dany/pdfio
1
0
Fork 0
mirror of https://github.com/michaelrsweet/pdfio.git synced 2024-12-26 13:28:22 +01:00
Code Issues Packages Projects Releases Wiki Activity
83bfb135c6
pdfio/pdfio-private.h

402 lines
17 KiB
C
Raw Normal View History

Initial checkin.
2021-04-10 14:00:52 +02:00
//
Rename as PDFio for consistency. Add PDFio logo image. Move documentation to "doc" subdirectory.
2021-05-30 13:10:44 +02:00
// Private header file for PDFio.
Initial checkin.
2021-04-10 14:00:52 +02:00
//
Clean up private header.
2023-10-02 11:27:40 +02:00
// Copyright © 2021-2023 by Michael R Sweet.
Initial checkin.
2021-04-10 14:00:52 +02:00
//
// Licensed under Apache License v2.0. See the file "LICENSE" for more
// information.
//
#ifndef PDFIO_PRIVATE_H
# define PDFIO_PRIVATE_H
Initial Visual Studio support.
2021-06-21 17:39:06 +02:00
# ifdef _WIN32
Clean up private header.
2023-10-02 11:27:40 +02:00
# define _CRT_SECURE_NO_WARNINGS // Disable bogus VS warnings/errors...
Initial Visual Studio support.
2021-06-21 17:39:06 +02:00
# endif // _WIN32
Start filling out internal APIs.
2021-04-17 03:09:43 +02:00
# include "pdfio.h"
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
# include <stdarg.h>
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
# include <stdint.h>
Initial Visual Studio support.
2021-06-21 17:39:06 +02:00
# include <string.h>
Cleanup spacing and comments.
2023-10-06 20:39:42 +02:00
# include <ctype.h>
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
# include <errno.h>
Implement pdfioFileOpen reading of xref table.
2021-05-01 17:41:43 +02:00
# include <inttypes.h>
Initial Visual Studio support.
2021-06-21 17:39:06 +02:00
# include <fcntl.h>
Implement JPEG image object support.
2021-05-25 22:31:32 +02:00
# ifdef _WIN32
# include <io.h>
# include <direct.h>
added windows.h header for GetTempPathA
2023-10-10 09:12:03 +02:00
# include <windows.h> // GetTempPathA
Clean up private header.
2023-10-02 11:27:40 +02:00
# define access _access // Map standard POSIX/C99 names
Implement JPEG image object support.
2021-05-25 22:31:32 +02:00
# define close _close
# define fileno _fileno
# define lseek _lseek
# define mkdir(d,p) _mkdir(d)
# define open _open
# define read _read
# define rmdir _rmdir
# define snprintf _snprintf
# define strdup _strdup
# define unlink _unlink
# define vsnprintf _vsnprintf
# define write _write
Clean up private header.
2023-10-02 11:27:40 +02:00
# define F_OK 00 // POSIX parameters/flags
Implement JPEG image object support.
2021-05-25 22:31:32 +02:00
# define W_OK 02
# define R_OK 04
Cleanup spacing and comments.
2023-10-06 20:39:42 +02:00
# define O_RDONLY _O_RDONLY // Map standard POSIX open flags
Implement JPEG image object support.
2021-05-25 22:31:32 +02:00
# define O_WRONLY _O_WRONLY
# define O_CREAT _O_CREAT
# define O_TRUNC _O_TRUNC
# define O_BINARY _O_BINARY
# else // !_WIN32
# include <unistd.h>
Cleanup spacing and comments.
2023-10-06 20:39:42 +02:00
# define O_BINARY 0 // Map Windows-specific open flag
Implement JPEG image object support.
2021-05-25 22:31:32 +02:00
# endif // _WIN32
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
# include <zlib.h>
Start filling out internal APIs.
2021-04-17 03:09:43 +02:00
Initial checkin.
2021-04-10 14:00:52 +02:00
//
// Visibility and other annotations...
//
# if defined(__has_extension) || defined(__GNUC__)
Save work on documentation.
2021-07-24 19:08:46 +02:00
# define _PDFIO_INTERNAL __attribute__ ((visibility("hidden")))
# define _PDFIO_PRIVATE __attribute__ ((visibility("default")))
# define _PDFIO_NONNULL(...) __attribute__ ((nonnull(__VA_ARGS__)))
# define _PDFIO_NORETURN __attribute__ ((noreturn))
Initial checkin.
