summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorTor Andersson <tor@ghostscript.com>2011-02-07 12:13:50 +0000
committerTor Andersson <tor@ghostscript.com>2011-02-07 12:13:50 +0000
commit276f0a6b1175834dc32e0ab1dc52098f1e945395 (patch)
tree5a617c84429f97645f385b653632965704dce00c
parent1cd8d2d359ff0c8ea7ba7a5837bc4a4f2f4e02ad (diff)
downloadmupdf-276f0a6b1175834dc32e0ab1dc52098f1e945395.tar.xz
Add support for AESv3 encryption from SumatraPDF.
-rw-r--r--Makefile1
-rw-r--r--fitz/crypt_sha2.c182
-rw-r--r--fitz/fitz.h18
-rw-r--r--mupdf/mupdf.h8
-rw-r--r--mupdf/pdf_crypt.c124
-rw-r--r--win32/mupdf/mupdf.vcproj4
6 files changed, 320 insertions, 17 deletions
diff --git a/Makefile b/Makefile
index 19ca9adf..7d4254df 100644
--- a/Makefile
+++ b/Makefile
@@ -85,6 +85,7 @@ FITZ_SRC := \
fitz/crypt_aes.c \
fitz/crypt_arc4.c \
fitz/crypt_md5.c \
+ fitz/crypt_sha2.c \
fitz/dev_bbox.c \
fitz/dev_draw.c \
fitz/dev_list.c \
diff --git a/fitz/crypt_sha2.c b/fitz/crypt_sha2.c
new file mode 100644
index 00000000..6796c5ab
--- /dev/null
+++ b/fitz/crypt_sha2.c
@@ -0,0 +1,182 @@
+/*
+This code is based on the code found from 7-Zip, which has a modified
+version of the SHA-256 found from Crypto++ <http://www.cryptopp.com/>.
+The code was modified a little to fit into liblzma and fitz.
+
+This file has been put into the public domain.
+You can do whatever you want with this file.
+*/
+
+#include "fitz.h"
+
+static inline int isbigendian(void)
+{
+ static const int one = 1;
+ return *(char*)&one == 0;
+}
+
+static inline unsigned int bswap32(unsigned int num)
+{
+ if (!isbigendian())
+ {
+ return ( (((num) << 24))
+ | (((num) << 8) & 0x00FF0000)
+ | (((num) >> 8) & 0x0000FF00)
+ | (((num) >> 24)) );
+ }
+ return num;
+}
+
+/* At least on x86, GCC is able to optimize this to a rotate instruction. */
+#define rotr_32(num, amount) ((num) >> (amount) | (num) << (32 - (amount)))
+
+#define blk0(i) (W[i] = data[i])
+#define blk2(i) (W[i & 15] += s1(W[(i - 2) & 15]) + W[(i - 7) & 15] \
+ + s0(W[(i - 15) & 15]))
+
+#define Ch(x, y, z) (z ^ (x & (y ^ z)))
+#define Maj(x, y, z) ((x & y) | (z & (x | y)))
+
+#define a(i) T[(0 - i) & 7]
+#define b(i) T[(1 - i) & 7]
+#define c(i) T[(2 - i) & 7]
+#define d(i) T[(3 - i) & 7]
+#define e(i) T[(4 - i) & 7]
+#define f(i) T[(5 - i) & 7]
+#define g(i) T[(6 - i) & 7]
+#define h(i) T[(7 - i) & 7]
+
+#define R(i) \
+ h(i) += S1(e(i)) + Ch(e(i), f(i), g(i)) + SHA256_K[i + j] \
+ + (j ? blk2(i) : blk0(i)); \
