Compare commits
21 Commits
main
...
896a9ff085
| Author | SHA1 | Date | |
|---|---|---|---|
| 896a9ff085 | |||
| eece52600b | |||
| 7cb4b28da8 | |||
| bf958056bd | |||
| ec6df32839 | |||
| de57d9f14b | |||
| f1f90cf56b | |||
| 41d7cf2717 | |||
| 002a8e7064 | |||
| fc80095d1f | |||
| 642d000ebd | |||
| b4a4865870 | |||
| 1b4d6a6156 | |||
| 26fabd91b9 | |||
| adde1db0f4 | |||
| 7eec0e0730 | |||
| 6c454fbb8e | |||
| fd8231b02d | |||
| 09f2315b66 | |||
| 91162654b3 | |||
| 6e93d3ecd1 |
1
.gitignore
vendored
1
.gitignore
vendored
@@ -3,3 +3,4 @@
|
||||
src/ignore/**/*
|
||||
main
|
||||
compile_commands.json
|
||||
gentable
|
||||
|
||||
28
compile
28
compile
@@ -1,24 +1,28 @@
|
||||
#!/bin/bash
|
||||
|
||||
CC=clang
|
||||
CFLAGS="-g -Wall -Werror -pedantic -Iinclude"
|
||||
SRC="src/main.c"
|
||||
OUT=main
|
||||
CFLAGS="-g -Wall -Werror -pedantic -I./include"
|
||||
SRC="./src/main.c"
|
||||
OUT=./main
|
||||
|
||||
# STATE_TABLE
|
||||
(set -x;$CC $CFLAGS -I./include/lexer ./generate_state_table.c -o ./gentable)
|
||||
./gentable
|
||||
|
||||
# DSTRING
|
||||
CFLAGS+=" -Iinclude/dstring"
|
||||
SRC+=" src/dstring/*.c"
|
||||
CFLAGS+=" -I./include/dstring"
|
||||
SRC+=" ./src/dstring/*.c"
|
||||
|
||||
# JSON_ENTITIES
|
||||
CFLAGS+=" -Iinclude/json_entities"
|
||||
SRC+=" src/json_entities/*.c"
|
||||
CFLAGS+=" -I./include/json_entities"
|
||||
SRC+=" ./src/json_entities/*.c"
|
||||
|
||||
# LEXER
|
||||
CFLAGS+=" -Iinclude/lexer"
|
||||
SRC+=" src/lexer/*.c"
|
||||
CFLAGS+=" -I./include/lexer"
|
||||
SRC+=" ./src/lexer/*.c"
|
||||
|
||||
# PARSER
|
||||
CFLAGS+=" -Iinclude/parser"
|
||||
SRC+=" src/parser/*.c"
|
||||
CFLAGS+=" -I./include/parser"
|
||||
SRC+=" ./src/parser/*.c"
|
||||
|
||||
$CC $CFLAGS $SRC -o $OUT
|
||||
(set -x;$CC $CFLAGS $SRC -o $OUT)
|
||||
|
||||
639
generate_state_table.c
Normal file
639
generate_state_table.c
Normal file
@@ -0,0 +1,639 @@
|
||||
#include "aliases.h"
|
||||
#include "lexer_data.h"
|
||||
#include <ctype.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
|
||||
#define STRING_BUF_LENGTH 100
|
||||
|
||||
INTERNAL void fill_table(void);
|
||||
INTERNAL void write_table(FILE *fp);
|
||||
|
||||
INTERNAL void set_column_width(lexer_state_t state);
|
||||
INTERNAL const char *get_input_string(lexer_input_t input);
|
||||
INTERNAL const char *get_state_string(lexer_state_t state);
|
||||
|
||||
INTERNAL lexer_state_t lexer_state_machine(lexer_state_t state,
|
||||
lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_lexer_start(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_last_collection(char input);
|
||||
INTERNAL lexer_state_t handle_object(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_array(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_key(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_value(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_string(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_string_end(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_escape_sequence(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_unicode_hex(lexer_input_t input,
|
||||
lexer_state_t return_state);
|
||||
INTERNAL lexer_state_t handle_decimal(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_number(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_fraction(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_exponent(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_exp_sign(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_power(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_number_end(lexer_input_t input);
|
||||
INTERNAL lexer_state_t handle_true(lexer_input_t input,
|
||||
lexer_state_t start_state);
|
||||
INTERNAL lexer_state_t handle_false(lexer_input_t input,
|
||||
lexer_state_t start_state);
|
||||
INTERNAL lexer_state_t handle_null(lexer_input_t input,
|
||||
lexer_state_t start_state);
|
||||
INTERNAL lexer_state_t handle_keyword_end(lexer_input_t input);
|
||||
|
||||
INTERNAL i32 column_width = 0;
|
||||
|
||||
INTERNAL lexer_state_t state_table[COUNT_LEXER_STATES][COUNT_LEXER_INPUTS] = {
|
||||
0};
|
||||
|
||||
INTERNAL const char *input_strings[COUNT_LEXER_STATES] = {
|
||||
[LEXER_INPUT_NEWLINE] = "NEWLINE",
|
||||
[LEXER_INPUT_WHITE_SPACE] = "WHITE_SPACE",
|
||||
[LEXER_INPUT_OPEN_BRACE] = "OPEN_BRACE",
|
||||
[LEXER_INPUT_CLOSE_BRACE] = "CLOSE_BRACE",
|
||||
[LEXER_INPUT_OPEN_BRACKET] = "OPEN_BRACKET",
|
||||
[LEXER_INPUT_CLOSE_BRACKET] = "CLOSE_BRACKET",
|
||||
[LEXER_INPUT_COMMA] = "COMMA",
|
||||
[LEXER_INPUT_COLON] = "COLON",
|
||||
[LEXER_INPUT_DOUBLE_QUOTE] = "DOUBLE_QUOTE",
|
||||
[LEXER_INPUT_BACK_SLASH] = "BACK_SLASH",
|
||||
[LEXER_INPUT_FORWARD_SLASH] = "FORWARD_SLASH",
|
||||
[LEXER_INPUT_LOWER_A] = "LOWER_A",
|
||||
[LEXER_INPUT_LOWER_B] = "LOWER_B",
|
||||
[LEXER_INPUT_LOWER_C] = "LOWER_C",
|
||||
[LEXER_INPUT_LOWER_D] = "LOWER_D",
|
||||
[LEXER_INPUT_LOWER_E] = "LOWER_E",
|
||||
[LEXER_INPUT_LOWER_F] = "LOWER_F",
|
||||
[LEXER_INPUT_LOWER_L] = "LOWER_L",
|
||||
[LEXER_INPUT_LOWER_N] = "LOWER_N",
|
||||
[LEXER_INPUT_LOWER_R] = "LOWER_R",
|
||||
[LEXER_INPUT_LOWER_S] = "LOWER_S",
|
||||
[LEXER_INPUT_LOWER_T] = "LOWER_T",
|
||||
[LEXER_INPUT_LOWER_U] = "LOWER_U",
|
||||
[LEXER_INPUT_UPPER_A] = "UPPER_A",
|
||||
[LEXER_INPUT_UPPER_B] = "UPPER_B",
|
||||
[LEXER_INPUT_UPPER_C] = "UPPER_C",
|
||||
[LEXER_INPUT_UPPER_D] = "UPPER_D",
|
||||
[LEXER_INPUT_UPPER_E] = "UPPER_E",
|
||||
[LEXER_INPUT_UPPER_F] = "UPPER_F",
|
||||
[LEXER_INPUT_MINUS] = "MINUS",
|
||||
[LEXER_INPUT_PLUS] = "PLUS",
|
||||
[LEXER_INPUT_DECIMAL] = "DECIMAL",
|
