#include "config.h"
#include "mobi.h"
#include "memory.h"
#include "buffer.h"
#include "read.h"
#include "compression.h"
#include <zlib.h>

Macros
#define	strdup mobi_strdup
	strdup replacement

#define	m_uncompress uncompress

#define	m_crc32 crc32

#define	M_OK Z_OK

#define	UNUSED(x) (void)(x)

#define	AUDI_MAGIC "AUDI"
	Magic numbers of records.

#define	CDIC_MAGIC "CDIC"

#define	CMET_MAGIC "CMET"

#define	EXTH_MAGIC "EXTH"

#define	FDST_MAGIC "FDST"

#define	FONT_MAGIC "FONT"

#define	HUFF_MAGIC "HUFF"

#define	IDXT_MAGIC "IDXT"

#define	INDX_MAGIC "INDX"

#define	LIGT_MAGIC "LIGT"

#define	MOBI_MAGIC "MOBI"

#define	ORDT_MAGIC "ORDT"

#define	RESC_MAGIC "RESC"

#define	SRCS_MAGIC "SRCS"

#define	TAGX_MAGIC "TAGX"

#define	VIDE_MAGIC "VIDE"

#define	BOUNDARY_MAGIC "BOUNDARY"

#define	EOF_MAGIC "\xe9\x8e\r\n"

#define	REPLICA_MAGIC "%MOP"

#define	EPOCH_MAC_DIFF 2082844800UL
	Difference in seconds between epoch time and mac time.

#define	PALMDB_HEADER_LEN 78

#define	PALMDB_NAME_SIZE_MAX 32

#define	PALMDB_RECORD_INFO_SIZE 8

#define	PALMDB_ATTRIBUTE_DEFAULT 0

#define	PALMDB_VERSION_DEFAULT 0

#define	PALMDB_MODNUM_DEFAULT 0

#define	PALMDB_APPINFO_DEFAULT 0

#define	PALMDB_SORTINFO_DEFAULT 0

#define	PALMDB_TYPE_DEFAULT "BOOK"

#define	PALMDB_CREATOR_DEFAULT "MOBI"

#define	PALMDB_NEXTREC_DEFAULT 0

#define	RECORD0_HEADER_LEN 16

#define	RECORD0_TEXT_SIZE_MAX 4096

#define	RECORD0_FULLNAME_SIZE_MAX 1024

#define	CDIC_HEADER_LEN 16

#define	CDIC_RECORD_MAXCNT 1024

#define	HUFF_CODELEN_MAX 16

#define	HUFF_HEADER_LEN 24

#define	HUFF_RECORD_MAXCNT 1024

#define	HUFF_RECORD_MINSIZE 2584

#define	FONT_HEADER_LEN 24

#define	MEDIA_HEADER_LEN 12

#define	FONT_SIZEMAX (50 * 1024 * 1024)

#define	RAWTEXT_SIZEMAX 0xfffffff

#define	MOBI_HEADER_V2_SIZE 0x18

#define	MOBI_HEADER_V3_SIZE 0x74

#define	MOBI_HEADER_V4_SIZE 0xd0

#define	MOBI_HEADER_V5_SIZE 0xe4

#define	MOBI_HEADER_V6_SIZE 0xe4

#define	MOBI_HEADER_V6_EXT_SIZE 0xe8

#define	MOBI_HEADER_V7_SIZE 0xe4

#define	max(a, b) ((a) > (b) ? (a) : (b))

#define	min(a, b) ((a) < (b) ? (a) : (b))

#define	ARRAYSIZE(arr) (sizeof(arr) / sizeof(arr[0]))

#define	MOBI_TITLE_SIZEMAX 1024

Functions
int	mobi_bitcount (const uint8_t byte)
	Get number of bits set in a given byte. More...

MOBI_RET	mobi_delete_record_by_seqnumber (MOBIData *m, const size_t num)
	Delete palm database record with given sequential number from MOBIData structure. More...