+ d(i) += h(i); \
+ h(i) += S0(a(i)) + Maj(a(i), b(i), c(i))
+
+#define S0(x) (rotr_32(x, 2) ^ rotr_32(x, 13) ^ rotr_32(x, 22))
+#define S1(x) (rotr_32(x, 6) ^ rotr_32(x, 11) ^ rotr_32(x, 25))
+#define s0(x) (rotr_32(x, 7) ^ rotr_32(x, 18) ^ (x >> 3))
+#define s1(x) (rotr_32(x, 17) ^ rotr_32(x, 19) ^ (x >> 10))
+
+static const unsigned int SHA256_K[64] = {
+ 0x428A2F98, 0x71374491, 0xB5C0FBCF, 0xE9B5DBA5,
+ 0x3956C25B, 0x59F111F1, 0x923F82A4, 0xAB1C5ED5,
+ 0xD807AA98, 0x12835B01, 0x243185BE, 0x550C7DC3,
+ 0x72BE5D74, 0x80DEB1FE, 0x9BDC06A7, 0xC19BF174,
+ 0xE49B69C1, 0xEFBE4786, 0x0FC19DC6, 0x240CA1CC,
+ 0x2DE92C6F, 0x4A7484AA, 0x5CB0A9DC, 0x76F988DA,
+ 0x983E5152, 0xA831C66D, 0xB00327C8, 0xBF597FC7,
+ 0xC6E00BF3, 0xD5A79147, 0x06CA6351, 0x14292967,
+ 0x27B70A85, 0x2E1B2138, 0x4D2C6DFC, 0x53380D13,
+ 0x650A7354, 0x766A0ABB, 0x81C2C92E, 0x92722C85,
+ 0xA2BFE8A1, 0xA81A664B, 0xC24B8B70, 0xC76C51A3,
+ 0xD192E819, 0xD6990624, 0xF40E3585, 0x106AA070,
+ 0x19A4C116, 0x1E376C08, 0x2748774C, 0x34B0BCB5,
+ 0x391C0CB3, 0x4ED8AA4A, 0x5B9CCA4F, 0x682E6FF3,
+ 0x748F82EE, 0x78A5636F, 0x84C87814, 0x8CC70208,
+ 0x90BEFFFA, 0xA4506CEB, 0xBEF9A3F7, 0xC67178F2,
+};
+
+static void
+transform(unsigned int state[8], const unsigned int data_xe[16])
+{
+ unsigned int data[16];
+ unsigned int W[16];
+ unsigned int T[8];
+ unsigned int j;
+
+ /* ensure big-endian integers */
+ for (j = 0; j < 16; j++)
+ data[j] = bswap32(data_xe[j]);
+
+ /* Copy state[] to working vars. */
+ memcpy(T, state, sizeof(T));
+
+ /* 64 operations, partially loop unrolled */
+ for (j = 0; j < 64; j += 16) {
+ R( 0); R( 1); R( 2); R( 3);
+ R( 4); R( 5); R( 6); R( 7);
+ R( 8); R( 9); R(10); R(11);
+ R(12); R(13); R(14); R(15);
+ }
+
+ /* Add the working vars back into state[]. */
+ state[0] += a(0);
+ state[1] += b(0);
+ state[2] += c(0);
+ state[3] += d(0);
+ state[4] += e(0);
+ state[5] += f(0);
+ state[6] += g(0);
+ state[7] += h(0);
+}
+
+void fz_sha256init(fz_sha256 *context)
+{
+ context->count[0] = context->count[1] = 0;
+
+ context->state[0] = 0x6A09E667;
+ context->state[1] = 0xBB67AE85;
+ context->state[2] = 0x3C6EF372;
+ context->state[3] = 0xA54FF53A;
+ context->state[4] = 0x510E527F;
+ context->state[5] = 0x9B05688C;
+ context->state[6] = 0x1F83D9AB;
+ context->state[7] = 0x5BE0CD19;
+}
+
+void fz_sha256update(fz_sha256 *context, const unsigned char *input, unsigned int inlen)
+{
+ /* Copy the input data into a properly aligned temporary buffer.
