Allow ubiquitous tokens to also be used in grammar rules
This commit is contained in:
parent
a9dff20658
commit
9686c57e90
22 changed files with 49452 additions and 47887 deletions
|
|
@ -1,4 +1,5 @@
|
|||
#include "tree_sitter/parser/lr_parser.h"
|
||||
#include "runtime/tree.h"
|
||||
|
||||
/*
|
||||
* Private
|
||||
|
|
@ -8,7 +9,8 @@ static const ts_parse_action * actions_for_state(ts_lr_parser *parser, ts_state_
|
|||
return parser->config.parse_table + (state * parser->config.symbol_count);
|
||||
}
|
||||
|
||||
void shift(ts_lr_parser *parser, ts_state_id parse_state) {
|
||||
void shift(ts_lr_parser *parser, ts_state_id parse_state, int is_extra) {
|
||||
parser->lookahead->is_extra = is_extra;
|
||||
ts_stack_push(&parser->stack, parse_state, parser->lookahead);
|
||||
parser->lookahead = parser->next_lookahead;
|
||||
parser->next_lookahead = NULL;
|
||||
|
|
@ -20,7 +22,7 @@ void reduce(ts_lr_parser *parser, ts_symbol symbol, size_t child_count) {
|
|||
symbol,
|
||||
child_count,
|
||||
parser->config.hidden_symbol_flags,
|
||||
parser->config.ubiquitous_symbol_flags);
|
||||
1);
|
||||
}
|
||||
|
||||
static size_t breakdown_stack(ts_lr_parser *parser, ts_input_edit *edit) {
|
||||
|
|
@ -129,6 +131,7 @@ ts_tree * get_tree_root(ts_lr_parser *parser) {
|
|||
stack->size--;
|
||||
for (size_t i = 0; i < immediate_child_count; i++) {
|
||||
ts_tree *child = immedate_children[i];
|
||||
child->is_extra = 0;
|
||||
ts_tree_retain(child);
|
||||
ts_state_id state = ts_stack_top_state(stack);
|
||||
ts_state_id next_state = actions_for_state(parser, state)[ts_tree_symbol(child)].data.to_state;
|
||||
|
|
@ -136,10 +139,10 @@ ts_tree * get_tree_root(ts_lr_parser *parser) {
|
|||
}
|
||||
|
||||
ts_tree *new_node = ts_stack_reduce(stack,
|
||||
ts_tree_symbol(top_node),
|
||||
top_node->symbol,
|
||||
stack->size,
|
||||
parser->config.hidden_symbol_flags,
|
||||
NULL);
|
||||
0);
|
||||
ts_tree_release(top_node);
|
||||
return new_node;
|
||||
}
|
||||
|
|
@ -159,8 +162,7 @@ ts_lr_parser * ts_lr_parser_make(size_t symbol_count,
|
|||
const ts_parse_action *parse_table,
|
||||
const ts_state_id *lex_states,
|
||||
ts_tree * (* lex_fn)(ts_lexer *, ts_state_id),
|
||||
const int *hidden_symbol_flags,
|
||||
const int *ubiquitous_symbol_flags) {
|
||||
const int *hidden_symbol_flags) {
|
||||
ts_lr_parser *result = malloc(sizeof(ts_lr_parser));
|
||||
*result = (ts_lr_parser) {
|
||||
.lexer = ts_lexer_make(),
|
||||
|
|
@ -171,7 +173,6 @@ ts_lr_parser * ts_lr_parser_make(size_t symbol_count,
|
|||
.lex_states = lex_states,
|
||||
.lex_fn = lex_fn,
|
||||
.hidden_symbol_flags = hidden_symbol_flags,
|
||||
.ubiquitous_symbol_flags = ubiquitous_symbol_flags,
|
||||
},
|
||||
};
|
||||
return result;
|
||||
|
|
@ -198,6 +199,8 @@ void ts_lr_parser_initialize(ts_lr_parser *parser, ts_input input, ts_input_edit
|
|||
ts_lexer_advance(&parser->lexer);
|
||||
}
|
||||
|
||||
/* #define TS_DEBUG_PARSE */
|
||||
|
||||
#ifdef TS_DEBUG_PARSE
|
||||
#include <stdio.h>
|
||||
#define DEBUG_PARSE(...) fprintf(stderr, "\n" __VA_ARGS__)
|
||||
|
|
@ -211,7 +214,11 @@ ts_tree * ts_lr_parser_parse(ts_lr_parser *parser, const char **symbol_names) {
|
|||
switch (action.type) {
|
||||
case ts_parse_action_type_shift:
