command-line tool to run GNU libextractor More...

#include "platform.h"
#include "extractor.h"
#include "getopt.h"
#include <signal.h>

Include dependency graph for extract.c:

Data Structures
struct	Help

struct	BibTexMap

Macros
#define	YES 1

#define	NO 0

#define	BORDER 29

Functions
static void	ignore_sigpipe ()

static void	format_help (const char general, const char description, const struct Help *opt)

static void	print_help ()

static int	print_selected_keywords (void cls, const char plugin_name, enum EXTRACTOR_MetaType type, enum EXTRACTOR_MetaFormat format, const char data_mime_type, const char data, size_t data_len)

static int	print_selected_keywords_grep_friendly (void cls, const char plugin_name, enum EXTRACTOR_MetaType type, enum EXTRACTOR_MetaFormat format, const char data_mime_type, const char data, size_t data_len)

static void	cleanup_bibtex ()

static int	print_bibtex (void cls, const char plugin_name, enum EXTRACTOR_MetaType type, enum EXTRACTOR_MetaFormat format, const char data_mime_type, const char data, size_t data_len)

static void	finish_bibtex (const char *fn)

static char **	_make_continuous_arg_copy (int argc, char const argv)

static int	_get_utf8_args (int argc, char const argv, int u8argc, char **u8argv)

int	main (int argc, char *argv[])

Variables
static int *	print

static int	verbose

static int	in_process

static int	from_memory

static char *	entry_type

static struct BibTexMap	btm []

Detailed Description

command-line tool to run GNU libextractor

Author: Christian Grothoff

Definition in file extract.c.

Macro Definition Documentation

◆ BORDER

#define BORDER 29

Indentation for descriptions.

Definition at line 111 of file extract.c.

◆ NO

#define NO 0

Definition at line 31 of file extract.c.

◆ YES

#define YES 1

Definition at line 30 of file extract.c.

Function Documentation

◆ _get_utf8_args()

static int _get_utf8_args	(	int	argc,
		char const	argv,
		int *	u8argc,
		char ***	u8argv
	)

static

Returns utf-8 encoded arguments. Returned argv has u8argv[u8argc] == NULL. Returned argv is a single memory block, and can be freed with a single free () call.

Parameters

argc	argc (as given by main())
argv	argv (as given by main())
u8argc	a location to store new argc in (though it's th same as argc)
u8argv	a location to store new argv in

Returns: 0 on success, -1 on failure

Definition at line 690 of file extract.c.

References _make_continuous_arg_copy(), and NULL.

Referenced by main().

◆ _make_continuous_arg_copy()

static char** _make_continuous_arg_copy	(	int	argc,
		char const	argv
	)

static

Makes a copy of argv that consists of a single memory chunk that can be freed with a single call to free ();

Definition at line 654 of file extract.c.

References NULL.

Referenced by _get_utf8_args().

◆ cleanup_bibtex()

static void cleanup_bibtex ( )

static

Clean up the bibtex processor in preparation for the next round.

Definition at line 498 of file extract.c.

References BibTexMap::bibTexName, btm, entry_type, NULL, and BibTexMap::value.

Referenced by main().

◆ finish_bibtex()

static void finish_bibtex ( const char * fn )

static

Print the computed bibTeX entry.

Parameters

fn	file for which the entry was created.

Definition at line 564 of file extract.c.

References BibTexMap::bibTexName, btm, entry_type, NULL, and BibTexMap::value.

Referenced by main().

◆ format_help()

static void format_help	(	const char *	general,
		const char *	description,
		const struct Help *	opt
	)

static

Display help text (–help).

Parameters

general	binary name
description	program description
opt	program options (NULL-terminated array)

Definition at line 122 of file extract.c.

References _, BORDER, gettext, and NULL.

Referenced by print_help().

◆ ignore_sigpipe()

static void ignore_sigpipe ( )

static

Install a signal handler to ignore SIGPIPE.

Definition at line 59 of file extract.c.

Referenced by main().

◆ main()

int main	(	int	argc,
		char *	argv[]
	)

Main function for the 'extract' tool. Invoke with a list of filenames to extract keywords from.