+ * This way we can be called with arbitrarily sized buffers
+ * (no need to be multiple of 64 bytes), and the code works also
+ * on architectures that don't allow unaligned memory access. */
+ while (inlen > 0)
+ {
+ const unsigned int copy_start = context->count[0] & 0x3F;
+ unsigned int copy_size = 64 - copy_start;
+ if (copy_size > inlen)
+ copy_size = inlen;
+
+ memcpy(context->buffer.u8 + copy_start, input, copy_size);
+
+ input += copy_size;
+ inlen -= copy_size;
+ context->count[0] += copy_size;
+ /* carry overflow from low to high */
+ if (context->count[0] < copy_size)
+ context->count[1]++;
+
+ if ((context->count[0] & 0x3F) == 0)
+ transform(context->state, context->buffer.u32);
+ }
+}
+
+void fz_sha256final(fz_sha256 *context, unsigned char digest[32])
+{
+ /* Add padding as described in RFC 3174 (it describes SHA-1 but
+ * the same padding style is used for SHA-256 too). */
+ unsigned int j = context->count[0] & 0x3F;
+ context->buffer.u8[j++] = 0x80;
+
+ while (j != 56)
+ {
+ if (j == 64)
+ {
+ transform(context->state, context->buffer.u32);
+ j = 0;
+ }
+ context->buffer.u8[j++] = 0x00;
+ }
+
+ /* Convert the message size from bytes to bits. */
+ context->count[1] = (context->count[1] << 3) + (context->count[0] >> 29);
+ context->count[0] = context->count[0] << 3;
+
+ context->buffer.u32[14] = bswap32(context->count[1]);
+ context->buffer.u32[15] = bswap32(context->count[0]);
+ transform(context->state, context->buffer.u32);
+
+ for (j = 0; j < 8; j++)
+ ((unsigned int *)digest)[j] = bswap32(context->state[j]);
+ memset(context, 0, sizeof(fz_sha256));
+}
diff --git a/fitz/fitz.h b/fitz/fitz.h
index 8f756895..6a339303 100644
--- a/fitz/fitz.h
+++ b/fitz/fitz.h
@@ -285,6 +285,24 @@ void fz_md5init(fz_md5 *state);
void fz_md5update(fz_md5 *state, const unsigned char *input, const unsigned inlen);
void fz_md5final(fz_md5 *state, unsigned char digest[16]);
+/* sha-256 digests */
+
+typedef struct fz_sha256_s fz_sha256;
+
+struct fz_sha256_s
+{
+ unsigned int state[8];
+ unsigned int count[2];
+ union {
+ unsigned char u8[64];
+ unsigned int u32[16];
+ } buffer;
+};
+
+void fz_sha256init(fz_sha256 *state);
+void fz_sha256update(fz_sha256 *state, const unsigned char *input, unsigned int inlen);
+void fz_sha256final(fz_sha256 *state, unsigned char digest[32]);
+
/* arc4 crypto */
typedef struct fz_arc4_s fz_arc4;
diff --git a/mupdf/mupdf.h b/mupdf/mupdf.h
index e53ea90a..4bbea33b 100644
--- a/mupdf/mupdf.h
+++ b/mupdf/mupdf.h
@@ -70,6 +70,7 @@ enum
PDF_CRYPT_NONE,
PDF_CRYPT_RC4,
PDF_CRYPT_AESV2,
+ PDF_CRYPT_AESV3,
PDF_CRYPT_UNKNOWN,
};
@@ -77,7 +78,6 @@ struct pdf_cryptfilter_s
{
int method;
int length;
- unsigned char key[16];
};
struct pdf_crypt_s
@@ -92,8 +92,10 @@ struct pdf_crypt_s
pdf_cryptfilter strf;
int r;
- unsigned char o[32];
- unsigned char u[32];
+ unsigned char o[48];
+ unsigned char u[48];
+ unsigned char oe[32];
