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 ([email protected])
9 Greg Stein ([email protected])
10 Trevor Perrin ([email protected])
11
12 Copyright (C) 2005-2007 Gregory P. Smith ([email protected])
13 Licensed to PSF under a Contributor Agreement.
14
15 */
16
17 /* MD5 objects */
18 #ifndef Py_BUILD_CORE_BUILTIN
19 # define Py_BUILD_CORE_MODULE 1
20 #endif
21
22 #include "Python.h"
23 #include "hashlib.h"
24 #include "pycore_strhex.h" // _Py_strhex()
25
26 /*[clinic input]
27 module _md5
28 class MD5Type "MD5object *" "&PyType_Type"
29 [clinic start generated code]*/
30 /*[clinic end generated code: output=da39a3ee5e6b4b0d input=6e5261719957a912]*/
31
32 /* Some useful types */
33
34 #if SIZEOF_INT == 4
35 typedef unsigned int MD5_INT32; /* 32-bit integer */
36 typedef long long MD5_INT64; /* 64-bit integer */
37 #else
38 /* not defined. compilation will die. */
39 #endif
40
41 /* The MD5 block size and message digest sizes, in bytes */
42
43 #define MD5_BLOCKSIZE 64
44 #define MD5_DIGESTSIZE 16
45
46 /* The structure for storing MD5 info */
47
48 struct md5_state {
49 MD5_INT64 length;
50 MD5_INT32 state[4], curlen;
51 unsigned char buf[MD5_BLOCKSIZE];
52 };
53
54 typedef struct {
55 PyObject_HEAD
56
57 struct md5_state hash_state;
58 } MD5object;
59
60 #include "clinic/md5module.c.h"
61
62 /* ------------------------------------------------------------------------
63 *
64 * This code for the MD5 algorithm was noted as public domain. The
65 * original headers are pasted below.
66 *
67 * Several changes have been made to make it more compatible with the
68 * Python environment and desired interface.
69 *
70 */
71
72 /* LibTomCrypt, modular cryptographic library -- Tom St Denis
73 *
74 * LibTomCrypt is a library that provides various cryptographic
75 * algorithms in a highly modular and flexible manner.
76 *
77 * The library is free for all purposes without any express
78 * guarantee it works.
79 *
80 * Tom St Denis, [email protected], https://www.libtom.net
81 */
82
83 /* rotate the hard way (platform optimizations could be done) */
84 #define ROLc(x, y) ( (((unsigned long)(x)<<(unsigned long)((y)&31)) | (((unsigned long)(x)&0xFFFFFFFFUL)>>(unsigned long)(32-((y)&31)))) & 0xFFFFFFFFUL)
85
86 /* Endian Neutral macros that work on all platforms */
87
88 #define STORE32L(x, y) \
89 { (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255); \
90 (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
91
92 #define LOAD32L(x, y) \
93 { x = ((unsigned long)((y)[3] & 255)<<24) | \
94 ((unsigned long)((y)[2] & 255)<<16) | \
95 ((unsigned long)((y)[1] & 255)<<8) | \
96 ((unsigned long)((y)[0] & 255)); }
97
98 #define STORE64L(x, y) \
99 { (y)[7] = (unsigned char)(((x)>>56)&255); (y)[6] = (unsigned char)(((x)>>48)&255); \
100 (y)[5] = (unsigned char)(((x)>>40)&255); (y)[4] = (unsigned char)(((x)>>32)&255); \
101 (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255); \