2021-04-10 14:00:52 +02:00
# else
Save work on documentation.
2021-07-24 19:08:46 +02:00
# define _PDFIO_INTERNAL
# define _PDFIO_PRIVATE
# define _PDFIO_NONNULL(...)
# define _PDFIO_NORETURN
Initial checkin.
2021-04-10 14:00:52 +02:00
# endif // __has_extension || __GNUC__
//
// Debug macro...
//
# ifdef DEBUG
Increase token buffer to 256 bytes, to avoid overflow scenarios. Add _pdfioTokenFlush to manually flush the token buffer. Add more debug printfs.
2021-05-10 23:37:57 +02:00
# define PDFIO_DEBUG(...) fprintf(stderr, __VA_ARGS__)
# define PDFIO_DEBUG_ARRAY(array) _pdfioArrayDebug(array, stderr)
# define PDFIO_DEBUG_DICT(dict) _pdfioDictDebug(dict, stderr)
# define PDFIO_DEBUG_VALUE(value) _pdfioValueDebug(value, stderr)
Initial checkin.
2021-04-10 14:00:52 +02:00
# else
# define PDFIO_DEBUG(...)
Rework debug functions, start xref stream support.
2021-05-06 15:51:48 +02:00
# define PDFIO_DEBUG_ARRAY(array)
# define PDFIO_DEBUG_DICT(dict)
# define PDFIO_DEBUG_VALUE(value)
Initial checkin.
2021-04-10 14:00:52 +02:00
# endif // DEBUG
//
// Types and constants...
//
Fix a few stack/buffer overflow bugs discovered by Bart, Steffan, and Mark from the Radboud University NL (thanks!) - Add depth argument to all value read functions that recurse - Add depth argument to page tree loading code - Validate xref stream sizes individually to avoid out-of-bounds access to local xref buffer.
2021-11-29 23:46:56 +01:00
# define PDFIO_MAX_DEPTH 32 // Maximum nesting depth for values
Finish implementation of pdfioContentTextMeasure (Issue #17)
2023-12-05 03:20:51 +01:00
typedef void (*_pdfio_extfree_t)(void *);
// Extension data free function
Save work.
2021-04-28 03:22:34 +02:00
typedef enum _pdfio_mode_e // Read/write mode
{
_PDFIO_MODE_READ, // Read a PDF file
_PDFIO_MODE_WRITE // Write a PDF file
} _pdfio_mode_t;
Safe work on streams - still need to implement predictors
2021-05-07 14:47:49 +02:00
typedef enum _pdfio_predictor_e // PNG predictor constants
{
_PDFIO_PREDICTOR_NONE = 1, // No predictor (default)
_PDFIO_PREDICTOR_TIFF2 = 2, // TIFF2 predictor (???)
_PDFIO_PREDICTOR_PNG_NONE = 10, // PNG None predictor (same as `_PDFIO_PREDICTOR_NONE`)
_PDFIO_PREDICTOR_PNG_SUB = 11, // PNG Sub predictor
_PDFIO_PREDICTOR_PNG_UP = 12, // PNG Up predictor
_PDFIO_PREDICTOR_PNG_AVERAGE = 13, // PNG Average predictor
Initial writing support.
2021-05-16 17:39:05 +02:00
_PDFIO_PREDICTOR_PNG_PAETH = 14, // PNG Paeth predictor
_PDFIO_PREDICTOR_PNG_AUTO = 15 // PNG "auto" predictor (currently mapped to Paeth)
Safe work on streams - still need to implement predictors
2021-05-07 14:47:49 +02:00
} _pdfio_predictor_t;
Rework token reading to be separate from the PDF file so we can add support for compressed object streams.
2021-05-08 13:38:44 +02:00
typedef ssize_t (*_pdfio_tconsume_cb_t)(void *data, size_t bytes);
typedef ssize_t (*_pdfio_tpeek_cb_t)(void *data, void *buffer, size_t bytes);
typedef struct _pdfio_token_s // Token buffer/stack
{
pdfio_file_t *pdf; // PDF file
_pdfio_tconsume_cb_t consume_cb; // Consume callback
_pdfio_tpeek_cb_t peek_cb; // Peek callback
void *cb_data; // Callback data
Increase token buffer to 256 bytes, to avoid overflow scenarios. Add _pdfioTokenFlush to manually flush the token buffer. Add more debug printfs.
