1 /* MD5 module */
2
3 /* This module provides an interface to the MD5 algorithm */
4
5 /* See below for information about the original code this module was
6 based upon. Additional work performed by:
7
8 Andrew Kuchling (amk@amk.ca)
9 Greg Stein (gstein@lyra.org)
10 Trevor Perrin (trevp@trevp.net)
11
12 Copyright (C) 2005-2007 Gregory P. Smith (greg@krypto.org)
13 Licensed to PSF under a Contributor Agreement.
14
15 */
16
17 /* MD5 objects */
18
19 #include "Python.h"
20 #include "hashlib.h"
21 #include "pystrhex.h"
22
23 /*[clinic input]
24 module _md5
25 class MD5Type "MD5object *" "&PyType_Type"
26 [clinic start generated code]*/
27 /*[clinic end generated code: output=da39a3ee5e6b4b0d input=6e5261719957a912]*/
28
29 /* Some useful types */
30
31 #if SIZEOF_INT == 4
32 typedef unsigned int MD5_INT32; /* 32-bit integer */
33 typedef long long MD5_INT64; /* 64-bit integer */
34 #else
35 /* not defined. compilation will die. */
36 #endif
37
38 /* The MD5 block size and message digest sizes, in bytes */
39
40 #define MD5_BLOCKSIZE 64
41 #define MD5_DIGESTSIZE 16
42
43 /* The structure for storing MD5 info */
44
45 struct md5_state {
46 MD5_INT64 length;
47 MD5_INT32 state[4], curlen;
48 unsigned char buf[MD5_BLOCKSIZE];
49 };
50
51 typedef struct {
52 PyObject_HEAD
53
54 struct md5_state hash_state;
55 } MD5object;
56
57 #include "clinic/md5module.c.h"
58
59 /* ------------------------------------------------------------------------
60 *
61 * This code for the MD5 algorithm was noted as public domain. The
62 * original headers are pasted below.
63 *
64 * Several changes have been made to make it more compatible with the
65 * Python environment and desired interface.
66 *
67 */
68
69 /* LibTomCrypt, modular cryptographic library -- Tom St Denis
70 *
71 * LibTomCrypt is a library that provides various cryptographic
72 * algorithms in a highly modular and flexible manner.
73 *
74 * The library is free for all purposes without any express
75 * guarantee it works.
76 *
77 * Tom St Denis, tomstdenis@gmail.com, http://libtom.org
78 */
79
80 /* rotate the hard way (platform optimizations could be done) */
81 #define ROLc(x, y) ( (((unsigned long)(x)<<(unsigned long)((y)&31)) | (((unsigned long)(x)&0xFFFFFFFFUL)>>(unsigned long)(32-((y)&31)))) & 0xFFFFFFFFUL)
82
83 /* Endian Neutral macros that work on all platforms */
84
85 #define STORE32L(x, y) \
86 { (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255); \
87 (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
88
89 #define LOAD32L(x, y) \
90 { x = ((unsigned long)((y)[3] & 255)<<24) | \
91 ((unsigned long)((y)[2] & 255)<<16) | \
92 ((unsigned long)((y)[1] & 255)<<8) | \
93 ((unsigned long)((y)[0] & 255)); }
94
95 #define STORE64L(x, y) \
96 { (y)[7] = (unsigned char)(((x)>>56)&255); (y)[6] = (unsigned char)(((x)>>48)&255); \
97 (y)[5] = (unsigned char)(((x)>>40)&255); (y)[4] = (unsigned char)(((x)>>32)&255); \
98 (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255); \
99 (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
100
101
102 /* MD5 macros */
103
104 #define F(x,y,z) (z ^ (x & (y ^ z)))
105 #define G(x,y,z) (y ^ (z & (y ^ x)))
