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
  21
  22 /* Some useful types */
  23
  24 #if SIZEOF_INT == 4
  25 typedef unsigned int MD5_INT32; /* 32-bit integer */
  26 typedef PY_LONG_LONG MD5_INT64; /* 64-bit integer */
  27 #else
  28 /* not defined. compilation will die. */
  29 #endif
  30
  31 /* The MD5 block size and message digest sizes, in bytes */
  32
  33 #define MD5_BLOCKSIZE    64
  34 #define MD5_DIGESTSIZE   16
  35
  36 /* The structure for storing MD5 info */
  37
  38 struct md5_state {
  39     MD5_INT64 length;
  40     MD5_INT32 state[4], curlen;
  41     unsigned char buf[MD5_BLOCKSIZE];
  42 };
  43
  44 typedef struct {
  45     PyObject_HEAD
  46
  47     struct md5_state hash_state;
  48 } MD5object;
  49
  50
  51 /* ------------------------------------------------------------------------
  52  *
  53  * This code for the MD5 algorithm was noted as public domain. The
  54  * original headers are pasted below.
  55  *
  56  * Several changes have been made to make it more compatible with the
  57  * Python environment and desired interface.
  58  *
  59  */
  60
  61 /* LibTomCrypt, modular cryptographic library -- Tom St Denis
  62  *
  63  * LibTomCrypt is a library that provides various cryptographic
  64  * algorithms in a highly modular and flexible manner.
  65  *
  66  * The library is free for all purposes without any express
  67  * guarantee it works.
  68  *
  69  * Tom St Denis, tomstdenis@gmail.com, http://libtom.org
  70  */
  71
  72 /* rotate the hard way (platform optimizations could be done) */
  73 #define ROLc(x, y) ( (((unsigned long)(x)<<(unsigned long)((y)&31)) | (((unsigned long)(x)&0xFFFFFFFFUL)>>(unsigned long)(32-((y)&31)))) & 0xFFFFFFFFUL)
  74
  75 /* Endian Neutral macros that work on all platforms */
  76
  77 #define STORE32L(x, y)                                                                     \
  78      { (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255);   \
  79        (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
  80
  81 #define LOAD32L(x, y)                            \
  82      { x = ((unsigned long)((y)[3] & 255)<<24) | \
  83            ((unsigned long)((y)[2] & 255)<<16) | \
  84            ((unsigned long)((y)[1] & 255)<<8)  | \
  85            ((unsigned long)((y)[0] & 255)); }
  86
  87 #define STORE64L(x, y)                                                                     \
  88      { (y)[7] = (unsigned char)(((x)>>56)&255); (y)[6] = (unsigned char)(((x)>>48)&255);   \
  89        (y)[5] = (unsigned char)(((x)>>40)&255); (y)[4] = (unsigned char)(((x)>>32)&255);   \
  90        (y)[3] = (unsigned char)(((x)>>24)&255); (y)[2] = (unsigned char)(((x)>>16)&255);   \
  91        (y)[1] = (unsigned char)(((x)>>8)&255); (y)[0] = (unsigned char)((x)&255); }
  92
  93 #ifndef MIN
  94    #define MIN(x, y) ( ((x)<(y))?(x):(y) )
  95 #endif
  96
  97
  98 /* MD5 macros */
  99
 100 #define F(x,y,z)  (z ^ (x & (y ^ z)))
 101 #define G(x,y,z)  (y ^ (z & (y ^ x)))
 102 #define H(x,y,z)  (x^y^z)
 103 #define I(x,y,z)  (y^(x|(~z)))
 104
 105 #define FF(a,b,c,d,M,s,t) \
 106     a = (a + F(b,c,d) + M + t); a = ROLc(a, s) + b;
 107
 108 #define GG(a,b,c,d,M,s,t) \
 109     a = (a + G(b,c,d) + M + t); a = ROLc(a, s) + b;
 110
 111 #define HH(a,b,c,d,M,s,t) \
 112     a = (a + H(b,c,d) + M + t); a = ROLc(a, s) + b;