||||
[LEXER_INPUT_ZERO] = "ZERO",
|
||||
[LEXER_INPUT_NON_ZERO] = "NON_ZERO",
|
||||
[LEXER_INPUT_OTHER] = "OTHER",
|
||||
};
|
||||
|
||||
INTERNAL const char *state_strings[COUNT_LEXER_STATES] = {
|
||||
[LEXER_STATE_ERROR] = "LEXER_STATE_ERROR",
|
||||
[LEXER_STATE_START] = "LEXER_STATE_START",
|
||||
[LEXER_STATE_VALUE] = "LEXER_STATE_VALUE",
|
||||
[LEXER_STATE_VALUE_END] = "LEXER_STATE_VALUE_END",
|
||||
[LEXER_STATE_OBJECT_START] = "LEXER_STATE_OBJECT_START",
|
||||
[LEXER_STATE_OBJECT] = "LEXER_STATE_OBJECT",
|
||||
[LEXER_STATE_OBJECT_END] = "LEXER_STATE_OBJECT_END",
|
||||
[LEXER_STATE_ARRAY_START] = "LEXER_STATE_ARRAY_START",
|
||||
[LEXER_STATE_ARRAY] = "LEXER_STATE_ARRAY",
|
||||
[LEXER_STATE_ARRAY_END] = "LEXER_STATE_ARRAY_END",
|
||||
[LEXER_STATE_LAST_COLLECTION] = "LEXER_STATE_LAST_COLLECTION",
|
||||
[LEXER_STATE_KEY] = "LEXER_STATE_KEY",
|
||||
[LEXER_STATE_KEY_END] = "LEXER_STATE_KEY_END",
|
||||
[LEXER_STATE_DECIMAL] = "LEXER_STATE_DECIMAL",
|
||||
[LEXER_STATE_NUMBER] = "LEXER_STATE_NUMBER",
|
||||
[LEXER_STATE_FRACTION] = "LEXER_STATE_FRACTION",
|
||||
[LEXER_STATE_EXPONENT] = "LEXER_STATE_EXPONENT",
|
||||
[LEXER_STATE_EXP_SIGN] = "LEXER_STATE_EXP_SIGN",
|
||||
[LEXER_STATE_POWER] = "LEXER_STATE_POWER",
|
||||
[LEXER_STATE_NUMBER_END] = "LEXER_STATE_NUMBER_END",
|
||||
[LEXER_STATE_STRING] = "LEXER_STATE_STRING",
|
||||
[LEXER_STATE_STRING_END] = "LEXER_STATE_STRING_END",
|
||||
[LEXER_STATE_ESCAPE_SEQUENCE] = "LEXER_STATE_ESCAPE_SEQUENCE",
|
||||
[LEXER_STATE_UNICODE_HEX1] = "LEXER_STATE_UNICODE_HEX1",
|
||||
[LEXER_STATE_UNICODE_HEX2] = "LEXER_STATE_UNICODE_HEX2",
|
||||
[LEXER_STATE_UNICODE_HEX3] = "LEXER_STATE_UNICODE_HEX3",
|
||||
[LEXER_STATE_UNICODE_HEX4] = "LEXER_STATE_UNICODE_HEX4",
|
||||
[LEXER_STATE_T] = "LEXER_STATE_T",
|
||||
[LEXER_STATE_TR] = "LEXER_STATE_TR",
|
||||
[LEXER_STATE_TRU] = "LEXER_STATE_TRU",
|
||||
[LEXER_STATE_TRUE] = "LEXER_STATE_TRUE",
|
||||
[LEXER_STATE_F] = "LEXER_STATE_F",
|
||||
[LEXER_STATE_FA] = "LEXER_STATE_FA",
|
||||
[LEXER_STATE_FAL] = "LEXER_STATE_FAL",
|
||||
[LEXER_STATE_FALS] = "LEXER_STATE_FALS",
|
||||
[LEXER_STATE_FALSE] = "LEXER_STATE_FALSE",
|
||||
[LEXER_STATE_N] = "LEXER_STATE_N",
|
||||
[LEXER_STATE_NU] = "LEXER_STATE_NU",
|
||||
[LEXER_STATE_NUL] = "LEXER_STATE_NUL",
|
||||
[LEXER_STATE_NULL] = "LEXER_STATE_NULL",
|
||||
[LEXER_STATE_KEYWORD_END] = "LEXER_STATE_KEYWORD_END",
|
||||
};
|
||||
|
||||
INTERNAL const char *filename = "./include/lexer/lexer_state_transitions.table";
|
||||
|
||||
int main(void) {
|
||||
FILE *fp = fopen(filename, "w");
|
||||
if (!fp) {
|
||||
printf("Failed to open file\n");
|
||||
|
||||
return EXIT_FAILURE;
|
||||
}
|
||||
|
||||
fill_table();
|
||||
|
||||
write_table(fp);
|
||||
|
||||
fclose(fp);
|
||||
|
||||
return EXIT_SUCCESS;
|
||||
}
|
||||
|
||||
INTERNAL void fill_table(void) {
|
||||
lexer_state_t state;
|
||||
lexer_input_t input;
|
||||
|
||||
for (u64 i = 0; i < COUNT_LEXER_STATES; ++i) {
|
||||
state = (lexer_state_t)i;
|
||||
|
||||
for (u64 j = 0; j < COUNT_LEXER_INPUTS; ++j) {
|
||||
input = (lexer_input_t)j;
|
||||
|
||||
state_table[i][j] = (lexer_state_t)lexer_state_machine(state, input);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void write_table(FILE *fp) {
|
||||
const char *array_open = "{ ";
|
||||
const u64 array_open_length = strlen(array_open);
|
||||
|
||||
const char *array_close = "},";
|
||||
const u64 array_close_length = strlen(array_close);
|
||||
|
||||
const char *comment = "//";
|
||||
const u64 comment_length = strlen(comment);
|
||||
|
||||
const char *table_header_end = " \n";
|
||||
const u64 table_header_end_length = strlen(table_header_end);
|
||||
|
||||
char output[STRING_BUF_LENGTH] = {0};
|
||||
u64 length = 0;
|
||||
|
||||
for (u64 i = 0; i < COUNT_LEXER_STATES; ++i) {
|
||||
set_column_width((lexer_state_t)i);
|
||||
}
|
||||
|
||||
fwrite(comment, sizeof(char), comment_length, fp);
|
||||
|
||||
i32 column_white_space = 0;
|
||||
|
||||
for (u64 i = 0; i < COUNT_LEXER_INPUTS; ++i) {
|
||||
const char *input_str = get_input_string((lexer_input_t)i);
|
||||
column_white_space = (column_width - strlen(input_str)) / 2;
|
||||
|
||||
sprintf(output, "%*s%*s| ", column_width - column_white_space, input_str,
|
||||
column_white_space, " ");
|
||||
length = strlen(output);
|
||||
fwrite(output, sizeof(char), length, fp);
|
||||
}
|
||||
|
||||
fwrite(table_header_end, sizeof(char), table_header_end_length, fp);
|
||||
|
||||
for (u64 i = 0; i < COUNT_LEXER_STATES; ++i) {
|
||||
fwrite(array_open, sizeof(char), array_open_length, fp);
|
||||
|
||||
for (u64 j = 0; j < COUNT_LEXER_INPUTS; ++j) {
|
||||
sprintf(output, "%*s, ", column_width,
|
||||
get_state_string((lexer_state_t)state_table[i][j]));
|
||||
length = strlen(output);
|
||||
fwrite(output, sizeof(char), length, fp);
|
||||
}
|
||||
|
||||
fwrite(array_close, sizeof(char), array_close_length, fp);
|
||||
|
||||
sprintf(output, " // %s\n", get_state_string((lexer_state_t)i));
|
||||
length = strlen(output);
|
||||
fwrite(output, sizeof(char), length, fp);
|
||||
|
||||
memset(output, 0, STRING_BUF_LENGTH);
|
||||
}
|
||||
}
|
||||
|
||||
INTERNAL void set_column_width(lexer_state_t state) {
|
||||
if (state >= COUNT_LEXER_STATES) {
|
||||
return;
|
||||
}
|
||||
|
||||
const char *output = state_strings[state];
|
||||
i32 length = strlen(output);
|
||||
|
||||
column_width = length > column_width ? length : column_width;
|
||||
}
|
||||
|
||||
INTERNAL const char *get_input_string(lexer_input_t input) {
|
||||
if (input >= COUNT_LEXER_INPUTS) {
