dany/pdfio
1
0
Fork 0
mirror of https://github.com/michaelrsweet/pdfio.git synced 2024-06-29 10:19:47 +02:00
Code Issues Packages Projects Releases Wiki Activity

Save work on resolving PDF loading issues with random PDFs using different encryption methods and line endings.

Browse Source
This commit is contained in:
Michael R Sweet 2021-11-01 21:30:46 -04:00
parent 6432187dea
commit 2f0d622873
No known key found for this signature in database
GPG Key ID: BE67C75EC81F3244
5 changed files with 103 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
pdfio-array.c
View File

@ -396,20 +396,27 @@ pdfioArrayGetBinary(
size_t n, // I - Index
size_t *length) // O - Length of string
{
if (!a || n >= a->num_values || a->values[n].type != PDFIO_VALTYPE_BINARY)
if (!a || n >= a->num_values || (a->values[n].type != PDFIO_VALTYPE_BINARY && a->values[n].type != PDFIO_VALTYPE_STRING))
{
if (length)
*length = 0;
return (NULL);
}
else
else if (a->values[n].type == PDFIO_VALTYPE_BINARY)
{
if (length)
*length = a->values[n].value.binary.datalen;
return (a->values[n].value.binary.data);
}
else
{
if (length)
*length = strlen(a->values[n].value.string);
return ((unsigned char *)a->values[n].value.string);
}
}

