index.c File Reference

Functions to parse index records. More...

#include <string.h>
#include <stdlib.h>
#include <stdint.h>
#include "index.h"
#include "util.h"
#include "memory.h"
#include "debug.h"
#include "buffer.h"

Macros
#define	_GNU_SOURCE   1
#define	__USE_BSD   /* for strdup on linux/glibc */

Functions
size_t	mobi_indx_get_label (unsigned char *output, MOBIBuffer *buf, const size_t length, const size_t has_ligatures)
	Read index entry label from buffer pointing at index record data. More...
size_t	mobi_ordt_getbuffer (const MOBIOrdt *ordt, MOBIBuffer *buf, uint16_t *offset)
	Get encoded character from dictionary index The characters are offsets into ORDT table. More...
uint16_t	mobi_ordt_lookup (const MOBIOrdt *ordt, const uint16_t offset)
	Fetch UTF-16 value from ORDT2 table. More...
size_t	mobi_getstring_ordt (const MOBIOrdt *ordt, MOBIBuffer *buf, unsigned char *output, size_t length)
	Get UTF-8 string from buffer, decoded by lookups in ORDT2 table. More...
MOBI_RET	mobi_parse_indx (const MOBIPdbRecord *indx_record, MOBIIndx *indx, MOBITagx *tagx, MOBIOrdt *ordt)
	Parser of INDX record. More...
MOBI_RET	mobi_parse_index (const MOBIData *m, MOBIIndx *indx, const size_t indx_record_number)
	Parser of a set of index records. More...
MOBI_RET	mobi_get_indxentry_tagvalue (uint32_t *tagvalue, const MOBIIndexEntry *entry, const unsigned tag_arr[])
	Get a value of tag[tagid][tagindex] for given index entry. More...
size_t	mobi_get_indxentry_tagarray (uint32_t **tagarr, const MOBIIndexEntry *entry, const size_t tagid)
	Get array of tagvalues of tag[tagid] for given index entry. More...
uint32_t	mobi_get_orth_entry_offset (const MOBIIndexEntry *entry)
	Get entry start offset for the orth entry. More...
uint32_t	mobi_get_orth_entry_length (const MOBIIndexEntry *entry)
	Get text length for the orth entry. More...
bool	mobi_indx_has_tag (const MOBIIndx *indx, const size_t tagid)
	Check if given tagid is present in the index. More...
char *	mobi_get_cncx_string (const MOBIPdbRecord *cncx_record, const uint32_t cncx_offset)
	Get compiled index entry string. More...
char *	mobi_get_cncx_string_utf8 (const MOBIPdbRecord *cncx_record, const uint32_t cncx_offset, MOBIEncoding cncx_encoding)
	Get compiled index entry string, converted to utf8 encoding. More...
char *	mobi_get_cncx_string_flat (const MOBIPdbRecord *cncx_record, const uint32_t cncx_offset, const size_t length)
	Get flat index entry string. More...
MOBI_RET	mobi_decode_infl (unsigned char *decoded, int *decoded_size, const unsigned char *rule)
	Decode compiled infl index entry. More...
size_t	mobi_trie_get_inflgroups (char **infl_strings, MOBITrie *const root, const char *string)
	Get all matches for given string from trie structure. More...
MOBI_RET	mobi_trie_insert_infl (MOBITrie **root, const MOBIIndx *indx, size_t i)
	Insert inversed inlection string for given entry into trie structure. More...

Detailed Description

Functions to parse index records.

Copyright (c) 2020 Bartek Fabiszewski http://www.fabiszewski.net

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

Function Documentation

mobi_decode_infl()

MOBI_RET mobi_decode_infl	(	unsigned char *	decoded,
		int *	decoded_size,
		const unsigned char *	rule
	)

Decode compiled infl index entry.

Buffer decoded must be initialized with basic index entry. Basic index entry will be transformed into inflected form, based on compiled rule. Min. size of input buffer (decoded) must be INDX_INFLBUF_SIZEMAX + 1

Parameters

[in,out]	decoded	Decoded entry string
[in,out]	decoded_size	Decoded entry size
[in]	rule	Compiled rule

Returns

MOBI_RET status code (on success MOBI_SUCCESS)

mobi_get_cncx_string()

char* mobi_get_cncx_string	(	const MOBIPdbRecord *	cncx_record,
		const uint32_t	cncx_offset
	)

Get compiled index entry string.

Allocates memory for the string. Must be freed by caller.

Parameters

[in]	cncx_record	MOBIPdbRecord structure with cncx record
[in]	cncx_offset	Offset of string entry from the beginning of the record

Returns

Entry string or null if malloc failed

mobi_get_cncx_string_flat()

char* mobi_get_cncx_string_flat	(	const MOBIPdbRecord *	cncx_record,
		const uint32_t	cncx_offset,
		const size_t	length
	)

Get flat index entry string.

Allocates memory for the string. Must be freed by caller.

Parameters

[in]	cncx_record	MOBIPdbRecord structure with cncx record
[in]	cncx_offset	Offset of string entry from the beginning of the record
[in]	length	Length of the string to be extracted

Returns

Entry string

mobi_get_cncx_string_utf8()

char* mobi_get_cncx_string_utf8	(	const MOBIPdbRecord *	cncx_record,
		const uint32_t	cncx_offset,
		MOBIEncoding	cncx_encoding
	)

Get compiled index entry string, converted to utf8 encoding.

Allocates memory for the string. Must be freed by caller.