|
||||
DEBUG_PARSE("SHIFT %d", action.data.to_state);
|
||||
shift(parser, action.data.to_state);
|
||||
shift(parser, action.data.to_state, 0);
|
||||
return NULL;
|
||||
case ts_parse_action_type_shift_extra:
|
||||
DEBUG_PARSE("SHIFT EXTRA");
|
||||
shift(parser, ts_stack_top_state(&parser->stack), 1);
|
||||
return NULL;
|
||||
case ts_parse_action_type_reduce:
|
||||
DEBUG_PARSE("REDUCE %s %d", symbol_names[action.data.symbol], action.data.child_count);
|
||||
|
|
|
|||
|
|
@ -1,5 +1,6 @@
|
|||
#include "tree_sitter/runtime.h"
|
||||
#include "tree_sitter/parser/stack.h"
|
||||
#include "runtime/tree.h"
|
||||
#include <string.h>
|
||||
|
||||
static size_t INITIAL_STACK_SIZE = 100;
|
||||
|
|
@ -54,12 +55,12 @@ ts_tree * ts_stack_reduce(ts_stack *stack,
|
|||
ts_symbol symbol,
|
||||
size_t immediate_child_count,
|
||||
const int *hidden_symbol_flags,
|
||||
const int *ubiquitous_symbol_flags) {
|
||||
int gather_extra) {
|
||||
|
||||
// First, walk down the stack to determine which symbols will be reduced.
|
||||
// The child node count is known ahead of time, but some of the
|
||||
// nodes at the top of the stack might be hidden nodes, in which
|
||||
// case we 'collapse' them. Some may also be ubiquitous tokens,
|
||||
// case we 'collapse' them. Some may also be extra tokens,
|
||||
// which don't count towards the child node count.
|
||||
static int collapse_flags[100];
|
||||
int child_count = 0;
|
||||
|
|
@ -77,7 +78,7 @@ ts_tree * ts_stack_reduce(ts_stack *stack,
|
|||
|
||||
child_count += collapse_flags[i] ? grandchild_count : 1;
|
||||
|
||||
if (ubiquitous_symbol_flags && ubiquitous_symbol_flags[child_symbol])
|
||||
if (gather_extra && child->is_extra)
|
||||
immediate_child_count++;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,25 +1,7 @@
|
|||
#include "tree_sitter/runtime.h"
|
||||
#include <string.h>
|
||||
#include <stdio.h>
|
||||
|
||||
struct ts_tree {
|
||||
ts_symbol symbol;
|
||||
size_t ref_count;
|
||||
size_t offset;
|
||||
size_t size;
|
||||
union {
|
||||
struct {
|
||||
size_t count;
|
||||
size_t immediate_count;
|
||||
struct ts_tree **contents;
|
||||
} children;
|
||||
struct {
|
||||
char lookahead_char;
|
||||
size_t expected_input_count;
|
||||
const ts_symbol *expected_inputs;
|
||||
} error;
|
||||
} data;
|
||||
};
|
||||
#include "runtime/tree.h"
|
||||
|
||||
static ts_tree * ts_tree_make(ts_symbol symbol, size_t size, size_t offset) {
|
||||
ts_tree *result = malloc(sizeof(ts_tree));
|
||||
|
|
|
|||
26
src/runtime/tree.h
Normal file
26
src/runtime/tree.h
Normal file
|
|
@ -0,0 +1,26 @@
|
|||
#ifndef RUNTIME_TREE_H_
|
||||
#define RUNTIME_TREE_H_
|
||||
|
||||
#include "tree_sitter/runtime.h"
|
||||
|
||||
struct ts_tree {
|
||||
ts_symbol symbol;
|
||||
size_t ref_count;
|
||||
size_t offset;
|
||||
size_t size;
|
||||
int is_extra;
|
||||
union {
|
||||
struct {
|
||||
size_t count;
|
||||
size_t immediate_count;
|
||||
struct ts_tree **contents;
|
||||
} children;
|
||||
struct {
|
||||
char lookahead_char;
|
||||
size_t expected_input_count;
|
||||
const ts_symbol *expected_inputs;
|
||||
} error;
|
||||
} data;
|
||||
};
|
||||
|
||||
#endif // RUNTIME_TREE_H_
|
||||
Loading…
Add table
Add a link
Reference in a new issue