author | Pierre-Yves David <pierre-yves.david@fb.com> |
Sat, 27 Feb 2016 18:02:12 +0100 | |
changeset 28280 | dc6032a1d888 |
parent 27060 | 4613a89bea42 |
child 32356 | 7948adb53e28 |
permissions | -rw-r--r-- |
3283 | 1 |
/* |
2 |
base85 codec |
|
3 |
||
4 |
Copyright 2006 Brendan Cully <brendan@kublai.com> |
|
5 |
||
6 |
This software may be used and distributed according to the terms of |
|
7 |
the GNU General Public License, incorporated herein by reference. |
|
8 |
||
9 |
Largely based on git's implementation |
|
10 |
*/ |
|
11 |
||
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
12 |
#define PY_SSIZE_T_CLEAN |
3283 | 13 |
#include <Python.h> |
14 |
||
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
15 |
#include "util.h" |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
16 |
|
3283 | 17 |
static const char b85chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ" |
18 |
"abcdefghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~"; |
|
19 |
static char b85dec[256]; |
|
20 |
||
27060
4613a89bea42
base85: clean up function definition style
Augie Fackler <augie@google.com>
parents:
26074
diff
changeset
|
21 |
static void b85prep(void) |
3283 | 22 |
{ |
26074
c1aefe57cf4e
base85: fix comparison of an int and a long
Augie Fackler <augie@google.com>
parents:
16848
diff
changeset
|
23 |
unsigned i; |
3283 | 24 |
|
25 |
memset(b85dec, 0, sizeof(b85dec)); |
|
26 |
for (i = 0; i < sizeof(b85chars); i++) |
|
27 |
b85dec[(int)(b85chars[i])] = i + 1; |
|
28 |
} |
|
29 |
||
27060
4613a89bea42
base85: clean up function definition style
Augie Fackler <augie@google.com>
parents:
26074
diff
changeset
|
30 |
static PyObject *b85encode(PyObject *self, PyObject *args) |
3283 | 31 |
{ |
32 |
const unsigned char *text; |
|
33 |
PyObject *out; |
|
34 |
char *dst; |
|
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
35 |
Py_ssize_t len, olen, i; |
3283 | 36 |
unsigned int acc, val, ch; |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
37 |
int pad = 0; |
3283 | 38 |
|
3369
4bad632913d8
python2.5 PyArg_ParseTuple fix
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
3332
diff
changeset
|
39 |
if (!PyArg_ParseTuple(args, "s#|i", &text, &len, &pad)) |
3283 | 40 |
return NULL; |
41 |
||
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
42 |
if (pad) |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
43 |
olen = ((len + 3) / 4 * 5) - 3; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
44 |
else { |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
45 |
olen = len % 4; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
46 |
if (olen) |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
47 |
olen++; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
48 |
olen += len / 4 * 5; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
49 |
} |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
50 |
if (!(out = PyBytes_FromStringAndSize(NULL, olen + 3))) |
3283 | 51 |
return NULL; |
52 |
||
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
53 |
dst = PyBytes_AsString(out); |
3283 | 54 |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
55 |
while (len) { |
3283 | 56 |
acc = 0; |
57 |
for (i = 24; i >= 0; i -= 8) { |
|
58 |
ch = *text++; |
|
59 |
acc |= ch << i; |
|
60 |
if (--len == 0) |
|
61 |
break; |
|
62 |
} |
|
63 |
for (i = 4; i >= 0; i--) { |
|
64 |
val = acc % 85; |
|
65 |
acc /= 85; |
|
66 |
dst[i] = b85chars[val]; |
|
67 |
} |
|
68 |
dst += 5; |
|
69 |
} |
|
70 |
||
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
71 |
if (!pad) |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
72 |
_PyBytes_Resize(&out, olen); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
73 |
|
3283 | 74 |
return out; |
75 |
} |
|
76 |
||
27060
4613a89bea42
base85: clean up function definition style
Augie Fackler <augie@google.com>
parents:
26074
diff
changeset
|
77 |
static PyObject *b85decode(PyObject *self, PyObject *args) |
3283 | 78 |
{ |
79 |
PyObject *out; |
|
80 |
const char *text; |
|
81 |
char *dst; |
|
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
82 |
Py_ssize_t len, i, j, olen, cap; |
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
83 |
int c; |
3283 | 84 |
unsigned int acc; |
85 |
||
3369
4bad632913d8
python2.5 PyArg_ParseTuple fix
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
3332
diff
changeset
|
86 |
if (!PyArg_ParseTuple(args, "s#", &text, &len)) |
3283 | 87 |
return NULL; |
88 |
||
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
89 |
olen = len / 5 * 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
90 |
i = len % 5; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
91 |
if (i) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
92 |
olen += i - 1; |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
93 |
if (!(out = PyBytes_FromStringAndSize(NULL, olen))) |
3283 | 94 |
return NULL; |
95 |
||
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
96 |
dst = PyBytes_AsString(out); |
3283 | 97 |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
98 |