68
pdfio-crypto.c
View File

@ -430,6 +430,24 @@ _pdfio_crypto_cb_t // O - Decryption callback or `NULL` for none
*ivlen = 0;
return (NULL);
case PDFIO_ENCRYPTION_RC4_40 :
// Copy the key data for the MD5 hash.
memcpy(data, pdf->file_key, sizeof(pdf->file_key));
data[16] = (uint8_t)obj->number;
data[17] = (uint8_t)(obj->number >> 8);
data[18] = (uint8_t)(obj->number >> 16);
data[19] = (uint8_t)obj->generation;
data[20] = (uint8_t)(obj->generation >> 8);
// Hash it...
_pdfioCryptoMD5Init(&md5);
_pdfioCryptoMD5Append(&md5, data, sizeof(data));
_pdfioCryptoMD5Finish(&md5, digest);
// Initialize the RC4 context using 40 bits of the digest...
_pdfioCryptoRC4Init(&ctx->rc4, digest, 5);
return ((_pdfio_crypto_cb_t)_pdfioCryptoRC4Crypt);
case PDFIO_ENCRYPTION_RC4_128 :
case PDFIO_ENCRYPTION_AES_128 :
// Copy the key data for the MD5 hash.
@ -635,7 +653,7 @@ _pdfioCryptoUnlock(
}
}
}
else if (version == 2)
else if (version == 1 || version == 2)
{
if (revision == 2)
{
@ -651,6 +669,8 @@ _pdfioCryptoUnlock(
}
// TODO: Implement AES-256 - V6 R6
PDFIO_DEBUG("_pdfioCryptoUnlock: encryption=%d, length=%d\n", pdf->encryption, length);
if (pdf->encryption == PDFIO_ENCRYPTION_NONE)
{
_pdfioFileError(pdf, "Unsupported encryption V%d R%d.", version, revision);
@ -661,23 +681,39 @@ _pdfioCryptoUnlock(
pdf->file_keylen = (size_t)(length / 8);
pdf->permissions = (pdfio_permission_t)pdfioDictGetNumber(encrypt_dict, "P");
PDFIO_DEBUG("_pdfioCryptoUnlock: permissions=%d\n", pdf->permissions);
owner_key = pdfioDictGetBinary(encrypt_dict, "O", &owner_keylen);
user_key = pdfioDictGetBinary(encrypt_dict, "U", &user_keylen);
if (!owner_key || owner_keylen < 32 || owner_keylen > sizeof(pdf->owner_key))
if (!owner_key)
{
_pdfioFileError(pdf, "Missing or bad owner key, unable to unlock file.");
_pdfioFileError(pdf, "Missing owner key, unable to unlock file.");
return (false);
}
else if (owner_keylen < 32 || owner_keylen > sizeof(pdf->owner_key))
{
_pdfioFileError(pdf, "Bad %d bytes owner key, unable to unlock file.", (int)owner_keylen);
return (false);
}
PDFIO_DEBUG("_pdfioCryptoUnlock: owner_key[%d]=%02X%02X%02X%02X...%02X%02X%02X%02X\n", (int)owner_keylen, owner_key[0], owner_key[1], owner_key[2], owner_key[3], owner_key[28], owner_key[29], owner_key[30], owner_key[31]);
memcpy(pdf->owner_key, owner_key, owner_keylen);
pdf->owner_keylen = owner_keylen;
if (!user_key || user_keylen < 32 || user_keylen > sizeof(pdf->user_key))
if (!user_key)
{
_pdfioFileError(pdf, "Missing or bad user key, unable to unlock file.");
_pdfioFileError(pdf, "Missing user key, unable to unlock file.");
return (false);
}
else if (user_keylen < 32 || user_keylen > sizeof(pdf->user_key))
{
_pdfioFileError(pdf, "Bad %d byte user key, unable to unlock file.", (int)user_keylen);
return (false);
}
PDFIO_DEBUG("_pdfioCryptoUnlock: user_key[%d]=%02X%02X%02X%02X...%02X%02X%02X%02X\n", (int)user_keylen, user_key[0], user_key[1], user_key[2], user_key[3], user_key[28], user_key[29], user_key[30], user_key[31]);
memcpy(pdf->user_key, user_key, user_keylen);
pdf->user_keylen = user_keylen;
@ -709,20 +745,20 @@ _pdfioCryptoUnlock(
pad_password(password, pad);
// Generate keys to see if things match...
PDFIO_DEBUG("\nTrying %02X%02X%02X%02X...%02X%02X%02X%02X\n", pad[0], pad[1], pad[2], pad[3], pad[28], pad[29], pad[30], pad[31]);
PDFIO_DEBUG("P=%d\n", pdf->permissions);
PDFIO_DEBUG("Fid(%d)=%02X%02X%02X%02X...%02X%02X%02X%02X\n", (int)file_idlen, file_id[0], file_id[1], file_id[2], file_id[3], file_id[12], file_id[13], file_id[14], file_id[15]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Trying %02X%02X%02X%02X...%02X%02X%02X%02X\n", pad[0], pad[1], pad[2], pad[3], pad[28], pad[29], pad[30], pad[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: P=%d\n", pdf->permissions);
PDFIO_DEBUG("_pdfioCryptoUnlock: Fid(%d)=%02X%02X%02X%02X...%02X%02X%02X%02X\n", (int)file_idlen, file_id[0], file_id[1], file_id[2], file_id[3], file_id[12], file_id[13], file_id[14], file_id[15]);
make_owner_key(pdf->encryption, pad, pdf->owner_key, user_pad);
PDFIO_DEBUG("Upad=%02X%02X%02X%02X...%02X%02X%02X%02X\n", user_pad[0], user_pad[1], user_pad[2], user_pad[3], user_pad[28], user_pad[29], user_pad[30], user_pad[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Upad=%02X%02X%02X%02X...%02X%02X%02X%02X\n", user_pad[0], user_pad[1], user_pad[2], user_pad[3], user_pad[28], user_pad[29], user_pad[30], user_pad[31]);
make_file_key(pdf->encryption, pdf->permissions, file_id, file_idlen, user_pad, pdf->owner_key, file_key);
PDFIO_DEBUG("Fown=%02X%02X%02X%02X...%02X%02X%02X%02X\n", file_key[0], file_key[1], file_key[2], file_key[3], file_key[12], file_key[13], file_key[14], file_key[15]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Fown=%02X%02X%02X%02X...%02X%02X%02X%02X\n", file_key[0], file_key[1], file_key[2], file_key[3], file_key[12], file_key[13], file_key[14], file_key[15]);
make_user_key(file_id, file_idlen, own_user_key);
PDFIO_DEBUG("U=%02X%02X%02X%02X...%02X%02X%02X%02X\n", pdf->user_key[0], pdf->user_key[1], pdf->user_key[2], pdf->user_key[3], pdf->user_key[28], pdf->user_key[29], pdf->user_key[30], pdf->user_key[31]);
PDFIO_DEBUG("Uown=%02X%02X%02X%02X...%02X%02X%02X%02X\n", own_user_key[0], own_user_key[1], own_user_key[2], own_user_key[3], own_user_key[28], own_user_key[29], own_user_key[30], own_user_key[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: U=%02X%02X%02X%02X...%02X%02X%02X%02X\n", pdf->user_key[0], pdf->user_key[1], pdf->user_key[2], pdf->user_key[3], pdf->user_key[28], pdf->user_key[29], pdf->user_key[30], pdf->user_key[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Uown=%02X%02X%02X%02X...%02X%02X%02X%02X\n", own_user_key[0], own_user_key[1], own_user_key[2], own_user_key[3], own_user_key[28], own_user_key[29], own_user_key[30], own_user_key[31]);
if (!memcmp(own_user_key, pdf->user_key, sizeof(own_user_key)))
{
@ -736,17 +772,17 @@ _pdfioCryptoUnlock(
*/
make_file_key(pdf->encryption, pdf->permissions, file_id, file_idlen, pad, pdf->owner_key, file_key);
PDFIO_DEBUG("Fuse=%02X%02X%02X%02X...%02X%02X%02X%02X\n", file_key[0], file_key[1], file_key[2], file_key[3], file_key[12], file_key[13], file_key[14], file_key[15]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Fuse=%02X%02X%02X%02X...%02X%02X%02X%02X\n", file_key[0], file_key[1], file_key[2], file_key[3], file_key[12], file_key[13], file_key[14], file_key[15]);
make_user_key(file_id, file_idlen, user_key);
memcpy(pdf_user_key, pdf->user_key, sizeof(pdf_user_key));
decrypt_user_key(pdf->encryption, file_key, pdf_user_key);
PDFIO_DEBUG("Uuse=%02X%02X%02X%02X...%02X%02X%02X%02X\n", user_key[0], user_key[1], user_key[2], user_key[3], user_key[28], user_key[29], user_key[30], user_key[31]);
PDFIO_DEBUG("Updf=%02X%02X%02X%02X...%02X%02X%02X%02X\n", pdf_user_key[0], pdf_user_key[1], pdf_user_key[2], pdf_user_key[3], pdf_user_key[28], pdf_user_key[29], pdf_user_key[30], pdf_user_key[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Uuse=%02X%02X%02X%02X...%02X%02X%02X%02X\n", user_key[0], user_key[1], user_key[2], user_key[3], user_key[28], user_key[29], user_key[30], user_key[31]);
PDFIO_DEBUG("_pdfioCryptoUnlock: Updf=%02X%02X%02X%02X...%02X%02X%02X%02X\n", pdf_user_key[0], pdf_user_key[1], pdf_user_key[2], pdf_user_key[3], pdf_user_key[28], pdf_user_key[29], pdf_user_key[30], pdf_user_key[31]);
if (!memcmp(user_key, pdf_user_key, 16))
if (!memcmp(pad, pdf_user_key, 32) || !memcmp(own_user_key, user_key, 32) || !memcmp(user_key, pdf_user_key, 16))
{
// Matches!
memcpy(pdf->file_key, file_key, sizeof(pdf->file_key));

