wash/ivy
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

lang: replace stack-based parser with a recursive parser

Browse Source
This commit is contained in:
max 🍓
2026-04-06 18:25:43 +01:00
parent 3a549aa6df
commit b904813cef
77 changed files with 2984 additions and 6499 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lang/ast/cond.c
+1 -353
View File
@@ -1,337 +1,5 @@
#include "block.h"
#include "expr/expr.h"
#include "iterate.h"
struct cond_group_parser_state {
struct parser_state s_base;
bool s_inline;
unsigned int s_prev_token;
struct ivy_ast_cond_node *s_cur_branch;
fx_queue s_branches;
struct ivy_ast_node *s_prev_node;
};
static void init_state(struct ivy_parser *ctx, struct parser_state *sp, uintptr_t arg)
{
struct cond_group_parser_state *state
= (struct cond_group_parser_state *)sp;
state->s_prev_node = (struct ivy_ast_node *)arg;
}
static enum ivy_status flush_current_branch(struct cond_group_parser_state *state)
{
if (!state->s_cur_branch) {
return IVY_ERR_INTERNAL_FAILURE;
}
fx_queue_push_back(&state->s_branches, &state->s_cur_branch->n_base.n_entry);
state->s_cur_branch
= (struct ivy_ast_cond_node *)ast_node_create(IVY_AST_COND);
if (!state->s_cur_branch) {
return IVY_ERR_NO_MEMORY;
}
return IVY_OK;
}
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);
if (state->s_cur_branch) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
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);
}
if (state->s_prev_node) {
/* this is an inline if-else */
state->s_inline = true;
state->s_cur_branch->n_body = state->s_prev_node;
state->s_prev_node = NULL;
} else {
state->s_inline = false;
}
struct expr_parser_state *expr
= (struct expr_parser_state *)parser_push_state(
ctx, IVY_AST_EXPR, 0);
state->s_prev_token = IVY_KW_IF;
expr->s_subexpr_depth = 1;
return PARSE_RESULT(IVY_OK, 0);
}
struct token_parse_result parse_elif(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_cur_branch) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_inline) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_prev_token != IVY_KW_THEN) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
state->s_cur_branch->n_body = state->s_prev_node;
state->s_prev_node = NULL;
enum ivy_status status = flush_current_branch(state);
if (status != IVY_OK) {
return PARSE_RESULT(status, 0);
}
struct expr_parser_state *expr
= (struct expr_parser_state *)parser_push_state(
ctx, IVY_AST_EXPR, 0);
state->s_prev_token = IVY_KW_ELIF;
expr->s_subexpr_depth = 1;
return PARSE_RESULT(IVY_OK, 0);
}
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);
if (!state->s_cur_branch) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_inline) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_prev_token != IVY_KW_IF && state->s_prev_token != IVY_KW_ELIF) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
/* previous component was the if-condition. */
if (!state->s_prev_node) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
state->s_cur_branch->n_cond = state->s_prev_node;
state->s_prev_node = NULL;
/* next component will be a block. */
struct block_parser_state *block
= (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_add_terminator(block, IVY_KW_ELIF);
block_add_terminator(block, IVY_KW_ELSE);
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)
{
enum ivy_status status;
struct cond_group_parser_state *state
= parser_get_state(ctx, struct cond_group_parser_state);
if (!state->s_cur_branch) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (!state->s_inline) {
/* previous component was either an expression or a block, and is the if branch body. */
if (!state->s_prev_node) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_prev_token != IVY_KW_THEN) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
state->s_cur_branch->n_body = state->s_prev_node;
state->s_prev_node = NULL;
status = flush_current_branch(state);
if (status != IVY_OK) {
return PARSE_RESULT(status, 0);
}
/* next component should be a block. */
} else {
/* previous component was the if-condition. */
if (!state->s_prev_node) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
if (state->s_prev_token != IVY_KW_IF) {
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
