Add keyword support
This commit is contained in:
@@ -3,28 +3,28 @@
|
||||
#include <ctype.h>
|
||||
#include <stdbool.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
|
||||
#define MAX_KEYWORD_LENGTH 5
|
||||
#define MAX_STACK_CAPACITY 1024
|
||||
|
||||
// clang-format off
|
||||
typedef enum {
|
||||
// GENERAL STATES
|
||||
// GENERAL STATES
|
||||
LEXER_STATE_START,
|
||||
LEXER_STATE_ERROR,
|
||||
LEXER_STATE_VALUE,
|
||||
LEXER_STATE_KEYWORD,
|
||||
// COLLECTION STATES
|
||||
LEXER_STATE_OBJECT_START,
|
||||
LEXER_STATE_OBJECT,
|
||||
LEXER_STATE_OBJECT_END,
|
||||
LEXER_STATE_ARRAY_START,
|
||||
LEXER_STATE_ARRAY,
|
||||
LEXER_STATE_ARRAY_END,
|
||||
LEXER_STATE_LAST_COLLECTION,
|
||||
// OBJECT STATES
|
||||
LEXER_STATE_KEY,
|
||||
// NUMBER STATES
|
||||
// COLLECTION STATES
|
||||
LEXER_STATE_OBJECT_START,
|
||||
LEXER_STATE_OBJECT,
|
||||
LEXER_STATE_OBJECT_END,
|
||||
LEXER_STATE_ARRAY_START,
|
||||
LEXER_STATE_ARRAY,
|
||||
LEXER_STATE_ARRAY_END,
|
||||
LEXER_STATE_LAST_COLLECTION,
|
||||
// OBJECT STATES
|
||||
LEXER_STATE_KEY,
|
||||
// NUMBER STATES
|
||||
LEXER_STATE_DECIMAL,
|
||||
LEXER_STATE_NUMBER,
|
||||
LEXER_STATE_FRACTION,
|
||||
@@ -32,32 +32,45 @@ typedef enum {
|
||||
LEXER_STATE_EXP_SIGN,
|
||||
LEXER_STATE_POWER,
|
||||
LEXER_STATE_NUMBER_END,
|
||||
// STRING STATES
|
||||
LEXER_STATE_STRING,
|
||||
LEXER_STATE_STRING_END,
|
||||
LEXER_STATE_ESCAPE_SEQUENCE,
|
||||
LEXER_STATE_UNICODE_HEX,
|
||||
// STRING STATES
|
||||
LEXER_STATE_STRING,
|
||||
LEXER_STATE_STRING_END,
|
||||
LEXER_STATE_ESCAPE_SEQUENCE,
|
||||
LEXER_STATE_UNICODE_HEX,
|
||||
// KEYWORD STATES
|
||||
LEXER_STATE_TRUE,
|
||||
LEXER_STATE_FALSE,
|
||||
LEXER_STATE_NULL,
|
||||
LEXER_STATE_KEYWORD_END,
|
||||
|
||||
COUNT_LEXER_STATES,
|
||||
} lexer_state_t;
|
||||
// clang-format on
|
||||
|
||||
typedef struct {
|
||||
lexer_state_t stack[MAX_STACK_CAPACITY];
|
||||
u64 size;
|
||||
} state_stack_t;
|
||||
|
||||
typedef struct {
|
||||
u64 size;
|
||||
char str[MAX_KEYWORD_LENGTH + 1];
|
||||
} keyword_t;
|
||||
|
||||
struct lexer {
|
||||
lexer_state_t current;
|
||||
state_stack_t stack;
|
||||
u64 line;
|
||||
u64 column;
|
||||
char current_keyword[MAX_KEYWORD_LENGTH + 1];
|
||||
keyword_t keyword;
|
||||
};
|
||||
|
||||
void stack_push(state_stack_t *stack, lexer_state_t value);
|
||||
lexer_state_t stack_pop(state_stack_t *stack);
|
||||
|
||||
void append_to_keyword(keyword_t *kw, char input);
|
||||
void clear_keyword(keyword_t *kw);
|
||||
bool strequal(const char *first, const char *second);
|
||||
|
||||
void lexer_state_machine(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_lexer_start(char input);
|
||||
lexer_state_t handle_last_collection(char input);
|
||||
@@ -76,6 +89,11 @@ lexer_state_t handle_exponent(char input);
|
||||
lexer_state_t handle_exp_sign(char input);
|
||||
lexer_state_t handle_power(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_number_end(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_keyword(char input);
|
||||
lexer_state_t handle_true(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_false(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_null(lexer_t *lexer, char input);
|
||||
lexer_state_t handle_keyword_end(lexer_t *lexer, char input);
|
||||
|
||||
bool validate_json(char *json) {
|
||||
lexer_t lexer = {0};
|
||||
@@ -112,6 +130,23 @@ lexer_state_t stack_pop(state_stack_t *stack) {
|
||||
return state;
|
||||
}
|
||||
|
||||
void append_to_keyword(keyword_t *kw, char input) {
|
||||
if (kw->size + 1 > MAX_KEYWORD_LENGTH) {
|
||||
return;
|
||||
}
|
||||
|
||||
kw->str[(kw->size)++] = input;
|
||||
}
|
||||
|
||||
void clear_keyword(keyword_t *kw) {
|
||||
memset(kw->str, 0, MAX_KEYWORD_LENGTH + 1);
|
||||
kw->size = 0;
|
||||
}
|
||||
|
||||
bool strequal(const char *first, const char *second) {
|
||||
return strcmp(first, second) == 0;
|
||||
}
|
||||
|
||||
void lexer_state_machine(lexer_t *lexer, char input) {
|
||||
switch (lexer->current) {
|
||||
case LEXER_STATE_START:
|
||||
@@ -120,19 +155,17 @@ void lexer_state_machine(lexer_t *lexer, char input) {
|
||||
case LEXER_STATE_VALUE:
|
||||
lexer->current = handle_value(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_KEYWORD:
|
||||
break;
|
||||
case LEXER_STATE_OBJECT_START:
|
||||
stack_push(&(lexer->stack), LEXER_STATE_OBJECT);
|
||||
// break is left intentionally here to utilise the fallthrough behaviour of