106 #define H(x,y,z) (x^y^z)
107 #define I(x,y,z) (y^(x|(~z)))
108
109 #define FF(a,b,c,d,M,s,t) \
110 a = (a + F(b,c,d) + M + t); a = ROLc(a, s) + b;
111
112 #define GG(a,b,c,d,M,s,t) \
113 a = (a + G(b,c,d) + M + t); a = ROLc(a, s) + b;
114
115 #define HH(a,b,c,d,M,s,t) \
116 a = (a + H(b,c,d) + M + t); a = ROLc(a, s) + b;
117
118 #define II(a,b,c,d,M,s,t) \
119 a = (a + I(b,c,d) + M + t); a = ROLc(a, s) + b;
120
121
md5_compress(struct md5_state * md5,const unsigned char * buf)122 static void md5_compress(struct md5_state *md5, const unsigned char *buf)
123 {
124 MD5_INT32 i, W[16], a, b, c, d;
125
126 assert(md5 != NULL);
127 assert(buf != NULL);
128
129 /* copy the state into 512-bits into W[0..15] */
130 for (i = 0; i < 16; i++) {
131 LOAD32L(W[i], buf + (4*i));
132 }
133
134 /* copy state */
135 a = md5->state[0];
136 b = md5->state[1];
137 c = md5->state[2];
138 d = md5->state[3];
139
140 FF(a,b,c,d,W[0],7,0xd76aa478UL)
141 FF(d,a,b,c,W[1],12,0xe8c7b756UL)
142 FF(c,d,a,b,W[2],17,0x242070dbUL)
143 FF(b,c,d,a,W[3],22,0xc1bdceeeUL)
144 FF(a,b,c,d,W[4],7,0xf57c0fafUL)
145 FF(d,a,b,c,W[5],12,0x4787c62aUL)
146 FF(c,d,a,b,W[6],17,0xa8304613UL)
147 FF(b,c,d,a,W[7],22,0xfd469501UL)
148 FF(a,b,c,d,W[8],7,0x698098d8UL)
149 FF(d,a,b,c,W[9],12,0x8b44f7afUL)
150 FF(c,d,a,b,W[10],17,0xffff5bb1UL)
151 FF(b,c,d,a,W[11],22,0x895cd7beUL)
152 FF(a,b,c,d,W[12],7,0x6b901122UL)
153 FF(d,a,b,c,W[13],12,0xfd987193UL)
154 FF(c,d,a,b,W[14],17,0xa679438eUL)
155 FF(b,c,d,a,W[15],22,0x49b40821UL)
156 GG(a,b,c,d,W[1],5,0xf61e2562UL)
157 GG(d,a,b,c,W[6],9,0xc040b340UL)
158 GG(c,d,a,b,W[11],14,0x265e5a51UL)
159 GG(b,c,d,a,W[0],20,0xe9b6c7aaUL)
160 GG(a,b,c,d,W[5],5,0xd62f105dUL)
161 GG(d,a,b,c,W[10],9,0x02441453UL)
162 GG(c,d,a,b,W[15],14,0xd8a1e681UL)
163 GG(b,c,d,a,W[4],20,0xe7d3fbc8UL)
164 GG(a,b,c,d,W[9],5,0x21e1cde6UL)
165 GG(d,a,b,c,W[14],9,0xc33707d6UL)
166 GG(c,d,a,b,W[3],14,0xf4d50d87UL)
167 GG(b,c,d,a,W[8],20,0x455a14edUL)
168 GG(a,b,c,d,W[13],5,0xa9e3e905UL)
169 GG(d,a,b,c,W[2],9,0xfcefa3f8UL)
170 GG(c,d,a,b,W[7],14,0x676f02d9UL)
171 GG(b,c,d,a,W[12],20,0x8d2a4c8aUL)
172 HH(a,b,c,d,W[5],4,0xfffa3942UL)
173 HH(d,a,b,c,W[8],11,0x8771f681UL)
174 HH(c,d,a,b,W[11],16,0x6d9d6122UL)
175 HH(b,c,d,a,W[14],23,0xfde5380cUL)
176 HH(a,b,c,d,W[1],4,0xa4beea44UL)
177 HH(d,a,b,c,W[4],11,0x4bdecfa9UL)
178 HH(c,d,a,b,W[7],16,0xf6bb4b60UL)
179 HH(b,c,d,a,W[10],23,0xbebfbc70UL)
180 HH(a,b,c,d,W[13],4,0x289b7ec6UL)
181 HH(d,a,b,c,W[0],11,0xeaa127faUL)
182 HH(c,d,a,b,W[3],16,0xd4ef3085UL)
183 HH(b,c,d,a,W[6],23,0x04881d05UL)
184 HH(a,b,c,d,W[9],4,0xd9d4d039UL)
185 HH(d,a,b,c,W[12],11,0xe6db99e5UL)
186 HH(c,d,a,b,W[15],16,0x1fa27cf8UL)
187 HH(b,c,d,a,W[2],23,0xc4ac5665UL)
188 II(a,b,c,d,W[0],6,0xf4292244UL)
189 II(d,a,b,c,W[7],10,0x432aff97UL)
190 II(c,d,a,b,W[14],15,0xab9423a7UL)
191 II(b,c,d,a,W[5],21,0xfc93a039UL)
192 II(a,b,c,d,W[12],6,0x655b59c3UL)
193 II(d,a,b,c,W[3],10,0x8f0ccc92UL)
194 II(c,d,a,b,W[10],15,0xffeff47dUL)
195 II(b,c,d,a,W[1],21,0x85845dd1UL)
196 II(a,b,c,d,W[8],6,0x6fa87e4fUL)
197 II(d,a,b,c,W[15],10,0xfe2ce6e0UL)
198 II(c,d,a,b,W[6],15,0xa3014314UL)
199 II(b,c,d,a,W[13],21,0x4e0811a1UL)