102 (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
103
104
105 /* MD5 macros */
106
107 #define F(x,y,z) (z ^ (x & (y ^ z)))
108 #define G(x,y,z) (y ^ (z & (y ^ x)))
109 #define H(x,y,z) (x^y^z)
110 #define I(x,y,z) (y^(x|(~z)))
111
112 #define FF(a,b,c,d,M,s,t) \
113 a = (a + F(b,c,d) + M + t); a = ROLc(a, s) + b;
114
115 #define GG(a,b,c,d,M,s,t) \
116 a = (a + G(b,c,d) + M + t); a = ROLc(a, s) + b;
117
118 #define HH(a,b,c,d,M,s,t) \
119 a = (a + H(b,c,d) + M + t); a = ROLc(a, s) + b;
120
121 #define II(a,b,c,d,M,s,t) \
122 a = (a + I(b,c,d) + M + t); a = ROLc(a, s) + b;
123
124
md5_compress(struct md5_state * md5,const unsigned char * buf)125 static void md5_compress(struct md5_state *md5, const unsigned char *buf)
126 {
127 MD5_INT32 i, W[16], a, b, c, d;
128
129 assert(md5 != NULL);
130 assert(buf != NULL);
131
132 /* copy the state into 512-bits into W[0..15] */
133 for (i = 0; i < 16; i++) {
134 LOAD32L(W[i], buf + (4*i));
135 }
136
137 /* copy state */
138 a = md5->state[0];
139 b = md5->state[1];
140 c = md5->state[2];
141 d = md5->state[3];
142
143 FF(a,b,c,d,W[0],7,0xd76aa478UL)
144 FF(d,a,b,c,W[1],12,0xe8c7b756UL)
145 FF(c,d,a,b,W[2],17,0x242070dbUL)
146 FF(b,c,d,a,W[3],22,0xc1bdceeeUL)
147 FF(a,b,c,d,W[4],7,0xf57c0fafUL)
148 FF(d,a,b,c,W[5],12,0x4787c62aUL)
149 FF(c,d,a,b,W[6],17,0xa8304613UL)
150 FF(b,c,d,a,W[7],22,0xfd469501UL)
151 FF(a,b,c,d,W[8],7,0x698098d8UL)
152 FF(d,a,b,c,W[9],12,0x8b44f7afUL)
153 FF(c,d,a,b,W[10],17,0xffff5bb1UL)
154 FF(b,c,d,a,W[11],22,0x895cd7beUL)
155 FF(a,b,c,d,W[12],7,0x6b901122UL)
156 FF(d,a,b,c,W[13],12,0xfd987193UL)
157 FF(c,d,a,b,W[14],17,0xa679438eUL)
158 FF(b,c,d,a,W[15],22,0x49b40821UL)
159 GG(a,b,c,d,W[1],5,0xf61e2562UL)
160 GG(d,a,b,c,W[6],9,0xc040b340UL)
161 GG(c,d,a,b,W[11],14,0x265e5a51UL)
162 GG(b,c,d,a,W[0],20,0xe9b6c7aaUL)
163 GG(a,b,c,d,W[5],5,0xd62f105dUL)
164 GG(d,a,b,c,W[10],9,0x02441453UL)
165 GG(c,d,a,b,W[15],14,0xd8a1e681UL)
166 GG(b,c,d,a,W[4],20,0xe7d3fbc8UL)
167 GG(a,b,c,d,W[9],5,0x21e1cde6UL)
168 GG(d,a,b,c,W[14],9,0xc33707d6UL)
169 GG(c,d,a,b,W[3],14,0xf4d50d87UL)
170 GG(b,c,d,a,W[8],20,0x455a14edUL)
171 GG(a,b,c,d,W[13],5,0xa9e3e905UL)
172 GG(d,a,b,c,W[2],9,0xfcefa3f8UL)
173 GG(c,d,a,b,W[7],14,0x676f02d9UL)
174 GG(b,c,d,a,W[12],20,0x8d2a4c8aUL)
175 HH(a,b,c,d,W[5],4,0xfffa3942UL)
176 HH(d,a,b,c,W[8],11,0x8771f681UL)
177 HH(c,d,a,b,W[11],16,0x6d9d6122UL)
178 HH(b,c,d,a,W[14],23,0xfde5380cUL)
179 HH(a,b,c,d,W[1],4,0xa4beea44UL)
180 HH(d,a,b,c,W[4],11,0x4bdecfa9UL)
181 HH(c,d,a,b,W[7],16,0xf6bb4b60UL)
182 HH(b,c,d,a,W[10],23,0xbebfbc70UL)
183 HH(a,b,c,d,W[13],4,0x289b7ec6UL)
184 HH(d,a,b,c,W[0],11,0xeaa127faUL)
185 HH(c,d,a,b,W[3],16,0xd4ef3085UL)
186 HH(b,c,d,a,W[6],23,0x04881d05UL)
187 HH(a,b,c,d,W[9],4,0xd9d4d039UL)
188 HH(d,a,b,c,W[12],11,0xe6db99e5UL)
189 HH(c,d,a,b,W[15],16,0x1fa27cf8UL)
190 HH(b,c,d,a,W[2],23,0xc4ac5665UL)