|
||||
return "";
|
||||
}
|
||||
|
||||
return input_strings[input];
|
||||
}
|
||||
|
||||
INTERNAL const char *get_state_string(lexer_state_t state) {
|
||||
if (state >= COUNT_LEXER_STATES) {
|
||||
return "";
|
||||
}
|
||||
|
||||
return state_strings[state];
|
||||
}
|
||||
|
||||
lexer_state_t lexer_state_machine(lexer_state_t state, lexer_input_t input) {
|
||||
switch (state) {
|
||||
case LEXER_STATE_START:
|
||||
return handle_lexer_start(input);
|
||||
case LEXER_STATE_VALUE:
|
||||
return handle_value(input);
|
||||
case LEXER_STATE_OBJECT:
|
||||
return handle_object(input);
|
||||
case LEXER_STATE_ARRAY:
|
||||
return handle_array(input);
|
||||
case LEXER_STATE_KEY:
|
||||
return handle_key(input);
|
||||
case LEXER_STATE_DECIMAL:
|
||||
return handle_decimal(input);
|
||||
case LEXER_STATE_NUMBER:
|
||||
return handle_number(input);
|
||||
case LEXER_STATE_FRACTION:
|
||||
return handle_fraction(input);
|
||||
case LEXER_STATE_EXPONENT:
|
||||
return handle_exponent(input);
|
||||
case LEXER_STATE_EXP_SIGN:
|
||||
return handle_exp_sign(input);
|
||||
case LEXER_STATE_POWER:
|
||||
return handle_power(input);
|
||||
case LEXER_STATE_NUMBER_END:
|
||||
return handle_number_end(input);
|
||||
case LEXER_STATE_STRING:
|
||||
return handle_string(input);
|
||||
case LEXER_STATE_STRING_END:
|
||||
return handle_string_end(input);
|
||||
case LEXER_STATE_ESCAPE_SEQUENCE:
|
||||
return handle_escape_sequence(input);
|
||||
case LEXER_STATE_UNICODE_HEX1:
|
||||
return handle_unicode_hex(input, LEXER_STATE_UNICODE_HEX2);
|
||||
case LEXER_STATE_UNICODE_HEX2:
|
||||
return handle_unicode_hex(input, LEXER_STATE_UNICODE_HEX3);
|
||||
case LEXER_STATE_UNICODE_HEX3:
|
||||
return handle_unicode_hex(input, LEXER_STATE_UNICODE_HEX4);
|
||||
case LEXER_STATE_UNICODE_HEX4:
|
||||
return handle_unicode_hex(input, LEXER_STATE_STRING);
|
||||
case LEXER_STATE_T:
|
||||
return handle_true(input, LEXER_STATE_T);
|
||||
case LEXER_STATE_TR:
|
||||
return handle_true(input, LEXER_STATE_TR);
|
||||
case LEXER_STATE_TRU:
|
||||
return handle_true(input, LEXER_STATE_TRU);
|
||||
case LEXER_STATE_F:
|
||||
return handle_false(input, LEXER_STATE_F);
|
||||
case LEXER_STATE_FA:
|
||||
return handle_false(input, LEXER_STATE_FA);
|
||||
case LEXER_STATE_FAL:
|
||||
return handle_false(input, LEXER_STATE_FAL);
|
||||
case LEXER_STATE_FALS:
|
||||
return handle_false(input, LEXER_STATE_FALS);
|
||||
case LEXER_STATE_N:
|
||||
return handle_null(input, LEXER_STATE_N);
|
||||
case LEXER_STATE_NU:
|
||||
return handle_null(input, LEXER_STATE_NU);
|
||||
case LEXER_STATE_NUL:
|
||||
return handle_null(input, LEXER_STATE_NUL);
|
||||
case LEXER_STATE_KEYWORD_END:
|
||||
return handle_keyword_end(input);
|
||||
case LEXER_STATE_LAST_COLLECTION:
|
||||
return handle_last_collection(input);
|
||||
case LEXER_STATE_TRUE:
|
||||
case LEXER_STATE_FALSE:
|
||||
case LEXER_STATE_NULL:
|
||||
case LEXER_STATE_OBJECT_START:
|
||||
case LEXER_STATE_ARRAY_START:
|
||||
case LEXER_STATE_OBJECT_END:
|
||||
case LEXER_STATE_ARRAY_END:
|
||||
case LEXER_STATE_KEY_END:
|
||||
case LEXER_STATE_VALUE_END:
|
||||
case LEXER_STATE_ERROR:
|
||||
case COUNT_LEXER_STATES:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_lexer_start(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_START;
|
||||
case LEXER_INPUT_OPEN_BRACE:
|
||||
return LEXER_STATE_OBJECT_START;
|
||||
case LEXER_INPUT_OPEN_BRACKET:
|
||||
return LEXER_STATE_ARRAY_START;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_last_collection(char input) {
|
||||
if (input == LEXER_INPUT_NEWLINE || input == LEXER_INPUT_WHITE_SPACE) {
|
||||
return LEXER_STATE_LAST_COLLECTION;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_object(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_OBJECT;
|
||||
case LEXER_INPUT_DOUBLE_QUOTE:
|
||||
return LEXER_STATE_KEY;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_array(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_ARRAY;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
default:
|
||||
return handle_value(input);
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_key(lexer_input_t input) { return LEXER_STATE_STRING; }
|
||||
|
||||
lexer_state_t handle_value(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_VALUE;
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
case LEXER_INPUT_MINUS:
|
||||
return LEXER_STATE_NUMBER;
|
||||
case LEXER_INPUT_ZERO:
|
||||
return LEXER_STATE_DECIMAL;
|
||||
case LEXER_INPUT_DOUBLE_QUOTE:
|
||||
return LEXER_STATE_STRING;
|
||||
case LEXER_INPUT_OPEN_BRACE:
|
||||
return LEXER_STATE_OBJECT_START;
|
||||
case LEXER_INPUT_OPEN_BRACKET:
|
||||
return LEXER_STATE_ARRAY_START;
|
||||
case LEXER_INPUT_LOWER_T:
|
||||
return LEXER_STATE_T;
|
||||
case LEXER_INPUT_LOWER_F:
|
||||
return LEXER_STATE_F;
|
||||
case LEXER_INPUT_LOWER_N:
|
||||
return LEXER_STATE_N;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_string(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_BACK_SLASH:
|
||||
return LEXER_STATE_ESCAPE_SEQUENCE;
|
||||
case LEXER_INPUT_DOUBLE_QUOTE:
|
||||
return LEXER_STATE_STRING_END;
|
||||
default:
|
||||
return LEXER_STATE_STRING;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_string_end(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_STRING_END;
|
||||
case LEXER_INPUT_COLON:
|
||||
return LEXER_STATE_KEY_END;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_escape_sequence(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_DOUBLE_QUOTE:
|
||||
case LEXER_INPUT_FORWARD_SLASH:
|
||||
