asprise_ocr_api.h File Reference

#include <stdio.h>
#include <string.h>
#include <dlfcn.h>

Go to the source code of this file.

Macros
#define	EXPORT_API
#define	CALL_CONVENTION
#define	OCR_SPEED_FASTEST   "fastest"
#define	OCR_SPEED_FAST   "fast"
#define	OCR_SPEED_SLOW   "slow"
#define	OCR_RECOGNIZE_TYPE_TEXT   "text"
#define	OCR_RECOGNIZE_TYPE_BARCODE   "barcode"
#define	OCR_RECOGNIZE_TYPE_ALL   "all"
#define	OCR_OUTPUT_FORMAT_PLAINTEXT   "text"
#define	OCR_OUTPUT_FORMAT_XML   "xml"
#define	OCR_OUTPUT_FORMAT_PDF   "pdf"
#define	OCR_OUTPUT_FORMAT_RTF   "rtf"
#define	OCR_PROP_LIMIT_TO_CHARSET   "PROP_LIMIT_TO_CHARSET"
#define	OCR_PROP_PAGE_TYPE   "PROP_PAGE_TYPE"
#define	OCR_PROP_PAGE_TYPE_AUTO_DETECT   "auto"
#define	OCR_PROP_PAGE_TYPE_SINGLE_BLOCK   "single_block"
#define	OCR_PROP_PAGE_TYPE_SINGLE_COLUMN   "single_column"
#define	OCR_PROP_PAGE_TYPE_SINGLE_LINE   "single_line"
#define	OCR_PROP_PAGE_TYPE_SINGLE_WORD   "single_word"
#define	OCR_PROP_PAGE_TYPE_SINGLE_CHARACTOR   "single_char"
#define	OCR_PROP_PAGE_TYPE_SCATTERED   "scattered"
#define	OCR_PROP_OUTPUT_SEPARATE_WORDS   "PROP_OUTPUT_SEPARATE_WORDS"
#define	OCR_PROP_INPUT_PDF_DPI   "PROP_INPUT_PDF_DPI"
#define	OCR_START_PROP_DICT_SKIP_BUILT_IN_DEFAULT   "START_PROP_DICT_SKIP_BUILT_IN_DEFAULT"
#define	OCR_START_PROP_DICT_SKIP_BUILT_IN_ALL   "START_PROP_DICT_SKIP_BUILT_IN_ALL"
#define	OCR_START_PROP_DICT_CUSTOM_DICT_FILE   "START_PROP_DICT_CUSTOM_DICT_FILE"
#define	OCR_START_PROP_DICT_CUSTOM_TEMPLATES_FILE   "START_PROP_DICT_CUSTOM_TEMPLATES_FILE"
#define	OCR_PROP_DICT_DICT_IMPORTANCE   "PROP_DICT_DICT_IMPORTANCE"
#define	OCR_PROP_PDF_OUTPUT_FILE   "PROP_PDF_OUTPUT_FILE"
#define	OCR_PROP_PDF_OUTPUT_IMAGE_DPI   "PROP_PDF_OUTPUT_IMAGE_DPI"
#define	OCR_PROP_PDF_OUTPUT_FONT   "PROP_PDF_OUTPUT_FONT"
#define	OCR_PROP_PDF_OUTPUT_TEXT_VISIBLE   "PROP_PDF_OUTPUT_TEXT_VISIBLE"