2021-05-10 23:37:57 +02:00
unsigned char buffer[256], // Buffer
Rework token reading to be separate from the PDF file so we can add support for compressed object streams.
2021-05-08 13:38:44 +02:00
*bufptr, // Pointer into buffer
*bufend; // Last valid byte in buffer
size_t num_tokens; // Number of tokens in stack
char *tokens[4]; // Token stack
} _pdfio_token_t;
Save work.
2021-04-28 03:22:34 +02:00
typedef struct _pdfio_value_s // Value structure
{
pdfio_valtype_t type; // Type of value
union
{
pdfio_array_t *array; // Array value
struct
{
unsigned char *data; // Data
size_t datalen; // Length
} binary; // Binary ("Hex String") data
bool boolean; // Boolean value
time_t date; // Date/time value
pdfio_dict_t *dict; // Dictionary value
Implement value/array/dict reading, refactor indirect references.
2021-05-04 16:34:17 +02:00
struct
{
size_t number; // Object number
unsigned short generation; // Generation number
} indirect; // Indirect object reference
Save work.
2021-04-28 03:22:34 +02:00
const char *name; // Name value
Switch to using double for numbers, since they are also used to record lengths and other potentially large contents.
2021-05-30 03:16:21 +02:00
double number; // Number value
Save work.
2021-04-28 03:22:34 +02:00
const char *string; // String value
} value; // Value union
} _pdfio_value_t;
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
typedef struct _pdfio_aes_s // AES encryption state
{
size_t round_size; // Size of round key
uint8_t round_key[240], // Round key
iv[16]; // Initialization vector
} _pdfio_aes_t;
typedef struct _pdfio_md5_s // MD5 hash state
{
uint32_t count[2]; // Message length in bits, lsw first
uint32_t abcd[4]; // Digest buffer
uint8_t buf[64]; // Accumulate block
} _pdfio_md5_t;
typedef struct _pdfio_rc4_s // RC4 encryption state
{
uint8_t sbox[256]; // S boxes for encryption
uint8_t i, j; // Current indices into S boxes
} _pdfio_rc4_t;
Add SHA-256 code from RFC 6234.
2021-10-09 00:55:25 +02:00
typedef struct _pdfio_sha265_s // SHA-256 hash state
{
uint32_t Intermediate_Hash[8]; // Message Digest
uint32_t Length_High; // Message length in bits
uint32_t Length_Low; // Message length in bits
int Message_Block_Index; // Message_Block array index
uint8_t Message_Block[64]; // 512-bit message blocks
int Computed; // Is the hash computed?
int Corrupted; // Cumulative corruption code
} _pdfio_sha256_t;
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
Move AES code to separate file, prep private API for making keys/contexts/callbacks for encryption.
2021-10-09 16:49:22 +02:00
typedef union _pdfio_crypto_ctx_u // Cryptographic contexts
{
_pdfio_aes_t aes; // AES-128/256 context
_pdfio_rc4_t rc4; // RC4-40/128 context
} _pdfio_crypto_ctx_t;
Update crypto callback to return the number of output bytes (to account for AES expansion).
2021-10-23 20:37:25 +02:00
typedef size_t (*_pdfio_crypto_cb_t)(_pdfio_crypto_ctx_t *ctx, uint8_t *outbuffer, const uint8_t *inbuffer, size_t len);
Move AES code to separate file, prep private API for making keys/contexts/callbacks for encryption.
2021-10-09 16:49:22 +02:00
Save work.
2021-04-28 03:22:34 +02:00
struct _pdfio_array_s
{
pdfio_file_t *pdf; // PDF file
size_t num_values, // Number of values in use
alloc_values; // Number of allocated values
_pdfio_value_t *values; // Array of values
};
typedef struct _pdfio_pair_s // Key/value pair
{
const char *key; // Key string
_pdfio_value_t value; // Value
} _pdfio_pair_t;
Implement object mapping framework.
2021-05-13 19:18:01 +02:00
struct _pdfio_dict_s // Dictionary
Save work.
