/home/runner/work/decancer/decancer/bindings/native/decancer.h

Go to the documentation of this file.

// 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 0x040000

#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_DISABLE_ALPHABETICAL_LEETSPEAK (1 << 3)

#define DECANCER_OPTION_RETAIN_DIACRITICS (1 << 4)

#define DECANCER_OPTION_RETAIN_GREEK (1 << 5)

#define DECANCER_OPTION_RETAIN_CYRILLIC (1 << 6)

#define DECANCER_OPTION_RETAIN_HEBREW (1 << 7)

#define DECANCER_OPTION_RETAIN_ARABIC (1 << 8)

#define DECANCER_OPTION_RETAIN_DEVANAGARI (1 << 9)

#define DECANCER_OPTION_RETAIN_BENGALI (1 << 10)

#define DECANCER_OPTION_RETAIN_ARMENIAN (1 << 11)

#define DECANCER_OPTION_RETAIN_GUJARATI (1 << 12)

#define DECANCER_OPTION_RETAIN_TAMIL (1 << 13)

#define DECANCER_OPTION_RETAIN_THAI (1 << 14)

#define DECANCER_OPTION_RETAIN_LAO (1 << 15)

#define DECANCER_OPTION_RETAIN_BURMESE (1 << 16)

#define DECANCER_OPTION_RETAIN_KHMER (1 << 17)

#define DECANCER_OPTION_RETAIN_MONGOLIAN (1 << 18)

#define DECANCER_OPTION_RETAIN_CHINESE (1 << 19)

#define DECANCER_OPTION_RETAIN_JAPANESE (1 << 20)

#define DECANCER_OPTION_RETAIN_KOREAN (1 << 21)

#define DECANCER_OPTION_RETAIN_BRAILLE (1 << 22)

#define DECANCER_OPTION_RETAIN_EMOJIS (1 << 23)

#define DECANCER_OPTION_RETAIN_TURKISH (1 << 24)

#define DECANCER_OPTION_ASCII_ONLY (1 << 25)

#define DECANCER_OPTION_ALPHANUMERIC_ONLY (1 << 26)

#define DECANCER_OPTION_ALL 0x7ffffff

#define DECANCER_OPTION_PURE_HOMOGLYPH 0xfffff0

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;
 
#ifndef DECANCER_NO_SUGGESTIONS
#define DECANCER_SUGGESTION_REMOVED_INDEX ((size_t)0xffffffff)

typedef struct {
  size_t old_index;

  size_t new_index;

  decancer_translation_t translation;
} decancer_suggestion_t;
#endif

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

  DECANCER_EXPORT void decancer_disable_alphabetical_leetspeak(decancer_cured_t cured, const bool switch_);
#endif
 
#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_NO_SUGGESTIONS
  DECANCER_EXPORT size_t decancer_get_suggestion_length(decancer_cured_t cured);

  DECANCER_EXPORT void decancer_get_suggestion(decancer_cured_t cured, const size_t index, decancer_suggestion_t* suggestion);
#endif
 
#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