case LEXER_INPUT_BACK_SLASH:
|
||||
case LEXER_INPUT_LOWER_B:
|
||||
case LEXER_INPUT_LOWER_F:
|
||||
case LEXER_INPUT_LOWER_N:
|
||||
case LEXER_INPUT_LOWER_R:
|
||||
case LEXER_INPUT_LOWER_T:
|
||||
return LEXER_STATE_STRING;
|
||||
case LEXER_INPUT_LOWER_U:
|
||||
return LEXER_STATE_UNICODE_HEX1;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
INTERNAL lexer_state_t handle_unicode_hex(lexer_input_t input,
|
||||
lexer_state_t return_state) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_LOWER_A:
|
||||
case LEXER_INPUT_LOWER_B:
|
||||
case LEXER_INPUT_LOWER_C:
|
||||
case LEXER_INPUT_LOWER_D:
|
||||
case LEXER_INPUT_LOWER_E:
|
||||
case LEXER_INPUT_LOWER_F:
|
||||
case LEXER_INPUT_UPPER_A:
|
||||
case LEXER_INPUT_UPPER_B:
|
||||
case LEXER_INPUT_UPPER_C:
|
||||
case LEXER_INPUT_UPPER_D:
|
||||
case LEXER_INPUT_UPPER_E:
|
||||
case LEXER_INPUT_UPPER_F:
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return return_state;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_decimal(lexer_input_t input) {
|
||||
if (input == LEXER_INPUT_DECIMAL) {
|
||||
return LEXER_STATE_FRACTION;
|
||||
}
|
||||
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
|
||||
lexer_state_t handle_number(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return LEXER_STATE_NUMBER;
|
||||
case LEXER_INPUT_DECIMAL:
|
||||
return LEXER_STATE_FRACTION;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_NUMBER_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_fraction(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return LEXER_STATE_FRACTION;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
case LEXER_INPUT_LOWER_E:
|
||||
case LEXER_INPUT_UPPER_E:
|
||||
return LEXER_STATE_EXPONENT;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_NUMBER_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_exponent(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return LEXER_STATE_POWER;
|
||||
case LEXER_INPUT_PLUS:
|
||||
case LEXER_INPUT_MINUS:
|
||||
return LEXER_STATE_EXP_SIGN;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_exp_sign(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return LEXER_STATE_POWER;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_power(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_ZERO:
|
||||
case LEXER_INPUT_NON_ZERO:
|
||||
return LEXER_STATE_POWER;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_NUMBER_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_number_end(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_NUMBER_END;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_true(lexer_input_t input, lexer_state_t start_state) {
|
||||
switch (start_state) {
|
||||
case LEXER_STATE_T:
|
||||
return input == LEXER_INPUT_LOWER_R ? LEXER_STATE_TR : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_TR:
|
||||
return input == LEXER_INPUT_LOWER_U ? LEXER_STATE_TRU : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_TRU:
|
||||
return input == LEXER_INPUT_LOWER_E ? LEXER_STATE_TRUE : LEXER_STATE_ERROR;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_false(lexer_input_t input, lexer_state_t start_state) {
|
||||
switch (start_state) {
|
||||
case LEXER_STATE_F:
|
||||
return input == LEXER_INPUT_LOWER_A ? LEXER_STATE_FA : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_FA:
|
||||
return input == LEXER_INPUT_LOWER_L ? LEXER_STATE_FAL : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_FAL:
|
||||
return input == LEXER_INPUT_LOWER_S ? LEXER_STATE_FALS : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_FALS:
|
||||
return input == LEXER_INPUT_LOWER_E ? LEXER_STATE_FALSE : LEXER_STATE_ERROR;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_null(lexer_input_t input, lexer_state_t start_state) {
|
||||
switch (start_state) {
|
||||
case LEXER_STATE_N:
|
||||
return input == LEXER_INPUT_LOWER_U ? LEXER_STATE_NU : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_NU:
|
||||
return input == LEXER_INPUT_LOWER_L ? LEXER_STATE_NUL : LEXER_STATE_ERROR;
|
||||
case LEXER_STATE_NUL:
|
||||
return input == LEXER_INPUT_LOWER_L ? LEXER_STATE_NULL : LEXER_STATE_ERROR;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
|
||||
lexer_state_t handle_keyword_end(lexer_input_t input) {
|
||||
switch (input) {
|
||||
case LEXER_INPUT_NEWLINE:
|
||||
case LEXER_INPUT_WHITE_SPACE:
|
||||
return LEXER_STATE_KEYWORD_END;
|
||||
case LEXER_INPUT_CLOSE_BRACE:
|
||||
return LEXER_STATE_OBJECT_END;
|
||||
case LEXER_INPUT_CLOSE_BRACKET:
|
||||
return LEXER_STATE_ARRAY_END;
|
||||
case LEXER_INPUT_COMMA:
|
||||
return LEXER_STATE_VALUE_END;
|
||||
default:
|
||||
return LEXER_STATE_ERROR;
|
||||
}
|
||||
}
|
||||
@@ -4,9 +4,6 @@
|
||||
#include "aliases.h"
|
||||
#include <stdbool.h>
|
||||
|
||||
#define VALID_JSON true
|
||||
#define INVALID_JSON false
|
||||
|
||||
typedef const char *str_view_t;
|
||||
|
||||
typedef enum {
|
||||
|
||||
96
include/lexer/lexer_data.h
Normal file
96
include/lexer/lexer_data.h
Normal file
@@ -0,0 +1,96 @@
|
||||
#ifndef LEXER_DATA_H
|
||||
#define LEXER_DATA_H
|
||||
|
||||
typedef enum {
|
||||
// GENERAL STATES
|
||||
LEXER_STATE_ERROR,
|
||||
LEXER_STATE_START,
|
||||
LEXER_STATE_VALUE,
|
||||
LEXER_STATE_VALUE_END,
|
||||
// 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,
|
||||
LEXER_STATE_KEY_END,
|
||||
// NUMBER STATES
|
||||
LEXER_STATE_DECIMAL,
|
||||
LEXER_STATE_NUMBER,
|
||||
LEXER_STATE_FRACTION,
|
||||
LEXER_STATE_EXPONENT,
|
||||
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_HEX1,
|
||||
LEXER_STATE_UNICODE_HEX2,
|
||||
LEXER_STATE_UNICODE_HEX3,
|
||||