2021-04-28 03:22:34 +02:00
{
pdfio_file_t *pdf; // PDF file
size_t num_pairs, // Number of pairs in use
alloc_pairs; // Number of allocated pairs
_pdfio_pair_t *pairs; // Array of pairs
};
Implement object mapping framework.
2021-05-13 19:18:01 +02:00
typedef struct _pdfio_objmap_s // PDF object map
{
pdfio_obj_t *obj; // Object for this file
pdfio_file_t *src_pdf; // Source PDF file
size_t src_number; // Source object number
} _pdfio_objmap_t;
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
struct _pdfio_file_s // PDF file structure
{
Save work.
2021-04-28 03:22:34 +02:00
char *filename; // Filename
char *version; // Version number
Initial writing support.
2021-05-16 17:39:05 +02:00
pdfio_rect_t media_box, // Default MediaBox value
crop_box; // Default CropBox value
Save work.
2021-04-28 03:22:34 +02:00
_pdfio_mode_t mode; // Read/write mode
Add pdfioFileCreateOutput API (Issue #21)
2021-09-27 13:41:50 +02:00
pdfio_output_cb_t output_cb; // Output callback
void *output_ctx; // Context for output callback
pdfio_dict_t implementation
2021-04-26 02:12:52 +02:00
pdfio_error_cb_t error_cb; // Error callback
void *error_data; // Data for error callback
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
Implement pdfioFileSetPermissions.
2021-10-11 04:27:09 +02:00
pdfio_encryption_t encryption; // Encryption mode
pdfio_permission_t permissions; // Access permissions (encrypted PDF files)
Save work on encrypted PDF reading.
2021-10-26 01:36:39 +02:00
uint8_t file_key[16], // File encryption key
Implement pdfioFileSetPermissions.
2021-10-11 04:27:09 +02:00
owner_key[32], // Owner encryption key
Support per-object file IDs (Issue #42)
2023-12-13 03:48:58 +01:00
user_key[32], // User encryption key
password[32]; // Padded password
Save work on encrypted PDF reading.
2021-10-26 01:36:39 +02:00
size_t file_keylen, // Length of file encryption key
Do some reorganization and start the implementation of decryption.
2021-10-24 16:59:25 +02:00
owner_keylen, // Length of owner encryption key
user_keylen; // Length of user encryption key
Implement pdfioFileSetPermissions.
2021-10-11 04:27:09 +02:00
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
// Active file data
int fd; // File descriptor
char buffer[8192], // Read/write buffer
*bufptr, // Pointer into buffer
*bufend; // End of buffer
off_t bufpos; // Position in file for start of buffer
New member names to specify type of value.
2021-10-11 04:40:42 +02:00
pdfio_dict_t *trailer_dict; // Trailer dictionary
pdfio_obj_t *root_obj; // Root object/dictionary
pdfio_obj_t *info_obj; // Information object
pdfio_obj_t *pages_obj; // Root pages object
Implement pdfioFileSetPermissions.
2021-10-11 04:27:09 +02:00
pdfio_obj_t *encrypt_obj; // De/Encryption object/dictionary
Test both CP1252 and Unicode font rendering. Add cached CP1252 and Unicode font encoding objects. Clean up CP1252 ("simple") TrueType font embedding.
2021-06-24 03:26:01 +02:00
pdfio_obj_t *cp1252_obj, // CP1252 font encoding object
*unicode_obj; // Unicode font encoding object
Load trailer dict.
2021-05-04 18:24:29 +02:00
pdfio_array_t *id_array; // ID array
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
// Allocated data elements
size_t num_arrays, // Number of arrays
alloc_arrays; // Allocated arrays
pdfio_array_t **arrays; // Arrays
size_t num_dicts, // Number of dictionaries
alloc_dicts; // Allocated dictionaries
pdfio_dict_t **dicts; // Dictionaries
size_t num_objs, // Number of objects
Increase the maximum number of object streams in a file (Issue #58) - most files only contain 1 or 2... Change the implementation of add/find object to use a custom binary insertion sort algorithm rather than doing a qsort after every addition. This results in a significant improvement in open speed - from 2371 seconds (about 39.5 minutes) to 3.1 seconds for one large test file (an ESRI standard).
2023-12-13 18:21:59 +01:00
alloc_objs, // Allocated objects
last_obj; // Last object added
Add some protection against opening multiple streams in the same file at the same time.