i = 0; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
99 |
while (i < len) |
3283 | 100 |
{ |
101 |
acc = 0; |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
102 |
cap = len - i - 1; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
103 |
if (cap > 4) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
104 |
cap = 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
105 |
for (j = 0; j < cap; i++, j++) |
3283 | 106 |
{ |
107 |
c = b85dec[(int)*text++] - 1; |
|
108 |
if (c < 0) |
|
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
109 |
return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
110 |
PyExc_ValueError, |
16848
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
111 |
"bad base85 character at position %d", |
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
112 |
(int)i); |
3283 | 113 |
acc = acc * 85 + c; |
114 |
} |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
115 |
if (i++ < len) |
3283 | 116 |
{ |
117 |
c = b85dec[(int)*text++] - 1; |
|
118 |
if (c < 0) |
|
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
119 |
return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
120 |
PyExc_ValueError, |
16848
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
121 |
"bad base85 character at position %d", |
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
122 |
(int)i); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
123 |
/* overflow detection: 0xffffffff == "|NsC0", |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
124 |
* "|NsC" == 0x03030303 */ |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
125 |
if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c) |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
126 |
return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
127 |
PyExc_ValueError, |
16848
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
128 |
"bad base85 sequence at position %d", |
19a915d43a68
base85: cast Py_ssize_t values to int (issue3481)
Adrian Buehlmann <adrian@cadifra.com>
parents:
16837
diff
changeset
|
129 |
(int)i); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
130 |
acc += c; |
3283 | 131 |
} |
132 |
||
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
133 |
cap = olen < 4 ? olen : 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
134 |
olen -= cap; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
135 |
for (j = 0; j < 4 - cap; j++) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
136 |
acc *= 85; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
137 |
if (cap && cap < 4) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
138 |
acc += 0xffffff >> (cap - 1) * 8; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
139 |
for (j = 0; j < cap; j++) |
3283 | 140 |
{ |
141 |
acc = (acc << 8) | (acc >> 24); |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
142 |
*dst++ = acc; |
3283 | 143 |
} |
144 |
} |
|
145 |
||
146 |
return out; |
|
147 |
} |
|
148 |
||
149 |
static char base85_doc[] = "Base85 Data Encoding"; |
|
150 |
||
151 |
static PyMethodDef methods[] = { |
|
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
152 |
{"b85encode", b85encode, METH_VARARGS, |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
153 |
"Encode text in base85.\n\n" |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
154 |
"If the second parameter is true, pad the result to a multiple of " |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
155 |
"five characters.\n"}, |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
156 |
{"b85decode", b85decode, METH_VARARGS, "Decode base85 text.\n"}, |
3283 | 157 |
{NULL, NULL} |
158 |
}; |
|
159 |
||
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
160 |
#ifdef IS_PY3K |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
161 |
static struct PyModuleDef base85_module = { |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
162 |
PyModuleDef_HEAD_INIT, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
163 |
"base85", |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
164 |
base85_doc, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
165 |
-1, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
166 |
methods |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
167 |
}; |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
168 |
|
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
169 |
PyMODINIT_FUNC PyInit_base85(void) |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
170 |
{ |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
171 |
b85prep(); |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
172 |
|
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
173 |
return PyModule_Create(&base85_module); |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
174 |
} |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
175 |
#else |
3283 | 176 |
PyMODINIT_FUNC initbase85(void) |
177 |
{ |
|
178 |
Py_InitModule3("base85", methods, base85_doc); |
|
179 |
||
180 |
b85prep(); |
|
181 |
} |
|
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
182 |
#endif |