Mercurial > hg > mercurial-crew-with-dirclash

comparison mercurial/base85.c @ 3287:e93c926e069e

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Handle odd-sized base85 input and output
author Brendan Cully <brendan@kublai.com>
date Sat, 07 Oct 2006 16:21:33 -0700
parents 1f2c3983a6c5
children a5209a1e53d8
comparison
equal deleted inserted replaced
3279:1f2c3983a6c5 3287:e93c926e069e
31 const unsigned char *text; 31 const unsigned char *text;
32 PyObject *out; 32 PyObject *out;
33 char *dst; 33 char *dst;
34 int len, olen, i; 34 int len, olen, i;
35 unsigned int acc, val, ch; 35 unsigned int acc, val, ch;
36 int pad = 0;
36 37
37 if (!PyArg_ParseTuple(args, "s#", &text, &len)) 38 if (!PyArg_ParseTuple(args, "s#|i", &text, &len, &pad))
38 return NULL; 39 return NULL;
39 40
40 olen = (len + 3) / 4 * 5; 41 if (pad)
41 if (!(out = PyString_FromStringAndSize(NULL, olen))) 42 olen = ((len + 3) / 4 * 5) - 3;
43 else {
44 olen = len % 4;
45 if (olen)
46 olen++;
47 olen += len / 4 * 5;
48 }
49 if (!(out = PyString_FromStringAndSize(NULL, olen + 3)))
42 return NULL; 50 return NULL;
43 51
44 dst = PyString_AS_STRING(out); 52 dst = PyString_AS_STRING(out);
45 53
46 while (len) 54 while (len) {
47 {
48 acc = 0; 55 acc = 0;
49 for (i = 24; i >= 0; i -= 8) { 56 for (i = 24; i >= 0; i -= 8) {
50 ch = *text++; 57 ch = *text++;
51 acc |= ch << i; 58 acc |= ch << i;
52 if (--len == 0) 59 if (--len == 0)
58 dst[i] = b85chars[val]; 65 dst[i] = b85chars[val];
59 } 66 }
60 dst += 5; 67 dst += 5;
61 } 68 }
62 69
70 if (!pad)
71 _PyString_Resize(&out, olen);
72
63 return out; 73 return out;
64 } 74 }
65 75
66 static PyObject * 76 static PyObject *
67 b85decode(PyObject *self, PyObject *args) 77 b85decode(PyObject *self, PyObject *args)
68 { 78 {
69 PyObject *out; 79 PyObject *out;
70 const char *text; 80 const char *text;
71 char *dst; 81 char *dst;
72 int len, i, j, olen, c; 82 int len, i, j, olen, c, cap;
73 unsigned int acc; 83 unsigned int acc;
74 84
75 if (!PyArg_ParseTuple(args, "s#", &text, &len)) 85 if (!PyArg_ParseTuple(args, "s#", &text, &len))
76 return NULL; 86 return NULL;
77 87
78 olen = (len + 4) / 5 * 4; 88 olen = len / 5 * 4;
89 i = len % 5;
90 if (i)
91 olen += i - 1;
79 if (!(out = PyString_FromStringAndSize(NULL, olen))) 92 if (!(out = PyString_FromStringAndSize(NULL, olen)))
80 return NULL; 93 return NULL;
81 94
82 dst = PyString_AS_STRING(out); 95 dst = PyString_AS_STRING(out);
83 96
84 for (i = 1; len; i++) 97 i = 0;
98 while (i < len)
85 { 99 {
86 acc = 0; 100 acc = 0;
87 for (j = 0; j < 4 && --len; j++) 101 cap = len - i - 1;
102 if (cap > 4)
103 cap = 4;
104 for (j = 0; j < cap; i++, j++)
88 { 105 {
89 c = b85dec[(int)*text++] - 1; 106 c = b85dec[(int)*text++] - 1;
90 if (c < 0) 107 if (c < 0)
91 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i); 108 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i);
92 acc = acc * 85 + c; 109 acc = acc * 85 + c;
93 } 110 }
94 if (len--) 111 if (i++ < len)
95 { 112 {
96 c = b85dec[(int)*text++] - 1; 113 c = b85dec[(int)*text++] - 1;
97 if (c < 0) 114 if (c < 0)
98 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i); 115 return PyErr_Format(PyExc_ValueError, "Bad base85 character at position %d", i);
116 /* overflow detection: 0xffffffff == "|NsC0",
117 * "|NsC" == 0x03030303 */
118 if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c)
119 return PyErr_Format(PyExc_ValueError, "Bad base85 sequence at position %d", i);
120 acc += c;
99 } 121 }
100 else
101 c = 0;
102 /* overflow detection: 0xffffffff == "|NsC0",
103 * "|NsC" == 0x03030303 */
104 if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c)
105 return PyErr_Format(PyExc_ValueError, "Bad base85 sequence at position %d", i);
106
107 acc += c;
108 122
109 for (j = 0; j < 4; j++) 123 cap = olen < 4 ? olen : 4;
124 olen -= cap;
125 for (j = 0; j < 4 - cap; j++)
126 acc *= 85;
127 if (cap && cap < 4)
128 acc += 0xffffff >> (cap - 1) * 8;
129 for (j = 0; j < cap; j++)
110 { 130 {
111 acc = (acc << 8) | (acc >> 24); 131 acc = (acc << 8) | (acc >> 24);
112 *dst++ = (char)acc; 132 *dst++ = acc;
113 } 133 }
114 } 134 }
115 135
116 return out; 136 return out;
117 } 137 }
118 138
119 static char base85_doc[] = "Base85 Data Encoding"; 139 static char base85_doc[] = "Base85 Data Encoding";
120 140
121 static PyMethodDef methods[] = { 141 static PyMethodDef methods[] = {
122 {"b85encode", b85encode, METH_VARARGS, "encode text in base85\n"}, 142 {"b85encode", b85encode, METH_VARARGS,
123 {"b85decode", b85decode, METH_VARARGS, "decode base85 text\n"}, 143 "Encode text in base85.\n\n"
144 "If the second parameter is true, pad the result to a multiple of "
145 "five characters.\n"},
146 {"b85decode", b85decode, METH_VARARGS, "Decode base85 text.\n"},
124 {NULL, NULL} 147 {NULL, NULL}
125 }; 148 };
126 149
127 PyMODINIT_FUNC initbase85(void) 150 PyMODINIT_FUNC initbase85(void)
128 { 151 {