decancer/native_docs/decancer_8h_source.html

// SPDX-License-Identifier: MIT

// SPDX-FileCopyrightText: 2021-2026 null8626


#ifndef __DECANCER_H__

#define __DECANCER_H__


#include <stdbool.h>

#include <stddef.h>

#include <stdint.h>


#if defined(_WIN32) && !defined(DECANCER_STATIC)

#define DECANCER_EXPORT __declspec(dllimport)

#else

#define DECANCER_EXPORT

#endif


#define DECANCER_VERSION 0x030303


#define DECANCER_VERSION_MAJOR ((DECANCER_VERSION & 0xff0000) >> 16)


#define DECANCER_VERSION_MINOR ((DECANCER_VERSION & 0xff00) >> 8)


#define DECANCER_VERSION_PATCH (DECANCER_VERSION & 0xff)


#define DECANCER_TRANSLATION_KIND_CHARACTER 0


#define DECANCER_TRANSLATION_KIND_STRING 1


#define DECANCER_TRANSLATION_KIND_NONE 2


#define DECANCER_OPTION_DEFAULT 0


#define DECANCER_OPTION_RETAIN_CAPITALIZATION (1 << 0)


#define DECANCER_OPTION_DISABLE_BIDI (1 << 1)


#define DECANCER_OPTION_DISABLE_LEETSPEAK (1 << 2)


#define DECANCER_OPTION_RETAIN_DIACRITICS (1 << 3)


#define DECANCER_OPTION_RETAIN_GREEK (1 << 4)


#define DECANCER_OPTION_RETAIN_CYRILLIC (1 << 5)


#define DECANCER_OPTION_RETAIN_HEBREW (1 << 6)


#define DECANCER_OPTION_RETAIN_ARABIC (1 << 7)


#define DECANCER_OPTION_RETAIN_DEVANAGARI (1 << 8)


#define DECANCER_OPTION_RETAIN_BENGALI (1 << 9)


#define DECANCER_OPTION_RETAIN_ARMENIAN (1 << 10)


#define DECANCER_OPTION_RETAIN_GUJARATI (1 << 11)


#define DECANCER_OPTION_RETAIN_TAMIL (1 << 12)


#define DECANCER_OPTION_RETAIN_THAI (1 << 13)


#define DECANCER_OPTION_RETAIN_LAO (1 << 14)


#define DECANCER_OPTION_RETAIN_BURMESE (1 << 15)


#define DECANCER_OPTION_RETAIN_KHMER (1 << 16)


#define DECANCER_OPTION_RETAIN_MONGOLIAN (1 << 17)


#define DECANCER_OPTION_RETAIN_CHINESE (1 << 18)


#define DECANCER_OPTION_RETAIN_JAPANESE (1 << 19)


#define DECANCER_OPTION_RETAIN_KOREAN (1 << 20)


#define DECANCER_OPTION_RETAIN_BRAILLE (1 << 21)


#define DECANCER_OPTION_RETAIN_EMOJIS (1 << 22)


#define DECANCER_OPTION_RETAIN_TURKISH (1 << 23)


#define DECANCER_OPTION_ASCII_ONLY (1 << 24)


#define DECANCER_OPTION_ALPHANUMERIC_ONLY (1 << 25)


#define DECANCER_OPTION_ALL 0x3ffffff


#define DECANCER_OPTION_PURE_HOMOGLYPH 0x7ffff8


typedef struct {

  const char* message;


  uint8_t message_length;

} decancer_error_t;


#ifndef DECANCER_UTF16_ONLY


typedef struct {

  const uint8_t* string;


  size_t size;

} decancer_keyword_t;


#endif


#ifndef DECANCER_UTF8_ONLY


typedef struct {

  const uint16_t* string;


  size_t length;

} decancer_keyword_utf16_t;


typedef void* decancer_cured_raw_utf16_t;

#endif


#ifndef DECANCER_UTF16_ONLY

typedef void* decancer_matcher_t;

#endif


#ifndef DECANCER_UTF8_ONLY

typedef void* decancer_matcher_utf16_t;

#endif


typedef void* decancer_matches_t;


typedef struct {

  uint8_t kind;


  union {

    uint32_t character;


    struct {

      const uint8_t* contents;


      size_t size;


      void* __heap;

