lang: ast: add support for multiple block termination tokens

This commit is contained in:
2025-01-16 13:15:18 +00:00
parent fa33336ed7
commit 143d61e329
7 changed files with 102 additions and 31 deletions

View File

@@ -8,6 +8,34 @@
#include <blue/object/string.h>
#include <ivy/lang/lex.h>
void block_add_terminator(struct block_parser_state *state, unsigned short tok)
{
if (state->s_nr_terminators < BLOCK_TERMINATOR_MAX) {
state->s_terminators[state->s_nr_terminators++] = tok;
}
}
void block_copy_terminators(
const struct block_parser_state *src, struct block_parser_state *dest)
{
dest->s_nr_terminators = src->s_nr_terminators;
for (unsigned int i = 0; i < src->s_nr_terminators; i++) {
dest->s_terminators[i] = src->s_terminators[i];
}
}
bool block_terminates_at_token(struct block_parser_state *state, unsigned short tok)
{
for (unsigned int i = 0; i < BLOCK_TERMINATOR_MAX; i++) {
if (state->s_terminators[i] == tok) {
return true;
}
}
return false;
}
static struct token_parse_result parse_end(
struct ivy_parser *ctx, struct ivy_token *tok)
{
@@ -18,7 +46,7 @@ static struct token_parse_result parse_end(
= (struct ivy_ast_block_node *)(state->s_base.s_node);
int flags = 0;
if (state->s_terminator == IVY_KW_END) {
if (block_terminates_at_token(state, IVY_KW_END)) {
flags = PARSE_REPEAT_TOKEN;
}
@@ -39,6 +67,23 @@ static struct token_parse_result parse_else(
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
static struct token_parse_result parse_keyword(
struct ivy_parser *ctx, struct ivy_token *tok)
{
struct block_parser_state *state
= parser_get_state(ctx, struct block_parser_state);
struct ivy_ast_block_node *node
= (struct ivy_ast_block_node *)(state->s_base.s_node);
if (block_terminates_at_token(state, tok->t_keyword)) {
parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
static struct token_parse_result parse_symbol(
struct ivy_parser *ctx, struct ivy_token *tok)
{
@@ -48,7 +93,7 @@ static struct token_parse_result parse_symbol(
struct ivy_ast_block_node *node
= (struct ivy_ast_block_node *)(state->s_base.s_node);
if (state->s_terminator == tok->t_symbol) {
if (block_terminates_at_token(state, tok->t_symbol)) {
parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
@@ -69,6 +114,10 @@ static struct token_parse_result parse_expr_begin(
= (struct expr_parser_state *)parser_push_state(
ctx, IVY_AST_EXPR, 0);
memcpy(expr->s_terminators, state->s_terminators,
sizeof expr->s_terminators);
expr->s_nr_terminators = state->s_nr_terminators;
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
@@ -103,6 +152,7 @@ struct ast_node_type block_node_ops = {
.n_keyword_parsers = {
KW_PARSER(END, parse_end),
KW_PARSER(ELSE, parse_else),
KW_PARSER_FALLBACK(parse_keyword),
},
.n_symbol_parsers = {
SYM_PARSER_FALLBACK(parse_symbol),

View File

@@ -3,10 +3,24 @@
#include "ctx.h"
#define BLOCK_TERMINATOR_MAX 8
struct block_parser_state {
struct parser_state s_base;
bool s_single_expr;
unsigned int s_terminator;
/* the block will be terminated when any token in this list
* is encountered. the token that terminated the block will
* not be consumed. */
unsigned short s_terminators[BLOCK_TERMINATOR_MAX];
unsigned short s_nr_terminators;
};
extern void block_add_terminator(
struct block_parser_state *state, unsigned short tok);
extern void block_copy_terminators(
const struct block_parser_state *src, struct block_parser_state *dest);
extern bool block_terminates_at_token(
struct block_parser_state *state, unsigned short tok);
#endif