 113
 114 #define II(a,b,c,d,M,s,t) \
 115     a = (a + I(b,c,d) + M + t); a = ROLc(a, s) + b;
 116
 117
 118 static void md5_compress(struct md5_state *md5, unsigned char *buf)
 119 {
 120     MD5_INT32 i, W[16], a, b, c, d;
 121
 122     assert(md5 != NULL);
 123     assert(buf != NULL);
 124
 125     /* copy the state into 512-bits into W[0..15] */
 126     for (i = 0; i < 16; i++) {
 127         LOAD32L(W[i], buf + (4*i));
 128     }
 129
 130     /* copy state */
 131     a = md5->state[0];
 132     b = md5->state[1];
 133     c = md5->state[2];
 134     d = md5->state[3];
 135
 136     FF(a,b,c,d,W[0],7,0xd76aa478UL)
 137     FF(d,a,b,c,W[1],12,0xe8c7b756UL)
 138     FF(c,d,a,b,W[2],17,0x242070dbUL)
 139     FF(b,c,d,a,W[3],22,0xc1bdceeeUL)
 140     FF(a,b,c,d,W[4],7,0xf57c0fafUL)
 141     FF(d,a,b,c,W[5],12,0x4787c62aUL)
 142     FF(c,d,a,b,W[6],17,0xa8304613UL)
 143     FF(b,c,d,a,W[7],22,0xfd469501UL)
 144     FF(a,b,c,d,W[8],7,0x698098d8UL)
 145     FF(d,a,b,c,W[9],12,0x8b44f7afUL)
 146     FF(c,d,a,b,W[10],17,0xffff5bb1UL)
 147     FF(b,c,d,a,W[11],22,0x895cd7beUL)
 148     FF(a,b,c,d,W[12],7,0x6b901122UL)
 149     FF(d,a,b,c,W[13],12,0xfd987193UL)
 150     FF(c,d,a,b,W[14],17,0xa679438eUL)
 151     FF(b,c,d,a,W[15],22,0x49b40821UL)
 152     GG(a,b,c,d,W[1],5,0xf61e2562UL)
 153     GG(d,a,b,c,W[6],9,0xc040b340UL)
 154     GG(c,d,a,b,W[11],14,0x265e5a51UL)
 155     GG(b,c,d,a,W[0],20,0xe9b6c7aaUL)
 156     GG(a,b,c,d,W[5],5,0xd62f105dUL)
 157     GG(d,a,b,c,W[10],9,0x02441453UL)
 158     GG(c,d,a,b,W[15],14,0xd8a1e681UL)
 159     GG(b,c,d,a,W[4],20,0xe7d3fbc8UL)
 160     GG(a,b,c,d,W[9],5,0x21e1cde6UL)
 161     GG(d,a,b,c,W[14],9,0xc33707d6UL)
 162     GG(c,d,a,b,W[3],14,0xf4d50d87UL)
 163     GG(b,c,d,a,W[8],20,0x455a14edUL)
 164     GG(a,b,c,d,W[13],5,0xa9e3e905UL)
 165     GG(d,a,b,c,W[2],9,0xfcefa3f8UL)
 166     GG(c,d,a,b,W[7],14,0x676f02d9UL)
 167     GG(b,c,d,a,W[12],20,0x8d2a4c8aUL)
 168     HH(a,b,c,d,W[5],4,0xfffa3942UL)
 169     HH(d,a,b,c,W[8],11,0x8771f681UL)
 170     HH(c,d,a,b,W[11],16,0x6d9d6122UL)
 171     HH(b,c,d,a,W[14],23,0xfde5380cUL)
 172     HH(a,b,c,d,W[1],4,0xa4beea44UL)
 173     HH(d,a,b,c,W[4],11,0x4bdecfa9UL)
 174     HH(c,d,a,b,W[7],16,0xf6bb4b60UL)
 175     HH(b,c,d,a,W[10],23,0xbebfbc70UL)
 176     HH(a,b,c,d,W[13],4,0x289b7ec6UL)
 177     HH(d,a,b,c,W[0],11,0xeaa127faUL)
 178     HH(c,d,a,b,W[3],16,0xd4ef3085UL)
 179     HH(b,c,d,a,W[6],23,0x04881d05UL)
 180     HH(a,b,c,d,W[9],4,0xd9d4d039UL)
 181     HH(d,a,b,c,W[12],11,0xe6db99e5UL)
 182     HH(c,d,a,b,W[15],16,0x1fa27cf8UL)
 183     HH(b,c,d,a,W[2],23,0xc4ac5665UL)
 184     II(a,b,c,d,W[0],6,0xf4292244UL)
 185     II(d,a,b,c,W[7],10,0x432aff97UL)
 186     II(c,d,a,b,W[14],15,0xab9423a7UL)
 187     II(b,c,d,a,W[5],21,0xfc93a039UL)
 188     II(a,b,c,d,W[12],6,0x655b59c3UL)
 189     II(d,a,b,c,W[3],10,0x8f0ccc92UL)
 190     II(c,d,a,b,W[10],15,0xffeff47dUL)
 191     II(b,c,d,a,W[1],21,0x85845dd1UL)
 192     II(a,b,c,d,W[8],6,0x6fa87e4fUL)
 193     II(d,a,b,c,W[15],10,0xfe2ce6e0UL)
 194     II(c,d,a,b,W[6],15,0xa3014314UL)
 195     II(b,c,d,a,W[13],21,0x4e0811a1UL)
 196     II(a,b,c,d,W[4],6,0xf7537e82UL)
 197     II(d,a,b,c,W[11],10,0xbd3af235UL)
 198     II(c,d,a,b,W[2],15,0x2ad7d2bbUL)
 199     II(b,c,d,a,W[9],21,0xeb86d391UL)
 200
 201     md5->state[0] = md5->state[0] + a;
 202     md5->state[1] = md5->state[1] + b;