2022-07-04 19:03:11 +02:00
pdfio_obj_t **objs, // Objects
*current_obj; // Current object being written/read
Implement object mapping framework.
2021-05-13 19:18:01 +02:00
size_t num_objmaps, // Number of object maps
alloc_objmaps; // Allocated object maps
_pdfio_objmap_t *objmaps; // Object maps
Save work.
2021-04-28 03:22:34 +02:00
size_t num_pages, // Number of pages
alloc_pages; // Allocated pages
pdfio_obj_t **pages; // Pages
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
size_t num_strings, // Number of strings
alloc_strings; // Allocated strings
char **strings; // Nul-terminated strings
};
struct _pdfio_obj_s // Object
{
pdfio_dict_t implementation
2021-04-26 02:12:52 +02:00
pdfio_file_t *pdf; // PDF file
Implement pdfioFileOpen reading of xref table.
2021-05-01 17:41:43 +02:00
size_t number; // Number
unsigned short generation; // Generation
Save work.
2021-04-28 03:22:34 +02:00
off_t offset, // Offset to object in file
length_offset, // Offset to /Length in object dict
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
stream_offset; // Offset to start of stream in file
size_t stream_length; // Length of stream, if any
Implement object loading.
2021-05-04 23:04:09 +02:00
_pdfio_value_t value; // Dictionary/number/etc. value
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
pdfio_stream_t *stream; // Open stream, if any
Save work on string measure support (Issue #17)
2023-12-05 00:54:33 +01:00
void *data; // Extension data, if any
Finish implementation of pdfioContentTextMeasure (Issue #17)
2023-12-05 03:20:51 +01:00
_pdfio_extfree_t datafree; // Free callback for extension data
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
};
struct _pdfio_stream_s // Stream
{
pdfio_file_t *pdf; // PDF file
pdfio_obj_t *obj; // Object
Add pdfioFileCreateOutput API (Issue #21)
2021-09-27 13:41:50 +02:00
pdfio_obj_t *length_obj; // Length object, if any
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
pdfio_filter_t filter; // Compression/decompression filter
Safe work on streams - still need to implement predictors
2021-05-07 14:47:49 +02:00
size_t remaining; // Remaining bytes in stream
char buffer[8192], // Read/write buffer
*bufptr, // Current position in buffer
*bufend; // End of buffer
Save work on low-level file I/O.
2021-04-26 16:42:01 +02:00
z_stream flate; // Flate filter state
Save work on PNG predictors. Apparently object streams are also a common thing for newer Acrobat so we'll need to implement those, too.
2021-05-08 01:51:38 +02:00
_pdfio_predictor_t predictor; // Predictor function, if any
size_t pbpixel, // Size of a pixel in bytes
pbsize; // Predictor buffer size, if any
unsigned char cbuffer[4096], // Compressed data buffer
Fix PNG predictor 12, 14, and 15.
2021-06-01 23:48:36 +02:00
*prbuffer, // Raw buffer (previous line), as needed
*psbuffer; // PNG filter buffer, as needed
Move AES code to separate file, prep private API for making keys/contexts/callbacks for encryption.
2021-10-09 16:49:22 +02:00
_pdfio_crypto_cb_t crypto_cb; // Encryption/descryption callback, if any
_pdfio_crypto_ctx_t crypto_ctx; // Cryptographic context
Add string array, have the pdfio_file_t object manage allocated data.
2021-04-25 17:28:56 +02:00
};
Initial checkin.
2021-04-10 14:00:52 +02:00
//
// Functions...
//
Defer object/value decryption to after the object is loaded (Issue #42)
2023-12-14 22:02:26 +01:00
extern bool _pdfioArrayDecrypt(pdfio_file_t *pdf, pdfio_obj_t *obj, pdfio_array_t *a, size_t depth) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern void _pdfioArrayDebug(pdfio_array_t *a, FILE *fp) _PDFIO_INTERNAL;
extern void _pdfioArrayDelete(pdfio_array_t *a) _PDFIO_INTERNAL;
extern _pdfio_value_t *_pdfioArrayGetValue(pdfio_array_t *a, size_t n) _PDFIO_INTERNAL;
Fix a few stack/buffer overflow bugs discovered by Bart, Steffan, and Mark from the Radboud University NL (thanks!) - Add depth argument to all value read functions that recurse - Add depth argument to page tree loading code - Validate xref stream sizes individually to avoid out-of-bounds access to local xref buffer.