    } string;

  } contents;

} decancer_translation_t;


typedef void* decancer_cured_t;


typedef struct {

  size_t start;


  size_t end;

} decancer_match_t;


typedef uint32_t decancer_options_t;


#ifdef __cplusplus

extern "C" {

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT decancer_cured_t decancer_cure(const uint8_t* input_str, const size_t input_size, const decancer_options_t options, decancer_error_t* error);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT decancer_cured_t decancer_cure_utf16(const uint16_t* input_str, const size_t input_length, const decancer_options_t options, decancer_error_t* error);

#endif


  DECANCER_EXPORT void decancer_cure_char(const uint32_t input, const decancer_options_t options, decancer_translation_t* translation);


  DECANCER_EXPORT void decancer_disable_leetspeak(decancer_cured_t cured, const bool switch_);


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT const uint8_t* decancer_cured_raw(decancer_cured_t cured, const decancer_match_t* match, size_t* output_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT decancer_cured_raw_utf16_t decancer_cured_raw_utf16(decancer_cured_t cured, const decancer_match_t* match, uint16_t** output_ptr, size_t* output_length);

#endif


  DECANCER_EXPORT const decancer_match_t* decancer_matches_raw(decancer_matches_t matches, size_t* output_size);


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT decancer_matcher_t decancer_find(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT decancer_matcher_utf16_t decancer_find_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT decancer_matches_t decancer_find_multiple(decancer_cured_t cured, const decancer_keyword_t* other, const size_t other_length);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT decancer_matches_t decancer_find_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t* other, const size_t other_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_matcher_next(decancer_matcher_t matcher, decancer_match_t* match);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_matcher_utf16_next(decancer_matcher_utf16_t matcher, decancer_match_t* match);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_censor(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size, const uint32_t replacement_char);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_censor_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length, const uint32_t replacement_char);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_replace(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size, const uint8_t* replacement_str, const size_t replacement_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_replace_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length, const uint16_t* replacement_str, const size_t replacement_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_censor_multiple(decancer_cured_t cured, const decancer_keyword_t* other, const size_t other_length, const uint32_t replacement_char);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_censor_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t* other, const size_t other_length, const uint32_t replacement_char);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_replace_multiple(decancer_cured_t cured, const decancer_keyword_t* other, const size_t other_length, const uint8_t* replacement_str, const size_t replacement_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_replace_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t* other, const size_t other_length, const uint16_t* replacement_str, const size_t replacement_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_contains(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_contains_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_starts_with(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_starts_with_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_ends_with(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_ends_with_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT bool decancer_equals(decancer_cured_t cured, const uint8_t* other_str, const size_t other_size);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT bool decancer_equals_utf16(decancer_cured_t cured, const uint16_t* other_str, const size_t other_length);


  DECANCER_EXPORT decancer_cured_raw_utf16_t decancer_cured_raw_utf16_clone(decancer_cured_raw_utf16_t raw_utf16_handle);


  DECANCER_EXPORT void decancer_cured_raw_utf16_free(decancer_cured_raw_utf16_t raw_utf16_handle);

#endif


#ifndef DECANCER_UTF16_ONLY

  DECANCER_EXPORT decancer_matches_t decancer_matcher_consume(decancer_matcher_t matcher);


  DECANCER_EXPORT void decancer_matcher_free(decancer_matcher_t matcher);

#endif


#ifndef DECANCER_UTF8_ONLY

  DECANCER_EXPORT decancer_matches_t decancer_matcher_utf16_consume(decancer_matcher_utf16_t matcher);


  DECANCER_EXPORT void decancer_matcher_utf16_free(decancer_matcher_utf16_t matcher);

#endif


  DECANCER_EXPORT decancer_matches_t decancer_matches_clone(decancer_matches_t matches);


  DECANCER_EXPORT void decancer_matches_free(decancer_matches_t matches);


  DECANCER_EXPORT void decancer_translation_init(decancer_translation_t* translation);


  DECANCER_EXPORT void decancer_translation_clone(const decancer_translation_t* translation_in, decancer_translation_t* translation_out);


