Modules/md5module.c

   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
  22
  23 /* Some useful types */
  24
  25 #if SIZEOF_INT == 4
  26 typedef unsigned int MD5_INT32; /* 32-bit integer */
  27 typedef PY_LONG_LONG MD5_INT64; /* 64-bit integer */
  28 #else
  29 /* not defined. compilation will die. */
  30 #endif
  31
  32 /* The MD5 block size and message digest sizes, in bytes */
  33
  34 #define MD5_BLOCKSIZE    64
  35 #define MD5_DIGESTSIZE   16
  36
  37 /* The structure for storing MD5 info */
  38
  39 struct md5_state {
  40     MD5_INT64 length;
  41     MD5_INT32 state[4], curlen;
  42     unsigned char buf[MD5_BLOCKSIZE];
  43 };
  44
  45 typedef struct {
  46     PyObject_HEAD
  47
  48     struct md5_state hash_state;
  49 } MD5object;
  50
  51
  52 /* ------------------------------------------------------------------------
  53  *
  54  * This code for the MD5 algorithm was noted as public domain. The
  55  * original headers are pasted below.
  56  *
  57  * Several changes have been made to make it more compatible with the
  58  * Python environment and desired interface.
  59  *
  60  */
  61
  62 /* LibTomCrypt, modular cryptographic library -- Tom St Denis
  63  *
  64  * LibTomCrypt is a library that provides various cryptographic
  65  * algorithms in a highly modular and flexible manner.
  66  *
  67  * The library is free for all purposes without any express
  68  * guarantee it works.
  69  *
  70  * Tom St Denis, tomstdenis@gmail.com, http://libtom.org
  71  */
  72
  73 /* rotate the hard way (platform optimizations could be done) */
  74 #define ROLc(x, y) ( (((unsigned long)(x)<<(unsigned long)((y)&31)) | (((unsigned long)(x)&0xFFFFFFFFUL)>>(unsigned long)(32-((y)&31)))) & 0xFFFFFFFFUL)
  75
  76 /* Endian Neutral macros that work on all platforms */
  77
  78 #define STORE32L(x, y)                                                                     \
  79      { (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255);   \
  80        (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
  81
  82 #define LOAD32L(x, y)                            \
  83      { x = ((unsigned long)((y)[3] & 255)<<24) | \
  84            ((unsigned long)((y)[2] & 255)<<16) | \
  85            ((unsigned long)((y)[1] & 255)<<8)  | \
  86            ((unsigned long)((y)[0] & 255)); }
  87
  88 #define STORE64L(x, y)                                                                     \
  89      { (y)[7] = (unsigned char)(((x)>>56)&255); (y)[6] = (unsigned char)(((x)>>48)&255);   \
  90        (y)[5] = (unsigned char)(((x)>>40)&255); (y)[4] = (unsigned char)(((x)>>32)&255);   \
  91        (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255);   \
  92        (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
  93
  94 #ifndef MIN
  95    #define MIN(x, y) ( ((x)<(y))?(x):(y) )
  96 #endif
  97
  98
  99 /* MD5 macros */
 100
 101 #define F(x,y,z)  (z ^ (x & (y ^ z)))
 102 #define G(x,y,z)  (y ^ (z & (y ^ x)))
 103 #define H(x,y,z)  (x^y^z)
 104 #define I(x,y,z)  (y^(x|(~z)))
 105
