == 'H': |
| 10 | 31 self.buffer = based_on |
| 32 else: | |
| 33 self.buffer = array.array('H', based_on) | |
| 34 elif isinstance(based_on, str): | |
| 35 # A string should always be able to encode to runes. | |
| 36 self.buffer = array.array('H', self.codec.encode(based_on, 'strict')[0]) | |
| 37 elif based_on is None: | |
| 38 self.buffer = array.array('H', bytes()) | |
| 12 | 39 elif isinstance(based_on, Runes): |
| 40 self.buffer = array.array('H', based_on.buffer) | |
| 10 | 41 else: |
| 42 self.buffer = array.array('H', based_on) | |
| 43 | |
| 44 def __str__(self): | |
| 45 """ | |
| 46 Convert this object to a string. We deliberately do not have a | |
| 47 __repr__ method, to underscore that runes are not strings. | |
| 48 """ | |
| 49 # Runes might not always be able to decode to a string. | |
| 50 return self.codec.decode(self.buffer, 'replace')[0] | |
| 51 | |
| 52 def __bytes__(self): | |
| 53 return bytes(self.buffer) | |
| 54 | |
| 55 def __len__(self): | |
| 56 return len(self.buffer) | |
| 57 | |
| 58 def __lt__(self, other): | |
| 59 return self.buffer < other.buffer | |
| 60 | |
| 61 def __le__(self, other): | |
| 62 return self.buffer <= other.buffer | |
| 63 | |
| 64 def __gt__(self, other): | |
| 65 return self.buffer > other.buffer | |
| 66 | |
| 67 def __ge__(self, other): | |
| 68 return self.buffer >= other.buffer | |
| 69 | |
| 70 def __eq__(self, other): | |
| 71 return self.buffer == other.buffer | |
| 72 | |
| 73 def __ne__(self, other): | |
| 74 return self.buffer != other.buffer | |
| 75 | |
| 76 def __hash__(self): | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
77 raise TypeError("unhashable type") |
| 10 | 78 |
| 79 def __bool__(self): | |
| 80 return bool(self.buffer) | |
| 81 | |
| 82 def __getitem__(self, key): | |
| 83 ret = self.buffer[key] | |
| 84 if isinstance(ret, array.array): | |
| 85 return Runes(ret) | |
| 86 else: | |
| 87 return ret | |
| 88 | |
| 89 def __setitem__(self, key, value): | |
| 90 if isinstance(key, int): | |
| 91 if isinstance(value, int): | |
| 92 self.buffer[key] = value | |
| 93 else: | |
| 94 raise TypeError("integer required") | |
| 95 elif isinstance(value, Runes): | |
| 96 self.buffer[key] = value.buffer | |
| 97 else: | |
| 98 raise TypeError("runes required") | |
| 99 | |
| 100 def __delitem__(self, key): | |
| 101 del self.buffer[key] | |
| 102 | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
103 def __del__(self): |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
104 # Paranoid |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
105 if isinstance(self.buffer, memoryview): |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
106 self.buffer.release() |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
107 |
| 10 | 108 def clear(self): |
| 109 del self[:] | |
| 110 | |
| 111 def __iter__(self): | |
| 112 return iter(self.buffer) | |
| 113 | |
| 114 def __reversed__(self): | |
| 115 return reversed(self.buffer) | |
| 116 | |
| 117 def append(self, value): | |
| 118 if isinstance(value, int): | |
| 119 self.buffer.append(value) | |
| 120 elif isinstance(value, Runes): | |
| 121 self.buffer.extend(value.buffer) | |
| 122 else: | |
| 123 raise TypeError("integer or runes required") | |
| 124 | |
| 125 def __contains__(self, value): | |
| 126 return value in self.buffer | |
| 127 | |
| 128 def index(self, value): | |
| 129 return self.buffer.index(value) | |
| 130 | |
| 131 def find(self, value): | |
| 132 try: | |
| 133 return self.index(value) | |
| 134 except ValueError: | |
| 135 return -1 | |
| 136 | |
| 137 class Workspace(Runes): | |
| 138 """ | |
| 139 A Runes object that acts a bit more string-like, in that __setitem__ | |
| 140 also accepts a string as an argument and __getitem__ always returns | |
| 141 a string. We also return empty strings instead of throwing IndexError | |
| 142 when attempting to read out-of-range values, because that makes life | |
| 143 easier for us when curling quotes. | |
| 144 """ | |
| 145 def __setitem__(self, key, value): | |
| 146 if isinstance(value, str): | |
| 147 if isinstance(key, int): | |
| 148 Runes.__setitem__(self, key, self._ord(value)) | |
| 149 else: | |
| 150 Runes.__setitem__(self, key, Runes(value)) | |
| 151 else: | |
| 152 Runes.__setitem__(self, key, value) | |
| 153 | |
| 154 def __getitem__(self, key): | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
155 view = memoryview(self.buffer) |
| 10 | 156 try: |
| 15 | 157 result = view[key] |
| 158 if isinstance (result, int): | |
| 159 return chr(result) | |
| 160 if isinstance(result, memoryview): | |
| 161 ret = self.codec.decode(result, 'replace')[0] | |
| 162 result.release() | |
| 163 return ret | |
| 164 else: | |
| 165 raise AssertionError("this shouldn't happen") | |
| 10 | 166 except IndexError: |
| 167 return "" | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
168 finally: |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
169 view.release() |
| 10 | 170 |
| 171 def append(self, value): | |
| 172 if isinstance(value, str): | |
| 173 Runes.append(self, Runes(value)) | |
| 174 else: | |
| 175 Runes.append(self, value) | |
| 176 | |
| 177 def index(self, value): | |
| 178 if isinstance(value, str): | |
| 179 return Runes.index(self, self._ord(value)) | |
| 180 else: | |
| 181 return Runes.index(self, value) | |
| 182 | |
| 183 def find(self, value): | |
| 184 try: | |
| 185 return self.index(value) | |
| 186 except ValueError: | |
| 187 return -1 | |
| 188 | |
| 189 def _ord(self, string): | |
| 190 length = len(string) | |
| 191 if length != 1: | |
| 192 raise ValueError("expected a character, but string of length {0} found".format(length)) | |
| 193 raw = Runes(string) | |
| 194 if len(raw) != 1: | |
| 195 raise ValueError("character not in BMP") | |
| 196 return raw[0] |