2014-07-10 13:14:52 -07:00
|
|
|
#include <stdio.h>
|
|
|
|
|
#include "tree_sitter/runtime.h"
|
|
|
|
|
#include "tree_sitter/parser.h"
|
|
|
|
|
#include "runtime/tree.h"
|
2014-07-30 23:40:02 -07:00
|
|
|
#include "runtime/lexer.h"
|
|
|
|
|
#include "runtime/stack.h"
|
|
|
|
|
#include "runtime/parser.h"
|
2014-07-10 13:14:52 -07:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Private
|
|
|
|
|
*/
|
|
|
|
|
|
2014-07-30 23:40:02 -07:00
|
|
|
static const TSParseAction *actions_for_state(const TSLanguage *language,
|
2014-07-20 21:43:27 -07:00
|
|
|
TSStateId state) {
|
2014-07-30 23:40:02 -07:00
|
|
|
return language->parse_table + (state * language->symbol_count);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static size_t breakdown_stack(TSParser *parser, TSInputEdit *edit) {
|
2014-07-20 21:43:27 -07:00
|
|
|
if (!edit)
|
|
|
|
|
return 0;
|
2014-07-20 20:27:33 -07:00
|
|
|
|
|
|
|
|
TSStack *stack = &parser->stack;
|
|
|
|
|
size_t position = 0;
|
|
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
|
TSTree *node = ts_stack_top_node(stack);
|
2014-07-20 21:43:27 -07:00
|
|
|
if (!node)
|
|
|
|
|
break;
|
2014-07-20 20:27:33 -07:00
|
|
|
|
|
|
|
|
position = ts_stack_right_position(stack);
|
|
|
|
|
size_t child_count;
|
|
|
|
|
TSTree **children = ts_tree_children(node, &child_count);
|
2014-07-20 21:43:27 -07:00
|
|
|
if (position <= edit->position && !children)
|
|
|
|
|
break;
|
2014-07-20 20:27:33 -07:00
|
|
|
|
|
|
|
|
stack->size--;
|
|
|
|
|
position -= ts_tree_total_size(node);
|
|
|
|
|
|
|
|
|
|
for (size_t i = 0; i < child_count && position < edit->position; i++) {
|
|
|
|
|
TSTree *child = children[i];
|
|
|
|
|
TSStateId state = ts_stack_top_state(stack);
|
2014-07-31 13:11:39 -07:00
|
|
|
TSStateId next_state = actions_for_state(
|
|
|
|
|
parser->language, state)[child->symbol].data.to_state;
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_stack_push(stack, next_state, child);
|
|
|
|
|
ts_tree_retain(child);
|
|
|
|
|
position += ts_tree_total_size(child);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_tree_release(node);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return position;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-20 21:43:27 -07:00
|
|
|
static TSSymbol *expected_symbols(TSParser *parser, size_t *count) {
|
2014-07-20 20:27:33 -07:00
|
|
|
*count = 0;
|
2014-07-20 21:43:27 -07:00
|
|
|
const TSParseAction *actions =
|
2014-07-30 23:40:02 -07:00
|
|
|
actions_for_state(parser->language, ts_stack_top_state(&parser->stack));
|
|
|
|
|
for (size_t i = 0; i < parser->language->symbol_count; i++)
|
2014-07-20 20:27:33 -07:00
|
|
|
if (actions[i].type != TSParseActionTypeError)
|
|
|
|
|
(*count)++;
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
size_t n = 0;
|
|
|
|
|
TSSymbol *result = malloc(*count * sizeof(*result));
|
2014-07-30 23:40:02 -07:00
|
|
|
for (TSSymbol i = 0; i < parser->language->symbol_count; i++)
|
2014-07-20 20:27:33 -07:00
|
|
|
if (actions[i].type != TSParseActionTypeError)
|
|
|
|
|
result[n++] = i;
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
return result;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Public
|
|
|
|
|
*/
|
|
|
|
|
|
2014-07-30 23:40:02 -07:00
|
|
|
TSParser ts_parser_make(const TSLanguage *language) {
|
|
|
|
|
return (TSParser) { .lexer = ts_lexer_make(),
|
|
|
|
|
.stack = ts_stack_make(),
|
|
|
|
|
.debug = 0,
|
|
|
|
|
.language = language, };
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-30 23:40:02 -07:00
|
|
|