2021-11-29 23:46:56 +01:00
extern pdfio_array_t *_pdfioArrayRead(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_token_t *ts, size_t depth) _PDFIO_INTERNAL;
Need object when reading/writing encrypted PDFs (to decrypt/encrypt strings), RC4 writing is now working, AES-128 needs work, AES-256 hasn't been done yet.
2021-10-24 00:08:16 +02:00
extern bool _pdfioArrayWrite(pdfio_array_t *a, pdfio_obj_t *obj) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
extern void _pdfioCryptoAESInit(_pdfio_aes_t *ctx, const uint8_t *key, size_t keylen, const uint8_t *iv) _PDFIO_INTERNAL;
Update crypto callback to return the number of output bytes (to account for AES expansion).
2021-10-23 20:37:25 +02:00
extern size_t _pdfioCryptoAESDecrypt(_pdfio_aes_t *ctx, uint8_t *outbuffer, const uint8_t *inbuffer, size_t len) _PDFIO_INTERNAL;
extern size_t _pdfioCryptoAESEncrypt(_pdfio_aes_t *ctx, uint8_t *outbuffer, const uint8_t *inbuffer, size_t len) _PDFIO_INTERNAL;
Do some reorganization and start the implementation of decryption.
2021-10-24 16:59:25 +02:00
extern bool _pdfioCryptoLock(pdfio_file_t *pdf, pdfio_permission_t permissions, pdfio_encryption_t encryption, const char *owner_password, const char *user_password) _PDFIO_INTERNAL;
Add random number generation support.
2021-10-10 05:05:39 +02:00
extern void _pdfioCryptoMakeRandom(uint8_t *buffer, size_t bytes) _PDFIO_INTERNAL;
Move AES code to separate file, prep private API for making keys/contexts/callbacks for encryption.
2021-10-09 16:49:22 +02:00
extern _pdfio_crypto_cb_t _pdfioCryptoMakeReader(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_crypto_ctx_t *ctx, uint8_t *iv, size_t *ivlen) _PDFIO_INTERNAL;
extern _pdfio_crypto_cb_t _pdfioCryptoMakeWriter(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_crypto_ctx_t *ctx, uint8_t *iv, size_t *ivlen) _PDFIO_INTERNAL;
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
extern void _pdfioCryptoMD5Append(_pdfio_md5_t *pms, const uint8_t *data, size_t nbytes) _PDFIO_INTERNAL;
extern void _pdfioCryptoMD5Finish(_pdfio_md5_t *pms, uint8_t digest[16]) _PDFIO_INTERNAL;
extern void _pdfioCryptoMD5Init(_pdfio_md5_t *pms) _PDFIO_INTERNAL;
extern void _pdfioCryptoRC4Init(_pdfio_rc4_t *ctx, const uint8_t *key, size_t keylen) _PDFIO_INTERNAL;
Update crypto callback to return the number of output bytes (to account for AES expansion).
2021-10-23 20:37:25 +02:00
extern size_t _pdfioCryptoRC4Crypt(_pdfio_rc4_t *ctx, uint8_t *outbuffer, const uint8_t *inbuffer, size_t len) _PDFIO_INTERNAL;
Add SHA-256 code from RFC 6234.
2021-10-09 00:55:25 +02:00
extern void _pdfioCryptoSHA256Append(_pdfio_sha256_t *, const uint8_t *bytes, size_t bytecount) _PDFIO_INTERNAL;
extern void _pdfioCryptoSHA256Init(_pdfio_sha256_t *ctx) _PDFIO_INTERNAL;
extern void _pdfioCryptoSHA256Finish(_pdfio_sha256_t *ctx, uint8_t *Message_Digest) _PDFIO_INTERNAL;
Do some reorganization and start the implementation of decryption.
2021-10-24 16:59:25 +02:00
extern bool _pdfioCryptoUnlock(pdfio_file_t *pdf, pdfio_password_cb_t password_cb, void *password_data) _PDFIO_INTERNAL;
Save work on AES and RC4.