MOBI_RET	mobi_swap_mobidata (MOBIData *m)
	Swap KF7 and KF8 MOBIData structures in a hybrid file. More...

char *	mobi_strdup (const char *s)
	strdup replacement More...

bool	mobi_is_cp1252 (const MOBIData *m)
	Check if document's text is cp1252 encoded. More...

bool	mobi_has_drmkey (const MOBIData *m)
	Check if DRM key is set for the document. More...

bool	mobi_has_drmcookies (const MOBIData *m)
	Check if DRM cookies are set for the document. More...

MOBI_RET	mobi_cp1252_to_utf8 (char output, const char input, size_t *outsize, const size_t insize)
	Convert cp1252 encoded string to utf-8. More...

MOBI_RET	mobi_utf8_to_cp1252 (char output, const char input, size_t *outsize, const size_t insize)
	Convert utf-8 encoded string to cp1252. More...

uint8_t	mobi_ligature_to_cp1252 (const uint8_t byte1, const uint8_t byte2)
	Decode ligature to cp1252. More...

uint16_t	mobi_ligature_to_utf16 (const uint32_t byte1, const uint32_t byte2)
	Decode ligature to utf-16. More...

MOBIFiletype	mobi_determine_resource_type (const MOBIPdbRecord *record)
	Get resource type (image, font) by checking its magic header. More...

MOBIFiletype	mobi_determine_flowpart_type (const MOBIRawml *rawml, const size_t part_number)
	Get file type of given part with number [part_number]. More...

MOBI_RET	mobi_base32_decode (uint32_t decoded, const char encoded)
	Decode positive number from base 32 to base 10. More...

MOBIFiletype	mobi_get_resourcetype_by_uid (const MOBIRawml *rawml, const size_t uid)
	Get MOBIFiletype type of MOBIPart resource record with given unique id. More...

uint32_t	mobi_get_exthsize (const MOBIData *m)
	Get size of serialized exth record including padding. More...

uint32_t	mobi_get_drmsize (const MOBIData *m)
	Get size of serialized DRM data. More...

uint16_t	mobi_get_records_count (const MOBIData *m)
	Get count of palm database records. More...

void	mobi_remove_zeros (unsigned char buffer, size_t len)
	Remove null characters from char buffer. More...

MOBI_RET	mobi_add_audio_resource (MOBIPart *part)
	Replace part data with decoded audio data. More...

MOBI_RET	mobi_add_video_resource (MOBIPart *part)
	Replace part data with decoded video data. More...

MOBI_RET	mobi_add_font_resource (MOBIPart *part)
	Replace part data with decoded font data. More...

MOBI_RET	mobi_set_fullname (MOBIData m, const char fullname)
	Set ebook full name stored in Record 0 at offset given in MOBI header. More...

MOBI_RET	mobi_set_pdbname (MOBIData m, const char name)
	Set palm database name. More...

void	mobi_free_internals (MOBIData *m)
	Free internals. More...

uint32_t	mobi_get32be (const unsigned char buf[4])
	Convert char buffer to 32-bit unsigned integer big endian. More...

uint32_t	mobi_get32le (const unsigned char buf[4])
	Convert char buffer to 32-bit unsigned integer little endian. More...

Detailed Description

This file is part of libmobi. Licensed under LGPL, either version 3, or any later. See http://www.gnu.org/licenses/

Function Documentation

◆ mobi_add_audio_resource()

MOBI_RET mobi_add_audio_resource ( MOBIPart * part )

Replace part data with decoded audio data.

Parameters

[in,out] part MOBIPart structure containing font resource, decoded part type will be set in the structure

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_add_font_resource()

MOBI_RET mobi_add_font_resource ( MOBIPart * part )

Replace part data with decoded font data.

Parameters

[in,out] part MOBIPart structure containing font resource, decoded part type will be set in the structure

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_add_video_resource()

MOBI_RET mobi_add_video_resource ( MOBIPart * part )

Replace part data with decoded video data.

Parameters

[in,out] part MOBIPart structure containing font resource, decoded part type will be set in the structure

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_base32_decode()

MOBI_RET mobi_base32_decode	(	uint32_t *	decoded,
		const char *	encoded
	)

Decode positive number from base 32 to base 10.

Base 32 characters must be upper case. Maximal supported value is VVVVVV.

