Modules/_hashopenssl.c

   1 /* Module that wraps all OpenSSL hash algorithms */
   2
   3 /*
   4  * Copyright (C) 2005   Gregory P. Smith (greg@krypto.org)
   5  * Licensed to PSF under a Contributor Agreement.
   6  *
   7  * Derived from a skeleton of shamodule.c containing work performed by:
   8  *
   9  * Andrew Kuchling (amk@amk.ca)
  10  * Greg Stein (gstein@lyra.org)
  11  *
  12  */
  13
  14 #define PY_SSIZE_T_CLEAN
  15
  16 #include "Python.h"
  17 #include "structmember.h"
  18
  19 /* EVP is the preferred interface to hashing in OpenSSL */
  20 #include <openssl/evp.h>
  21
  22 #define MUNCH_SIZE INT_MAX
  23
  24
  25 #ifndef HASH_OBJ_CONSTRUCTOR
  26 #define HASH_OBJ_CONSTRUCTOR 0
  27 #endif
  28
  29 typedef struct {
  30     PyObject_HEAD
  31     PyObject            *name;  /* name of this hash algorithm */
  32     EVP_MD_CTX          ctx;    /* OpenSSL message digest context */
  33 } EVPobject;
  34
  35
  36 static PyTypeObject EVPtype;
  37
  38
  39 #define DEFINE_CONSTS_FOR_NEW(Name)  \
  40     static PyObject *CONST_ ## Name ## _name_obj; \
  41     static EVP_MD_CTX CONST_new_ ## Name ## _ctx; \
  42     static EVP_MD_CTX *CONST_new_ ## Name ## _ctx_p = NULL;
  43
  44 DEFINE_CONSTS_FOR_NEW(md5)
  45 DEFINE_CONSTS_FOR_NEW(sha1)
  46 DEFINE_CONSTS_FOR_NEW(sha224)
  47 DEFINE_CONSTS_FOR_NEW(sha256)
  48 DEFINE_CONSTS_FOR_NEW(sha384)
  49 DEFINE_CONSTS_FOR_NEW(sha512)
  50
  51
  52 static EVPobject *
  53 newEVPobject(PyObject *name)
  54 {
  55     EVPobject *retval = (EVPobject *)PyObject_New(EVPobject, &EVPtype);
  56
  57     /* save the name for .name to return */
  58     if (retval != NULL) {
  59         Py_INCREF(name);
  60         retval->name = name;
  61     }
  62
  63     return retval;
  64 }
  65
  66 /* Internal methods for a hash object */
  67
  68 static void
  69 EVP_dealloc(PyObject *ptr)
  70 {
  71     EVP_MD_CTX_cleanup(&((EVPobject *)ptr)->ctx);
  72     Py_XDECREF(((EVPobject *)ptr)->name);
  73     PyObject_Del(ptr);
  74 }
  75
  76
  77 /* External methods for a hash object */
  78
  79 PyDoc_STRVAR(EVP_copy__doc__, "Return a copy of the hash object.");
  80
  81 static PyObject *
  82 EVP_copy(EVPobject *self, PyObject *unused)
  83 {
  84     EVPobject *newobj;
  85
  86     if ( (newobj = newEVPobject(self->name))==NULL)
  87         return NULL;
  88
  89     EVP_MD_CTX_copy(&newobj->ctx, &self->ctx);
  90     return (PyObject *)newobj;
  91 }
  92
  93 PyDoc_STRVAR(EVP_digest__doc__,
  94 "Return the digest value as a string of binary data.");
  95
  96 static PyObject *
  97 EVP_digest(EVPobject *self, PyObject *unused)
  98 {
  99     unsigned char digest[EVP_MAX_MD_SIZE];
 100     EVP_MD_CTX temp_ctx;
 101     PyObject *retval;
 102     unsigned int digest_size;
 103
 104     EVP_MD_CTX_copy(&temp_ctx, &self->ctx);
 105     digest_size = EVP_MD_CTX_size(&temp_ctx);
 106     EVP_DigestFinal(&temp_ctx, digest, NULL);
 107
 108     retval = PyString_FromStringAndSize((const char *)digest, digest_size);
 109     EVP_MD_CTX_cleanup(&temp_ctx);
 110     return retval;
 111 }
 112
 113 PyDoc_STRVAR(EVP_hexdigest__doc__,
 114 "Return the digest value as a string of hexadecimal digits.");
 115
 116 static PyObject *
 117 EVP_hexdigest(EVPobject *self, PyObject *unused)
 118 {
 119     unsigned char digest[EVP_MAX_MD_SIZE];
 120     EVP_MD_CTX temp_ctx;
 121     PyObject *retval;