+ unsigned char ue[32];
int p;
int encryptmetadata;
diff --git a/mupdf/pdf_crypt.c b/mupdf/pdf_crypt.c
index 019f889c..25cb1d8d 100644
--- a/mupdf/pdf_crypt.c
+++ b/mupdf/pdf_crypt.c
@@ -35,7 +35,7 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
obj = fz_dictgets(dict, "V");
if (fz_isint(obj))
crypt->v = fz_toint(obj);
- if (crypt->v != 1 && crypt->v != 2 && crypt->v != 4)
+ if (crypt->v != 1 && crypt->v != 2 && crypt->v != 4 && crypt->v != 5)
{
pdf_freecrypt(crypt);
return fz_throw("unknown encryption version");
@@ -64,6 +64,9 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
}
}
+ if (crypt->v == 5)
+ crypt->length = 256;
+
if (crypt->v == 1 || crypt->v == 2)
{
crypt->stmf.method = PDF_CRYPT_RC4;
@@ -73,7 +76,7 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
crypt->strf.length = crypt->length;
}
- if (crypt->v == 4)
+ if (crypt->v == 4 || crypt->v == 5)
{
crypt->stmf.method = PDF_CRYPT_NONE;
crypt->stmf.length = crypt->length;
@@ -134,6 +137,9 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
obj = fz_dictgets(dict, "O");
if (fz_isstring(obj) && fz_tostrlen(obj) == 32)
memcpy(crypt->o, fz_tostrbuf(obj), 32);
+ /* /O and /U are supposed to be 48 bytes long for revision 5, they're often longer, though */
+ else if (crypt->r == 5 && fz_isstring(obj) && fz_tostrlen(obj) >= 48)
+ memcpy(crypt->o, fz_tostrbuf(obj), 48);
else
{
pdf_freecrypt(crypt);
@@ -143,6 +149,8 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
obj = fz_dictgets(dict, "U");
if (fz_isstring(obj) && fz_tostrlen(obj) == 32)
memcpy(crypt->u, fz_tostrbuf(obj), 32);
+ else if (fz_isstring(obj) && fz_tostrlen(obj) >= 48 && crypt->r == 5)
+ memcpy(crypt->u, fz_tostrbuf(obj), 48);
else
{
pdf_freecrypt(crypt);
@@ -158,6 +166,25 @@ pdf_newcrypt(pdf_crypt **cryptp, fz_obj *dict, fz_obj *id)
return fz_throw("encryption dictionary missing permissions value");
}
+ if (crypt->r == 5)
+ {
+ obj = fz_dictgets(dict, "OE");
+ if (!fz_isstring(obj) || fz_tostrlen(obj) != 32)
+ {
+ pdf_freecrypt(crypt);
+ return fz_throw("encryption dictionary missing owner encryption key");
+ }
+ memcpy(crypt->oe, fz_tostrbuf(obj), 32);
+
+ obj = fz_dictgets(dict, "UE");
+ if (!fz_isstring(obj) || fz_tostrlen(obj) != 32)
+ {
+ pdf_freecrypt(crypt);
+ return fz_throw("encryption dictionary missing user encryption key");
+ }
+ memcpy(crypt->ue, fz_tostrbuf(obj), 32);
+ }
+
crypt->encryptmetadata = 1;
obj = fz_dictgets(dict, "EncryptMetadata");
if (fz_isbool(obj))
@@ -214,6 +241,8 @@ pdf_parsecryptfilter(pdf_cryptfilter *cf, fz_obj *dict, int defaultlength)
cf->method = PDF_CRYPT_RC4;
else if (!strcmp(fz_toname(obj), "AESV2"))
cf->method = PDF_CRYPT_AESV2;
+ else if (!strcmp(fz_toname(obj), "AESV3"))
+ cf->method = PDF_CRYPT_AESV3;
else
fz_throw("unknown encryption method: %s", fz_toname(obj));
}
@@ -306,6 +335,51 @@ pdf_computeencryptionkey(pdf_crypt *crypt, unsigned char *password, int pwlen, u