 106 #define FF(a,b,c,d,M,s,t) \
 107     a = (a + F(b,c,d) + M + t); a = ROLc(a, s) + b;
 108
 109 #define GG(a,b,c,d,M,s,t) \
 110     a = (a + G(b,c,d) + M + t); a = ROLc(a, s) + b;
 111
 112 #define HH(a,b,c,d,M,s,t) \
 113     a = (a + H(b,c,d) + M + t); a = ROLc(a, s) + b;
 114
 115 #define II(a,b,c,d,M,s,t) \
 116     a = (a + I(b,c,d) + M + t); a = ROLc(a, s) + b;
 117
 118
 119 static void md5_compress(struct md5_state *md5, unsigned char *buf)
 120 {
 121     MD5_INT32 i, W[16], a, b, c, d;
 122
 123     assert(md5 != NULL);
 124     assert(buf != NULL);
 125
 126     /* copy the state into 512-bits into W[0..15] */
 127     for (i = 0; i < 16; i++) {
 128         LOAD32L(W[i], buf + (4*i));
 129     }
 130
 131     /* copy state */
 132     a = md5->state[0];
 133     b = md5->state[1];
 134     c = md5->state[2];
 135     d = md5->state[3];
 136
 137     FF(a,b,c,d,W[0],7,0xd76aa478UL)
 138     FF(d,a,b,c,W[1],12,0xe8c7b756UL)
 139     FF(c,d,a,b,W[2],17,0x242070dbUL)
 140     FF(b,c,d,a,W[3],22,0xc1bdceeeUL)
 141     FF(a,b,c,d,W[4],7,0xf57c0fafUL)
 142     FF(d,a,b,c,W[5],12,0x4787c62aUL)
 143     FF(c,d,a,b,W[6],17,0xa8304613UL)
 144     FF(b,c,d,a,W[7],22,0xfd469501UL)
 145     FF(a,b,c,d,W[8],7,0x698098d8UL)
 146     FF(d,a,b,c,W[9],12,0x8b44f7afUL)
 147     FF(c,d,a,b,W[10],17,0xffff5bb1UL)
 148     FF(b,c,d,a,W[11],22,0x895cd7beUL)
 149     FF(a,b,c,d,W[12],7,0x6b901122UL)
 150     FF(d,a,b,c,W[13],12,0xfd987193UL)
 151     FF(c,d,a,b,W[14],17,0xa679438eUL)
 152     FF(b,c,d,a,W[15],22,0x49b40821UL)
 153     GG(a,b,c,d,W[1],5,0xf61e2562UL)
 154     GG(d,a,b,c,W[6],9,0xc040b340UL)
 155     GG(c,d,a,b,W[11],14,0x265e5a51UL)
 156     GG(b,c,d,a,W[0],20,0xe9b6c7aaUL)
 157     GG(a,b,c,d,W[5],5,0xd62f105dUL)
 158     GG(d,a,b,c,W[10],9,0x02441453UL)
 159     GG(c,d,a,b,W[15],14,0xd8a1e681UL)
 160     GG(b,c,d,a,W[4],20,0xe7d3fbc8UL)
 161     GG(a,b,c,d,W[9],5,0x21e1cde6UL)
 162     GG(d,a,b,c,W[14],9,0xc33707d6UL)
 163     GG(c,d,a,b,W[3],14,0xf4d50d87UL)
 164     GG(b,c,d,a,W[8],20,0x455a14edUL)
 165     GG(a,b,c,d,W[13],5,0xa9e3e905UL)
 166     GG(d,a,b,c,W[2],9,0xfcefa3f8UL)
 167     GG(c,d,a,b,W[7],14,0x676f02d9UL)
 168     GG(b,c,d,a,W[12],20,0x8d2a4c8aUL)
 169     HH(a,b,c,d,W[5],4,0xfffa3942UL)
 170     HH(d,a,b,c,W[8],11,0x8771f681UL)
 171     HH(c,d,a,b,W[11],16,0x6d9d6122UL)
 172     HH(b,c,d,a,W[14],23,0xfde5380cUL)
 173     HH(a,b,c,d,W[1],4,0xa4beea44UL)
 174     HH(d,a,b,c,W[4],11,0x4bdecfa9UL)
 175     HH(c,d,a,b,W[7],16,0xf6bb4b60UL)
 176     HH(b,c,d,a,W[10],23,0xbebfbc70UL)
 177     HH(a,b,c,d,W[13],4,0x289b7ec6UL)
 178     HH(d,a,b,c,W[0],11,0xeaa127faUL)
 179     HH(c,d,a,b,W[3],16,0xd4ef3085UL)
 180     HH(b,c,d,a,W[6],23,0x04881d05UL)
 181     HH(a,b,c,d,W[9],4,0xd9d4d039UL)
 182     HH(d,a,b,c,W[12],11,0xe6db99e5UL)
 183     HH(c,d,a,b,W[15],16,0x1fa27cf8UL)
 184     HH(b,c,d,a,W[2],23,0xc4ac5665UL)
 185     II(a,b,c,d,W[0],6,0xf4292244UL)
 186     II(d,a,b,c,W[7],10,0x432aff97UL)
 187     II(c,d,a,b,W[14],15,0xab9423a7UL)
 188     II(b,c,d,a,W[5],21,0xfc93a039UL)
 189     II(a,b,c,d,W[12],6,0x655b59c3UL)