200 II(a,b,c,d,W[4],6,0xf7537e82UL)
201 II(d,a,b,c,W[11],10,0xbd3af235UL)
202 II(c,d,a,b,W[2],15,0x2ad7d2bbUL)
203 II(b,c,d,a,W[9],21,0xeb86d391UL)
204
205 md5->state[0] = md5->state[0] + a;
206 md5->state[1] = md5->state[1] + b;
207 md5->state[2] = md5->state[2] + c;
208 md5->state[3] = md5->state[3] + d;
209 }
210
211
212 /**
213 Initialize the hash state
214 @param sha1 The hash state you wish to initialize
215 */
216 static void
md5_init(struct md5_state * md5)217 md5_init(struct md5_state *md5)
218 {
219 assert(md5 != NULL);
220 md5->state[0] = 0x67452301UL;
221 md5->state[1] = 0xefcdab89UL;
222 md5->state[2] = 0x98badcfeUL;
223 md5->state[3] = 0x10325476UL;
224 md5->curlen = 0;
225 md5->length = 0;
226 }
227
228 /**
229 Process a block of memory though the hash
230 @param sha1 The hash state
231 @param in The data to hash
232 @param inlen The length of the data (octets)
233 */
234 static void
md5_process(struct md5_state * md5,const unsigned char * in,Py_ssize_t inlen)235 md5_process(struct md5_state *md5, const unsigned char *in, Py_ssize_t inlen)
236 {
237 Py_ssize_t n;
238
239 assert(md5 != NULL);
240 assert(in != NULL);
241 assert(md5->curlen <= sizeof(md5->buf));
242
243 while (inlen > 0) {
244 if (md5->curlen == 0 && inlen >= MD5_BLOCKSIZE) {
245 md5_compress(md5, in);
246 md5->length += MD5_BLOCKSIZE * 8;
247 in += MD5_BLOCKSIZE;
248 inlen -= MD5_BLOCKSIZE;
249 } else {
250 n = Py_MIN(inlen, (Py_ssize_t)(MD5_BLOCKSIZE - md5->curlen));
251 memcpy(md5->buf + md5->curlen, in, (size_t)n);
252 md5->curlen += (MD5_INT32)n;
253 in += n;
254 inlen -= n;
255 if (md5->curlen == MD5_BLOCKSIZE) {
256 md5_compress(md5, md5->buf);
257 md5->length += 8*MD5_BLOCKSIZE;
258 md5->curlen = 0;
259 }
260 }
261 }
262 }
263
264 /**
265 Terminate the hash to get the digest
266 @param sha1 The hash state
267 @param out [out] The destination of the hash (16 bytes)
268 */
269 static void
md5_done(struct md5_state * md5,unsigned char * out)270 md5_done(struct md5_state *md5, unsigned char *out)
271 {
272 int i;
273
274 assert(md5 != NULL);
275 assert(out != NULL);
276 assert(md5->curlen < sizeof(md5->buf));
277
278 /* increase the length of the message */
279 md5->length += md5->curlen * 8;
280
281 /* append the '1' bit */
282 md5->buf[md5->curlen++] = (unsigned char)0x80;
283
284 /* if the length is currently above 56 bytes we append zeros
285 * then compress. Then we can fall back to padding zeros and length
286 * encoding like normal.
287 */
288 if (md5->curlen > 56) {
289 while (md5->curlen < 64) {
290 md5->buf[md5->curlen++] = (unsigned char)0;
291 }
292 md5_compress(md5, md5->buf);
293 md5->curlen = 0;
294 }
295
296 /* pad up to 56 bytes of zeroes */
297 while (md5->curlen < 56) {
298 md5->buf[md5->curlen++] = (unsigned char)0;
299 }
300
301 /* store length */
302 STORE64L(md5->length, md5->buf+56);
303 md5_compress(md5, md5->buf);
304
305 /* copy output */
306 for (i = 0; i < 4; i++) {
307 STORE32L(md5->state[i], out+(4*i));
308 }
309 }
310
311 /* .Source: /cvs/libtom/libtomcrypt/src/hashes/md5.c,v $ */
312 /* .Revision: 1.10 $ */
313 /* .Date: 2007/05/12 14:25:28 $ */
314
315 /*
316 * End of copied MD5 code.