Parameters

[in,out]	decoded	Base 10 output number
[in]	encoded	Base 32 input number

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_bitcount()

int mobi_bitcount ( const uint8_t byte )

Get number of bits set in a given byte.

Parameters

[in] byte A byte

Returns: Number of bits set

◆ mobi_cp1252_to_utf8()

MOBI_RET mobi_cp1252_to_utf8	(	char *	output,
		const char *	input,
		size_t *	outsize,
		const size_t	insize
	)

Convert cp1252 encoded string to utf-8.

Maximum length of output string is 3 * (input string length) + 1 Output string will be null terminated (even if truncated)

Parameters

[in,out]	output	Output string
[in,out]	input	Input string
[in,out]	outsize	Size of the allocated output buffer, will be set to output string length (without null terminator) on return
[in]	insize	Length of the input string.

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_delete_record_by_seqnumber()

MOBI_RET mobi_delete_record_by_seqnumber	(	MOBIData *	m,
		const size_t	num
	)

Delete palm database record with given sequential number from MOBIData structure.

Parameters

[in,out]	m	MOBIData structure with loaded data
[in]	num	Sequential number

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_determine_flowpart_type()

MOBIFiletype mobi_determine_flowpart_type	(	const MOBIRawml *	rawml,
		const size_t	part_number
	)

Get file type of given part with number [part_number].

Parameters

[in]	rawml	MOBIRawml parsed records structure
[in]	part_number	Sequential number of the part within rawml structure

Returns: MOBIFiletype file type

◆ mobi_determine_resource_type()

MOBIFiletype mobi_determine_resource_type ( const MOBIPdbRecord * record )

Get resource type (image, font) by checking its magic header.

Parameters

[in] record MOBIPdbRecord structure containing unknown record type

Returns: MOBIFiletype file type, T_UNKNOWN if not determined, T_BREAK if end of records mark found

◆ mobi_free_internals()

void mobi_free_internals ( MOBIData * m )

Free internals.

Parameters

[in,out] m MOBIData structure with raw data and metadata

◆ mobi_get32be()

uint32_t mobi_get32be ( const unsigned char buf[4] )

Convert char buffer to 32-bit unsigned integer big endian.

Parameters

[in] buf Input buffer

Returns: Converted value

◆ mobi_get32le()

uint32_t mobi_get32le ( const unsigned char buf[4] )

Convert char buffer to 32-bit unsigned integer little endian.

Parameters

[in] buf Input buffer

Returns: Converted value

◆ mobi_get_drmsize()

uint32_t mobi_get_drmsize ( const MOBIData * m )

Get size of serialized DRM data.

Parameters

[in] m MOBIData structure

Returns: Size of DRM data, zero on failure

◆ mobi_get_exthsize()

uint32_t mobi_get_exthsize ( const MOBIData * m )

Get size of serialized exth record including padding.

Parameters

[in] m MOBIData structure

Returns: Size of exth record, zero on failure

◆ mobi_get_records_count()

uint16_t mobi_get_records_count ( const MOBIData * m )

Get count of palm database records.

Parameters

[in] m MOBIData structure

Returns: Count of records, zero on failure

◆ mobi_get_resourcetype_by_uid()

MOBIFiletype mobi_get_resourcetype_by_uid	(	const MOBIRawml *	rawml,
		const size_t	uid
	)

Get MOBIFiletype type of MOBIPart resource record with given unique id.

Parameters

[in]	rawml	MOBIRawml structure with loaded data
[in]	uid	Unique id

Returns: Pointer to MOBIPart resource structure, NULL on failure

◆ mobi_has_drmcookies()

bool mobi_has_drmcookies ( const MOBIData * m )

Check if DRM cookies are set for the document.

Parameters

[in] m MOBIData structure with loaded Record(s) 0 headers

Returns: true or false

◆ mobi_has_drmkey()

bool mobi_has_drmkey ( const MOBIData * m )

Check if DRM key is set for the document.

Parameters

[in] m MOBIData structure with loaded Record(s) 0 headers

Returns: true or false

◆ mobi_is_cp1252()

bool mobi_is_cp1252 ( const MOBIData * m )

Check if document's text is cp1252 encoded.