LEXER_STATE_UNICODE_HEX4,
|
||||
// KEYWORD STATES
|
||||
LEXER_STATE_T,
|
||||
LEXER_STATE_TR,
|
||||
LEXER_STATE_TRU,
|
||||
LEXER_STATE_TRUE,
|
||||
LEXER_STATE_F,
|
||||
LEXER_STATE_FA,
|
||||
LEXER_STATE_FAL,
|
||||
LEXER_STATE_FALS,
|
||||
LEXER_STATE_FALSE,
|
||||
LEXER_STATE_N,
|
||||
LEXER_STATE_NU,
|
||||
LEXER_STATE_NUL,
|
||||
LEXER_STATE_NULL,
|
||||
LEXER_STATE_KEYWORD_END,
|
||||
|
||||
COUNT_LEXER_STATES,
|
||||
} lexer_state_t;
|
||||
|
||||
typedef enum {
|
||||
LEXER_INPUT_NEWLINE,
|
||||
LEXER_INPUT_WHITE_SPACE,
|
||||
LEXER_INPUT_OPEN_BRACE,
|
||||
LEXER_INPUT_CLOSE_BRACE,
|
||||
LEXER_INPUT_OPEN_BRACKET,
|
||||
LEXER_INPUT_CLOSE_BRACKET,
|
||||
LEXER_INPUT_COMMA,
|
||||
LEXER_INPUT_COLON,
|
||||
LEXER_INPUT_DOUBLE_QUOTE,
|
||||
LEXER_INPUT_BACK_SLASH,
|
||||
LEXER_INPUT_FORWARD_SLASH,
|
||||
LEXER_INPUT_LOWER_A,
|
||||
LEXER_INPUT_LOWER_B,
|
||||
LEXER_INPUT_LOWER_C,
|
||||
LEXER_INPUT_LOWER_D,
|
||||
LEXER_INPUT_LOWER_E,
|
||||
LEXER_INPUT_LOWER_F,
|
||||
LEXER_INPUT_LOWER_L,
|
||||
LEXER_INPUT_LOWER_N,
|
||||
LEXER_INPUT_LOWER_R,
|
||||
LEXER_INPUT_LOWER_S,
|
||||
LEXER_INPUT_LOWER_T,
|
||||
LEXER_INPUT_LOWER_U,
|
||||
LEXER_INPUT_UPPER_A,
|
||||
LEXER_INPUT_UPPER_B,
|
||||
LEXER_INPUT_UPPER_C,
|
||||
LEXER_INPUT_UPPER_D,
|
||||
LEXER_INPUT_UPPER_E,
|
||||
LEXER_INPUT_UPPER_F,
|
||||
LEXER_INPUT_MINUS,
|
||||
LEXER_INPUT_PLUS,
|
||||
LEXER_INPUT_DECIMAL,
|
||||
LEXER_INPUT_ZERO,
|
||||
LEXER_INPUT_NON_ZERO,
|
||||
LEXER_INPUT_OTHER,
|
||||
|
||||
COUNT_LEXER_INPUTS,
|
||||
} lexer_input_t;
|
||||
|
||||
#endif // !LEXER_DATA_H
|
||||
42
include/lexer/lexer_state_transitions.table
Normal file
42
include/lexer/lexer_state_transitions.table
Normal file
@@ -0,0 +1,42 @@
|
||||
// NEWLINE | WHITE_SPACE | OPEN_BRACE | CLOSE_BRACE | OPEN_BRACKET | CLOSE_BRACKET | COMMA | COLON | DOUBLE_QUOTE | BACK_SLASH | FORWARD_SLASH | LOWER_A | LOWER_B | LOWER_C | LOWER_D | LOWER_E | LOWER_F | LOWER_L | LOWER_N | LOWER_R | LOWER_S | LOWER_T | LOWER_U | UPPER_A | UPPER_B | UPPER_C | UPPER_D | UPPER_E | UPPER_F | MINUS | PLUS | DECIMAL | ZERO | NON_ZERO | OTHER |
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_ERROR
|
||||
{ LEXER_STATE_START, LEXER_STATE_START, LEXER_STATE_OBJECT_START, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_START, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_START
|
||||
{ LEXER_STATE_VALUE, LEXER_STATE_VALUE, LEXER_STATE_OBJECT_START, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_START, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_F, LEXER_STATE_ERROR, LEXER_STATE_N, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_T, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_NUMBER, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_DECIMAL, LEXER_STATE_NUMBER, LEXER_STATE_ERROR, }, // LEXER_STATE_VALUE
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_VALUE_END
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_OBJECT_START
|
||||
{ LEXER_STATE_OBJECT, LEXER_STATE_OBJECT, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_KEY, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_OBJECT
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_OBJECT_END
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_ARRAY_START
|
||||
{ LEXER_STATE_ARRAY, LEXER_STATE_ARRAY, LEXER_STATE_OBJECT_START, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_START, LEXER_STATE_ARRAY_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_F, LEXER_STATE_ERROR, LEXER_STATE_N, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_T, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_NUMBER, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_DECIMAL, LEXER_STATE_NUMBER, LEXER_STATE_ERROR, }, // LEXER_STATE_ARRAY
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_ARRAY_END
|
||||
{ LEXER_STATE_LAST_COLLECTION, LEXER_STATE_LAST_COLLECTION, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_LAST_COLLECTION
|
||||
{ LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, }, // LEXER_STATE_KEY
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_KEY_END
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FRACTION, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_DECIMAL
|
||||
{ LEXER_STATE_NUMBER_END, LEXER_STATE_NUMBER_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FRACTION, LEXER_STATE_NUMBER, LEXER_STATE_NUMBER, LEXER_STATE_ERROR, }, // LEXER_STATE_NUMBER
|
||||
{ LEXER_STATE_NUMBER_END, LEXER_STATE_NUMBER_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_EXPONENT, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_EXPONENT, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FRACTION, LEXER_STATE_FRACTION, LEXER_STATE_ERROR, }, // LEXER_STATE_FRACTION
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_EXP_SIGN, LEXER_STATE_EXP_SIGN, LEXER_STATE_ERROR, LEXER_STATE_POWER, LEXER_STATE_POWER, LEXER_STATE_ERROR, }, // LEXER_STATE_EXPONENT
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_POWER, LEXER_STATE_POWER, LEXER_STATE_ERROR, }, // LEXER_STATE_EXP_SIGN
|
||||
{ LEXER_STATE_NUMBER_END, LEXER_STATE_NUMBER_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_POWER, LEXER_STATE_POWER, LEXER_STATE_ERROR, }, // LEXER_STATE_POWER
|
||||
{ LEXER_STATE_NUMBER_END, LEXER_STATE_NUMBER_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_NUMBER_END
|
||||
{ LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING_END, LEXER_STATE_ESCAPE_SEQUENCE, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, }, // LEXER_STATE_STRING
|
||||