317 *
318 * ------------------------------------------------------------------------
319 */
320
321 static PyTypeObject MD5type;
322
323
324 static MD5object *
newMD5object(void)325 newMD5object(void)
326 {
327 return (MD5object *)PyObject_New(MD5object, &MD5type);
328 }
329
330
331 /* Internal methods for a hash object */
332
333 static void
MD5_dealloc(PyObject * ptr)334 MD5_dealloc(PyObject *ptr)
335 {
336 PyObject_Del(ptr);
337 }
338
339
340 /* External methods for a hash object */
341
342 /*[clinic input]
343 MD5Type.copy
344
345 Return a copy of the hash object.
346 [clinic start generated code]*/
347
348 static PyObject *
MD5Type_copy_impl(MD5object * self)349 MD5Type_copy_impl(MD5object *self)
350 /*[clinic end generated code: output=596eb36852f02071 input=2c09e6d2493f3079]*/
351 {
352 MD5object *newobj;
353
354 if ((newobj = newMD5object())==NULL)
355 return NULL;
356
357 newobj->hash_state = self->hash_state;
358 return (PyObject *)newobj;
359 }
360
361 /*[clinic input]
362 MD5Type.digest
363
364 Return the digest value as a bytes object.
365 [clinic start generated code]*/
366
367 static PyObject *
MD5Type_digest_impl(MD5object * self)368 MD5Type_digest_impl(MD5object *self)
369 /*[clinic end generated code: output=eb691dc4190a07ec input=bc0c4397c2994be6]*/
370 {
371 unsigned char digest[MD5_DIGESTSIZE];
372 struct md5_state temp;
373
374 temp = self->hash_state;
375 md5_done(&temp, digest);
376 return PyBytes_FromStringAndSize((const char *)digest, MD5_DIGESTSIZE);
377 }
378
379 /*[clinic input]
380 MD5Type.hexdigest
381
382 Return the digest value as a string of hexadecimal digits.
383 [clinic start generated code]*/
384
385 static PyObject *
MD5Type_hexdigest_impl(MD5object * self)386 MD5Type_hexdigest_impl(MD5object *self)
387 /*[clinic end generated code: output=17badced1f3ac932 input=b60b19de644798dd]*/
388 {
389 unsigned char digest[MD5_DIGESTSIZE];
390 struct md5_state temp;
391
392 /* Get the raw (binary) digest value */
393 temp = self->hash_state;
394 md5_done(&temp, digest);
395
396 return _Py_strhex((const char*)digest, MD5_DIGESTSIZE);
397 }
398
399 /*[clinic input]
400 MD5Type.update
401
402 obj: object
403 /
404
405 Update this hash object's state with the provided string.
406 [clinic start generated code]*/
407
408 static PyObject *
MD5Type_update(MD5object * self,PyObject * obj)409 MD5Type_update(MD5object *self, PyObject *obj)
410 /*[clinic end generated code: output=f6ad168416338423 input=6e1efcd9ecf17032]*/
411 {
412 Py_buffer buf;
413
414 GET_BUFFER_VIEW_OR_ERROUT(obj, &buf);
415
416 md5_process(&self->hash_state, buf.buf, buf.len);
417
418 PyBuffer_Release(&buf);
419 Py_RETURN_NONE;
420 }
421
422 static PyMethodDef MD5_methods[] = {
423 MD5TYPE_COPY_METHODDEF
424 MD5TYPE_DIGEST_METHODDEF
425 MD5TYPE_HEXDIGEST_METHODDEF
426 MD5TYPE_UPDATE_METHODDEF
427 {NULL, NULL} /* sentinel */
428 };
429
430 static PyObject *
MD5_get_block_size(PyObject * self,void * closure)431 MD5_get_block_size(PyObject *self, void *closure)
432 {
433 return PyLong_FromLong(MD5_BLOCKSIZE);
434 }
435
436 static PyObject *
MD5_get_name(PyObject * self,void * closure)437 MD5_get_name(PyObject *self, void *closure)
438 {
439 return PyUnicode_FromStringAndSize("md5", 3);