5
pdfio-dict.c
View File

@ -245,6 +245,11 @@ pdfioDictGetBinary(pdfio_dict_t *dict, // I - Dictionary
*length = value->value.binary.datalen;
return (value->value.binary.data);
}
else if (value && value->type == PDFIO_VALTYPE_STRING)
{
*length = strlen(value->value.string);
return ((unsigned char *)value->value.string);
}
else
{
*length = 0;

7
pdfio-file.c
View File

@ -1414,6 +1414,7 @@ load_xref(
offset; // Offset in file
int generation; // Generation number
_pdfio_token_t tb; // Token buffer/stack
off_t line_offset; // Offset to start of line
while (!done)
@ -1426,6 +1427,8 @@ load_xref(
do
{
line_offset = _pdfioFileTell(pdf);
if (!_pdfioFileGets(pdf, line, sizeof(line)))
{
_pdfioFileError(pdf, "Unable to read start of xref table.");
@ -1434,7 +1437,7 @@ load_xref(
}
while (!line[0]);
PDFIO_DEBUG("load_xref: xref_offset=%lu, line='%s'\n", (unsigned long)xref_offset, line);
PDFIO_DEBUG("load_xref: line_offset=%lu, line='%s'\n", (unsigned long)line_offset, line);
if (isdigit(line[0] & 255) && strlen(line) > 4 && (!strcmp(line + strlen(line) - 4, " obj") || ((ptr = strstr(line, " obj")) != NULL && ptr[4] == '<')))
{
@ -1477,7 +1480,7 @@ load_xref(
return (false);
}
if (_pdfioFileSeek(pdf, xref_offset + ptr + 3 - line, SEEK_SET) < 0)
if (_pdfioFileSeek(pdf, line_offset + ptr + 3 - line, SEEK_SET) < 0)
{
_pdfioFileError(pdf, "Unable to seek to xref object %lu %u.", (unsigned long)number, (unsigned)generation);
return (false);

32
pdfio-token.c
View File

@ -200,6 +200,7 @@ _pdfioTokenRead(_pdfio_token_t *tb, // I - Token buffer/stack
char *bufptr, // Pointer into buffer
*bufend, // End of buffer
state = '\0'; // Current state
bool saw_nul = false; // Did we see a nul character?
//
@ -260,6 +261,9 @@ _pdfioTokenRead(_pdfio_token_t *tb, // I - Token buffer/stack
case '(' : // Literal string
while ((ch = get_char(tb)) != EOF)
{
if (ch == 0)
saw_nul = true;
if (ch == '\\')
{
// Quoted character...
@ -350,6 +354,34 @@ _pdfioTokenRead(_pdfio_token_t *tb, // I - Token buffer/stack
_pdfioFileError(tb->pdf, "Unterminated string literal.");
return (false);
}
if (saw_nul)
{
// Convert to a hex (binary) string...
char *litptr, // Pointer to literal character
*hexptr; // Pointer to hex character
size_t bytes = (size_t)(bufptr - buffer - 1);
// Bytes of data...
static const char *hexchars = "0123456789ABCDEF";
// Hex digits
PDFIO_DEBUG("_pdfioTokenRead: Converting nul-containing string to binary.\n");
if ((2 * (bytes + 1)) > bufsize)
{
// Out of space...
_pdfioFileError(tb->pdf, "Token too large.");
return (false);
}
*buffer = '<';
for (litptr = bufptr - 1, hexptr = buffer + 2 * bytes - 1; litptr > buffer; litptr --, hexptr -= 2)
{
hexptr[0] = hexchars[(*litptr >> 4) & 15];
hexptr[1] = hexchars[*litptr & 15];
}
bufptr = buffer + 2 * bytes + 1;
}
break;
case 'K' : // keyword