{ LEXER_STATE_STRING_END, LEXER_STATE_STRING_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_KEY_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_STRING_END
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_UNICODE_HEX1, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_ESCAPE_SEQUENCE
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_UNICODE_HEX2, LEXER_STATE_ERROR, }, // LEXER_STATE_UNICODE_HEX1
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_UNICODE_HEX3, LEXER_STATE_ERROR, }, // LEXER_STATE_UNICODE_HEX2
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_UNICODE_HEX4, LEXER_STATE_ERROR, }, // LEXER_STATE_UNICODE_HEX3
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_STRING, LEXER_STATE_STRING, LEXER_STATE_ERROR, }, // LEXER_STATE_UNICODE_HEX4
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_TR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_T
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_TRU, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_TR
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_TRUE, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_TRU
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_TRUE
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FA, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_F
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FAL, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_FA
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FALS, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_FAL
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_FALSE, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_FALS
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_FALSE
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_NU, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_N
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_NUL, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_NU
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_NULL, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_NUL
|
||||
{ LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_NULL
|
||||
{ LEXER_STATE_KEYWORD_END, LEXER_STATE_KEYWORD_END, LEXER_STATE_ERROR, LEXER_STATE_OBJECT_END, LEXER_STATE_ERROR, LEXER_STATE_ARRAY_END, LEXER_STATE_VALUE_END, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, LEXER_STATE_ERROR, }, // LEXER_STATE_KEYWORD_END
|
||||
@@ -1,6 +1,7 @@
|
||||
#include "lexer.h"
|
||||
#include "aliases.h"
|
||||
#include "dstring.h"
|
||||
#include "lexer_data.h"
|
||||
#include <assert.h>
|
||||
#include <ctype.h>
|
||||
#include <stdbool.h>
|
||||
@@ -13,43 +14,6 @@
|
||||
#define MAX_STACK_CAPACITY 1024
|
||||
#define STRING_BUF_START_CAPACITY 1024
|
||||
|
||||
typedef enum {
|
||||
// GENERAL STATES
|
||||
LEXER_STATE_START,
|
||||
LEXER_STATE_ERROR,
|
||||
LEXER_STATE_VALUE,
|
||||
// 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,
|
||||
LEXER_STATE_EXPONENT,
|
||||
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,
|
||||
// KEYWORD STATES
|
||||
LEXER_STATE_TRUE,
|
||||
LEXER_STATE_FALSE,
|
||||
LEXER_STATE_NULL,
|
||||
LEXER_STATE_KEYWORD_END,
|
||||
|
||||
COUNT_LEXER_STATES,
|
||||
} lexer_state_t;
|
||||
|
||||
typedef struct {
|
||||
lexer_state_t stack[MAX_STACK_CAPACITY];
|
||||
u64 size;
|
||||
@@ -84,6 +48,7 @@ struct lexer_s {
|
||||
u64 text_length;
|
||||
const char *text;
|
||||
lexer_state_t current;
|
||||
lexer_state_t next;
|
||||
state_stack_t stack;
|
||||
lexer_string_t keyword;
|
||||
lexer_string_t codepoint;
|
||||
@@ -93,8 +58,11 @@ struct lexer_s {
|
||||
bool has_extra_token;
|
||||
token_t extra_token;
|
||||
dstr_t *error_message;
|
||||
char current_char;
|
||||
};
|
||||
|
||||
INTERNAL lexer_input_t char_type(char input);
|
||||
|
||||
INTERNAL void stack_push(state_stack_t *stack, lexer_state_t value);
|
||||
INTERNAL lexer_state_t stack_pop(state_stack_t *stack);
|
||||
|
||||
@@ -108,31 +76,14 @@ INTERNAL token_t dstr_to_numerical_token(const dstr_t *str);
|
||||
INTERNAL void set_token(token_t *token, u64 line, u64 column, token_type type,
|
||||
token_value_t value);
|
||||
|
||||
INTERNAL void lexer_state_machine(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_lexer_start(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_last_collection(char input);
|
||||
INTERNAL lexer_state_t handle_collection_end(lexer_t *lexer, char input);
|
||||
INTERNAL void handle_input_after_collection_end(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_object(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_array(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_key(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_value(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_string(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_string_end(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_escape_sequence(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_unicode_sequence(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_decimal(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_number(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_fraction(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_exponent(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_exp_sign(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_power(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_number_end(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_keyword(char input);
|
||||
INTERNAL lexer_state_t handle_true(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_false(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_null(lexer_t *lexer, char input);