|
||||
// the switch statement
|
||||
// break is left out intentionally here to utilise the fallthrough behaviour
|
||||
// of the switch statement
|
||||
case LEXER_STATE_OBJECT:
|
||||
lexer->current = handle_object(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_ARRAY_START:
|
||||
stack_push(&(lexer->stack), LEXER_STATE_ARRAY);
|
||||
// break is left intentionally here to utilise the fallthrough behaviour of
|
||||
// the switch statement
|
||||
// break is left out intentionally here to utilise the fallthrough behaviour
|
||||
// of the switch statement
|
||||
case LEXER_STATE_ARRAY:
|
||||
lexer->current = handle_array(lexer, input);
|
||||
break;
|
||||
@@ -182,6 +215,18 @@ void lexer_state_machine(lexer_t *lexer, char input) {
|
||||
break;
|
||||
case LEXER_STATE_UNICODE_HEX:
|
||||
break;
|
||||
case LEXER_STATE_TRUE:
|
||||
lexer->current = handle_true(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_FALSE:
|
||||
lexer->current = handle_false(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_NULL:
|
||||
lexer->current = handle_null(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_KEYWORD_END:
|
||||
lexer->current = handle_keyword_end(lexer, input);
|
||||
break;
|
||||
case LEXER_STATE_LAST_COLLECTION:
|
||||
lexer->current = handle_last_collection(input);
|
||||
break;
|
||||
@@ -282,7 +327,9 @@ lexer_state_t handle_value(lexer_t *lexer, char input) {
|
||||
case 't':
|
||||
case 'f':
|
||||
case 'n':
|
||||
return LEXER_STATE_KEYWORD;
|
||||
append_to_keyword(&(lexer->keyword), input);
|
||||
|
||||
return handle_keyword(input);
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
@@ -426,3 +473,92 @@ lexer_state_t handle_number_end(lexer_t *lexer, char input) {
|
||||
return collection_end ? handle_collection_end(lexer, input)
|
||||
: LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_keyword(char input) {
|
||||
switch (input) {
|
||||
case 't':
|
||||
return LEXER_STATE_TRUE;
|
||||
case 'f':
|
||||
return LEXER_STATE_FALSE;
|
||||
case 'n':
|
||||
return LEXER_STATE_NULL;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_true(lexer_t *lexer, char input) {
|
||||
char current[MAX_KEYWORD_LENGTH + 1];
|
||||
strcpy(current, lexer->keyword.str);
|
||||
|
||||
append_to_keyword(&(lexer->keyword), input);
|
||||
|
||||
bool return_state_true = (strequal(current, "t") && input == 'r') ||
|
||||
(strequal(current, "tr") && input == 'u');
|
||||
|
||||
bool return_state_end = strequal(current, "tru") && input == 'e';
|
||||
|
||||
if (return_state_true) {
|
||||
return LEXER_STATE_TRUE;
|
||||
} else if (return_state_end) {
|
||||
return LEXER_STATE_KEYWORD_END;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_false(lexer_t *lexer, char input) {
|
||||
char current[MAX_KEYWORD_LENGTH + 1];
|
||||
strcpy(current, lexer->keyword.str);
|
||||
|
||||
append_to_keyword(&(lexer->keyword), input);
|
||||
|
||||
bool return_state_false = (strequal(current, "f") && input == 'a') ||
|
||||
(strequal(current, "fa") && input == 'l') ||
|
||||
(strequal(current, "fal") && input == 's');
|
||||
|
||||
bool return_state_end = strequal(current, "fals") && input == 'e';
|
||||
|
||||
if (return_state_false) {
|
||||
return LEXER_STATE_FALSE;
|
||||
} else if (return_state_end) {
|
||||
return LEXER_STATE_KEYWORD_END;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_null(lexer_t *lexer, char input) {
|
||||
char current[MAX_KEYWORD_LENGTH + 1];
|
||||
strcpy(current, lexer->keyword.str);
|
||||
|
||||
append_to_keyword(&(lexer->keyword), input);
|
||||
|
||||
bool return_state_null = (strequal(current, "n") && input == 'u') ||
|
||||
(strequal(current, "nu") && input == 'l');
|
||||
|
||||
bool return_state_end = strequal(current, "nul") && input == 'l';
|
||||
|
||||
if (return_state_null) {
|
||||
return LEXER_STATE_NULL;
|
||||
} else if (return_state_end) {
|
||||
return LEXER_STATE_KEYWORD_END;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_keyword_end(lexer_t *lexer, char input) {
|
||||
clear_keyword(&(lexer->keyword));
|
||||
|
||||
if (isspace(input)) {
|
||||
return LEXER_STATE_KEYWORD_END;
|
||||
} else if (input == ',') {
|
||||
return lexer->stack.stack[lexer->stack.size - 1];
|
||||
}
|
||||
|
||||
bool collection_end = input == '}' || input == ']';
|
||||
|
||||
return collection_end ? handle_collection_end(lexer, input)
|
||||
: LEXER_STATE_ERROR;
|
||||
}
|
||||
|
Reference in New Issue
Block a user