Parameters

[in] m MOBIData structure holding document data and metadata

Returns: True or false

◆ mobi_ligature_to_cp1252()

uint8_t mobi_ligature_to_cp1252	(	const uint8_t	byte1,
		const uint8_t	byte2
	)

Decode ligature to cp1252.

Some latin ligatures are encoded in indices to facilitate search They are listed in LIGT header, but it seems every LIGT header contains same 5 ligatures (even if not all of them are used). So, instead of parsing header, we use static replacements. Invalid control characters are skipped

Parameters

[in]	byte1	First byte - control character
[in]	byte2	Second byte of the ligature

Returns: Ligature in cp1252 encoding, zero if not found

◆ mobi_ligature_to_utf16()

uint16_t mobi_ligature_to_utf16	(	const uint32_t	byte1,
		const uint32_t	byte2
	)

Decode ligature to utf-16.

Parameters

[in]	byte1	First byte - control character, should be <= 5
[in]	byte2	Second byte of the ligature

Returns: Ligature in utf-16 encoding, uni_replacement if not found

◆ mobi_remove_zeros()

void mobi_remove_zeros	(	unsigned char *	buffer,
		size_t *	len
	)

Remove null characters from char buffer.

Parameters

[in,out]	buffer	Character buffer
[in,out]	len	Size of buffer, will be updated with new length

◆ mobi_set_fullname()

MOBI_RET mobi_set_fullname	(	MOBIData *	m,
		const char *	fullname
	)

Set ebook full name stored in Record 0 at offset given in MOBI header.

Parameters

[in,out]	m	MOBIData structure with loaded data
[in]	fullname	Memory area to be filled with zero terminated full name string

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_set_pdbname()

MOBI_RET mobi_set_pdbname	(	MOBIData *	m,
		const char *	name
	)

Set palm database name.

Parameters

[in,out]	m	MOBIData structure with loaded data
[in]	name	Name

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_strdup()

char* mobi_strdup ( const char * s )

strdup replacement

Returned pointer must be freed by caller

Parameters

[in] s Input string

Returns: Duplicated string

◆ mobi_swap_mobidata()

MOBI_RET mobi_swap_mobidata ( MOBIData * m )

Swap KF7 and KF8 MOBIData structures in a hybrid file.

MOBIData structures form a circular linked list in case of hybrid files. By default KF8 structure is first one in the list. This function puts KF7 structure on the first place, so that it starts to be used by default.

Parameters

[in,out] m MOBIData structure

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

◆ mobi_utf8_to_cp1252()

MOBI_RET mobi_utf8_to_cp1252	(	char *	output,
		const char *	input,
		size_t *	outsize,
		const size_t	insize
	)

Convert utf-8 encoded string to cp1252.

Characters out of range will be replaced with substitute character

Parameters

[in,out]	output	Output string
[in,out]	input	Input string
[in,out]	outsize	Size of the allocated output buffer, will be set to output string length (without null terminator) on return
[in]	insize	Length of the input string.

Returns: MOBI_RET status code (on success MOBI_SUCCESS)

Macros

Functions

Detailed Description

Function Documentation

◆ mobi_add_audio_resource()

◆ mobi_add_font_resource()

◆ mobi_add_video_resource()

◆ mobi_base32_decode()

◆ mobi_bitcount()

◆ mobi_cp1252_to_utf8()

◆ mobi_delete_record_by_seqnumber()

◆ mobi_determine_flowpart_type()

◆ mobi_determine_resource_type()

◆ mobi_free_internals()

◆ mobi_get32be()

◆ mobi_get32le()

◆ mobi_get_drmsize()

◆ mobi_get_exthsize()

◆ mobi_get_records_count()

◆ mobi_get_resourcetype_by_uid()

◆ mobi_has_drmcookies()

◆ mobi_has_drmkey()

◆ mobi_is_cp1252()

◆ mobi_ligature_to_cp1252()

◆ mobi_ligature_to_utf16()

◆ mobi_remove_zeros()

◆ mobi_set_fullname()

◆ mobi_set_pdbname()

◆ mobi_strdup()

◆ mobi_swap_mobidata()

◆ mobi_utf8_to_cp1252()