  DECANCER_EXPORT void decancer_translation_free(decancer_translation_t* translation);


  DECANCER_EXPORT decancer_cured_t decancer_cured_clone(decancer_cured_t cured);


  DECANCER_EXPORT void decancer_cured_free(decancer_cured_t cured);


#undef DECANCER_EXPORT


#ifdef __cplusplus

} // extern "C"

#endif

#endif

decancer_replace_utf16
DECANCER_EXPORT bool decancer_replace_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length, const uint16_t *replacement_str, const size_t replacement_length)
Replaces every similar-looking match of the specified UTF-16 encoded string with another UTF-16 encod...

decancer_contains_utf16
DECANCER_EXPORT bool decancer_contains_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length)
Checks if the cured string similarly contains the specified UTF-16 encoded string.

decancer_matches_raw
DECANCER_EXPORT const decancer_match_t * decancer_matches_raw(decancer_matches_t matches, size_t *output_size)
Returns the raw list of every similar-looking match from a decancer_matches_t object.

decancer_cured_raw_utf16_free
DECANCER_EXPORT void decancer_cured_raw_utf16_free(decancer_cured_raw_utf16_t raw_utf16_handle)
Frees the rust object created by decancer_cured_raw_utf16.

decancer_equals_utf16
DECANCER_EXPORT bool decancer_equals_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length)
Checks if the cured string is similar with the specified UTF-16 encoded string.

decancer_cured_clone
DECANCER_EXPORT decancer_cured_t decancer_cured_clone(decancer_cured_t cured)
Clones the cured string object created by decancer_cure and decancer_cure_utf16.

decancer_find
DECANCER_EXPORT decancer_matcher_t decancer_find(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size)
Finds every similar-looking match of a UTF-8 encoded string in the cured string.

decancer_replace_multiple
DECANCER_EXPORT bool decancer_replace_multiple(decancer_cured_t cured, const decancer_keyword_t *other, const size_t other_length, const uint8_t *replacement_str, const size_t replacement_size)
Replaces every similar-looking match of the specified list of UTF-8 keywords with another UTF-8 encod...

decancer_censor_multiple_utf16
DECANCER_EXPORT bool decancer_censor_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t *other, const size_t other_length, const uint32_t replacement_char)
Censors every similar-looking match of the specified list of UTF-16 keywords. Unlike decancer_censor_...

decancer_translation_init
DECANCER_EXPORT void decancer_translation_init(decancer_translation_t *translation)
Initiates a newly created translation struct for use.

decancer_find_utf16
DECANCER_EXPORT decancer_matcher_utf16_t decancer_find_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length)
Finds every similar-looking match of a UTF-16 encoded string in the cured string.

decancer_replace_multiple_utf16
DECANCER_EXPORT bool decancer_replace_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t *other, const size_t other_length, const uint16_t *replacement_str, const size_t replacement_length)
Replaces every similar-looking match of the specified list of UTF-16 keywords with another UTF-16 enc...

decancer_cured_free
DECANCER_EXPORT void decancer_cured_free(decancer_cured_t cured)
Frees the cured string object created by decancer_cure and decancer_cure_utf16.

decancer_options_t
uint32_t decancer_options_t
An unsigned 32-bit bitflags that lets you customize decancer's behavior in its curing functions.
Definition decancer.h:584

decancer_cured_raw_utf16
DECANCER_EXPORT decancer_cured_raw_utf16_t decancer_cured_raw_utf16(decancer_cured_t cured, const decancer_match_t *match, uint16_t **output_ptr, size_t *output_length)
Retrieves the raw UTF-16 bytes from a cured string object.

decancer_matcher_t
void * decancer_matcher_t
Represents a UTF-8 matcher iterator object returned from decancer_find.
Definition decancer.h:400

decancer_matches_clone
DECANCER_EXPORT decancer_matches_t decancer_matches_clone(decancer_matches_t matches)
Clones the matches object created by decancer_find_multiple and decancer_find_multiple_utf16.

decancer_find_multiple
DECANCER_EXPORT decancer_matches_t decancer_find_multiple(decancer_cured_t cured, const decancer_keyword_t *other, const size_t other_length)
Finds every similar-looking match from a list of UTF-8 keywords in the cured string....

