annotate mercurial/base85.c @ 3279:1f2c3983a6c5

Add a base85 codec
author Brendan Cully <brendan@kublai.com>
date Fri, 06 Oct 2006 13:01:54 -0700
parents
children e93c926e069e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3279
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
1 /*
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
2 base85 codec
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
3
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
4 Copyright 2006 Brendan Cully <brendan@kublai.com>
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
5
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
6 This software may be used and distributed according to the terms of
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
7 the GNU General Public License, incorporated herein by reference.
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
8
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
9 Largely based on git's implementation
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
10 */
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
11
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
12 #include <Python.h>
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
13
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
14 static const char b85chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
15 "abcdefghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~";
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
16 static char b85dec[256];
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
17
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
18 static void
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
19 b85prep(void)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
20 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
21 int i;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
22
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
23 memset(b85dec, 0, sizeof(b85dec));
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
24 for (i = 0; i < sizeof(b85chars); i++)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
25 b85dec[(int)(b85chars[i])] = i + 1;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
26 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
27
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
28 static PyObject *
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
29 b85encode(PyObject *self, PyObject *args)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
30 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
31 const unsigned char *text;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
32 PyObject *out;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
33 char *dst;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
34 int len, olen, i;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
35 unsigned int acc, val, ch;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
36
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
37 if (!PyArg_ParseTuple(args, "s#", &text, &len))
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
38 return NULL;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
39
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
40 olen = (len + 3) / 4 * 5;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
41 if (!(out = PyString_FromStringAndSize(NULL, olen)))
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
42 return NULL;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
43
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
44 dst = PyString_AS_STRING(out);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
45
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
46 while (len)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
47 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
48 acc = 0;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
49 for (i = 24; i >= 0; i -= 8) {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
50 ch = *text++;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
51 acc |= ch << i;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
52 if (--len == 0)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
53 break;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
54 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
55 for (i = 4; i >= 0; i--) {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
56 val = acc % 85;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
57 acc /= 85;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
58 dst[i] = b85chars[val];
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
59 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
60 dst += 5;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
61 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
62
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
63 return out;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
64 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
65
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
66 static PyObject *
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
67 b85decode(PyObject *self, PyObject *args)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
68 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
69 PyObject *out;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
70 const char *text;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
71 char *dst;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
72 int len, i, j, olen, c;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
73 unsigned int acc;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
74
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
75 if (!PyArg_ParseTuple(args, "s#", &text, &len))
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
76 return NULL;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
77
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
78 olen = (len + 4) / 5 * 4;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
79 if (!(out = PyString_FromStringAndSize(NULL, olen)))
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
80 return NULL;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
81
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
82 dst = PyString_AS_STRING(out);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
83
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
84 for (i = 1; len; i++)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
85 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
86 acc = 0;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
87 for (j = 0; j < 4 && --len; j++)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
88 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
89 c = b85dec[(int)*text++] - 1;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
90 if (c < 0)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
91 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
92 acc = acc * 85 + c;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
93 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
94 if (len--)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
95 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
96 c = b85dec[(int)*text++] - 1;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
97 if (c < 0)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
98 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
99 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
100 else
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
101 c = 0;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
102 /* overflow detection: 0xffffffff == "|NsC0",
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
103 * "|NsC" == 0x03030303 */
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
104 if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
105 return PyErr_Format(PyExc_ValueError, "Bad base85 sequence at position %d", i);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
106
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
107 acc += c;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
108
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
109 for (j = 0; j < 4; j++)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
110 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
111 acc = (acc << 8) | (acc >> 24);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
112 *dst++ = (char)acc;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
113 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
114 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
115
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
116 return out;
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
117 }
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
118
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
119 static char base85_doc[] = "Base85 Data Encoding";
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
120
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
121 static PyMethodDef methods[] = {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
122 {"b85encode", b85encode, METH_VARARGS, "encode text in base85\n"},
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
123 {"b85decode", b85decode, METH_VARARGS, "decode base85 text\n"},
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
124 {NULL, NULL}
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
125 };
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
126
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
127 PyMODINIT_FUNC initbase85(void)
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
128 {
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
129 Py_InitModule3("base85", methods, base85_doc);
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
130
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
131 b85prep();
1f2c3983a6c5 Add a base85 codec
Brendan Cully <brendan@kublai.com>
parents:
diff changeset
132 }