 122     char *hex_digest;
 123     unsigned int i, j, digest_size;
 124
 125     /* Get the raw (binary) digest value */
 126     EVP_MD_CTX_copy(&temp_ctx, &self->ctx);
 127     digest_size = EVP_MD_CTX_size(&temp_ctx);
 128     EVP_DigestFinal(&temp_ctx, digest, NULL);
 129
 130     EVP_MD_CTX_cleanup(&temp_ctx);
 131
 132     /* Create a new string */
 133     /* NOTE: not thread safe! modifying an already created string object */
 134     /* (not a problem because we hold the GIL by default) */
 135     retval = PyString_FromStringAndSize(NULL, digest_size * 2);
 136     if (!retval)
 137             return NULL;
 138     hex_digest = PyString_AsString(retval);
 139     if (!hex_digest) {
 140             Py_DECREF(retval);
 141             return NULL;
 142     }
 143
 144     /* Make hex version of the digest */
 145     for(i=j=0; i<digest_size; i++) {
 146         char c;
 147         c = (digest[i] >> 4) & 0xf;
 148         c = (c>9) ? c+'a'-10 : c + '0';
 149         hex_digest[j++] = c;
 150         c = (digest[i] & 0xf);
 151         c = (c>9) ? c+'a'-10 : c + '0';
 152         hex_digest[j++] = c;
 153     }
 154     return retval;
 155 }
 156
 157 PyDoc_STRVAR(EVP_update__doc__,
 158 "Update this hash object's state with the provided string.");
 159
 160 static PyObject *
 161 EVP_update(EVPobject *self, PyObject *args)
 162 {
 163     unsigned char *cp;
 164     Py_ssize_t len;
 165
 166     if (!PyArg_ParseTuple(args, "s#:update", &cp, &len))
 167         return NULL;
 168
 169     if (len > 0 && len <= MUNCH_SIZE) {
 170     EVP_DigestUpdate(&self->ctx, cp, Py_SAFE_DOWNCAST(len, Py_ssize_t,
 171                                                       unsigned int));
 172     } else {
 173         Py_ssize_t offset = 0;
 174         while (len) {
 175             unsigned int process = len > MUNCH_SIZE ? MUNCH_SIZE : len;
 176             EVP_DigestUpdate(&self->ctx, cp + offset, process);
 177             len -= process;
 178             offset += process;
 179         }
 180     }
 181     Py_INCREF(Py_None);
 182     return Py_None;
 183 }
 184
 185 static PyMethodDef EVP_methods[] = {
 186     {"update",    (PyCFunction)EVP_update,    METH_VARARGS, EVP_update__doc__},
 187     {"digest",    (PyCFunction)EVP_digest,    METH_NOARGS,  EVP_digest__doc__},
 188     {"hexdigest", (PyCFunction)EVP_hexdigest, METH_NOARGS,  EVP_hexdigest__doc__},
 189     {"copy",      (PyCFunction)EVP_copy,      METH_NOARGS,  EVP_copy__doc__},
 190     {NULL,        NULL}         /* sentinel */
 191 };
 192
 193 static PyObject *
 194 EVP_get_block_size(EVPobject *self, void *closure)
 195 {
 196     return PyInt_FromLong(EVP_MD_CTX_block_size(&((EVPobject *)self)->ctx));
 197 }
 198
 199 static PyObject *
 200 EVP_get_digest_size(EVPobject *self, void *closure)
 201 {
 202     return PyInt_FromLong(EVP_MD_CTX_size(&((EVPobject *)self)->ctx));
 203 }
 204
 205 static PyMemberDef EVP_members[] = {
 206     {"name", T_OBJECT, offsetof(EVPobject, name), READONLY, PyDoc_STR("algorithm name.")},
 207     {NULL}  /* Sentinel */
 208 };
 209
 210 static PyGetSetDef EVP_getseters[] = {
 211     {"digest_size",
 212      (getter)EVP_get_digest_size, NULL,
 213      NULL,
 214      NULL},
 215     {"block_size",
 216      (getter)EVP_get_block_size, NULL,
 217      NULL,
 218      NULL},
 219     /* the old md5 and sha modules support 'digest_size' as in PEP 247.