440 }
441
442 static PyObject *
md5_get_digest_size(PyObject * self,void * closure)443 md5_get_digest_size(PyObject *self, void *closure)
444 {
445 return PyLong_FromLong(MD5_DIGESTSIZE);
446 }
447
448
449 static PyGetSetDef MD5_getseters[] = {
450 {"block_size",
451 (getter)MD5_get_block_size, NULL,
452 NULL,
453 NULL},
454 {"name",
455 (getter)MD5_get_name, NULL,
456 NULL,
457 NULL},
458 {"digest_size",
459 (getter)md5_get_digest_size, NULL,
460 NULL,
461 NULL},
462 {NULL} /* Sentinel */
463 };
464
465 static PyTypeObject MD5type = {
466 PyVarObject_HEAD_INIT(NULL, 0)
467 "_md5.md5", /*tp_name*/
468 sizeof(MD5object), /*tp_basicsize*/
469 0, /*tp_itemsize*/
470 /* methods */
471 MD5_dealloc, /*tp_dealloc*/
472 0, /*tp_vectorcall_offset*/
473 0, /*tp_getattr*/
474 0, /*tp_setattr*/
475 0, /*tp_as_async*/
476 0, /*tp_repr*/
477 0, /*tp_as_number*/
478 0, /*tp_as_sequence*/
479 0, /*tp_as_mapping*/
480 0, /*tp_hash*/
481 0, /*tp_call*/
482 0, /*tp_str*/
483 0, /*tp_getattro*/
484 0, /*tp_setattro*/
485 0, /*tp_as_buffer*/
486 Py_TPFLAGS_DEFAULT, /*tp_flags*/
487 0, /*tp_doc*/
488 0, /*tp_traverse*/
489 0, /*tp_clear*/
490 0, /*tp_richcompare*/
491 0, /*tp_weaklistoffset*/
492 0, /*tp_iter*/
493 0, /*tp_iternext*/
494 MD5_methods, /* tp_methods */
495 NULL, /* tp_members */
496 MD5_getseters, /* tp_getset */
497 };
498
499
500 /* The single module-level function: new() */
501
502 /*[clinic input]
503 _md5.md5
504
505 string: object(c_default="NULL") = b''
506 *
507 usedforsecurity: bool = True
508
509 Return a new MD5 hash object; optionally initialized with a string.
510 [clinic start generated code]*/
511
512 static PyObject *
_md5_md5_impl(PyObject * module,PyObject * string,int usedforsecurity)513 _md5_md5_impl(PyObject *module, PyObject *string, int usedforsecurity)
514 /*[clinic end generated code: output=587071f76254a4ac input=7a144a1905636985]*/
515 {
516 MD5object *new;
517 Py_buffer buf;
518
519 if (string)
520 GET_BUFFER_VIEW_OR_ERROUT(string, &buf);
521
522 if ((new = newMD5object()) == NULL) {
523 if (string)
524 PyBuffer_Release(&buf);
525 return NULL;
526 }
527
528 md5_init(&new->hash_state);
529
530 if (PyErr_Occurred()) {
531 Py_DECREF(new);
532 if (string)
533 PyBuffer_Release(&buf);
534 return NULL;
535 }
536 if (string) {
537 md5_process(&new->hash_state, buf.buf, buf.len);
538 PyBuffer_Release(&buf);
539 }
540
541 return (PyObject *)new;
542 }
543
544
545 /* List of functions exported by this module */
546
547 static struct PyMethodDef MD5_functions[] = {
548 _MD5_MD5_METHODDEF
549 {NULL, NULL} /* Sentinel */
550 };
551
552
553 /* Initialize this module. */
554
555 static struct PyModuleDef _md5module = {
556 PyModuleDef_HEAD_INIT,
557 "_md5",
558 NULL,
559 -1,
560 MD5_functions,
561 NULL,
562 NULL,
563 NULL,
564 NULL
565 };
566
567 PyMODINIT_FUNC
PyInit__md5(void)568 PyInit__md5(void)
569 {
570 PyObject *m;
571
572 Py_SET_TYPE(&MD5type, &PyType_Type);
573 if (PyType_Ready(&MD5type) < 0) {
574 return NULL;
575 }
576
577 m = PyModule_Create(&_md5module);
578 if (m == NULL) {
579 return NULL;
580 }
581
582 Py_INCREF((PyObject *)&MD5type);
583 PyModule_AddObject(m, "MD5Type", (PyObject *)&MD5type);
584 return m;
585 }
586