2021-10-05 03:13:01 +02:00
Fix up copying objects from unencrypted to AES-encrypted documents (still looks like there are some issues with strings in dicts)
2021-10-26 03:22:59 +02:00
extern void _pdfioDictClear(pdfio_dict_t *dict, const char *key) _PDFIO_INTERNAL;
Defer object/value decryption to after the object is loaded (Issue #42)
2023-12-14 22:02:26 +01:00
extern bool _pdfioDictDecrypt(pdfio_file_t *pdf, pdfio_obj_t *obj, pdfio_dict_t *dict, size_t depth) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern void _pdfioDictDebug(pdfio_dict_t *dict, FILE *fp) _PDFIO_INTERNAL;
extern void _pdfioDictDelete(pdfio_dict_t *dict) _PDFIO_INTERNAL;
extern _pdfio_value_t *_pdfioDictGetValue(pdfio_dict_t *dict, const char *key) _PDFIO_INTERNAL;
Fix a few stack/buffer overflow bugs discovered by Bart, Steffan, and Mark from the Radboud University NL (thanks!) - Add depth argument to all value read functions that recurse - Add depth argument to page tree loading code - Validate xref stream sizes individually to avoid out-of-bounds access to local xref buffer.
2021-11-29 23:46:56 +01:00
extern pdfio_dict_t *_pdfioDictRead(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_token_t *ts, size_t depth) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern bool _pdfioDictSetValue(pdfio_dict_t *dict, const char *key, _pdfio_value_t *value) _PDFIO_INTERNAL;
Need object when reading/writing encrypted PDFs (to decrypt/encrypt strings), RC4 writing is now working, AES-128 needs work, AES-256 hasn't been done yet.
2021-10-24 00:08:16 +02:00
extern bool _pdfioDictWrite(pdfio_dict_t *dict, pdfio_obj_t *obj, off_t *length) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern bool _pdfioFileAddMappedObj(pdfio_file_t *pdf, pdfio_obj_t *dst_obj, pdfio_obj_t *src_obj) _PDFIO_INTERNAL;
extern bool _pdfioFileAddPage(pdfio_file_t *pdf, pdfio_obj_t *obj) _PDFIO_INTERNAL;
extern bool _pdfioFileConsume(pdfio_file_t *pdf, size_t bytes) _PDFIO_INTERNAL;
extern pdfio_obj_t *_pdfioFileCreateObj(pdfio_file_t *pdf, pdfio_file_t *srcpdf, _pdfio_value_t *value) _PDFIO_INTERNAL;
extern bool _pdfioFileDefaultError(pdfio_file_t *pdf, const char *message, void *data) _PDFIO_INTERNAL;
extern bool _pdfioFileError(pdfio_file_t *pdf, const char *format, ...) _PDFIO_FORMAT(2,3) _PDFIO_INTERNAL;
extern pdfio_obj_t *_pdfioFileFindMappedObj(pdfio_file_t *pdf, pdfio_file_t *src_pdf, size_t src_number) _PDFIO_INTERNAL;
extern bool _pdfioFileFlush(pdfio_file_t *pdf) _PDFIO_INTERNAL;
extern int _pdfioFileGetChar(pdfio_file_t *pdf) _PDFIO_INTERNAL;
extern bool _pdfioFileGets(pdfio_file_t *pdf, char *buffer, size_t bufsize) _PDFIO_INTERNAL;
extern ssize_t _pdfioFilePeek(pdfio_file_t *pdf, void *buffer, size_t bytes) _PDFIO_INTERNAL;
extern bool _pdfioFilePrintf(pdfio_file_t *pdf, const char *format, ...) _PDFIO_FORMAT(2,3) _PDFIO_INTERNAL;
extern bool _pdfioFilePuts(pdfio_file_t *pdf, const char *s) _PDFIO_INTERNAL;
extern ssize_t _pdfioFileRead(pdfio_file_t *pdf, void *buffer, size_t bytes) _PDFIO_INTERNAL;
extern off_t _pdfioFileSeek(pdfio_file_t *pdf, off_t offset, int whence) _PDFIO_INTERNAL;
extern off_t _pdfioFileTell(pdfio_file_t *pdf) _PDFIO_INTERNAL;
extern bool _pdfioFileWrite(pdfio_file_t *pdf, const void *buffer, size_t bytes) _PDFIO_INTERNAL;
extern void _pdfioObjDelete(pdfio_obj_t *obj) _PDFIO_INTERNAL;
Save work on string measure support (Issue #17)
2023-12-05 00:54:33 +01:00