191 II(a,b,c,d,W[0],6,0xf4292244UL)
192 II(d,a,b,c,W[7],10,0x432aff97UL)
193 II(c,d,a,b,W[14],15,0xab9423a7UL)
194 II(b,c,d,a,W[5],21,0xfc93a039UL)
195 II(a,b,c,d,W[12],6,0x655b59c3UL)
196 II(d,a,b,c,W[3],10,0x8f0ccc92UL)
197 II(c,d,a,b,W[10],15,0xffeff47dUL)
198 II(b,c,d,a,W[1],21,0x85845dd1UL)
199 II(a,b,c,d,W[8],6,0x6fa87e4fUL)
200 II(d,a,b,c,W[15],10,0xfe2ce6e0UL)
201 II(c,d,a,b,W[6],15,0xa3014314UL)
202 II(b,c,d,a,W[13],21,0x4e0811a1UL)
203 II(a,b,c,d,W[4],6,0xf7537e82UL)
204 II(d,a,b,c,W[11],10,0xbd3af235UL)
205 II(c,d,a,b,W[2],15,0x2ad7d2bbUL)
206 II(b,c,d,a,W[9],21,0xeb86d391UL)
207
208 md5->state[0] = md5->state[0] + a;
209 md5->state[1] = md5->state[1] + b;
210 md5->state[2] = md5->state[2] + c;
211 md5->state[3] = md5->state[3] + d;
212 }
213
214
215 /**
216 Initialize the hash state
217 @param md5 The hash state you wish to initialize
218 */
219 static void
md5_init(struct md5_state * md5)220 md5_init(struct md5_state *md5)
221 {
222 assert(md5 != NULL);
223 md5->state[0] = 0x67452301UL;
224 md5->state[1] = 0xefcdab89UL;
225 md5->state[2] = 0x98badcfeUL;
226 md5->state[3] = 0x10325476UL;
227 md5->curlen = 0;
228 md5->length = 0;
229 }
230
231 /**
232 Process a block of memory though the hash
233 @param md5 The hash state
234 @param in The data to hash
235 @param inlen The length of the data (octets)
236 */
237 static void
md5_process(struct md5_state * md5,const unsigned char * in,Py_ssize_t inlen)238 md5_process(struct md5_state *md5, const unsigned char *in, Py_ssize_t inlen)
239 {
240 Py_ssize_t n;
241
242 assert(md5 != NULL);
243 assert(in != NULL);
244 assert(md5->curlen <= sizeof(md5->buf));
245
246 while (inlen > 0) {
247 if (md5->curlen == 0 && inlen >= MD5_BLOCKSIZE) {
248 md5_compress(md5, in);
249 md5->length += MD5_BLOCKSIZE * 8;
250 in += MD5_BLOCKSIZE;
251 inlen -= MD5_BLOCKSIZE;
252 } else {
253 n = Py_MIN(inlen, (Py_ssize_t)(MD5_BLOCKSIZE - md5->curlen));
254 memcpy(md5->buf + md5->curlen, in, (size_t)n);
255 md5->curlen += (MD5_INT32)n;
256 in += n;
257 inlen -= n;
258 if (md5->curlen == MD5_BLOCKSIZE) {
259 md5_compress(md5, md5->buf);
260 md5->length += 8*MD5_BLOCKSIZE;
261 md5->curlen = 0;
262 }
263 }
264 }
265 }
266
267 /**
268 Terminate the hash to get the digest
269 @param md5 The hash state
270 @param out [out] The destination of the hash (16 bytes)
271 */
272 static void
md5_done(struct md5_state * md5,unsigned char * out)273 md5_done(struct md5_state *md5, unsigned char *out)
274 {
275 int i;
276
277 assert(md5 != NULL);
278 assert(out != NULL);
279 assert(md5->curlen < sizeof(md5->buf));
280
281 /* increase the length of the message */
282 md5->length += md5->curlen * 8;
283
284 /* append the '1' bit */
285 md5->buf[md5->curlen++] = (unsigned char)0x80;
286
287 /* if the length is currently above 56 bytes we append zeros
288 * then compress. Then we can fall back to padding zeros and length
289 * encoding like normal.