#define	OCR_PROP_PDF_OUTPUT_IMAGE_FORCE_BW   "PROP_PDF_OUTPUT_IMAGE_FORCE_BW"
#define	OCR_PROP_PDF_OUTPUT_PDFA   "PROP_PDF_OUTPUT_PDFA"
#define	OCR_PROP_PDF_OUTPUT_PDFA_FONT_FILE   "PROP_PDF_OUTPUT_PDFA_FONT_FILE"
#define	OCR_PROP_PDF_OUTPUT_CONF_THRESHOLD   "PROP_PDF_OUTPUT_CONF_THRESHOLD"
#define	OCR_PROP_PDF_OUTPUT_RETURN_TEXT   "PROP_PDF_OUTPUT_RETURN_TEXT"
#define	OCR_PROP_RTF_OUTPUT_FILE   "PROP_RTF_OUTPUT_FILE"
#define	OCR_PROP_RTF_OUTPUT_RETURN_TEXT   "PROP_RTF_OUTPUT_RETURN_TEXT"
#define	OCR_PROP_RTF_PAPER_SIZE   "PROP_RTF_PAPER_SIZE"
#define	OCR_PROP_RTF_SHOW_TABLE_CELL_BORDER   "PROP_RTF_SHOW_TABLE_CELL_BORDER"
#define	OCR_PROP_IMG_PREPROCESS_TYPE   "PROP_IMG_PREPROCESS_TYPE"
#define	OCR_PROP_IMG_PREPROCESS_TYPE_DEFAULT   "default"
#define	OCR_PROP_IMG_PREPROCESS_TYPE_DEFAULT_WITH_ORIENTATION_DETECTION   "default_with_orientation_detection"
#define	OCR_PROP_IMG_PREPROCESS_TYPE_CUSTOM   "custom"
#define	OCR_PROP_IMG_PREPROCESS_CUSTOM_CMDS   "PROP_IMG_PREPROCESS_CUSTOM_CMDS"
#define	OCR_PROP_TABLE_SKIP_DETECTION   "PROP_TABLE_SKIP_DETECTION"
#define	OCR_PROP_TABLE_MIN_SIDE_LENGTH   "PROP_TABLE_MIN_SIDE_LENGTH"
#define	OCR_PROP_SAVE_INTERMEDIATE_IMAGES_TO_DIR   "PROP_SAVE_INTERMEDIATE_IMAGES_TO_DIR"
#define	OCR_PROP_INPUT_PDF_SINGLE_IMG_PER_PAGE   "PROP_INPUT_PDF_SINGLE_IMG_PER_PAGE"
#define	FUNC_NAME_com_asprise_ocr_version   "com_asprise_ocr_version"
#define	FUNC_NAME_com_asprise_ocr_setup   "com_asprise_ocr_setup"
#define	FUNC_NAME_com_asprise_ocr_list_supported_langs   "com_asprise_ocr_list_supported_langs"
#define	FUNC_NAME_com_asprise_ocr_start   "com_asprise_ocr_start"
#define	FUNC_NAME_com_asprise_ocr_stop   "com_asprise_ocr_stop"
#define	FUNC_NAME_com_asprise_ocr_recognize   "com_asprise_ocr_recognize"
#define	FUNC_NAME_com_asprise_ocr_util_delete   "com_asprise_ocr_util_delete"
#define	FUNC_NAME_com_asprise_ocr_input_license   "com_asprise_ocr_input_license"