 220      * the old sha module also supported 'digestsize'.  ugh. */
 221     {"digestsize",
 222      (getter)EVP_get_digest_size, NULL,
 223      NULL,
 224      NULL},
 225     {NULL}  /* Sentinel */
 226 };
 227
 228
 229 static PyObject *
 230 EVP_repr(PyObject *self)
 231 {
 232     char buf[100];
 233     PyOS_snprintf(buf, sizeof(buf), "<%s HASH object @ %p>",
 234             PyString_AsString(((EVPobject *)self)->name), self);
 235     return PyString_FromString(buf);
 236 }
 237
 238 #if HASH_OBJ_CONSTRUCTOR
 239 static int
 240 EVP_tp_init(EVPobject *self, PyObject *args, PyObject *kwds)
 241 {
 242     static char *kwlist[] = {"name", "string", NULL};
 243     PyObject *name_obj = NULL;
 244     char *nameStr;
 245     unsigned char *cp = NULL;
 246     Py_ssize_t len = 0;
 247     const EVP_MD *digest;
 248
 249     if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s#:HASH", kwlist,
 250                                      &name_obj, &cp, &len)) {
 251         return -1;
 252     }
 253
 254     if (!PyArg_Parse(name_obj, "s", &nameStr)) {
 255         PyErr_SetString(PyExc_TypeError, "name must be a string");
 256         return -1;
 257     }
 258
 259     digest = EVP_get_digestbyname(nameStr);
 260     if (!digest) {
 261         PyErr_SetString(PyExc_ValueError, "unknown hash function");
 262         return -1;
 263     }
 264     EVP_DigestInit(&self->ctx, digest);
 265
 266     self->name = name_obj;
 267     Py_INCREF(self->name);
 268
 269     if (cp && len) {
 270         if (len > 0 && len <= MUNCH_SIZE) {
 271         EVP_DigestUpdate(&self->ctx, cp, Py_SAFE_DOWNCAST(len, Py_ssize_t,
 272                                                           unsigned int));
 273         } else {
 274             Py_ssize_t offset = 0;
 275             while (len) {
 276                 unsigned int process = len > MUNCH_SIZE ? MUNCH_SIZE : len;
 277                 EVP_DigestUpdate(&self->ctx, cp + offset, process);
 278                 len -= process;
 279                 offset += process;
 280             }
 281         }
 282     }
 283
 284     return 0;
 285 }
 286 #endif
 287
 288
 289 PyDoc_STRVAR(hashtype_doc,
 290 "A hash represents the object used to calculate a checksum of a\n\
 291 string of information.\n\
 292 \n\
 293 Methods:\n\
 294 \n\
 295 update() -- updates the current digest with an additional string\n\
 296 digest() -- return the current digest value\n\
 297 hexdigest() -- return the current digest as a string of hexadecimal digits\n\
 298 copy() -- return a copy of the current hash object\n\
 299 \n\
 300 Attributes:\n\
 301 \n\
 302 name -- the hash algorithm being used by this object\n\
 303 digest_size -- number of bytes in this hashes output\n");
 304
 305 static PyTypeObject EVPtype = {
 306     PyVarObject_HEAD_INIT(NULL, 0)
 307     "_hashlib.HASH",    /*tp_name*/
 308     sizeof(EVPobject),  /*tp_basicsize*/
 309     0,                  /*tp_itemsize*/
 310     /* methods */
 311     EVP_dealloc,        /*tp_dealloc*/
 312     0,                  /*tp_print*/
 313     0,                  /*tp_getattr*/
 314     0,                  /*tp_setattr*/
 315     0,                  /*tp_compare*/
 316     EVP_repr,           /*tp_repr*/
 317     0,                  /*tp_as_number*/
 318     0,                  /*tp_as_sequence*/
 319     0,                  /*tp_as_mapping*/
 320     0,                  /*tp_hash*/
 321     0,                  /*tp_call*/
 322     0,                  /*tp_str*/
 323     0,                  /*tp_getattro*/
 324     0,                  /*tp_setattro*/
 325     0,                  /*tp_as_buffer*/
