Mercurial > cgi-bin > hgweb.cgi > curlyq
annotate runes.py @ 20:be0fd5c8121d
Can't use memoryview to base Runes on; different methods inside.
| author | David Barts <n5jrn@me.com> |
|---|---|
| date | Sat, 28 Dec 2019 01:22:03 -0800 |
| parents | 0c8d787bc7e1 |
| children | 35f29952b51e |
| 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): | |
|
20
be0fd5c8121d
Can't use memoryview to base Runes on; different methods inside.
David Barts <n5jrn@me.com>
parents:
17
diff
changeset
|
28 if isinstance(based_on, array.array): |
|
be0fd5c8121d
Can't use memoryview to base Runes on; different methods inside.
David Barts <n5jrn@me.com>
parents:
17
diff
changeset
|
29 if based_on.typecode == 'H': |
| 10 | 30 self.buffer = based_on |
| 31 else: | |
| 32 self.buffer = array.array('H', based_on) | |
| 33 elif isinstance(based_on, str): | |
| 34 # A string should always be able to encode to runes. | |
| 35 self.buffer = array.array('H', self.codec.encode(based_on, 'strict')[0]) | |
| 36 elif based_on is None: | |
| 37 self.buffer = array.array('H', bytes()) | |
| 12 | 38 elif isinstance(based_on, Runes): |
| 39 self.buffer = array.array('H', based_on.buffer) | |
| 10 | 40 else: |
| 41 self.buffer = array.array('H', based_on) | |
| 42 | |
| 43 def __str__(self): | |
| 44 """ | |
| 45 Convert this object to a string. We deliberately do not have a | |
| 46 __repr__ method, to underscore that runes are not strings. | |
| 47 """ | |
| 48 # Runes might not always be able to decode to a string. | |
| 49 return self.codec.decode(self.buffer, 'replace')[0] | |
| 50 | |
| 51 def __bytes__(self): | |
| 52 return bytes(self.buffer) | |
| 53 | |
| 54 def __len__(self): | |
| 55 return len(self.buffer) | |
| 56 | |
| 57 def __lt__(self, other): | |
| 58 return self.buffer < other.buffer | |
| 59 | |
| 60 def __le__(self, other): | |
| 61 return self.buffer <= other.buffer | |
| 62 | |
| 63 def __gt__(self, other): | |
| 64 return self.buffer > other.buffer | |
| 65 | |
| 66 def __ge__(self, other): | |
| 67 return self.buffer >= other.buffer | |
| 68 | |
| 69 def __eq__(self, other): | |
| 70 return self.buffer == other.buffer | |
| 71 | |
| 72 def __ne__(self, other): | |
| 73 return self.buffer != other.buffer | |
| 74 | |
| 75 def __hash__(self): | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
76 raise TypeError("unhashable type") |
| 10 | 77 |
| 78 def __bool__(self): | |
| 79 return bool(self.buffer) | |
| 80 | |
| 81 def __getitem__(self, key): | |
| 82 ret = self.buffer[key] | |
| 83 if isinstance(ret, array.array): | |
| 84 return Runes(ret) | |
| 85 else: | |
| 86 return ret | |
| 87 | |
| 88 def __setitem__(self, key, value): | |
| 89 if isinstance(key, int): | |
| 90 if isinstance(value, int): | |
| 91 self.buffer[key] = value | |
| 92 else: | |
| 93 raise TypeError("integer required") | |
| 94 elif isinstance(value, Runes): | |
| 95 self.buffer[key] = value.buffer | |
| 96 else: | |
| 97 raise TypeError("runes required") | |
| 98 | |
| 99 def __delitem__(self, key): | |
| 100 del self.buffer[key] | |
| 101 | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
102 def __del__(self): |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
103 # Paranoid |
| 17 | 104 if hasattr(self, 'buffer') and isinstance(self.buffer, memoryview): |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
105 self.buffer.release() |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
106 |
| 10 | 107 def clear(self): |
| 108 del self[:] | |
| 109 | |
| 110 def __iter__(self): | |
| 111 return iter(self.buffer) | |
| 112 | |
| 113 def __reversed__(self): | |
| 114 return reversed(self.buffer) | |
| 115 | |
| 116 def append(self, value): | |
| 117 if isinstance(value, int): | |
| 118 self.buffer.append(value) | |
| 119 elif isinstance(value, Runes): | |
| 120 self.buffer.extend(value.buffer) | |
| 121 else: | |
| 122 raise TypeError("integer or runes required") | |
| 123 | |
| 124 def __contains__(self, value): | |
| 125 return value in self.buffer | |
| 126 | |
| 127 def index(self, value): | |
| 128 return self.buffer.index(value) | |
| 129 | |
| 130 def find(self, value): | |
| 131 try: | |
| 132 return self.index(value) | |
| 133 except ValueError: | |
| 134 return -1 | |
| 135 | |
| 136 class Workspace(Runes): | |
| 137 """ | |
| 138 A Runes object that acts a bit more string-like, in that __setitem__ | |
| 139 also accepts a string as an argument and __getitem__ always returns | |
| 140 a string. We also return empty strings instead of throwing IndexError | |
| 141 when attempting to read out-of-range values, because that makes life | |
| 142 easier for us when curling quotes. | |
| 143 """ | |
| 144 def __setitem__(self, key, value): | |
| 145 if isinstance(value, str): | |
| 146 if isinstance(key, int): | |
| 147 Runes.__setitem__(self, key, self._ord(value)) | |
| 148 else: | |
| 149 Runes.__setitem__(self, key, Runes(value)) | |
| 150 else: | |
| 151 Runes.__setitem__(self, key, value) | |
| 152 | |
| 153 def __getitem__(self, key): | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
154 view = memoryview(self.buffer) |
| 10 | 155 try: |
| 15 | 156 result = view[key] |
| 16 | 157 if isinstance(result, int): |
| 15 | 158 return chr(result) |
| 159 if isinstance(result, memoryview): | |
| 160 ret = self.codec.decode(result, 'replace')[0] | |
| 161 result.release() | |
| 162 return ret | |
| 163 else: | |
| 164 raise AssertionError("this shouldn't happen") | |
| 10 | 165 except IndexError: |
| 166 return "" | |
|
13
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
167 finally: |
|
6acda841a690
Add support for shared substrings.
David Barts <n5jrn@me.com>
parents:
12
diff
changeset
|
168 view.release() |
| 10 | 169 |
| 170 def append(self, value): | |
| 171 if isinstance(value, str): | |
| 172 Runes.append(self, Runes(value)) | |
| 173 else: | |
| 174 Runes.append(self, value) | |
| 175 | |
| 176 def index(self, value): | |
| 177 if isinstance(value, str): | |
| 178 return Runes.index(self, self._ord(value)) | |
| 179 else: | |
| 180 return Runes.index(self, value) | |
| 181 | |
| 182 def find(self, value): | |
| 183 try: | |
| 184 return self.index(value) | |
| 185 except ValueError: | |
| 186 return -1 | |
| 187 | |
| 188 def _ord(self, string): | |
| 189 length = len(string) | |
| 190 if length != 1: | |
| 191 raise ValueError("expected a character, but string of length {0} found".format(length)) | |
| 192 raw = Runes(string) | |
| 193 if len(raw) != 1: | |
| 194 raise ValueError("character not in BMP") | |
| 195 return raw[0] |