}
/*
+ * Compute an encryption key (PDF 1.7 ExtensionLevel 3 algorithm 3.2a)
+ */
+
+static void
+pdf_computeencryptionkey_r5(pdf_crypt *crypt, unsigned char *password, int pwlen, int ownerkey, unsigned char *validationkey)
+{
+ unsigned char buffer[128 + 8 + 48];
+ fz_sha256 sha256;
+ fz_aes aes;
+
+ /* Step 2 - truncate UTF-8 password to 127 characters */
+
+ if (pwlen > 127)
+ pwlen = 127;
+
+ /* Step 3/4 - test password against owner/user key and compute encryption key */
+
+ memcpy(buffer, password, pwlen);
+ if (ownerkey)
+ {
+ memcpy(buffer + pwlen, crypt->o + 32, 8);
+ memcpy(buffer + pwlen + 8, crypt->u, 48);
+ }
+ else
+ memcpy(buffer + pwlen, crypt->u + 32, 8);
+
+ fz_sha256init(&sha256);
+ fz_sha256update(&sha256, buffer, pwlen + 8 + (ownerkey ? 48 : 0));
+ fz_sha256final(&sha256, validationkey);
+
+ /* Step 3.5/4.5 - compute file encryption key from OE/UE */
+
+ memcpy(buffer + pwlen, crypt->u + 40, 8);
+
+ fz_sha256init(&sha256);
+ fz_sha256update(&sha256, buffer, pwlen + 8);
+ fz_sha256final(&sha256, buffer);
+
+ // clear password buffer and use it as iv
+ memset(buffer + 32, 0, sizeof(buffer) - 32);
+ aes_setkey_dec(&aes, buffer, crypt->length);
+ aes_crypt_cbc(&aes, AES_DECRYPT, 32, buffer + 32, ownerkey ? crypt->oe : crypt->ue, crypt->key);
+}
+
+/*
* Computing the user password (PDF 1.7 algorithm 3.4 and 3.5)
* Also save the generated key for decrypting objects and streams in crypt->key.
*/
@@ -322,7 +396,7 @@ pdf_computeuserpassword(pdf_crypt *crypt, unsigned char *password, int pwlen, un
fz_arc4encrypt(&arc4, output, padding, 32);
}
- if (crypt->r >= 3)
+ if (crypt->r == 3 || crypt->r == 4)
{
unsigned char xor[32];
unsigned char digest[16];
@@ -352,10 +426,16 @@ pdf_computeuserpassword(pdf_crypt *crypt, unsigned char *password, int pwlen, un
memcpy(output + 16, padding, 16);
}
+
+ if (crypt->r == 5)
+ {
+ pdf_computeencryptionkey_r5(crypt, password, pwlen, 0, output);
+ }
}
/*
- * Authenticating the user password (PDF 1.7 algorithm 3.6)
+ * Authenticating the user password (PDF 1.7 algorithm 3.6
+ * and ExtensionLevel 3 algorithm 3.11)
* This also has the side effect of saving a key generated
* from the password for decrypting objects and streams.
*/
@@ -365,15 +445,16 @@ pdf_authenticateuserpassword(pdf_crypt *crypt, unsigned char *password, int pwle
{
unsigned char output[32];
pdf_computeuserpassword(crypt, password, pwlen, output);
- if (crypt->r == 2)
+ if (crypt->r == 2 || crypt->r == 5)
return memcmp(output, crypt->u, 32) == 0;
- if (crypt->r >= 3)
+ if (crypt->r == 3 || crypt->r == 4)
return memcmp(output, crypt->u, 16) == 0;
return 0;
}
/*
- * Authenticating the owner password (PDF 1.7 algorithm 3.7)
+ * Authenticating the owner password (PDF 1.7 algorithm 3.7
+ * and ExtensionLevel 3 algorithm 3.12)
* Generates the user password from the owner password
* and calls pdf_authenticateuserpassword.