 203     md5->state[2] = md5->state[2] + c;
 204     md5->state[3] = md5->state[3] + d;
 205 }
 206
 207
 208 /**
 209    Initialize the hash state
 210    @param sha1   The hash state you wish to initialize
 211 */
 212 void md5_init(struct md5_state *md5)
 213 {
 214     assert(md5 != NULL);
 215     md5->state[0] = 0x67452301UL;
 216     md5->state[1] = 0xefcdab89UL;
 217     md5->state[2] = 0x98badcfeUL;
 218     md5->state[3] = 0x10325476UL;
 219     md5->curlen = 0;
 220     md5->length = 0;
 221 }
 222
 223 /**
 224    Process a block of memory though the hash
 225    @param sha1   The hash state
 226    @param in     The data to hash
 227    @param inlen  The length of the data (octets)
 228 */
 229 void md5_process(struct md5_state *md5,
 230                 const unsigned char *in, unsigned long inlen)
 231 {
 232     unsigned long n;
 233
 234     assert(md5 != NULL);
 235     assert(in != NULL);
 236     assert(md5->curlen <= sizeof(md5->buf));
 237
 238     while (inlen > 0) {
 239         if (md5->curlen == 0 && inlen >= MD5_BLOCKSIZE) {
 240            md5_compress(md5, (unsigned char *)in);
 241            md5->length    += MD5_BLOCKSIZE * 8;
 242            in             += MD5_BLOCKSIZE;
 243            inlen          -= MD5_BLOCKSIZE;
 244         } else {
 245            n = MIN(inlen, (MD5_BLOCKSIZE - md5->curlen));
 246            memcpy(md5->buf + md5->curlen, in, (size_t)n);
 247            md5->curlen    += n;
 248            in             += n;
 249            inlen          -= n;
 250            if (md5->curlen == MD5_BLOCKSIZE) {
 251               md5_compress(md5, md5->buf);
 252               md5->length += 8*MD5_BLOCKSIZE;
 253               md5->curlen = 0;
 254            }
 255        }
 256     }
 257 }
 258
 259 /**
 260    Terminate the hash to get the digest
 261    @param sha1  The hash state
 262    @param out [out] The destination of the hash (16 bytes)
 263 */
 264 void md5_done(struct md5_state *md5, unsigned char *out)
 265 {
 266     int i;
 267
 268     assert(md5 != NULL);
 269     assert(out != NULL);
 270     assert(md5->curlen < sizeof(md5->buf));
 271
 272     /* increase the length of the message */
 273     md5->length += md5->curlen * 8;
 274
 275     /* append the '1' bit */
 276     md5->buf[md5->curlen++] = (unsigned char)0x80;
 277
 278     /* if the length is currently above 56 bytes we append zeros
 279      * then compress.  Then we can fall back to padding zeros and length
 280      * encoding like normal.
 281      */
 282     if (md5->curlen > 56) {
 283         while (md5->curlen < 64) {
 284             md5->buf[md5->curlen++] = (unsigned char)0;
 285         }
 286         md5_compress(md5, md5->buf);
 287         md5->curlen = 0;
 288     }
 289
 290     /* pad upto 56 bytes of zeroes */
 291     while (md5->curlen < 56) {
 292         md5->buf[md5->curlen++] = (unsigned char)0;
 293     }
 294
 295     /* store length */
 296     STORE64L(md5->length, md5->buf+56);
 297     md5_compress(md5, md5->buf);
 298
 299     /* copy output */
 300     for (i = 0; i < 4; i++) {
 301         STORE32L(md5->state[i], out+(4*i));
 302     }
 303 }
 304
 305 /* .Source: /cvs/libtom/libtomcrypt/src/hashes/md5.c,v $ */
 306 /* .Revision: 1.10 $ */
 307 /* .Date: 2007/05/12 14:25:28 $ */
 308
 309 /*
 310  * End of copied MD5 code.