|
||||
INTERNAL lexer_state_t handle_keyword_end(lexer_t *lexer, char input);
|
||||
INTERNAL void finalise_state_transition(lexer_t *lexer);
|
||||
INTERNAL void post_keyword(lexer_t *lexer);
|
||||
INTERNAL void set_numerical_token(lexer_t *lexer);
|
||||
INTERNAL void handle_string_end(lexer_t *lexer);
|
||||
|
||||
INTERNAL lexer_state_t state_table[COUNT_LEXER_STATES][COUNT_LEXER_INPUTS] = {
|
||||
#include "lexer_state_transitions.table"
|
||||
};
|
||||
|
||||
void lexer_init(lexer_t **lexer) {
|
||||
if (*lexer) {
|
||||
@@ -151,6 +102,7 @@ void lexer_init(lexer_t **lexer) {
|
||||
(*lexer)->text_length = 0;
|
||||
(*lexer)->text = "";
|
||||
(*lexer)->current = LEXER_STATE_START;
|
||||
(*lexer)->next = LEXER_STATE_START;
|
||||
(*lexer)->keyword.type = LEXER_STRING_KEYWORD;
|
||||
(*lexer)->codepoint.type = LEXER_STRING_UNICODE;
|
||||
(*lexer)->current_string = dstr_with_capacity(STRING_BUF_START_CAPACITY);
|
||||
@@ -186,8 +138,6 @@ lex_result_t get_next_token(lexer_t *lexer, const char *text) {
|
||||
|
||||
dstr_clear(lexer->current_string);
|
||||
|
||||
char c;
|
||||
|
||||
while (lexer->cursor < lexer->text_length) {
|
||||
if (lexer->has_extra_token) {
|
||||
lexer->has_extra_token = false;
|
||||
@@ -198,11 +148,15 @@ lex_result_t get_next_token(lexer_t *lexer, const char *text) {
|
||||
};
|
||||
}
|
||||
|
||||
c = lexer->text[(lexer->cursor)++];
|
||||
lexer->current_char = lexer->text[(lexer->cursor)++];
|
||||
|
||||
lexer_state_machine(lexer, c);
|
||||
lexer_input_t input = char_type(lexer->current_char);
|
||||
|
||||
if (c == '\n') {
|
||||
lexer->next = state_table[lexer->current][input];
|
||||
|
||||
finalise_state_transition(lexer);
|
||||
|
||||
if (input == LEXER_INPUT_NEWLINE) {
|
||||
++(lexer->line);
|
||||
lexer->column = 0;
|
||||
} else {
|
||||
@@ -298,6 +252,83 @@ void print_token(token_t token) {
|
||||
printf("}\n");
|
||||
}
|
||||
|
||||
INTERNAL lexer_input_t char_type(char input) {
|
||||
if (input == '\n') {
|
||||
return LEXER_INPUT_NEWLINE;
|
||||
} else if (isspace(input)) {
|
||||
return LEXER_INPUT_WHITE_SPACE;
|
||||
} else if (input >= '1' && input <= '9') {
|
||||
return LEXER_INPUT_NON_ZERO;
|
||||
}
|
||||
|
||||
switch (input) {
|
||||
case '{':
|
||||
return LEXER_INPUT_OPEN_BRACE;
|
||||
case '}':
|
||||
return LEXER_INPUT_CLOSE_BRACE;
|
||||
case '[':
|
||||
return LEXER_INPUT_OPEN_BRACKET;
|
||||
case ']':
|
||||
return LEXER_INPUT_CLOSE_BRACKET;
|
||||
case ',':
|
||||
return LEXER_INPUT_COMMA;
|
||||
case ':':
|
||||
return LEXER_INPUT_COLON;
|
||||
case '"':
|
||||
return LEXER_INPUT_DOUBLE_QUOTE;
|
||||
case '\\':
|
||||
return LEXER_INPUT_BACK_SLASH;
|
||||
case '/':
|
||||
return LEXER_INPUT_FORWARD_SLASH;
|
||||
case 'a':
|
||||
return LEXER_INPUT_LOWER_A;
|
||||
case 'b':
|
||||
return LEXER_INPUT_LOWER_B;
|
||||
case 'c':
|
||||
return LEXER_INPUT_LOWER_C;
|
||||
case 'd':
|
||||
return LEXER_INPUT_LOWER_D;
|
||||
case 'e':
|
||||
return LEXER_INPUT_LOWER_E;
|
||||
case 'f':
|
||||
return LEXER_INPUT_LOWER_F;
|
||||
case 'l':
|
||||
return LEXER_INPUT_LOWER_L;
|
||||
case 'n':
|
||||
return LEXER_INPUT_LOWER_N;
|
||||
case 'r':
|
||||
return LEXER_INPUT_LOWER_R;
|
||||
case 's':
|
||||
return LEXER_INPUT_LOWER_S;
|
||||
case 't':
|
||||
return LEXER_INPUT_LOWER_T;
|
||||
case 'u':
|
||||
return LEXER_INPUT_LOWER_U;
|
||||
case 'A':
|
||||
return LEXER_INPUT_UPPER_A;
|
||||
case 'B':
|
||||
return LEXER_INPUT_UPPER_B;
|
||||
case 'C':
|
||||
return LEXER_INPUT_UPPER_C;
|
||||
case 'D':
|
||||
return LEXER_INPUT_UPPER_D;
|
||||
case 'E':
|
||||
return LEXER_INPUT_UPPER_E;
|
||||
case 'F':
|
||||
return LEXER_INPUT_UPPER_F;
|
||||
case '-':
|
||||
return LEXER_INPUT_MINUS;
|
||||
case '+':
|
||||
return LEXER_INPUT_PLUS;
|
||||
case '.':
|
||||
return LEXER_INPUT_DECIMAL;
|
||||
case '0':
|
||||
return LEXER_INPUT_ZERO;
|
||||
default:
|
||||
return LEXER_INPUT_OTHER;
|
||||
}
|
||||
}
|
||||
|
||||
void stack_push(state_stack_t *stack, lexer_state_t state) {
|
||||
if (stack->size + 1 >= MAX_STACK_CAPACITY) {
|
||||
return;
|
||||
@@ -403,6 +434,165 @@ void set_token(token_t *token, u64 line, u64 column, token_type type,
|
||||
};
|
||||
}
|
||||
|
||||
void finalise_state_transition(lexer_t *lexer) {
|
||||
switch (lexer->next) {
|
||||
case LEXER_STATE_OBJECT_START:
|
||||
lexer->token_ready = true;
|
||||
set_token(&(lexer->token), lexer->line, lexer->column, TK_L_BRACE,
|
||||
(token_value_t){0});
|
||||
|
||||
stack_push(&(lexer->stack), LEXER_STATE_OBJECT);
|
||||
|
||||
lexer->next = LEXER_STATE_OBJECT;
|
||||
|
||||
break;
|
||||
case LEXER_STATE_OBJECT_END:
|
||||
break;
|
||||
case LEXER_STATE_ARRAY_START:
|
||||
lexer->token_ready = true;
|
||||
set_token(&(lexer->token), lexer->line, lexer->column, TK_L_BRACKET,
|
||||
(token_value_t){0});
|
||||
|
||||
stack_push(&(lexer->stack), LEXER_STATE_ARRAY);
|
||||
|
||||
lexer->next = LEXER_STATE_ARRAY;
|
||||
|
||||
break;
|
||||
case LEXER_STATE_ARRAY_END:
|
||||
break;
|
||||
case LEXER_STATE_KEY:
|
||||
if (lexer->current == LEXER_STATE_OBJECT) {
|
||||
stack_push(&(lexer->stack), LEXER_STATE_KEY);
|
||||
|
||||
lexer->next = LEXER_STATE_STRING;
|
||||
}
|
||||
|
||||
break;
|
||||
case LEXER_STATE_KEY_END:
|
||||
lexer->next = LEXER_STATE_VALUE;
|
||||
|
||||
break;
|
||||
case LEXER_STATE_STRING:
|
||||
case LEXER_STATE_ESCAPE_SEQUENCE:
|
||||
case LEXER_STATE_UNICODE_HEX1:
|
||||
case LEXER_STATE_UNICODE_HEX2:
|
||||
case LEXER_STATE_UNICODE_HEX3:
|
||||
case LEXER_STATE_UNICODE_HEX4:
|
||||