 326     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
 327     hashtype_doc,       /*tp_doc*/
 328     0,                  /*tp_traverse*/
 329     0,                  /*tp_clear*/
 330     0,                  /*tp_richcompare*/
 331     0,                  /*tp_weaklistoffset*/
 332     0,                  /*tp_iter*/
 333     0,                  /*tp_iternext*/
 334     EVP_methods,        /* tp_methods */
 335     EVP_members,        /* tp_members */
 336     EVP_getseters,      /* tp_getset */
 337 #if 1
 338     0,                  /* tp_base */
 339     0,                  /* tp_dict */
 340     0,                  /* tp_descr_get */
 341     0,                  /* tp_descr_set */
 342     0,                  /* tp_dictoffset */
 343 #endif
 344 #if HASH_OBJ_CONSTRUCTOR
 345     (initproc)EVP_tp_init, /* tp_init */
 346 #endif
 347 };
 348
 349 static PyObject *
 350 EVPnew(PyObject *name_obj,
 351        const EVP_MD *digest, const EVP_MD_CTX *initial_ctx,
 352        const unsigned char *cp, Py_ssize_t len)
 353 {
 354     EVPobject *self;
 355
 356     if (!digest && !initial_ctx) {
 357         PyErr_SetString(PyExc_ValueError, "unsupported hash type");
 358         return NULL;
 359     }
 360
 361     if ((self = newEVPobject(name_obj)) == NULL)
 362         return NULL;
 363
 364     if (initial_ctx) {
 365         EVP_MD_CTX_copy(&self->ctx, initial_ctx);
 366     } else {
 367         EVP_DigestInit(&self->ctx, digest);
 368     }
 369
 370     if (cp && len) {
 371         if (len > 0 && len <= MUNCH_SIZE) {
 372             EVP_DigestUpdate(&self->ctx, cp, Py_SAFE_DOWNCAST(len, Py_ssize_t,
 373                                                               unsigned int));
 374         } else {
 375             Py_ssize_t offset = 0;
 376             while (len) {
 377                 unsigned int process = len > MUNCH_SIZE ? MUNCH_SIZE : len;
 378                 EVP_DigestUpdate(&self->ctx, cp + offset, process);
 379                 len -= process;
 380                 offset += process;
 381             }
 382         }
 383     }
 384
 385     return (PyObject *)self;
 386 }
 387
 388
 389 /* The module-level function: new() */
 390
 391 PyDoc_STRVAR(EVP_new__doc__,
 392 "Return a new hash object using the named algorithm.\n\
 393 An optional string argument may be provided and will be\n\
 394 automatically hashed.\n\
 395 \n\
 396 The MD5 and SHA1 algorithms are always supported.\n");
 397
 398 static PyObject *
 399 EVP_new(PyObject *self, PyObject *args, PyObject *kwdict)
 400 {
 401     static char *kwlist[] = {"name", "string", NULL};
 402     PyObject *name_obj = NULL;
 403     char *name;
 404     const EVP_MD *digest;
 405     unsigned char *cp = NULL;
 406     Py_ssize_t len = 0;
 407
 408     if (!PyArg_ParseTupleAndKeywords(args, kwdict, "O|s#:new", kwlist,
 409                                      &name_obj, &cp, &len)) {
 410         return NULL;
 411     }
 412
 413     if (!PyArg_Parse(name_obj, "s", &name)) {
 414         PyErr_SetString(PyExc_TypeError, "name must be a string");
 415         return NULL;
 416     }
 417
 418     digest = EVP_get_digestbyname(name);
 419
 420     return EVPnew(name_obj, digest, NULL, cp, len);
 421 }
 422
 423 /*
 424  *  This macro generates constructor function definitions for specific
 425  *  hash algorithms.  These constructors are much faster than calling
 426  *  the generic one passing it a python string and are noticably
 427  *  faster than calling a python new() wrapper.  Thats important for
 428  *  code that wants to make hashes of a bunch of small strings.