 311  *
 312  * ------------------------------------------------------------------------
 313  */
 314
 315 static PyTypeObject MD5type;
 316
 317
 318 static MD5object *
 319 newMD5object(void)
 320 {
 321     return (MD5object *)PyObject_New(MD5object, &MD5type);
 322 }
 323
 324
 325 /* Internal methods for a hash object */
 326
 327 static void
 328 MD5_dealloc(PyObject *ptr)
 329 {
 330     PyObject_Del(ptr);
 331 }
 332
 333
 334 /* External methods for a hash object */
 335
 336 PyDoc_STRVAR(MD5_copy__doc__, "Return a copy of the hash object.");
 337
 338 static PyObject *
 339 MD5_copy(MD5object *self, PyObject *unused)
 340 {
 341     MD5object *newobj;
 342
 343     if (Py_TYPE(self) == &MD5type) {
 344         if ( (newobj = newMD5object())==NULL)
 345             return NULL;
 346     } else {
 347         if ( (newobj = newMD5object())==NULL)
 348             return NULL;
 349     }
 350
 351     newobj->hash_state = self->hash_state;
 352     return (PyObject *)newobj;
 353 }
 354
 355 PyDoc_STRVAR(MD5_digest__doc__,
 356 "Return the digest value as a string of binary data.");
 357
 358 static PyObject *
 359 MD5_digest(MD5object *self, PyObject *unused)
 360 {
 361     unsigned char digest[MD5_DIGESTSIZE];
 362     struct md5_state temp;
 363
 364     temp = self->hash_state;
 365     md5_done(&temp, digest);
 366     return PyString_FromStringAndSize((const char *)digest, MD5_DIGESTSIZE);
 367 }
 368
 369 PyDoc_STRVAR(MD5_hexdigest__doc__,
 370 "Return the digest value as a string of hexadecimal digits.");
 371
 372 static PyObject *
 373 MD5_hexdigest(MD5object *self, PyObject *unused)
 374 {
 375     unsigned char digest[MD5_DIGESTSIZE];
 376     struct md5_state temp;
 377     PyObject *retval;
 378     Py_UNICODE *hex_digest;
 379     int i, j;
 380
 381     /* Get the raw (binary) digest value */
 382     temp = self->hash_state;
 383     md5_done(&temp, digest);
 384
 385     /* Create a new string */
 386     retval = PyUnicode_FromStringAndSize(NULL, MD5_DIGESTSIZE * 2);
 387     if (!retval)
 388             return NULL;
 389     hex_digest = PyUnicode_AS_UNICODE(retval);
 390     if (!hex_digest) {
 391             Py_DECREF(retval);
 392             return NULL;
 393     }
 394
 395     /* Make hex version of the digest */
 396     for(i=j=0; i<MD5_DIGESTSIZE; i++) {
 397         char c;
 398         c = (digest[i] >> 4) & 0xf;
 399         c = (c>9) ? c+'a'-10 : c + '0';
 400         hex_digest[j++] = c;
 401         c = (digest[i] & 0xf);
 402         c = (c>9) ? c+'a'-10 : c + '0';
 403         hex_digest[j++] = c;
 404     }
 405     return retval;
 406 }
 407
 408 PyDoc_STRVAR(MD5_update__doc__,
 409 "Update this hash object's state with the provided string.");
 410
 411 static PyObject *
 412 MD5_update(MD5object *self, PyObject *args)
 413 {
 414     unsigned char *cp;
 415     int len;
 416
 417     if (!PyArg_ParseTuple(args, "s#:update", &cp, &len))
 418         return NULL;
 419
 420     md5_process(&self->hash_state, cp, len);
 421
 422     Py_INCREF(Py_None);
 423     return Py_None;
 424 }
 425
 426 static PyMethodDef MD5_methods[] = {
 427     {"copy",      (PyCFunction)MD5_copy,      METH_NOARGS,  MD5_copy__doc__},
 428     {"digest",    (PyCFunction)MD5_digest,    METH_NOARGS,  MD5_digest__doc__},
 429     {"hexdigest", (PyCFunction)MD5_hexdigest, METH_NOARGS,  MD5_hexdigest__doc__},
 430     {"update",    (PyCFunction)MD5_update,    METH_VARARGS, MD5_update__doc__},
 431     {NULL,        NULL}         /* sentinel */
 432 };
 433
 434 static PyObject *
 435 MD5_get_block_size(PyObject *self, void *closure)
 436 {
 437     return PyLong_FromLong(MD5_BLOCKSIZE);