 190     II(d,a,b,c,W[3],10,0x8f0ccc92UL)
 191     II(c,d,a,b,W[10],15,0xffeff47dUL)
 192     II(b,c,d,a,W[1],21,0x85845dd1UL)
 193     II(a,b,c,d,W[8],6,0x6fa87e4fUL)
 194     II(d,a,b,c,W[15],10,0xfe2ce6e0UL)
 195     II(c,d,a,b,W[6],15,0xa3014314UL)
 196     II(b,c,d,a,W[13],21,0x4e0811a1UL)
 197     II(a,b,c,d,W[4],6,0xf7537e82UL)
 198     II(d,a,b,c,W[11],10,0xbd3af235UL)
 199     II(c,d,a,b,W[2],15,0x2ad7d2bbUL)
 200     II(b,c,d,a,W[9],21,0xeb86d391UL)
 201
 202     md5->state[0] = md5->state[0] + a;
 203     md5->state[1] = md5->state[1] + b;
 204     md5->state[2] = md5->state[2] + c;
 205     md5->state[3] = md5->state[3] + d;
 206 }
 207
 208
 209 /**
 210    Initialize the hash state
 211    @param sha1   The hash state you wish to initialize
 212 */
 213 void md5_init(struct md5_state *md5)
 214 {
 215     assert(md5 != NULL);
 216     md5->state[0] = 0x67452301UL;
 217     md5->state[1] = 0xefcdab89UL;
 218     md5->state[2] = 0x98badcfeUL;
 219     md5->state[3] = 0x10325476UL;
 220     md5->curlen = 0;
 221     md5->length = 0;
 222 }
 223
 224 /**
 225    Process a block of memory though the hash
 226    @param sha1   The hash state
 227    @param in     The data to hash
 228    @param inlen  The length of the data (octets)
 229 */
 230 void md5_process(struct md5_state *md5,
 231                 const unsigned char *in, unsigned long inlen)
 232 {
 233     unsigned long n;
 234
 235     assert(md5 != NULL);
 236     assert(in != NULL);
 237     assert(md5->curlen <= sizeof(md5->buf));
 238
 239     while (inlen > 0) {
 240         if (md5->curlen == 0 && inlen >= MD5_BLOCKSIZE) {
 241            md5_compress(md5, (unsigned char *)in);
 242            md5->length    += MD5_BLOCKSIZE * 8;
 243            in             += MD5_BLOCKSIZE;
 244            inlen          -= MD5_BLOCKSIZE;
 245         } else {
 246            n = MIN(inlen, (MD5_BLOCKSIZE - md5->curlen));
 247            memcpy(md5->buf + md5->curlen, in, (size_t)n);
 248            md5->curlen    += n;
 249            in             += n;
 250            inlen          -= n;
 251            if (md5->curlen == MD5_BLOCKSIZE) {
 252               md5_compress(md5, md5->buf);
 253               md5->length += 8*MD5_BLOCKSIZE;
 254               md5->curlen = 0;
 255            }
 256        }
 257     }
 258 }
 259
 260 /**
 261    Terminate the hash to get the digest
 262    @param sha1  The hash state
 263    @param out [out] The destination of the hash (16 bytes)
 264 */
 265 void md5_done(struct md5_state *md5, unsigned char *out)
 266 {
 267     int i;
 268
 269     assert(md5 != NULL);
 270     assert(out != NULL);
 271     assert(md5->curlen < sizeof(md5->buf));
 272
 273     /* increase the length of the message */
 274     md5->length += md5->curlen * 8;
 275
 276     /* append the '1' bit */
 277     md5->buf[md5->curlen++] = (unsigned char)0x80;
 278
 279     /* if the length is currently above 56 bytes we append zeros
 280      * then compress.  Then we can fall back to padding zeros and length
 281      * encoding like normal.