Parameters

argc	number of arguments in argv
argv	command line options and filename to run on

Returns: 0 on success

Definition at line 753 of file extract.c.

References _, _get_utf8_args(), cleanup_bibtex(), EXTRACTOR_extract(), EXTRACTOR_metatype_get_max(), EXTRACTOR_metatype_to_string(), EXTRACTOR_OPTION_DEFAULT_POLICY, EXTRACTOR_OPTION_IN_PROCESS, EXTRACTOR_plugin_add_config(), EXTRACTOR_plugin_add_defaults(), EXTRACTOR_plugin_remove_all(), finish_bibtex(), from_memory, getopt_long(), gettext, ignore_sigpipe(), in_process, NO, NULL, optarg, optind, print, print_bibtex(), print_help(), print_selected_keywords(), print_selected_keywords_grep_friendly(), textdomain, verbose, and YES.

◆ print_bibtex()

static int print_bibtex	(	void *	cls,
		const char *	plugin_name,
		enum EXTRACTOR_MetaType	type,
		enum EXTRACTOR_MetaFormat	format,
		const char *	data_mime_type,
		const char *	data,
		size_t	data_len
	)

static

Callback function for printing meta data in bibtex format.

Parameters

cls	closure, not used
plugin_name	name of the plugin that produced this value; special values can be used (i.e. '<zlib>' for zlib being used in the main libextractor library and yielding meta data).
type	libextractor-type describing the meta data
format	basic format information about data
data_mime_type	mime-type of data (not of the original file); can be NULL (if mime-type is not known)
data	actual meta-data found
data_len	number of bytes in data

Returns: 0 to continue extracting (always)

Definition at line 529 of file extract.c.

References BibTexMap::bibTexName, btm, entry_type, EXTRACTOR_METAFORMAT_UTF8, EXTRACTOR_METATYPE_BIBTEX_ENTRY_TYPE, BibTexMap::le_type, NULL, print, type, BibTexMap::value, and YES.

Referenced by main().

◆ print_help()

static void print_help ( )

static

Run –help.

Definition at line 215 of file extract.c.

References _, format_help(), gettext_noop, and NULL.

Referenced by main().

◆ print_selected_keywords()

static int print_selected_keywords	(	void *	cls,
		const char *	plugin_name,
		enum EXTRACTOR_MetaType	type,
		enum EXTRACTOR_MetaFormat	format,
		const char *	data_mime_type,
		const char *	data,
		size_t	data_len
	)

static

Print a keyword list to a file.

Parameters

cls	closure, not used
plugin_name	name of the plugin that produced this value; special values can be used (i.e. '<zlib>' for zlib being used in the main libextractor library and yielding meta data).
type	libextractor-type describing the meta data
format	basic format information about data
data_mime_type	mime-type of data (not of the original file); can be NULL (if mime-type is not known)
data	actual meta-data found
data_len	number of bytes in data

Returns: 0 to continue extracting, 1 to abort

Definition at line 275 of file extract.c.

References _, EXTRACTOR_METAFORMAT_BINARY, EXTRACTOR_METAFORMAT_C_STRING, EXTRACTOR_METAFORMAT_UNKNOWN, EXTRACTOR_METAFORMAT_UTF8, EXTRACTOR_metatype_to_string(), gettext, iconv_helper(), NULL, print, type, verbose, and YES.

Referenced by main().

◆ print_selected_keywords_grep_friendly()

static int print_selected_keywords_grep_friendly	(	void *	cls,
		const char *	plugin_name,
		enum EXTRACTOR_MetaType	type,
		enum EXTRACTOR_MetaFormat	format,
		const char *	data_mime_type,
		const char *	data,
		size_t	data_len
	)

static

Print a keyword list to a file without new lines.

Parameters

cls	closure, not used
plugin_name	name of the plugin that produced this value; special values can be used (i.e. '<zlib>' for zlib being used in the main libextractor library and yielding meta data).
type	libextractor-type describing the meta data
format	basic format information about data
data_mime_type	mime-type of data (not of the original file); can be NULL (if mime-type is not known)
data	actual meta-data found
data_len	number of bytes in data

Returns: 0 to continue extracting, 1 to abort

Definition at line 368 of file extract.c.