Parameters

[in]	cncx_record	MOBIPdbRecord structure with cncx record
[in]	cncx_offset	Offset of string entry from the beginning of the record
[in]	cncx_encoding	Encoding

Returns

Entry string or null if malloc failed

mobi_get_indxentry_tagarray()

size_t mobi_get_indxentry_tagarray	(	uint32_t **	tagarr,
		const MOBIIndexEntry *	entry,
		const size_t	tagid
	)

Get array of tagvalues of tag[tagid] for given index entry.

Parameters

[in,out]	tagarr	Pointer to tagvalues array
[in]	entry	Index entry to be search for the value
[in]	tagid	Id of the tag

Returns

Size of the array (zero on failure)

mobi_get_indxentry_tagvalue()

MOBI_RET mobi_get_indxentry_tagvalue	(	uint32_t *	tagvalue,
		const MOBIIndexEntry *	entry,
		const unsigned	tag_arr[]
	)

Get a value of tag[tagid][tagindex] for given index entry.

Parameters

[in,out]	tagvalue	Will be set to a tag value
[in]	entry	Index entry to be search for the value
[in]	tag_arr	Array: tag_arr[0] = tagid, tag_arr[1] = tagindex

Returns

MOBI_RET status code (on success MOBI_SUCCESS)

mobi_getstring_ordt()

size_t mobi_getstring_ordt	(	const MOBIOrdt *	ordt,
		MOBIBuffer *	buf,
		unsigned char *	output,
		size_t	length
	)

Get UTF-8 string from buffer, decoded by lookups in ORDT2 table.

Parameters

[in]	ordt	MOBIOrdt structure (ORDT data and metadata)
[in,out]	buf	MOBIBuffer structure with input string
[in,out]	output	Output buffer (INDX_LABEL_SIZEMAX + 1 bytes)
[in]	length	Length of input string contained in buf

Returns

Length of output string (without null terminator)

mobi_indx_get_label()

size_t mobi_indx_get_label	(	unsigned char *	output,
		MOBIBuffer *	buf,
		const size_t	length,
		const size_t	has_ligatures
	)

Read index entry label from buffer pointing at index record data.

Parameters

[in,out]	output	Output buffer (INDX_LABEL_SIZEMAX + 1 bytes)
[in,out]	buf	MOBIBuffer structure, offset pointing at index entry label
[in]	length	Number of bytes to be read
[in]	has_ligatures	Decode ligatures if true

Returns

Length of output string (without null terminator), on error buf->error set to MOBI_RET status

mobi_indx_has_tag()

bool mobi_indx_has_tag	(	const MOBIIndx *	indx,
		const size_t	tagid
	)

Check if given tagid is present in the index.

Parameters

[in]	indx	Index MOBIIndx structure
[in]	tagid	Id of the tag

Returns

True on success, false otherwise

mobi_ordt_getbuffer()

size_t mobi_ordt_getbuffer	(	const MOBIOrdt *	ordt,
		MOBIBuffer *	buf,
		uint16_t *	offset
	)

Get encoded character from dictionary index The characters are offsets into ORDT table.

Parameters

[in]	ordt	MOBIOrdt structure (ORDT data and metadata)
[in,out]	buf	MOBIBuffer structure with index data
[in,out]	offset	Value read from buffer

Returns

Number of bytes read (zero in case of error)

mobi_ordt_lookup()

uint16_t mobi_ordt_lookup	(	const MOBIOrdt *	ordt,
		const uint16_t	offset
	)

Fetch UTF-16 value from ORDT2 table.

Parameters

[in]	ordt	MOBIOrdt structure (ORDT data and metadata)
[in]	offset	Offset in ORDT2 table

Returns

UTF-16 code point

mobi_parse_index()

MOBI_RET mobi_parse_index	(	const MOBIData *	m,
		MOBIIndx *	indx,
		const size_t	indx_record_number
	)

Parser of a set of index records.

Parameters

[in]	m	MOBIData structure containing MOBI file metadata and data
[in,out]	indx	MOBIIndx structure to be filled with parsed entries
[in]	indx_record_number	Number of the first record of the set

Returns

MOBI_RET status code (on success MOBI_SUCCESS)

mobi_parse_indx()

MOBI_RET mobi_parse_indx	(	const MOBIPdbRecord *	indx_record,
		MOBIIndx *	indx,
		MOBITagx *	tagx,
		MOBIOrdt *	ordt
	)

Parser of INDX record.

Parameters

[in]	indx_record	MOBIPdbRecord structure with INDX record
[in,out]	indx	MOBIIndx structure to be filled with parsed entries
[in,out]	tagx	MOBITagx structure, will be filled with parsed TAGX section data if present in the INDX record, otherwise TAGX data will be used to parse the record
[in,out]	ordt	MOBIOrdt structure, will be filled with parsed ORDT sections

Returns

MOBI_RET status code (on success MOBI_SUCCESS)

mobi_trie_get_inflgroups()

size_t mobi_trie_get_inflgroups	(	char **	infl_strings,
		MOBITrie *const	root,
		const char *	string
	)

Get all matches for given string from trie structure.

Matches are made agains reversed string and all its substrings

Parameters

[in,out]	infl_strings	Array of returned strings
[in]	root	Root node of the tree
[in]	string	Index entry number

Returns

Number of returned strings

mobi_trie_insert_infl()

MOBI_RET mobi_trie_insert_infl	(	MOBITrie **	root,
		const MOBIIndx *	indx,
		size_t	i
	)

Insert inversed inlection string for given entry into trie structure.

Parameters

[in,out]	root	Root node of the tree, created if NULL
[in]	indx	MOBIIndx infl index records
[in]	i	Index entry number

Returns

MOBI_RET status code (on success MOBI_SUCCESS)