 282      */
 283     if (md5->curlen > 56) {
 284         while (md5->curlen < 64) {
 285             md5->buf[md5->curlen++] = (unsigned char)0;
 286         }
 287         md5_compress(md5, md5->buf);
 288         md5->curlen = 0;
 289     }
 290
 291     /* pad upto 56 bytes of zeroes */
 292     while (md5->curlen < 56) {
 293         md5->buf[md5->curlen++] = (unsigned char)0;
 294     }
 295
 296     /* store length */
 297     STORE64L(md5->length, md5->buf+56);
 298     md5_compress(md5, md5->buf);
 299
 300     /* copy output */
 301     for (i = 0; i < 4; i++) {
 302         STORE32L(md5->state[i], out+(4*i));
 303     }
 304 }
 305
 306 /* .Source: /cvs/libtom/libtomcrypt/src/hashes/md5.c,v $ */
 307 /* .Revision: 1.10 $ */
 308 /* .Date: 2007/05/12 14:25:28 $ */
 309
 310 /*
 311  * End of copied MD5 code.
 312  *
 313  * ------------------------------------------------------------------------
 314  */
 315
 316 static PyTypeObject MD5type;
 317
 318
 319 static MD5object *
 320 newMD5object(void)
 321 {
 322     return (MD5object *)PyObject_New(MD5object, &MD5type);
 323 }
 324
 325
 326 /* Internal methods for a hash object */
 327
 328 static void
 329 MD5_dealloc(PyObject *ptr)
 330 {
 331     PyObject_Del(ptr);
 332 }
 333
 334
 335 /* External methods for a hash object */
 336
 337 PyDoc_STRVAR(MD5_copy__doc__, "Return a copy of the hash object.");
 338
 339 static PyObject *
 340 MD5_copy(MD5object *self, PyObject *unused)
 341 {
 342     MD5object *newobj;
 343
 344     if (Py_TYPE(self) == &MD5type) {
 345         if ( (newobj = newMD5object())==NULL)
 346             return NULL;
 347     } else {
 348         if ( (newobj = newMD5object())==NULL)
 349             return NULL;
 350     }
 351
 352     newobj->hash_state = self->hash_state;
 353     return (PyObject *)newobj;
 354 }
 355
 356 PyDoc_STRVAR(MD5_digest__doc__,
 357 "Return the digest value as a string of binary data.");
 358
 359 static PyObject *
 360 MD5_digest(MD5object *self, PyObject *unused)
 361 {
 362     unsigned char digest[MD5_DIGESTSIZE];
 363     struct md5_state temp;
 364
 365     temp = self->hash_state;
 366     md5_done(&temp, digest);
 367     return PyBytes_FromStringAndSize((const char *)digest, MD5_DIGESTSIZE);
 368 }
 369
 370 PyDoc_STRVAR(MD5_hexdigest__doc__,
 371 "Return the digest value as a string of hexadecimal digits.");
 372
 373 static PyObject *
 374 MD5_hexdigest(MD5object *self, PyObject *unused)
 375 {
 376     unsigned char digest[MD5_DIGESTSIZE];
 377     struct md5_state temp;
 378     PyObject *retval;
 379     Py_UNICODE *hex_digest;
 380     int i, j;
 381
 382     /* Get the raw (binary) digest value */
 383     temp = self->hash_state;
 384     md5_done(&temp, digest);
 385
 386     /* Create a new string */
 387     retval = PyUnicode_FromStringAndSize(NULL, MD5_DIGESTSIZE * 2);
 388     if (!retval)
 389             return NULL;