290 */
291 if (md5->curlen > 56) {
292 while (md5->curlen < 64) {
293 md5->buf[md5->curlen++] = (unsigned char)0;
294 }
295 md5_compress(md5, md5->buf);
296 md5->curlen = 0;
297 }
298
299 /* pad up to 56 bytes of zeroes */
300 while (md5->curlen < 56) {
301 md5->buf[md5->curlen++] = (unsigned char)0;
302 }
303
304 /* store length */
305 STORE64L(md5->length, md5->buf+56);
306 md5_compress(md5, md5->buf);
307
308 /* copy output */
309 for (i = 0; i < 4; i++) {
310 STORE32L(md5->state[i], out+(4*i));
311 }
312 }
313
314 /* .Source: /cvs/libtom/libtomcrypt/src/hashes/md5.c,v $ */
315 /* .Revision: 1.10 $ */
316 /* .Date: 2007/05/12 14:25:28 $ */
317
318 /*
319 * End of copied MD5 code.
320 *
321 * ------------------------------------------------------------------------
322 */
323
324 typedef struct {
325 PyTypeObject* md5_type;
326 } MD5State;
327
328 static inline MD5State*
md5_get_state(PyObject * module)329 md5_get_state(PyObject *module)
330 {
331 void *state = PyModule_GetState(module);
332 assert(state != NULL);
333 return (MD5State *)state;
334 }
335
336 static MD5object *
newMD5object(MD5State * st)337 newMD5object(MD5State * st)
338 {
339 MD5object *md5 = (MD5object *)PyObject_GC_New(MD5object, st->md5_type);
340 PyObject_GC_Track(md5);
341 return md5;
342 }
343
344 /* Internal methods for a hash object */
345 static int
MD5_traverse(PyObject * ptr,visitproc visit,void * arg)346 MD5_traverse(PyObject *ptr, visitproc visit, void *arg)
347 {
348 Py_VISIT(Py_TYPE(ptr));
349 return 0;
350 }
351
352 static void
MD5_dealloc(PyObject * ptr)353 MD5_dealloc(PyObject *ptr)
354 {
355 PyTypeObject *tp = Py_TYPE(ptr);
356 PyObject_GC_UnTrack(ptr);
357 PyObject_GC_Del(ptr);
358 Py_DECREF(tp);
359 }
360
361
362 /* External methods for a hash object */
363
364 /*[clinic input]
365 MD5Type.copy
366
367 cls: defining_class
368
369 Return a copy of the hash object.
370 [clinic start generated code]*/
371
372 static PyObject *
MD5Type_copy_impl(MD5object * self,PyTypeObject * cls)373 MD5Type_copy_impl(MD5object *self, PyTypeObject *cls)
374 /*[clinic end generated code: output=bf055e08244bf5ee input=d89087dcfb2a8620]*/
375 {
376 MD5State *st = PyType_GetModuleState(cls);
377
378 MD5object *newobj;
379 if ((newobj = newMD5object(st))==NULL)
380 return NULL;
381
382 newobj->hash_state = self->hash_state;
383 return (PyObject *)newobj;
384 }
385
386 /*[clinic input]