decancer_starts_with_utf16
DECANCER_EXPORT bool decancer_starts_with_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length)
Checks if the cured string similarly starts with the specified UTF-16 encoded string.

decancer_censor_multiple
DECANCER_EXPORT bool decancer_censor_multiple(decancer_cured_t cured, const decancer_keyword_t *other, const size_t other_length, const uint32_t replacement_char)
Censors every similar-looking match of the specified list of UTF-8 keywords. Unlike decancer_censor,...

decancer_cured_t
void * decancer_cured_t
Represents a cured string returned from decancer_cure and decancer_cure_utf16.
Definition decancer.h:517

decancer_censor
DECANCER_EXPORT bool decancer_censor(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size, const uint32_t replacement_char)
Censors every similar-looking match of the specified UTF-8 encoded string.

decancer_cured_raw_utf16_t
void * decancer_cured_raw_utf16_t
Represents a rust object returned from decancer_cured_raw_utf16. This value has no use other than ret...
Definition decancer.h:384

decancer_matches_free
DECANCER_EXPORT void decancer_matches_free(decancer_matches_t matches)
Frees the matches object created by decancer_find_multiple and decancer_find_multiple_utf16.

decancer_find_multiple_utf16
DECANCER_EXPORT decancer_matches_t decancer_find_multiple_utf16(decancer_cured_t cured, const decancer_keyword_utf16_t *other, const size_t other_length)
Finds every similar-looking match from a list of UTF-16 keywords in the cured string....

decancer_matcher_utf16_free
DECANCER_EXPORT void decancer_matcher_utf16_free(decancer_matcher_utf16_t matcher)
Frees the UTF-16 matcher iterator object created by decancer_find_utf16.

decancer_cured_raw_utf16_clone
DECANCER_EXPORT decancer_cured_raw_utf16_t decancer_cured_raw_utf16_clone(decancer_cured_raw_utf16_t raw_utf16_handle)
Clones the rust object created by decancer_cured_raw_utf16.

decancer_translation_clone
DECANCER_EXPORT void decancer_translation_clone(const decancer_translation_t *translation_in, decancer_translation_t *translation_out)
Clones the translation struct used in decancer_cure_char.

decancer_matcher_consume
DECANCER_EXPORT decancer_matches_t decancer_matcher_consume(decancer_matcher_t matcher)
Consumes the UTF-8 matcher iterator object created by decancer_find and returns a matches object.

decancer_cure_utf16
DECANCER_EXPORT decancer_cured_t decancer_cure_utf16(const uint16_t *input_str, const size_t input_length, const decancer_options_t options, decancer_error_t *error)
Cures a UTF-16 encoded string.

decancer_matcher_utf16_next
DECANCER_EXPORT bool decancer_matcher_utf16_next(decancer_matcher_utf16_t matcher, decancer_match_t *match)
Iterates to the next element of a UTF-16 matcher iterator.

decancer_contains
DECANCER_EXPORT bool decancer_contains(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size)
Checks if the cured string similarly contains the specified UTF-8 encoded string.

decancer_disable_leetspeak
DECANCER_EXPORT void decancer_disable_leetspeak(decancer_cured_t cured, const bool switch_)
Prevents decancer from applying leetspeak comparisons in comparison methods.

decancer_replace
DECANCER_EXPORT bool decancer_replace(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size, const uint8_t *replacement_str, const size_t replacement_size)
Replaces every similar-looking match of the specified UTF-8 encoded string with another UTF-8 encoded...

decancer_ends_with_utf16
DECANCER_EXPORT bool decancer_ends_with_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length)
Checks if the cured string similarly ends with the specified UTF-16 encoded string.

decancer_translation_free
DECANCER_EXPORT void decancer_translation_free(decancer_translation_t *translation)
Frees the translation struct used in decancer_cure_char.

decancer_cure
DECANCER_EXPORT decancer_cured_t decancer_cure(const uint8_t *input_str, const size_t input_size, const decancer_options_t options, decancer_error_t *error)
Cures a UTF-8 encoded string.

decancer_cure_char
DECANCER_EXPORT void decancer_cure_char(const uint32_t input, const decancer_options_t options, decancer_translation_t *translation)
Cures a single unicode codepoint.