void ts_parser_destroy(TSParser *parser) {
|
2014-07-20 21:43:27 -07:00
|
|
|
if (parser->lookahead)
|
|
|
|
|
ts_tree_release(parser->lookahead);
|
|
|
|
|
if (parser->next_lookahead)
|
|
|
|
|
ts_tree_release(parser->next_lookahead);
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_stack_delete(&parser->stack);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void ts_parser_start(TSParser *parser, TSInput input, TSInputEdit *edit) {
|
2014-07-20 21:43:27 -07:00
|
|
|
if (!edit)
|
|
|
|
|
ts_stack_shrink(&parser->stack, 0);
|
2014-07-20 20:27:33 -07:00
|
|
|
parser->lookahead = NULL;
|
|
|
|
|
parser->next_lookahead = NULL;
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
size_t position = breakdown_stack(parser, edit);
|
|
|
|
|
input.seek_fn(input.data, position);
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
parser->lexer = ts_lexer_make();
|
|
|
|
|
parser->lexer.input = input;
|
|
|
|
|
ts_lexer_advance(&parser->lexer);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void ts_parser_shift(TSParser *parser, TSStateId parse_state) {
|
2014-07-20 20:27:33 -07:00
|
|
|
if (ts_tree_is_extra(parser->lookahead))
|
|
|
|
|
parse_state = ts_stack_top_state(&parser->stack);
|
|
|
|
|
ts_stack_push(&parser->stack, parse_state, parser->lookahead);
|
|
|
|
|
parser->lookahead = parser->next_lookahead;
|
|
|
|
|
parser->next_lookahead = NULL;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void ts_parser_shift_extra(TSParser *parser) {
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_tree_set_extra(parser->lookahead);
|
|
|
|
|
ts_parser_shift(parser, 0);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void ts_parser_reduce(TSParser *parser, TSSymbol symbol, size_t child_count) {
|
2014-07-20 20:27:33 -07:00
|
|
|
parser->next_lookahead = parser->lookahead;
|
2014-07-20 21:43:27 -07:00
|
|
|
parser->lookahead = ts_stack_reduce(&parser->stack, symbol, child_count,
|
2014-08-08 08:46:01 -07:00
|
|
|
parser->language->hidden_symbol_flags);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int ts_parser_reduce_extra(TSParser *parser, TSSymbol symbol) {
|
2014-07-20 20:27:33 -07:00
|
|
|
TSTree *top_node = ts_stack_top_node(&parser->stack);
|
|
|
|
|
if (top_node->symbol == symbol && !ts_tree_is_extra(top_node)) {
|
|
|
|
|
ts_parser_reduce(parser, symbol, 1);
|
|
|
|
|
ts_tree_set_extra(parser->lookahead);
|
|
|
|
|
return 1;
|
|
|
|
|
} else {
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int ts_parser_handle_error(TSParser *parser) {
|
2014-07-20 20:27:33 -07:00
|
|
|
size_t count = 0;
|
|
|
|
|
const TSSymbol *inputs = expected_symbols(parser, &count);
|
2014-07-20 21:43:27 -07:00
|
|
|
TSTree *error = ts_tree_make_error(ts_lexer_lookahead_char(&parser->lexer),
|
|
|
|
|
count, inputs, 0, 0);
|
2014-07-20 20:27:33 -07:00
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
|
ts_tree_release(parser->lookahead);
|
|
|
|
|
size_t position = ts_lexer_position(&parser->lexer);
|
2014-07-31 13:11:39 -07:00
|
|
|
parser->lookahead =
|
|
|
|
|
parser->language->lex_fn(&parser->lexer, ts_lex_state_error);
|
2014-07-20 20:27:33 -07:00
|
|
|
|
|
|
|
|
int at_end = 0;
|
|
|
|
|
if (ts_lexer_position(&parser->lexer) == position)
|
|
|
|
|
at_end = !ts_lexer_advance(&parser->lexer);
|
|
|
|
|
|
|
|
|
|
if (at_end || parser->lookahead->symbol == ts_builtin_sym_end) {
|
|
|
|
|
ts_stack_push(&parser->stack, 0, error);
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
/*
|
|
|
|
|
* Unwind the stack, looking for a state in which this token
|
|
|
|
|
* may appear after an error.