View File

@@ -1,6 +1,6 @@
#include "block.h"
#include "iterate.h"
#include "expr/expr.h"
#include "iterate.h"
struct cond_group_parser_state {
struct parser_state s_base;
@@ -20,9 +20,7 @@ static void init_state(struct ivy_parser *ctx, struct parser_state *sp, uintptr_
state->s_prev_node = (struct ivy_ast_node *)arg;
}
struct token_parse_result parse_if(
struct ivy_parser* ctx,
struct ivy_token* tok)
struct token_parse_result parse_if(struct ivy_parser *ctx, struct ivy_token *tok)
{
struct cond_group_parser_state *state
= parser_get_state(ctx, struct cond_group_parser_state);
@@ -31,7 +29,8 @@ struct token_parse_result parse_if(
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
state->s_cur_branch = (struct ivy_ast_cond_node *)ast_node_create(IVY_AST_COND);
state->s_cur_branch
= (struct ivy_ast_cond_node *)ast_node_create(IVY_AST_COND);
if (!state->s_cur_branch) {
return PARSE_RESULT(IVY_ERR_NO_MEMORY, 0);
}
@@ -73,7 +72,8 @@ static enum ivy_status flush_current_branch(struct cond_group_parser_state *stat
return IVY_OK;
}
static struct token_parse_result parse_then(struct ivy_parser *ctx, struct ivy_token *tok)
static struct token_parse_result parse_then(
struct ivy_parser *ctx, struct ivy_token *tok)
{
struct cond_group_parser_state *state
= parser_get_state(ctx, struct cond_group_parser_state);
@@ -103,13 +103,14 @@ static struct token_parse_result parse_then(struct ivy_parser *ctx, struct ivy_t
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
/* set the sub-expression depth to be non-zero so the expression parser doesn't consume the expression separator. */
block->s_terminator = IVY_KW_END;
block_add_terminator(block, IVY_KW_END);
state->s_prev_token = IVY_KW_THEN;
return PARSE_RESULT(IVY_OK, 0);
}
static struct token_parse_result parse_else(struct ivy_parser *ctx, struct ivy_token *tok)
static struct token_parse_result parse_else(
struct ivy_parser *ctx, struct ivy_token *tok)
{
enum ivy_status status;
struct cond_group_parser_state *state
@@ -168,18 +169,22 @@ static struct token_parse_result parse_else(struct ivy_parser *ctx, struct ivy_t
return PARSE_RESULT(IVY_OK, 0);
}
static struct token_parse_result parse_expr_begin(struct ivy_parser* ctx, struct ivy_token* tok)
static struct token_parse_result parse_expr_begin(
struct ivy_parser *ctx, struct ivy_token *tok)
{
struct cond_group_parser_state *state
= parser_get_state(ctx, struct cond_group_parser_state);
if (state->s_prev_token != IVY_KW_ELSE && state->s_prev_token != IVY_KW_THEN) {
if (state->s_prev_token != IVY_KW_ELSE
&& state->s_prev_token != IVY_KW_THEN) {
/* expression can only follow else and then keywords. */
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
struct block_parser_state *block = (struct block_parser_state *)parser_push_state(ctx, IVY_AST_BLOCK, 0);
block->s_terminator = IVY_KW_END;
struct block_parser_state *block
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block_add_terminator(block, IVY_KW_END);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
@@ -256,7 +261,8 @@ static struct token_parse_result parse_punct_terminator(
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
static struct token_parse_result parse_end(struct ivy_parser *ctx, struct ivy_token *tok)
static struct token_parse_result parse_end(
struct ivy_parser *ctx, struct ivy_token *tok)
{
struct cond_group_parser_state *state
= parser_get_state(ctx, struct cond_group_parser_state);
@@ -297,7 +303,8 @@ static void cond_group_collect_children(
b_queue_iterator it = {0};
b_queue_foreach (&it, &group->n_branches) {
struct ivy_ast_node *branch = b_unbox(struct ivy_ast_node, it.entry, n_entry);
struct ivy_ast_node *branch
= b_unbox(struct ivy_ast_node, it.entry, n_entry);
ast_node_iterator_enqueue_node(iterator, node, branch);
}
}

View File

@@ -101,7 +101,7 @@ static struct token_parse_result parse_do(
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block->s_terminator = IVY_KW_END;
block_add_terminator(block, IVY_KW_END);
state->s_prev_token = IVY_KW_DO;
return PARSE_RESULT(IVY_OK, 0);
}
@@ -120,7 +120,7 @@ static struct token_parse_result parse_expr_begin(
struct block_parser_state *block
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block->s_terminator = IVY_KW_END;
block_add_terminator(block, IVY_KW_END);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}

View File

@@ -29,7 +29,7 @@ static struct token_parse_result parse_expr_begin(
struct block_parser_state *block
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block->s_terminator = IVY_SYM_RIGHT_BRACKET;
block_add_terminator(block, IVY_SYM_RIGHT_BRACKET);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}

View File

@@ -40,7 +40,7 @@ static struct token_parse_result parse_linefeed(
struct block_parser_state *block_state
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block_state->s_terminator = IVY_SYM_BANG;
block_add_terminator(block_state, IVY_SYM_BANG);
return PARSE_RESULT(IVY_OK, 0);
}

View File

@@ -70,7 +70,7 @@ static struct token_parse_result parse_do(
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
/* set the sub-expression depth to be non-zero so the expression parser doesn't consume the expression separator. */
block->s_terminator = IVY_KW_END;
block_add_terminator(block, IVY_KW_END);
state->s_prev_token = IVY_KW_DO;
return PARSE_RESULT(IVY_OK, 0);
@@ -90,7 +90,7 @@ static struct token_parse_result parse_expr_begin(
struct block_parser_state *block
= (struct block_parser_state *)parser_push_state(
ctx, IVY_AST_BLOCK, 0);
block->s_terminator = IVY_KW_END;
block_add_terminator(block, IVY_KW_END);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}