decancer_equals
DECANCER_EXPORT bool decancer_equals(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size)
Checks if the cured string is similar with the specified UTF-8 encoded string.

decancer_ends_with
DECANCER_EXPORT bool decancer_ends_with(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size)
Checks if the cured string similarly ends with the specified UTF-8 encoded string.

decancer_matcher_next
DECANCER_EXPORT bool decancer_matcher_next(decancer_matcher_t matcher, decancer_match_t *match)
Iterates to the next element of a UTF-8 matcher iterator.

decancer_matcher_utf16_t
void * decancer_matcher_utf16_t
Represents a UTF-16 matcher iterator object returned from decancer_find_utf16.
Definition decancer.h:416

decancer_cured_raw
DECANCER_EXPORT const uint8_t * decancer_cured_raw(decancer_cured_t cured, const decancer_match_t *match, size_t *output_size)
Retrieves the raw UTF-8 bytes from a cured string object.

decancer_matcher_free
DECANCER_EXPORT void decancer_matcher_free(decancer_matcher_t matcher)
Frees the UTF-8 matcher iterator object created by decancer_find.

decancer_matcher_utf16_consume
DECANCER_EXPORT decancer_matches_t decancer_matcher_utf16_consume(decancer_matcher_utf16_t matcher)
Consumes the UTF-16 matcher iterator object created by decancer_find_utf16 and returns a matches obje...

decancer_starts_with
DECANCER_EXPORT bool decancer_starts_with(decancer_cured_t cured, const uint8_t *other_str, const size_t other_size)
Checks if the cured string similarly starts with the specified UTF-8 encoded string.

decancer_matches_t
void * decancer_matches_t
Represents a matcher iterator object returned from decancer_find_multiple and decancer_find_multiple_...
Definition decancer.h:431

decancer_censor_utf16
DECANCER_EXPORT bool decancer_censor_utf16(decancer_cured_t cured, const uint16_t *other_str, const size_t other_length, const uint32_t replacement_char)
Censors every similar-looking match of the specified UTF-16 encoded string.

decancer_error_t
Represents an error caused by decancer not being able to cure a string.
Definition decancer.h:300

decancer_error_t::message_length
uint8_t message_length
The length of the error message.
Definition decancer.h:311

decancer_error_t::message
const char * message
Null-terminated ASCII encoded error message.
Definition decancer.h:305

decancer_keyword_t
Represents a UTF-8 encoded keyword. This struct is often used inside an array.
Definition decancer.h:330

decancer_keyword_t::size
size_t size
UTF-8 size of the string, in bytes.
Definition decancer.h:341

decancer_keyword_t::string
const uint8_t * string
UTF-8 encoded string.
Definition decancer.h:335

decancer_keyword_utf16_t
Represents a UTF-16 encoded keyword. This struct is often used inside an array.
Definition decancer.h:361

decancer_keyword_utf16_t::length
size_t length
Length of the UTF-16 encoded string in units of uint16_t – or sizeof(string) / sizeof(uint16_t).
Definition decancer.h:372

decancer_keyword_utf16_t::string
const uint16_t * string
UTF-16 encoded string.
Definition decancer.h:366

decancer_match_t
Represents a match in UTF-8 indices.
Definition decancer.h:535

decancer_match_t::end
size_t end
End of the match in UTF-8 indices (non-inclusive).
Definition decancer.h:546

decancer_match_t::start
size_t start
Start of the match in UTF-8 indices.
Definition decancer.h:540

decancer_translation_t
Represents a translation of a unicode codepoint.
Definition decancer.h:459

decancer_translation_t::character
uint32_t character
The translation, as a unicode character.
Definition decancer.h:478

decancer_translation_t::kind
uint8_t kind
The type of the translation result. This can be any of the following values:
Definition decancer.h:467

decancer_translation_t::__heap
void * __heap
A pointer to a heap memory block, unused.
Definition decancer.h:502

decancer_translation_t::contents
const uint8_t * contents
Raw UTF-8 encoded string.
Definition decancer.h:489

decancer_translation_t::size
size_t size
UTF-8 size of the string, in bytes.
Definition decancer.h:495