Typedefs
typedef void(*	FUNC_TYPE) (void *)
typedef void *	LIBRARY_HANDLE
typedef const char *CALL_CONVENTION *	func_type_com_asprise_ocr_version()
typedef char *CALL_CONVENTION *	func_type_com_asprise_ocr_list_supported_langs()
typedef long	long(CALL_CONVENTION *func_type_com_asprise_ocr_start) (const char *
typedef long const char const char const char const char *typedef	void(CALL_CONVENTION *func_type_com_asprise_ocr_stop) (long long)
typedef char *CALL_CONVENTION *	func_type_com_asprise_ocr_recognize(long long, const char *, int, int, int, int, int, const char *, const char *, const char *, const char *, const char *)
typedef const char *func_type_com_asprise_ocr_input_license	func_com_asprise_ocr_input_license = NULL

Functions
EXPORT_API const char *CALL_CONVENTION	com_asprise_ocr_version ()
EXPORT_API int CALL_CONVENTION	com_asprise_ocr_setup (bool queryOnly)
EXPORT_API char *CALL_CONVENTION	com_asprise_ocr_list_supported_langs ()
EXPORT_API long long CALL_CONVENTION	com_asprise_ocr_start (const char *lang, const char *speed, const char *propSpec, const char *propSeparator, const char *propKeyValueSpeparator)
EXPORT_API char *CALL_CONVENTION	com_asprise_ocr_recognize (long long handle, const char *imgFiles, int pageIndex, int startX, int startY, int width, int height, const char *recognizeType, const char *outputFormat, const char *propSpec, const char *propSeparator, const char *propKeyValueSpeparator)
EXPORT_API void CALL_CONVENTION	com_asprise_ocr_stop (long long handle)
EXPORT_API void CALL_CONVENTION	com_asprise_ocr_input_license (const char *licenseeName, const char *licenseCode)
EXPORT_API void CALL_CONVENTION	com_asprise_ocr_util_delete (long long handle, bool isArray)
EXPORT_API void CALL_CONVENTION	initaocr (void)
EXPORT_API void CALL_CONVENTION	initaocr_x64 (void)
typedef	int (CALL_CONVENTION *func_type_com_asprise_ocr_setup)(bool)
typedef	void (CALL_CONVENTION *func_type_com_asprise_ocr_util_delete)(long long
bool	checkIfFuncFound (bool funcFound, const char *funcName)
LIBRARY_HANDLE	dynamic_load_aocr_library (const char *pathToFolderContainingAocrLibrary)
bool	dynamic_unload_aocr_library (LIBRARY_HANDLE handle)
bool	assertFuncExists (bool exists)
const char *	c_com_asprise_ocr_version ()
int CALL_CONVENTION	c_com_asprise_ocr_setup (bool queryOnly)
char *	c_com_asprise_ocr_list_supported_langs ()
long long	c_com_asprise_ocr_start (const char *lang, const char *speed, const char *propSpec, const char *propSeparator, const char *propKeyValueSpeparator)
char *	c_com_asprise_ocr_recognize (long long handle, const char *imgFiles, int pageIndex, int startX, int startY, int width, int height, const char *recognizeType, const char *outputFormat, const char *propSpec, const char *propSeparator, const char *propKeyValueSpeparator)
void	c_com_asprise_ocr_stop (long long handle)
void	c_com_asprise_ocr_input_license (const char *licenseeName, const char *licenseCode)
void	c_com_asprise_ocr_util_delete (long long handle, bool isArray)

Variables
typedef	bool
func_type_com_asprise_ocr_list_supported_langs	func_com_asprise_ocr_list_supported_langs = NULL
func_type_com_asprise_ocr_recognize	func_com_asprise_ocr_recognize = NULL
func_type_com_asprise_ocr_setup	func_com_asprise_ocr_setup = NULL
func_type_com_asprise_ocr_start	func_com_asprise_ocr_start = NULL
func_type_com_asprise_ocr_stop	func_com_asprise_ocr_stop = NULL
func_type_com_asprise_ocr_util_delete	func_com_asprise_ocr_util_delete = NULL
func_type_com_asprise_ocr_version	func_com_asprise_ocr_version = NULL

Macro Definition Documentation

#define CALL_CONVENTION

#define EXPORT_API

Asprise OCR API for C/C++: ALL RIGHTS RESERVED (c) Asprise.com

Read developer's guide to C/C++ OCR SDK API.

For evaluation only: 1) Using this file with or without modification in any application is strictly prohibited unless you have obtained a license from Asprise.

2) Distributing this file in any way is prohibited.

Visit Asprise OCR SDK C/C++ API

#define FUNC_NAME_com_asprise_ocr_input_license   "com_asprise_ocr_input_license"

#define FUNC_NAME_com_asprise_ocr_list_supported_langs   "com_asprise_ocr_list_supported_langs"

#define FUNC_NAME_com_asprise_ocr_recognize   "com_asprise_ocr_recognize"

#define FUNC_NAME_com_asprise_ocr_setup   "com_asprise_ocr_setup"

#define FUNC_NAME_com_asprise_ocr_start   "com_asprise_ocr_start"

#define FUNC_NAME_com_asprise_ocr_stop   "com_asprise_ocr_stop"

