index.h File Reference

#include "config.h"
#include "structure.h"
#include "mobi.h"

Go to the source code of this file.

Data Structures
struct	TAGXTags
	Tag entries in TAGX section (for internal INDX parsing) More...
struct	MOBITagx
	Parsed TAGX section (for internal INDX parsing) More...
struct	MOBIIdxt
	Parsed IDXT section (for internal INDX parsing) More...
struct	MOBIOrdt
	Parsed ORDT sections (for internal INDX parsing) More...

Macros
#define	INDX_TAG_GUIDE_TITLE_CNCX   (unsigned[]) {1, 0}
#define	INDX_TAG_NCX_FILEPOS   (unsigned[]) {1, 0}
#define	INDX_TAG_NCX_TEXT_CNCX   (unsigned[]) {3, 0}
#define	INDX_TAG_NCX_LEVEL   (unsigned[]) {4, 0}
#define	INDX_TAG_NCX_KIND_CNCX   (unsigned[]) {5, 0}
#define	INDX_TAG_NCX_POSFID   (unsigned[]) {6, 0}
#define	INDX_TAG_NCX_POSOFF   (unsigned[]) {6, 1}
#define	INDX_TAG_NCX_PARENT   (unsigned[]) {21, 0}
#define	INDX_TAG_NCX_CHILD_START   (unsigned[]) {22, 0}
#define	INDX_TAG_NCX_CHILD_END   (unsigned[]) {23, 0}
#define	INDX_TAG_SKEL_COUNT   (unsigned[]) {1, 0}
#define	INDX_TAG_SKEL_POSITION   (unsigned[]) {6, 0}
#define	INDX_TAG_SKEL_LENGTH   (unsigned[]) {6, 1}
#define	INDX_TAG_FRAG_AID_CNCX   (unsigned[]) {2, 0}
#define	INDX_TAG_FRAG_FILE_NR   (unsigned[]) {3, 0}
#define	INDX_TAG_FRAG_SEQUENCE_NR   (unsigned[]) {4, 0}
#define	INDX_TAG_FRAG_POSITION   (unsigned[]) {6, 0}
#define	INDX_TAG_FRAG_LENGTH   (unsigned[]) {6, 1}
#define	INDX_TAG_ORTH_POSITION   (unsigned[]) {1, 0}
#define	INDX_TAG_ORTH_LENGTH   (unsigned[]) {2, 0}
#define	INDX_TAGARR_ORTH_INFL   42
#define	INDX_TAGARR_INFL_GROUPS   5
#define	INDX_TAGARR_INFL_PARTS_V2   26
#define	INDX_TAGARR_INFL_PARTS_V1   7
#define	INDX_LABEL_SIZEMAX   1000
#define	INDX_INFLTAG_SIZEMAX   25000
#define	INDX_INFLBUF_SIZEMAX   500
#define	INDX_INFLSTRINGS_MAX   500
#define	ORDT_RECORD_MAXCNT   256 /* max entries count in old ordt */
#define	CNCX_RECORD_MAXCNT   0xf /* max entries count */
#define	INDX_RECORD_MAXCNT   6000 /* max index entries per record */
#define	INDX_TOTAL_MAXCNT   ((size_t) INDX_RECORD_MAXCNT * 0xffff) /* max total index entries */
#define	INDX_NAME_SIZEMAX   0xff
#define	INDX_TAGVALUES_MAX   100
	Maximum value of tag values in index entry (MOBIIndexTag)

Functions
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_parse_indx (const MOBIPdbRecord *indx_record, MOBIIndx *indx, MOBITagx *tagx, MOBIOrdt *ordt)
	Parser of INDX record. 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...
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...
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...
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...

Detailed Description

Copyright (c) 2014 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/

Macro Definition Documentation

INDX_INFLBUF_SIZEMAX

#define INDX_INFLBUF_SIZEMAX   500

Max size of index label

INDX_INFLSTRINGS_MAX

#define INDX_INFLSTRINGS_MAX   500

Max number of inflected strings

INDX_INFLTAG_SIZEMAX

#define INDX_INFLTAG_SIZEMAX   25000

Max size of inflections tags per entry

INDX_LABEL_SIZEMAX

#define INDX_LABEL_SIZEMAX   1000

Max size of index label

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_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_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)