dstr_append(&(lexer->current_string), lexer->current_char);
|
||||
|
||||
break;
|
||||
case LEXER_STATE_STRING_END:
|
||||
handle_string_end(lexer);
|
||||
|
||||
break;
|
||||
case LEXER_STATE_TRUE:
|
||||
case LEXER_STATE_FALSE:
|
||||
case LEXER_STATE_NULL:
|
||||
post_keyword(lexer);
|
||||
|
||||
break;
|
||||
case LEXER_STATE_VALUE_END:
|
||||
switch (lexer->current) {
|
||||
case LEXER_STATE_NUMBER:
|
||||
case LEXER_STATE_FRACTION:
|
||||
case LEXER_STATE_POWER:
|
||||
case LEXER_STATE_NUMBER_END:
|
||||
set_numerical_token(lexer);
|
||||
|
||||
break;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
|
||||
lexer->next = lexer->stack.stack[lexer->stack.size - 1];
|
||||
|
||||
break;
|
||||
case LEXER_STATE_NUMBER_END:
|
||||
switch (lexer->current) {
|
||||
case LEXER_STATE_NUMBER:
|
||||
case LEXER_STATE_FRACTION:
|
||||
case LEXER_STATE_POWER:
|
||||
set_numerical_token(lexer);
|
||||
|
||||
break;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
lexer->current = lexer->next;
|
||||
}
|
||||
|
||||
void post_keyword(lexer_t *lexer) {
|
||||
u64 keyword_char_count;
|
||||
u64 column;
|
||||
|
||||
token_t *token = &(lexer->token);
|
||||
|
||||
switch (lexer->current) {
|
||||
case LEXER_STATE_NULL:
|
||||
keyword_char_count = 4;
|
||||
|
||||
column = lexer->column - keyword_char_count;
|
||||
|
||||
set_token(token, lexer->line, column, TK_NULL, (token_value_t){0});
|
||||
|
||||
break;
|
||||
case LEXER_STATE_TRUE:
|
||||
keyword_char_count = 4;
|
||||
|
||||
column = lexer->column - keyword_char_count;
|
||||
|
||||
set_token(token, lexer->line, column, TK_BOOL,
|
||||
(token_value_t){.boolean = true});
|
||||
|
||||
break;
|
||||
case LEXER_STATE_FALSE:
|
||||
keyword_char_count = 5;
|
||||
|
||||
column = lexer->column - keyword_char_count;
|
||||
|
||||
set_token(token, lexer->line, column, TK_BOOL,
|
||||
(token_value_t){.boolean = false});
|
||||
|
||||
break;
|
||||
default:
|
||||
lexer->next = LEXER_STATE_ERROR;
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
lexer->token_ready = true;
|
||||
|
||||
lexer->next = LEXER_STATE_KEYWORD_END;
|
||||
}
|
||||
|
||||
void set_numerical_token(lexer_t *lexer) {
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
token_t token = dstr_to_numerical_token(lexer->current_string);
|
||||
|
||||
set_token(&(lexer->token), lexer->line, column, token.type, token.value);
|
||||
}
|
||||
|
||||
void handle_string_end(lexer_t *lexer) {
|
||||
lexer_state_t string_type = lexer->stack.stack[lexer->stack.size - 1];
|
||||
|
||||
lexer->token_ready = true;
|
||||
token_t *token = &(lexer->token);
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
token_value_t value = {.string = dstr_to_cstr(lexer->current_string)};
|
||||
|
||||
if (string_type == LEXER_STATE_KEY) {
|
||||
set_token(token, lexer->line, column, TK_STR_KEY, value);
|
||||
} else if (string_type == LEXER_STATE_VALUE) {
|
||||
set_token(token, lexer->line, column, TK_STR_VAL, value);
|
||||
}
|
||||
}
|
||||
|
||||
void lexer_state_machine(lexer_t *lexer, char input) {
|
||||
switch (lexer->current) {
|
||||
case LEXER_STATE_START:
|
||||
@@ -772,8 +962,6 @@ lexer_state_t handle_number(lexer_t *lexer, char input) {
|
||||
|
||||
return LEXER_STATE_FRACTION;
|
||||
} else if (input == '}' || input == ']') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -783,8 +971,6 @@ lexer_state_t handle_number(lexer_t *lexer, char input) {
|
||||
|
||||
return handle_collection_end(lexer, input);
|
||||
} else if (input == ',') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -794,8 +980,6 @@ lexer_state_t handle_number(lexer_t *lexer, char input) {
|
||||
|
||||
return lexer->stack.stack[lexer->stack.size - 1];
|
||||
} else if (isspace(input)) {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -815,8 +999,6 @@ lexer_state_t handle_fraction(lexer_t *lexer, char input) {
|
||||
|
||||
return LEXER_STATE_FRACTION;
|
||||
} else if (input == '}' || input == ']') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -830,8 +1012,6 @@ lexer_state_t handle_fraction(lexer_t *lexer, char input) {
|
||||
|
||||
return LEXER_STATE_EXPONENT;
|
||||
} else if (input == ',') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -841,8 +1021,6 @@ lexer_state_t handle_fraction(lexer_t *lexer, char input) {
|
||||
|
||||
return lexer->stack.stack[lexer->stack.size - 1];
|
||||
} else if (isspace(input)) {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -884,8 +1062,6 @@ lexer_state_t handle_power(lexer_t *lexer, char input) {
|
||||
|
||||
return LEXER_STATE_POWER;
|
||||
} else if (input == '}' || input == ']') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -895,8 +1071,6 @@ lexer_state_t handle_power(lexer_t *lexer, char input) {
|
||||
|
||||
return handle_collection_end(lexer, input);
|
||||
} else if (input == ',') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -906,8 +1080,6 @@ lexer_state_t handle_power(lexer_t *lexer, char input) {
|
||||
|
||||
return lexer->stack.stack[lexer->stack.size - 1];
|
||||
} else if (isspace(input)) {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
@@ -925,8 +1097,6 @@ lexer_state_t handle_number_end(lexer_t *lexer, char input) {
|
||||
if (isspace(input)) {
|
||||
return LEXER_STATE_NUMBER_END;
|
||||
} else if (input == ',') {
|
||||
// TODO (Abdelrahman): Set the token type correctly based on whether the
|
||||
// number is an integer or a double
|
||||
lexer->token_ready = true;
|
||||
u64 column = lexer->column - dstr_length(lexer->current_string);
|
||||
|
||||
|
||||
Reference in New Issue
Block a user