#define FUNC_NAME_com_asprise_ocr_util_delete   "com_asprise_ocr_util_delete"

#define FUNC_NAME_com_asprise_ocr_version   "com_asprise_ocr_version"

#define OCR_OUTPUT_FORMAT_PDF   "pdf"

#define OCR_OUTPUT_FORMAT_PLAINTEXT   "text"

#define OCR_OUTPUT_FORMAT_RTF   "rtf"

#define OCR_OUTPUT_FORMAT_XML   "xml"

#define OCR_PROP_DICT_DICT_IMPORTANCE   "PROP_DICT_DICT_IMPORTANCE"

Percentage measuring the importance of the dictionary (0: not at all; 100: extremely important; default: 10)

#define OCR_PROP_IMG_PREPROCESS_CUSTOM_CMDS   "PROP_IMG_PREPROCESS_CUSTOM_CMDS"

Image pre-processing command

#define OCR_PROP_IMG_PREPROCESS_TYPE   "PROP_IMG_PREPROCESS_TYPE"

Image pre-processing type

#define OCR_PROP_IMG_PREPROCESS_TYPE_CUSTOM   "custom"

#define OCR_PROP_IMG_PREPROCESS_TYPE_DEFAULT   "default"

#define OCR_PROP_IMG_PREPROCESS_TYPE_DEFAULT_WITH_ORIENTATION_DETECTION   "default_with_orientation_detection"

#define OCR_PROP_INPUT_PDF_DPI   "PROP_INPUT_PDF_DPI"

The DPI to be used to render the PDF file; default is 300 if not specified

#define OCR_PROP_INPUT_PDF_SINGLE_IMG_PER_PAGE   "PROP_INPUT_PDF_SINGLE_IMG_PER_PAGE"

Set to true if there is only one image per page on the input PDF file; potentially speed up of the process

#define OCR_PROP_LIMIT_TO_CHARSET   "PROP_LIMIT_TO_CHARSET"

#define OCR_PROP_OUTPUT_SEPARATE_WORDS   "PROP_OUTPUT_SEPARATE_WORDS"

#define OCR_PROP_PAGE_TYPE   "PROP_PAGE_TYPE"

#define OCR_PROP_PAGE_TYPE_AUTO_DETECT   "auto"

#define OCR_PROP_PAGE_TYPE_SCATTERED   "scattered"

#define OCR_PROP_PAGE_TYPE_SINGLE_BLOCK   "single_block"

#define OCR_PROP_PAGE_TYPE_SINGLE_CHARACTOR   "single_char"

#define OCR_PROP_PAGE_TYPE_SINGLE_COLUMN   "single_column"

#define OCR_PROP_PAGE_TYPE_SINGLE_LINE   "single_line"

#define OCR_PROP_PAGE_TYPE_SINGLE_WORD   "single_word"

#define OCR_PROP_PDF_OUTPUT_CONF_THRESHOLD   "PROP_PDF_OUTPUT_CONF_THRESHOLD"

Text recognized below or above confidence will be highlighted in different colors.

#define OCR_PROP_PDF_OUTPUT_FILE   "PROP_PDF_OUTPUT_FILE"

#define OCR_PROP_PDF_OUTPUT_FONT   "PROP_PDF_OUTPUT_FONT"

#define OCR_PROP_PDF_OUTPUT_IMAGE_DPI   "PROP_PDF_OUTPUT_IMAGE_DPI"

#define OCR_PROP_PDF_OUTPUT_IMAGE_FORCE_BW   "PROP_PDF_OUTPUT_IMAGE_FORCE_BW"

#define OCR_PROP_PDF_OUTPUT_PDFA   "PROP_PDF_OUTPUT_PDFA"

#define OCR_PROP_PDF_OUTPUT_PDFA_FONT_FILE   "PROP_PDF_OUTPUT_PDFA_FONT_FILE"

#define OCR_PROP_PDF_OUTPUT_RETURN_TEXT   "PROP_PDF_OUTPUT_RETURN_TEXT"