 390     hex_digest = PyUnicode_AS_UNICODE(retval);
 391     if (!hex_digest) {
 392             Py_DECREF(retval);
 393             return NULL;
 394     }
 395
 396     /* Make hex version of the digest */
 397     for(i=j=0; i<MD5_DIGESTSIZE; i++) {
 398         char c;
 399         c = (digest[i] >> 4) & 0xf;
 400         c = (c>9) ? c+'a'-10 : c + '0';
 401         hex_digest[j++] = c;
 402         c = (digest[i] & 0xf);
 403         c = (c>9) ? c+'a'-10 : c + '0';
 404         hex_digest[j++] = c;
 405     }
 406     return retval;
 407 }
 408
 409 PyDoc_STRVAR(MD5_update__doc__,
 410 "Update this hash object's state with the provided string.");
 411
 412 static PyObject *
 413 MD5_update(MD5object *self, PyObject *args)
 414 {
 415     PyObject *obj;
 416     Py_buffer buf;
 417
 418     if (!PyArg_ParseTuple(args, "O:update", &obj))
 419         return NULL;
 420
 421     GET_BUFFER_VIEW_OR_ERROUT(obj, &buf);
 422
 423     md5_process(&self->hash_state, buf.buf, buf.len);
 424
 425     PyBuffer_Release(&buf);
 426     Py_INCREF(Py_None);
 427     return Py_None;
 428 }
 429
 430 static PyMethodDef MD5_methods[] = {
 431     {"copy",      (PyCFunction)MD5_copy,      METH_NOARGS,  MD5_copy__doc__},
 432     {"digest",    (PyCFunction)MD5_digest,    METH_NOARGS,  MD5_digest__doc__},
 433     {"hexdigest", (PyCFunction)MD5_hexdigest, METH_NOARGS,  MD5_hexdigest__doc__},
 434     {"update",    (PyCFunction)MD5_update,    METH_VARARGS, MD5_update__doc__},
 435     {NULL,        NULL}         /* sentinel */
 436 };
 437
 438 static PyObject *
 439 MD5_get_block_size(PyObject *self, void *closure)
 440 {
 441     return PyLong_FromLong(MD5_BLOCKSIZE);
 442 }
 443
 444 static PyObject *
 445 MD5_get_name(PyObject *self, void *closure)
 446 {
 447     return PyUnicode_FromStringAndSize("MD5", 3);
 448 }
 449
 450 static PyObject *
 451 md5_get_digest_size(PyObject *self, void *closure)
 452 {
 453     return PyLong_FromLong(MD5_DIGESTSIZE);
 454 }
 455
 456
 457 static PyGetSetDef MD5_getseters[] = {
 458     {"block_size",
 459      (getter)MD5_get_block_size, NULL,
 460      NULL,
 461      NULL},
 462     {"name",
 463      (getter)MD5_get_name, NULL,
 464      NULL,
 465      NULL},
 466     {"digest_size",
 467      (getter)md5_get_digest_size, NULL,
 468      NULL,
 469      NULL},
 470     {NULL}  /* Sentinel */
 471 };
 472
 473 static PyTypeObject MD5type = {
 474     PyVarObject_HEAD_INIT(NULL, 0)
 475     "_md5.md5",         /*tp_name*/
 476     sizeof(MD5object),  /*tp_size*/
 477     0,                  /*tp_itemsize*/
 478     /* methods */
 479     MD5_dealloc,        /*tp_dealloc*/
 480     0,                  /*tp_print*/
 481     0,                  /*tp_getattr*/
 482     0,                  /*tp_setattr*/
 483     0,                  /*tp_reserved*/
 484     0,                  /*tp_repr*/
 485     0,                  /*tp_as_number*/