References EXTRACTOR_METAFORMAT_BINARY, EXTRACTOR_METAFORMAT_C_STRING, EXTRACTOR_METAFORMAT_UNKNOWN, EXTRACTOR_METAFORMAT_UTF8, EXTRACTOR_metatype_to_string(), gettext, gettext_noop, iconv_helper(), NULL, print, type, verbose, and YES.

Referenced by main().

Variable Documentation

◆ btm

struct BibTexMap btm[]

static

Initial value:

= {
  { "title", EXTRACTOR_METATYPE_TITLE, NULL},
  { "year", EXTRACTOR_METATYPE_PUBLICATION_YEAR, NULL },
  { "author", EXTRACTOR_METATYPE_AUTHOR_NAME, NULL },
  { "book", EXTRACTOR_METATYPE_BOOK_TITLE, NULL},
  { "edition", EXTRACTOR_METATYPE_BOOK_EDITION, NULL},
  { "chapter", EXTRACTOR_METATYPE_BOOK_CHAPTER_NUMBER, NULL},
  { "journal", EXTRACTOR_METATYPE_JOURNAL_NAME, NULL},
  { "volume", EXTRACTOR_METATYPE_JOURNAL_VOLUME, NULL},
  { "number", EXTRACTOR_METATYPE_JOURNAL_NUMBER, NULL},
  { "pages", EXTRACTOR_METATYPE_PAGE_COUNT, NULL },
  { "pages", EXTRACTOR_METATYPE_PAGE_RANGE, NULL },
  { "school", EXTRACTOR_METATYPE_AUTHOR_INSTITUTION, NULL},
  { "publisher", EXTRACTOR_METATYPE_PUBLISHER, NULL },
  { "address", EXTRACTOR_METATYPE_PUBLISHER_ADDRESS, NULL },
  { "institution", EXTRACTOR_METATYPE_PUBLISHER_INSTITUTION, NULL },
  { "series", EXTRACTOR_METATYPE_PUBLISHER_SERIES, NULL},
  { "month", EXTRACTOR_METATYPE_PUBLICATION_MONTH, NULL },
  { "url", EXTRACTOR_METATYPE_URL, NULL},
  { "note", EXTRACTOR_METATYPE_COMMENT, NULL},
  { "eprint", EXTRACTOR_METATYPE_BIBTEX_EPRINT, NULL },
  { "type", EXTRACTOR_METATYPE_PUBLICATION_TYPE, NULL },
  { NULL, 0, NULL }
}

Mapping between bibTeX strings, libextractor meta data types and values for the current document.

Definition at line 462 of file extract.c.

Referenced by cleanup_bibtex(), finish_bibtex(), and print_bibtex().

◆ entry_type

char* entry_type

static

Type of the entry for bibtex.

Definition at line 462 of file extract.c.

Referenced by cleanup_bibtex(), finish_bibtex(), and print_bibtex().

◆ from_memory

int from_memory

static

Read file contents into memory, then feed them to extractor.

Definition at line 52 of file extract.c.

Referenced by main().

◆ in_process

int in_process

static

Run plugins in-process.

Definition at line 47 of file extract.c.

Referenced by main().

◆ print

int* print

static

Which keyword types should we print?

Definition at line 37 of file extract.c.

Referenced by main(), print_bibtex(), print_selected_keywords(), and print_selected_keywords_grep_friendly().

◆ verbose

int verbose

static

How verbose are we supposed to be?

Definition at line 42 of file extract.c.

Referenced by main(), print_selected_keywords(), and print_selected_keywords_grep_friendly().

Data Structures

Macros

Functions

Variables

Detailed Description

Macro Definition Documentation

◆ BORDER

◆ NO

◆ YES

Function Documentation

◆ _get_utf8_args()

◆ _make_continuous_arg_copy()

◆ cleanup_bibtex()

◆ finish_bibtex()

◆ format_help()

◆ ignore_sigpipe()

◆ main()

◆ print_bibtex()

◆ print_help()

◆ print_selected_keywords()

◆ print_selected_keywords_grep_friendly()

Variable Documentation

◆ btm

◆ entry_type

◆ from_memory

◆ in_process

◆ print

◆ verbose