Return text in 'text' or 'xml' format

#define OCR_PROP_PDF_OUTPUT_TEXT_VISIBLE   "PROP_PDF_OUTPUT_TEXT_VISIBLE"

#define OCR_PROP_RTF_OUTPUT_FILE   "PROP_RTF_OUTPUT_FILE"

#define OCR_PROP_RTF_OUTPUT_RETURN_TEXT   "PROP_RTF_OUTPUT_RETURN_TEXT"

Return text in 'text' or 'xml' format

#define OCR_PROP_RTF_PAPER_SIZE   "PROP_RTF_PAPER_SIZE"

default is LETTER, may set to A4.

#define OCR_PROP_RTF_SHOW_TABLE_CELL_BORDER   "PROP_RTF_SHOW_TABLE_CELL_BORDER"

set to false to disable rendering table cells borders into black. Default is true.

#define OCR_PROP_SAVE_INTERMEDIATE_IMAGES_TO_DIR   "PROP_SAVE_INTERMEDIATE_IMAGES_TO_DIR"

Save intermediate images generated for debug purpose - don't specify or empty string to skip saving

#define OCR_PROP_TABLE_MIN_SIDE_LENGTH   "PROP_TABLE_MIN_SIDE_LENGTH"

default is 32 if not specified

#define OCR_PROP_TABLE_SKIP_DETECTION   "PROP_TABLE_SKIP_DETECTION"

table will be detected by default; set this property to true to skip detection.

#define OCR_RECOGNIZE_TYPE_ALL   "all"

#define OCR_RECOGNIZE_TYPE_BARCODE   "barcode"

#define OCR_RECOGNIZE_TYPE_TEXT   "text"

#define OCR_SPEED_FAST   "fast"

#define OCR_SPEED_FASTEST   "fastest"

#define OCR_SPEED_SLOW   "slow"

#define OCR_START_PROP_DICT_CUSTOM_DICT_FILE   "START_PROP_DICT_CUSTOM_DICT_FILE"

Path to your custom dictionary (words are separated using line breaks). Default value: null

#define OCR_START_PROP_DICT_CUSTOM_TEMPLATES_FILE   "START_PROP_DICT_CUSTOM_TEMPLATES_FILE"

Path to your custom templates (templates are separated using line breaks). Default value: null

#define OCR_START_PROP_DICT_SKIP_BUILT_IN_ALL   "START_PROP_DICT_SKIP_BUILT_IN_ALL"

set to 'true' to skip using all built-in dicts. Default value: 'false'

#define OCR_START_PROP_DICT_SKIP_BUILT_IN_DEFAULT   "START_PROP_DICT_SKIP_BUILT_IN_DEFAULT"

set to 'true' to skip using the default built in dict. Default value: 'false'

Typedef Documentation

typedef const char* func_type_com_asprise_ocr_input_license func_com_asprise_ocr_input_license = NULL

typedef void(* FUNC_TYPE) (void *)

typedef char* CALL_CONVENTION* func_type_com_asprise_ocr_list_supported_langs()

typedef char* CALL_CONVENTION* func_type_com_asprise_ocr_recognize(long long, const char *, int, int, int, int, int, const char *, const char *, const char *, const char *, const char *)

typedef const char* CALL_CONVENTION* func_type_com_asprise_ocr_version()

typedef void* LIBRARY_HANDLE

typedef long long(CALL_CONVENTION *func_type_com_asprise_ocr_start)(const char *

typedef void

Function Documentation

bool assertFuncExists

(

bool

exists

)

void c_com_asprise_ocr_input_license	(	const char *	licenseeName,
		const char *	licenseCode
	)

Inputs the license information

char* c_com_asprise_ocr_list_supported_langs

(

)

Call after setup; list all supported langs separated by comma; please delete returned char[] after use.

char* c_com_asprise_ocr_recognize	(	long long	handle,
		const char *	imgFiles,
		int	pageIndex,
		int	startX,
		int	startY,
		int	width,
		int	height,
		const char *	recognizeType,
		const char *	outputFormat,
		const char *	propSpec,
		const char *	propSeparator,
		const char *	propKeyValueSpeparator
	)

returns the recognized text or error message in format like '<error: ...'