*/
@@ -389,6 +470,15 @@ pdf_authenticateownerpassword(pdf_crypt *crypt, unsigned char *ownerpass, int pw
fz_md5 md5;
fz_arc4 arc4;
+ if (crypt->r == 5)
+ {
+ /* PDF 1.7 ExtensionLevel 3 algorithm 3.12 */
+
+ pdf_computeencryptionkey_r5(crypt, ownerpass, pwlen, 1, key);
+
+ return !memcmp(key, crypt->o, 32);
+ }
+
n = crypt->length / 8;
/* Step 1 -- steps 1 to 4 of PDF 1.7 algorithm 3.3 */
@@ -463,7 +553,7 @@ pdf_needspassword(pdf_xref *xref)
}
/*
- * PDF 1.7 algorithm 3.1
+ * PDF 1.7 algorithm 3.1 and ExtensionLevel 3 algorithm 3.1a
*
* Using the global encryption key that was generated from the
* password, create a new key that is used to decrypt indivual
@@ -477,6 +567,12 @@ pdf_computeobjectkey(pdf_crypt *crypt, pdf_cryptfilter *cf, int num, int gen, un
fz_md5 md5;
unsigned char message[5];
+ if (cf->method == PDF_CRYPT_AESV3)
+ {
+ memcpy(key, crypt->key, crypt->length / 8);
+ return crypt->length / 8;
+ }
+
fz_md5init(&md5);
fz_md5update(&md5, crypt->key, crypt->length / 8);
message[0] = (num) & 0xFF;
@@ -497,7 +593,7 @@ pdf_computeobjectkey(pdf_crypt *crypt, pdf_cryptfilter *cf, int num, int gen, un
}
/*
- * PDF 1.7 algorithm 3.1
+ * PDF 1.7 algorithm 3.1 and ExtensionLevel 3 algorithm 3.1a
*
* Decrypt all strings in obj modifying the data in-place.
* Recurse through arrays and dictionaries, but do not follow
@@ -525,7 +621,7 @@ pdf_cryptobjimp(pdf_crypt *crypt, fz_obj *obj, unsigned char *key, int keylen)
fz_arc4encrypt(&arc4, s, s, n);
}
- if (crypt->strf.method == PDF_CRYPT_AESV2)
+ if (crypt->strf.method == PDF_CRYPT_AESV2 || crypt->strf.method == PDF_CRYPT_AESV3)
{
if (n >= 32)
{
@@ -562,7 +658,7 @@ pdf_cryptobjimp(pdf_crypt *crypt, fz_obj *obj, unsigned char *key, int keylen)
void
pdf_cryptobj(pdf_crypt *crypt, fz_obj *obj, int num, int gen)
{
- unsigned char key[16];
+ unsigned char key[32];
int len;
len = pdf_computeobjectkey(crypt, &crypt->strf, num, gen, key);
@@ -571,14 +667,14 @@ pdf_cryptobj(pdf_crypt *crypt, fz_obj *obj, int num, int gen)
}
/*
- * PDF 1.7 algorithm 3.1
+ * PDF 1.7 algorithm 3.1 and ExtensionLevel 3 algorithm 3.1a
*
* Create filter suitable for de/encrypting a stream.
*/
fz_stream *
pdf_opencrypt(fz_stream *chain, pdf_crypt *crypt, pdf_cryptfilter *stmf, int num, int gen)
{
- unsigned char key[16];
+ unsigned char key[32];
int len;
len = pdf_computeobjectkey(crypt, stmf, num, gen, key);
@@ -586,7 +682,7 @@ pdf_opencrypt(fz_stream *chain, pdf_crypt *crypt, pdf_cryptfilter *stmf, int num
if (stmf->method == PDF_CRYPT_RC4)
return fz_openarc4(chain, key, len);
- if (stmf->method == PDF_CRYPT_AESV2)
+ if (stmf->method == PDF_CRYPT_AESV2 || stmf->method == PDF_CRYPT_AESV3)
return fz_openaesd(chain, key, len);
return fz_opencopy(chain);
diff --git a/win32/mupdf/mupdf.vcproj b/win32/mupdf/mupdf.vcproj
index ac0af2e9..8292a2a8 100644
--- a/win32/mupdf/mupdf.vcproj
+++ b/win32/mupdf/mupdf.vcproj
@@ -325,6 +325,10 @@
>
</File>
<File
+ RelativePath="..\..\fitz\crypt_sha2.c"
+ >
+ </File>
+ <File
RelativePath="..\..\fitz\dev_bbox.c"
>
</File>