libhl/source/hl.h

/* hl.h
 * Copyright 2023 Anon Anonson, Ognjen 'xolatile' Milan Robovic, Emil Williams
 * SPDX Identifier: GPL-3.0-only / NO WARRANTY / NO GUARANTEE */

#include <stdio.h>
#include <uthash.h>
#include <ctype.h>
#include <string.h>
#include "chad.h"
#include "regex.h"

// -------------------
// ### Definitions ###
// -------------------

typedef enum {
	KEYSYMBOL,
	KEYWORD,
	MATCH,
	REGION
} token_type_t;

typedef void (*attribute_callback_t) (const char * const string,
                                      const int          length,
                                            void * const attributes);

typedef struct {
	char                 * key;
	attribute_callback_t   callback;
	UT_hash_handle         hh;
} display_t;

typedef struct {
	void              * attributes;
	struct hl_group_t * link;
} hl_group_t;

typedef struct {
	hl_group_t   * hl;
	token_type_t   t;
	regex_t      * syntax;
} token_t;

// GLOBALS

token_t * token_table[1000];
int token_table_top = 0;

display_t * display_table = NULL;

// --------------------------------
// ### Constructors/Destructors ###
// --------------------------------

void new_display_mode(display_t * mode) {
	HASH_ADD_STR(display_table,
	             key,
	             mode);
}

int free_token(token_t * token) {
	free(token->hl);
	regex_free(token->syntax);

	return 0;
}

int append_token(token_t * token) {
	token_table[token_table_top++] = token;

	return 0;
}

token_t * new_symbol_token(const char         * const c,
                                 hl_group_t   * const g) {
	char * new_word;
	if (is_magic(*c)) {
		new_word = (char *)malloc(sizeof(char)*3);
		new_word[0] = '\\';
		new_word[1] = *c;
		new_word[2] = '\00';
	} else {
		new_word = strdup(c);
	}

	token_t * mt = (token_t*)malloc(sizeof(token_t));

	mt->hl     = g;
	mt->t      = KEYSYMBOL;
	mt->syntax = regex_compile(new_word);

	append_token(mt);

	return mt;
}

int new_symbol_tokens(const char       * const *     symbols,
                            hl_group_t * const             g) {
	int i = 0;

	while (*symbols) {
		if(new_symbol_token(*symbols, g)) {
			++i;
		}
		++symbols;
	}

	return i;
}

int new_char_tokens(const char       *       characters,
                          hl_group_t * const          g) {
	int  i         = 0;
	char buffer[2] = "";

	buffer[1] = '\00';

	for(const char * s = characters; *s != '\0'; s++) {
		buffer[0] = *s;
		if(new_symbol_token(buffer, g)) {
			++i;
		}
	}

	return i;
}

token_t * new_keyword_token(const char         * const word,
                                  hl_group_t   * const    g) {
	size_t   word_length = strlen(word);
	char   * new_word    = (char*)malloc(word_length + 4 + 1);

	memcpy(new_word, "\\<", 2);
	memcpy(new_word + 2, word, word_length);
	strcpy(new_word + 2 + word_length, "\\>");

	token_t * mt = (token_t*)malloc(sizeof(token_t));

	mt->hl     = g;
	mt->t      = KEYWORD;
	mt->syntax = regex_compile(new_word);

	append_token(mt);

	return mt;
}

int new_keyword_tokens(const char       * const * words,
                             hl_group_t * const   g) {
	int i = 0;

	while (*words) {
		if(new_keyword_token(*words, g)) {
			++i;
		}
		++words;
	}

	return i;
}

token_t * new_token(const char         * const word,
                    const token_type_t            t,
                          hl_group_t   * const    g) {
	switch (t) {
		case KEYSYMBOL: {
			return new_symbol_token(word, g);
		}
		case KEYWORD: {
			return new_keyword_token(word, g);
		}
		case MATCH: {
		} break;
		case REGION: {
		} break;
	}

	return NULL;
}

// --------------------
// ### Highlighting ###
// --------------------

int token_fits(const token_t * const       token,
               const char    * const       to,
               const int                   string_offset,
			         int     *             match_offset) {

	//return regex_match(pattern, to, string_offset, match_offset);
	return regex_search(token->syntax, to + string_offset);
}

void render_string(const char * const string,
                   const char * const mode) {
	for (const char * s = string; *s != '\00';) {
		int f;
		int token_index = 0;
		int offset;

		for (; token_index < token_table_top; token_index++) {
			f = token_fits(token_table[token_index], string, s - string, &offset);
			if (f) {
				break;
			}
		}
		//
		display_t * display;
		HASH_FIND_STR(display_table,
		              mode,
		              display);
		//
		if (f) {
			for (int i = 0; i < offset; i++) {
				display->callback(s + i,
				                  0,
				                  token_table[token_index]->hl->attributes);
			}
			display->callback(s + offset,
			                  f,
			                  token_table[token_index]->hl->attributes);
			s += f + offset;
		} else {
			display->callback(s,
			                  0,
			                  NULL);
			++s;
		}
	}
}

// -------------------------
// ### Library Mangement ###
// -------------------------
hl_group_t * keyword_hl = NULL;
hl_group_t * preprocessor_hl = NULL;
hl_group_t * symbol_hl = NULL;

int hl_init(void) {
	return 0;
}

int hl_deinit(void) {
	for (int i = 0; i < token_table_top; i++) {
		free_token(token_table[i]);
	}

	return 0;
}