int CALL_CONVENTION c_com_asprise_ocr_setup

(

bool

queryOnly

)

Returns 0 if setup required; 1 if setup has been done; or negative error code.

long long c_com_asprise_ocr_start	(	const char *	lang,
		const char *	speed,
		const char *	propSpec,
		const char *	propSeparator,
		const char *	propKeyValueSpeparator
	)

returns the ocr engine handle or negative error code.

void c_com_asprise_ocr_stop

(

long long

handle

)

Destroys the OCR engine.

void c_com_asprise_ocr_util_delete	(	long long	handle,
		bool	isArray
	)

Util function that perform C delete/delete[]

const char* c_com_asprise_ocr_version

(

)

Returns the version of the library.

bool checkIfFuncFound	(	bool	funcFound,
		const char *	funcName
	)

EXPORT_API void CALL_CONVENTION com_asprise_ocr_input_license	(	const char *	licenseeName,
		const char *	licenseCode
	)

Inputs the license information

EXPORT_API char* CALL_CONVENTION com_asprise_ocr_list_supported_langs

(

)

Call after setup; list all supported langs separated by comma; please delete returned char[] after use.

EXPORT_API char* CALL_CONVENTION com_asprise_ocr_recognize	(	long long	handle,
		const char *	imgFiles,
		int	pageIndex,
		int	startX,
		int	startY,
		int	width,
		int	height,
		const char *	recognizeType,
		const char *	outputFormat,
		const char *	propSpec,
		const char *	propSeparator,
		const char *	propKeyValueSpeparator
	)

returns the recognized text or error message in format like '<error: ...'

EXPORT_API int CALL_CONVENTION com_asprise_ocr_setup

(

bool

queryOnly

)

Returns 0 if setup required; 1 if setup has been done; or negative error code.

EXPORT_API long long CALL_CONVENTION com_asprise_ocr_start	(	const char *	lang,
		const char *	speed,
		const char *	propSpec,
		const char *	propSeparator,
		const char *	propKeyValueSpeparator
	)

returns the ocr engine handle or negative error code.

EXPORT_API void CALL_CONVENTION com_asprise_ocr_stop

(

long long

handle

)

Destroies the OCR engine.

EXPORT_API void CALL_CONVENTION com_asprise_ocr_util_delete	(	long long	handle,
		bool	isArray
	)

Util function that perform C delete/delete[]

EXPORT_API const char* CALL_CONVENTION com_asprise_ocr_version

(

)

Returns the version of the library.

LIBRARY_HANDLE dynamic_load_aocr_library

(

const char *

pathToFolderContainingAocrLibrary

)

Dynamically load aocr library; return the loaded library handle or NULL if failed.

bool dynamic_unload_aocr_library

(

LIBRARY_HANDLE

handle

)

unload aocr library; return true if successful or false if failed.

EXPORT_API void CALL_CONVENTION initaocr

(

void

)

call setup

EXPORT_API void CALL_CONVENTION initaocr_x64

(

void

)

typedef int

(

CALL_CONVENTION *

func_type_com_asprise_ocr_setup

)

typedef void

(

CALL_CONVENTION *

func_type_com_asprise_ocr_util_delete

)

Variable Documentation

typedef bool

func_type_com_asprise_ocr_list_supported_langs func_com_asprise_ocr_list_supported_langs = NULL

func_type_com_asprise_ocr_recognize func_com_asprise_ocr_recognize = NULL

func_type_com_asprise_ocr_setup func_com_asprise_ocr_setup = NULL

func_type_com_asprise_ocr_start func_com_asprise_ocr_start = NULL

func_type_com_asprise_ocr_stop func_com_asprise_ocr_stop = NULL

func_type_com_asprise_ocr_util_delete func_com_asprise_ocr_util_delete = NULL

func_type_com_asprise_ocr_version func_com_asprise_ocr_version = NULL