 429  */
 430 #define GEN_CONSTRUCTOR(NAME)  \
 431     static PyObject * \
 432     EVP_new_ ## NAME (PyObject *self, PyObject *args) \
 433     { \
 434         unsigned char *cp = NULL; \
 435         Py_ssize_t len = 0; \
 436      \
 437         if (!PyArg_ParseTuple(args, "|s#:" #NAME , &cp, &len)) { \
 438             return NULL; \
 439         } \
 440      \
 441         return EVPnew( \
 442                 CONST_ ## NAME ## _name_obj, \
 443                 NULL, \
 444                 CONST_new_ ## NAME ## _ctx_p, \
 445                 cp, len); \
 446     }
 447
 448 /* a PyMethodDef structure for the constructor */
 449 #define CONSTRUCTOR_METH_DEF(NAME)  \
 450     {"openssl_" #NAME, (PyCFunction)EVP_new_ ## NAME, METH_VARARGS, \
 451         PyDoc_STR("Returns a " #NAME \
 452                   " hash object; optionally initialized with a string") \
 453     }
 454
 455 /* used in the init function to setup a constructor */
 456 #define INIT_CONSTRUCTOR_CONSTANTS(NAME)  do { \
 457     CONST_ ## NAME ## _name_obj = PyString_FromString(#NAME); \
 458     if (EVP_get_digestbyname(#NAME)) { \
 459         CONST_new_ ## NAME ## _ctx_p = &CONST_new_ ## NAME ## _ctx; \
 460         EVP_DigestInit(CONST_new_ ## NAME ## _ctx_p, EVP_get_digestbyname(#NAME)); \
 461     } \
 462 } while (0);
 463
 464 GEN_CONSTRUCTOR(md5)
 465 GEN_CONSTRUCTOR(sha1)
 466 GEN_CONSTRUCTOR(sha224)
 467 GEN_CONSTRUCTOR(sha256)
 468 GEN_CONSTRUCTOR(sha384)
 469 GEN_CONSTRUCTOR(sha512)
 470
 471 /* List of functions exported by this module */
 472
 473 static struct PyMethodDef EVP_functions[] = {
 474     {"new", (PyCFunction)EVP_new, METH_VARARGS|METH_KEYWORDS, EVP_new__doc__},
 475     CONSTRUCTOR_METH_DEF(md5),
 476     CONSTRUCTOR_METH_DEF(sha1),
 477     CONSTRUCTOR_METH_DEF(sha224),
 478     CONSTRUCTOR_METH_DEF(sha256),
 479     CONSTRUCTOR_METH_DEF(sha384),
 480     CONSTRUCTOR_METH_DEF(sha512),
 481     {NULL,      NULL}            /* Sentinel */
 482 };
 483
 484
 485 /* Initialize this module. */
 486
 487 PyMODINIT_FUNC
 488 init_hashlib(void)
 489 {
 490     PyObject *m;
 491
 492     OpenSSL_add_all_digests();
 493
 494     /* TODO build EVP_functions openssl_* entries dynamically based
 495      * on what hashes are supported rather than listing many
 496      * but having some be unsupported.  Only init appropriate
 497      * constants. */
 498
 499     Py_TYPE(&EVPtype) = &PyType_Type;
 500     if (PyType_Ready(&EVPtype) < 0)
 501         return;
 502
 503     m = Py_InitModule("_hashlib", EVP_functions);
 504     if (m == NULL)
 505         return;
 506
 507 #if HASH_OBJ_CONSTRUCTOR
 508     Py_INCREF(&EVPtype);
 509     PyModule_AddObject(m, "HASH", (PyObject *)&EVPtype);
 510 #endif
 511
 512     /* these constants are used by the convenience constructors */
 513     INIT_CONSTRUCTOR_CONSTANTS(md5);
 514     INIT_CONSTRUCTOR_CONSTANTS(sha1);
 515     INIT_CONSTRUCTOR_CONSTANTS(sha224);
 516     INIT_CONSTRUCTOR_CONSTANTS(sha256);
 517     INIT_CONSTRUCTOR_CONSTANTS(sha384);
 518     INIT_CONSTRUCTOR_CONSTANTS(sha512);
 519 }