extern void *_pdfioObjGetExtension(pdfio_obj_t *obj) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern bool _pdfioObjLoad(pdfio_obj_t *obj) _PDFIO_INTERNAL;
Finish implementation of pdfioContentTextMeasure (Issue #17)
2023-12-05 03:20:51 +01:00
extern void _pdfioObjSetExtension(pdfio_obj_t *obj, void *data, _pdfio_extfree_t datafree) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
Add pdfioFileCreateOutput API (Issue #21)
2021-09-27 13:41:50 +02:00
extern pdfio_stream_t *_pdfioStreamCreate(pdfio_obj_t *obj, pdfio_obj_t *length_obj, pdfio_filter_t compression) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern pdfio_stream_t *_pdfioStreamOpen(pdfio_obj_t *obj, bool decode) _PDFIO_INTERNAL;
extern bool _pdfioStringIsAllocated(pdfio_file_t *pdf, const char *s) _PDFIO_INTERNAL;
extern void _pdfioTokenClear(_pdfio_token_t *tb) _PDFIO_INTERNAL;
extern void _pdfioTokenFlush(_pdfio_token_t *tb) _PDFIO_INTERNAL;
extern bool _pdfioTokenGet(_pdfio_token_t *tb, char *buffer, size_t bufsize) _PDFIO_INTERNAL;
Increase token buffer to 256 bytes, to avoid overflow scenarios. Add _pdfioTokenFlush to manually flush the token buffer. Add more debug printfs.
2021-05-10 23:37:57 +02:00
extern void _pdfioTokenInit(_pdfio_token_t *tb, pdfio_file_t *pdf, _pdfio_tconsume_cb_t consume_cb, _pdfio_tpeek_cb_t peek_cb, void *cb_data);
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern void _pdfioTokenPush(_pdfio_token_t *tb, const char *token) _PDFIO_INTERNAL;
Increase token buffer to 256 bytes, to avoid overflow scenarios. Add _pdfioTokenFlush to manually flush the token buffer. Add more debug printfs.
2021-05-10 23:37:57 +02:00
extern bool _pdfioTokenRead(_pdfio_token_t *tb, char *buffer, size_t bufsize);
Work on token parsing.
2021-05-01 23:50:52 +02:00
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern _pdfio_value_t *_pdfioValueCopy(pdfio_file_t *pdfdst, _pdfio_value_t *vdst, pdfio_file_t *pdfsrc, _pdfio_value_t *vsrc) _PDFIO_INTERNAL;
Defer object/value decryption to after the object is loaded (Issue #42)
2023-12-14 22:02:26 +01:00
extern bool _pdfioValueDecrypt(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_value_t *v, size_t depth) _PDFIO_INTERNAL;
Save work on documentation.
2021-07-24 19:08:46 +02:00
extern void _pdfioValueDebug(_pdfio_value_t *v, FILE *fp) _PDFIO_INTERNAL;
extern void _pdfioValueDelete(_pdfio_value_t *v) _PDFIO_INTERNAL;
Fix a few stack/buffer overflow bugs discovered by Bart, Steffan, and Mark from the Radboud University NL (thanks!) - Add depth argument to all value read functions that recurse - Add depth argument to page tree loading code - Validate xref stream sizes individually to avoid out-of-bounds access to local xref buffer.
2021-11-29 23:46:56 +01:00
extern _pdfio_value_t *_pdfioValueRead(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_token_t *ts, _pdfio_value_t *v, size_t depth) _PDFIO_INTERNAL;
Need object when reading/writing encrypted PDFs (to decrypt/encrypt strings), RC4 writing is now working, AES-128 needs work, AES-256 hasn't been done yet.
2021-10-24 00:08:16 +02:00
extern bool _pdfioValueWrite(pdfio_file_t *pdf, pdfio_obj_t *obj, _pdfio_value_t *v, off_t *length) _PDFIO_INTERNAL;
Initial checkin.
2021-04-10 14:00:52 +02:00
Clean up private header.
2023-10-02 11:27:40 +02:00
Initial checkin.
2021-04-10 14:00:52 +02:00
#endif // !PDFIO_PRIVATE_H
Reference in New Issue Copy Permalink