387 MD5Type.digest
388
389 Return the digest value as a bytes object.
390 [clinic start generated code]*/
391
392 static PyObject *
MD5Type_digest_impl(MD5object * self)393 MD5Type_digest_impl(MD5object *self)
394 /*[clinic end generated code: output=eb691dc4190a07ec input=bc0c4397c2994be6]*/
395 {
396 unsigned char digest[MD5_DIGESTSIZE];
397 struct md5_state temp;
398
399 temp = self->hash_state;
400 md5_done(&temp, digest);
401 return PyBytes_FromStringAndSize((const char *)digest, MD5_DIGESTSIZE);
402 }
403
404 /*[clinic input]
405 MD5Type.hexdigest
406
407 Return the digest value as a string of hexadecimal digits.
408 [clinic start generated code]*/
409
410 static PyObject *
MD5Type_hexdigest_impl(MD5object * self)411 MD5Type_hexdigest_impl(MD5object *self)
412 /*[clinic end generated code: output=17badced1f3ac932 input=b60b19de644798dd]*/
413 {
414 unsigned char digest[MD5_DIGESTSIZE];
415 struct md5_state temp;
416
417 /* Get the raw (binary) digest value */
418 temp = self->hash_state;
419 md5_done(&temp, digest);
420
421 return _Py_strhex((const char*)digest, MD5_DIGESTSIZE);
422 }
423
424 /*[clinic input]
425 MD5Type.update
426
427 obj: object
428 /
429
430 Update this hash object's state with the provided string.
431 [clinic start generated code]*/
432
433 static PyObject *
MD5Type_update(MD5object * self,PyObject * obj)434 MD5Type_update(MD5object *self, PyObject *obj)
435 /*[clinic end generated code: output=f6ad168416338423 input=6e1efcd9ecf17032]*/
436 {
437 Py_buffer buf;
438
439 GET_BUFFER_VIEW_OR_ERROUT(obj, &buf);
440
441 md5_process(&self->hash_state, buf.buf, buf.len);
442
443 PyBuffer_Release(&buf);
444 Py_RETURN_NONE;
445 }
446
447 static PyMethodDef MD5_methods[] = {
448 MD5TYPE_COPY_METHODDEF
449 MD5TYPE_DIGEST_METHODDEF
450 MD5TYPE_HEXDIGEST_METHODDEF
451 MD5TYPE_UPDATE_METHODDEF
452 {NULL, NULL} /* sentinel */
453 };
454
455 static PyObject *
MD5_get_block_size(PyObject * self,void * closure)456 MD5_get_block_size(PyObject *self, void *closure)
457 {
458 return PyLong_FromLong(MD5_BLOCKSIZE);
459 }
460
461 static PyObject *
MD5_get_name(PyObject * self,void * closure)462 MD5_get_name(PyObject *self, void *closure)
463 {
464 return PyUnicode_FromStringAndSize("md5", 3);
465 }
466
467 static PyObject *
md5_get_digest_size(PyObject * self,void * closure)468 md5_get_digest_size(PyObject *self, void *closure)
469 {
470 return PyLong_FromLong(MD5_DIGESTSIZE);
471 }
472
473 static PyGetSetDef MD5_getseters[] = {
474 {"block_size",
475 (getter)MD5_get_block_size, NULL,
476 NULL,
477 NULL},
478 {"name",
479 (getter)MD5_get_name, NULL,
480 NULL,
481 NULL},
482 {"digest_size",
483 (getter)md5_get_digest_size, NULL,
484 NULL,
485 NULL},
486 {NULL} /* Sentinel */
487 };
488
489 static PyType_Slot md5_type_slots[] = {
490 {Py_tp_dealloc, MD5_dealloc},
491 {Py_tp_methods, MD5_methods},
492 {Py_tp_getset, MD5_getseters},
493 {Py_tp_traverse, MD5_traverse},
494 {0,0}
495 };
496
497 static PyType_Spec md5_type_spec = {
498 .name = "_md5.md5",
499 .basicsize = sizeof(MD5object),
500 .flags = (Py_TPFLAGS_DEFAULT | Py_TPFLAGS_DISALLOW_INSTANTIATION |
501 Py_TPFLAGS_IMMUTABLETYPE | Py_TPFLAGS_HAVE_GC),
502 .slots = md5_type_slots
503 };
504
505 /* The single module-level function: new() */
506
507 /*[clinic input]
508 _md5.md5
509
510 string: object(c_default="NULL") = b''