state->s_cur_branch->n_cond = state->s_prev_node;
state->s_prev_node = NULL;
status = flush_current_branch(state);
if (status != IVY_OK) {
return PARSE_RESULT(status, 0);
}
/* next component will be an expression. */
struct expr_parser_state *expr
= (struct expr_parser_state *)parser_push_state(
ctx, IVY_AST_EXPR, 0);
/* set the sub-expression depth to be non-zero so the expression
* parser doesn't consume the expression separator. */
expr->s_subexpr_depth = 1;
}
state->s_prev_token = IVY_KW_ELSE;
return PARSE_RESULT(IVY_OK, 0);
}
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) {
/* 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_add_terminator(block, IVY_KW_END);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
static enum ivy_status finalise_cond_group(struct cond_group_parser_state *state)
{
struct ivy_ast_cond_group_node *group
= (struct ivy_ast_cond_group_node *)state->s_base.s_node;
if (state->s_inline) {
/* we have just reached the end of either the 'if' or the 'else' expression. */
if (!state->s_cur_branch) {
/* not currently parsing a conditional branch. */
return IVY_ERR_BAD_SYNTAX;
}
if (!state->s_prev_node) {
/* else expression is empty. */
return IVY_ERR_BAD_SYNTAX;
}
if (state->s_prev_token == IVY_KW_ELSE) {
/* this is the else expression, s_prev_node is the else body. */
state->s_cur_branch->n_body = state->s_prev_node;
} else if (state->s_prev_token == IVY_KW_IF) {
/* this is the if expression, s_prev_node is the if condition. */
state->s_cur_branch->n_cond = state->s_prev_node;
}
state->s_prev_node = NULL;
flush_current_branch(state);
group->n_branches = state->s_branches;
state->s_branches = FX_QUEUE_INIT;
return IVY_OK;
} else {
/* we have just reached the 'end' keyword. */
if (!state->s_cur_branch) {
/* not currently parsing a conditional branch. */
return IVY_ERR_BAD_SYNTAX;
}
if (!state->s_prev_node) {
/* else expression is empty. */
return IVY_ERR_BAD_SYNTAX;
}
state->s_cur_branch->n_body = state->s_prev_node;
state->s_prev_node = NULL;
flush_current_branch(state);
group->n_branches = state->s_branches;
state->s_branches = FX_QUEUE_INIT;
return IVY_OK;
}
}
static struct token_parse_result parse_punct_terminator(
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_inline) {
/* only inline conditional can be ended with punctuation. */
return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
}
enum ivy_status status = finalise_cond_group(state);
if (status != IVY_OK) {
return PARSE_RESULT(status, 0);
}
parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}
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);
enum ivy_status status = finalise_cond_group(state);
if (status != IVY_OK) {
return PARSE_RESULT(status, 0);
}
int flags = 0;
if (state->s_inline) {
flags = PARSE_REPEAT_TOKEN;
}
parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
return PARSE_RESULT(IVY_OK, flags);
}
static enum ivy_status add_child(
struct parser_state *parent, struct ivy_ast_node *child)
{
struct cond_group_parser_state *state
= (struct cond_group_parser_state *)parent;
if (state->s_prev_node) {
return IVY_ERR_BAD_SYNTAX;
}
state->s_prev_node = child;
return IVY_OK;
}
#include "node.h"
static void cond_group_collect_children(
struct ivy_ast_node *node, struct ivy_ast_node_iterator *iterator)
@@ -363,31 +31,11 @@ static void cond_collect_children(
}
struct ast_node_type cond_group_node_ops = {
.n_init_state = init_state,
.n_add_child = add_child,
.n_collect_children = cond_group_collect_children,
.n_state_size = sizeof(struct cond_group_parser_state),
.n_node_size = sizeof(struct ivy_ast_cond_group_node),
.n_keyword_parsers = {
KW_PARSER(IF, parse_if),
KW_PARSER(THEN, parse_then),
KW_PARSER(ELIF, parse_elif),
KW_PARSER(ELSE, parse_else),
KW_PARSER(END, parse_end),
},
.n_symbol_parsers = {
SYM_PARSER(DOT, parse_punct_terminator),
SYM_PARSER(COMMA, parse_punct_terminator),
SYM_PARSER(RIGHT_PAREN, parse_punct_terminator),
SYM_PARSER(BANG, parse_punct_terminator),
},
.n_expr_parser = {
.expr_begin = parse_expr_begin,
},
};
struct ast_node_type cond_node_ops = {
.n_collect_children = cond_collect_children,
.n_state_size = sizeof(struct parser_state),
.n_node_size = sizeof(struct ivy_ast_cond_node),
};