 486     0,                  /*tp_as_sequence*/
 487     0,                  /*tp_as_mapping*/
 488     0,                  /*tp_hash*/
 489     0,                  /*tp_call*/
 490     0,                  /*tp_str*/
 491     0,                  /*tp_getattro*/
 492     0,                  /*tp_setattro*/
 493     0,                  /*tp_as_buffer*/
 494     Py_TPFLAGS_DEFAULT, /*tp_flags*/
 495     0,                  /*tp_doc*/
 496     0,                  /*tp_traverse*/
 497     0,                  /*tp_clear*/
 498     0,                  /*tp_richcompare*/
 499     0,                  /*tp_weaklistoffset*/
 500     0,                  /*tp_iter*/
 501     0,                  /*tp_iternext*/
 502     MD5_methods,        /* tp_methods */
 503     NULL,               /* tp_members */
 504     MD5_getseters,      /* tp_getset */
 505 };
 506
 507
 508 /* The single module-level function: new() */
 509
 510 PyDoc_STRVAR(MD5_new__doc__,
 511 "Return a new MD5 hash object; optionally initialized with a string.");
 512
 513 static PyObject *
 514 MD5_new(PyObject *self, PyObject *args, PyObject *kwdict)
 515 {
 516     static char *kwlist[] = {"string", NULL};
 517     MD5object *new;
 518     PyObject *data_obj = NULL;
 519     Py_buffer buf;
 520
 521     if (!PyArg_ParseTupleAndKeywords(args, kwdict, "|O:new", kwlist,
 522                                      &data_obj)) {
 523         return NULL;
 524     }
 525
 526     if (data_obj)
 527         GET_BUFFER_VIEW_OR_ERROUT(data_obj, &buf);
 528
 529     if ((new = newMD5object()) == NULL) {
 530         if (data_obj)
 531             PyBuffer_Release(&buf);
 532         return NULL;
 533     }
 534
 535     md5_init(&new->hash_state);
 536
 537     if (PyErr_Occurred()) {
 538         Py_DECREF(new);
 539         if (data_obj)
 540             PyBuffer_Release(&buf);
 541         return NULL;
 542     }
 543     if (data_obj) {
 544         md5_process(&new->hash_state, buf.buf, buf.len);
 545         PyBuffer_Release(&buf);
 546     }
 547
 548     return (PyObject *)new;
 549 }
 550
 551
 552 /* List of functions exported by this module */
 553
 554 static struct PyMethodDef MD5_functions[] = {
 555     {"md5", (PyCFunction)MD5_new, METH_VARARGS|METH_KEYWORDS, MD5_new__doc__},
 556     {NULL,      NULL}            /* Sentinel */
 557 };
 558
 559
 560 /* Initialize this module. */
 561
 562 #define insint(n,v) { PyModule_AddIntConstant(m,n,v); }
 563
 564
 565 static struct PyModuleDef _md5module = {
 566         PyModuleDef_HEAD_INIT,
 567         "_md5",
 568         NULL,
 569         -1,
 570         MD5_functions,
 571         NULL,
 572         NULL,
 573         NULL,
 574         NULL
 575 };
 576
 577 PyMODINIT_FUNC
 578 PyInit__md5(void)
 579 {
 580     Py_TYPE(&MD5type) = &PyType_Type;
 581     if (PyType_Ready(&MD5type) < 0)
 582         return NULL;
 583     return PyModule_Create(&_md5module);
 584 }