511 *
512 usedforsecurity: bool = True
513
514 Return a new MD5 hash object; optionally initialized with a string.
515 [clinic start generated code]*/
516
517 static PyObject *
_md5_md5_impl(PyObject * module,PyObject * string,int usedforsecurity)518 _md5_md5_impl(PyObject *module, PyObject *string, int usedforsecurity)
519 /*[clinic end generated code: output=587071f76254a4ac input=7a144a1905636985]*/
520 {
521 MD5object *new;
522 Py_buffer buf;
523
524 if (string)
525 GET_BUFFER_VIEW_OR_ERROUT(string, &buf);
526
527 MD5State *st = md5_get_state(module);
528 if ((new = newMD5object(st)) == NULL) {
529 if (string)
530 PyBuffer_Release(&buf);
531 return NULL;
532 }
533
534 md5_init(&new->hash_state);
535
536 if (PyErr_Occurred()) {
537 Py_DECREF(new);
538 if (string)
539 PyBuffer_Release(&buf);
540 return NULL;
541 }
542 if (string) {
543 md5_process(&new->hash_state, buf.buf, buf.len);
544 PyBuffer_Release(&buf);
545 }
546
547 return (PyObject *)new;
548 }
549
550
551 /* List of functions exported by this module */
552
553 static struct PyMethodDef MD5_functions[] = {
554 _MD5_MD5_METHODDEF
555 {NULL, NULL} /* Sentinel */
556 };
557
558 static int
_md5_traverse(PyObject * module,visitproc visit,void * arg)559 _md5_traverse(PyObject *module, visitproc visit, void *arg)
560 {
561 MD5State *state = md5_get_state(module);
562 Py_VISIT(state->md5_type);
563 return 0;
564 }
565
566 static int
_md5_clear(PyObject * module)567 _md5_clear(PyObject *module)
568 {
569 MD5State *state = md5_get_state(module);
570 Py_CLEAR(state->md5_type);
571 return 0;
572 }
573
574 static void
_md5_free(void * module)575 _md5_free(void *module)
576 {
577 _md5_clear((PyObject *)module);
578 }
579
580 /* Initialize this module. */
581 static int
md5_exec(PyObject * m)582 md5_exec(PyObject *m)
583 {
584 MD5State *st = md5_get_state(m);
585
586 st->md5_type = (PyTypeObject *)PyType_FromModuleAndSpec(
587 m, &md5_type_spec, NULL);
588
589 if (st->md5_type == NULL) {
590 return -1;
591 }
592
593 Py_INCREF((PyObject *)st->md5_type);
594 if (PyModule_AddObject(m, "MD5Type", (PyObject *)st->md5_type) < 0) {
595 Py_DECREF(st->md5_type);
596 return -1;
597 }
598
599 return 0;
600 }
601
602 static PyModuleDef_Slot _md5_slots[] = {
603 {Py_mod_exec, md5_exec},
604 {0, NULL}
605 };
606
607
608 static struct PyModuleDef _md5module = {
609 PyModuleDef_HEAD_INIT,
610 .m_name = "_md5",
611 .m_size = sizeof(MD5State),
612 .m_methods = MD5_functions,
613 .m_slots = _md5_slots,
614 .m_traverse = _md5_traverse,
615 .m_clear = _md5_clear,
616 .m_free = _md5_free,
617 };
618
619 PyMODINIT_FUNC
PyInit__md5(void)620 PyInit__md5(void)
621 {
622 return PyModuleDef_Init(&_md5module);
623 }
624