 438 }
 439
 440 static PyObject *
 441 MD5_get_name(PyObject *self, void *closure)
 442 {
 443     return PyUnicode_FromStringAndSize("MD5", 3);
 444 }
 445
 446 static PyObject *
 447 md5_get_digest_size(PyObject *self, void *closure)
 448 {
 449     return PyLong_FromLong(MD5_DIGESTSIZE);
 450 }
 451
 452
 453 static PyGetSetDef MD5_getseters[] = {
 454     {"block_size",
 455      (getter)MD5_get_block_size, NULL,
 456      NULL,
 457      NULL},
 458     {"name",
 459      (getter)MD5_get_name, NULL,
 460      NULL,
 461      NULL},
 462     {"digest_size",
 463      (getter)md5_get_digest_size, NULL,
 464      NULL,
 465      NULL},
 466     {NULL}  /* Sentinel */
 467 };
 468
 469 static PyTypeObject MD5type = {
 470     PyVarObject_HEAD_INIT(NULL, 0)
 471     "_md5.md5",         /*tp_name*/
 472     sizeof(MD5object),  /*tp_size*/
 473     0,                  /*tp_itemsize*/
 474     /* methods */
 475     MD5_dealloc,        /*tp_dealloc*/
 476     0,                  /*tp_print*/
 477     0,                  /*tp_getattr*/
 478     0,                  /*tp_setattr*/
 479     0,                  /*tp_compare*/
 480     0,                  /*tp_repr*/
 481     0,                  /*tp_as_number*/
 482     0,                  /*tp_as_sequence*/
 483     0,                  /*tp_as_mapping*/
 484     0,                  /*tp_hash*/
 485     0,                  /*tp_call*/
 486     0,                  /*tp_str*/
 487     0,                  /*tp_getattro*/
 488     0,                  /*tp_setattro*/
 489     0,                  /*tp_as_buffer*/
 490     Py_TPFLAGS_DEFAULT, /*tp_flags*/
 491     0,                  /*tp_doc*/
 492     0,                  /*tp_traverse*/
 493     0,                  /*tp_clear*/
 494     0,                  /*tp_richcompare*/
 495     0,                  /*tp_weaklistoffset*/
 496     0,                  /*tp_iter*/
 497     0,                  /*tp_iternext*/
 498     MD5_methods,        /* tp_methods */
 499     NULL,               /* tp_members */
 500     MD5_getseters,      /* tp_getset */
 501 };
 502
 503
 504 /* The single module-level function: new() */
 505
 506 PyDoc_STRVAR(MD5_new__doc__,
 507 "Return a new MD5 hash object; optionally initialized with a string.");
 508
 509 static PyObject *
 510 MD5_new(PyObject *self, PyObject *args, PyObject *kwdict)
 511 {
 512     static char *kwlist[] = {"string", NULL};
 513     MD5object *new;
 514     unsigned char *cp = NULL;
 515     int len;
 516
 517     if (!PyArg_ParseTupleAndKeywords(args, kwdict, "|s#:new", kwlist,
 518                                      &cp, &len)) {
 519         return NULL;
 520     }
 521
 522     if ((new = newMD5object()) == NULL)
 523         return NULL;
 524
 525     md5_init(&new->hash_state);
 526
 527     if (PyErr_Occurred()) {
 528         Py_DECREF(new);
 529         return NULL;
 530     }
 531     if (cp)
 532         md5_process(&new->hash_state, cp, len);
 533
 534     return (PyObject *)new;
 535 }
 536
 537
 538 /* List of functions exported by this module */
 539
 540 static struct PyMethodDef MD5_functions[] = {
 541     {"md5", (PyCFunction)MD5_new, METH_VARARGS|METH_KEYWORDS, MD5_new__doc__},
 542     {NULL,      NULL}            /* Sentinel */
 543 };
 544
 545
 546 /* Initialize this module. */
 547
 548 #define insint(n,v) { PyModule_AddIntConstant(m,n,v); }
 549
 550 PyMODINIT_FUNC
 551 init_md5(void)
 552 {
 553     PyObject *m;
 554
 555     Py_TYPE(&MD5type) = &PyType_Type;
 556     if (PyType_Ready(&MD5type) < 0)
 557         return;
 558     m = Py_InitModule("_md5", MD5_functions);
 559     if (m == NULL)
 560         return;
 561 }