|
|
|
|
|
*/
|
|
|
|
|
for (size_t j = 0; j < parser->stack.size; j++) {
|
|
|
|
|
size_t i = parser->stack.size - 1 - j;
|
|
|
|
|
TSStateId stack_state = parser->stack.entries[i].state;
|
2014-07-31 13:11:39 -07:00
|
|
|
TSParseAction action_on_error = actions_for_state(
|
|
|
|
|
parser->language, stack_state)[ts_builtin_sym_error];
|
2014-07-20 20:27:33 -07:00
|
|
|
if (action_on_error.type == TSParseActionTypeShift) {
|
|
|
|
|
TSStateId state_after_error = action_on_error.data.to_state;
|
2014-07-30 23:40:02 -07:00
|
|
|
if (actions_for_state(parser->language,
|
2014-07-20 21:43:27 -07:00
|
|
|
state_after_error)[parser->lookahead->symbol]
|
|
|
|
|
.type != TSParseActionTypeError) {
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_stack_shrink(&parser->stack, i + 1);
|
|
|
|
|
ts_stack_push(&parser->stack, state_after_error, error);
|
|
|
|
|
return 1;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
2014-07-20 20:27:33 -07:00
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
2014-07-20 20:27:33 -07:00
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-20 21:43:27 -07:00
|
|
|
TSTree *ts_parser_tree_root(TSParser *parser) {
|
2014-08-08 23:58:59 -07:00
|
|
|
TSStack *stack = &parser->stack;
|
|
|
|
|
if (stack->size == 0)
|
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
|
|
TSTree *tree = ts_stack_reduce(stack, ts_builtin_sym_document,
|
|
|
|
|
stack->size, parser->language->hidden_symbol_flags);
|
|
|
|
|
tree->options = 0;
|
|
|
|
|
ts_stack_push(stack, 0, tree);
|
|
|
|
|
return tree;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TSParseAction ts_parser_next_action(TSParser *parser) {
|
2014-07-20 20:27:33 -07:00
|
|
|
TSStateId state = ts_stack_top_state(&parser->stack);
|
|
|
|
|
if (!parser->lookahead)
|
2014-07-31 13:11:39 -07:00
|
|
|
parser->lookahead = parser->language->lex_fn(
|
|
|
|
|
&parser->lexer, parser->language->lex_states[state]);
|
2014-07-30 23:40:02 -07:00
|
|
|
return actions_for_state(parser->language, state)[parser->lookahead->symbol];
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-20 21:43:27 -07:00
|
|
|
#define DEBUG_PARSE(...) \
|
|
|
|
|
if (parser->debug) { \
|
|
|
|
|
fprintf(stderr, "\n" __VA_ARGS__); \
|
|
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 21:43:27 -07:00
|
|
|
TSTree *ts_parser_step(TSParser *parser) {
|
2014-07-20 20:27:33 -07:00
|
|
|
TSParseAction action = ts_parser_next_action(parser);
|
2014-07-20 21:43:27 -07:00
|
|
|
DEBUG_PARSE("LOOKAHEAD %s",
|
2014-07-30 23:40:02 -07:00
|
|
|
parser->language->symbol_names[parser->lookahead->symbol]);
|
2014-07-20 20:27:33 -07:00
|
|
|
switch (action.type) {
|
|
|
|
|
case TSParseActionTypeShift:
|
|
|
|
|
DEBUG_PARSE("SHIFT %d", action.data.to_state);
|
|
|
|
|
ts_parser_shift(parser, action.data.to_state);
|
|
|
|
|
return NULL;
|
|
|
|
|
case TSParseActionTypeShiftExtra:
|
|
|
|
|
DEBUG_PARSE("SHIFT EXTRA");
|
|
|
|
|
ts_parser_shift_extra(parser);
|
|
|
|
|
return NULL;
|
|
|
|
|
case TSParseActionTypeReduce:
|
2014-07-20 21:43:27 -07:00
|
|
|
DEBUG_PARSE("REDUCE %s %d",
|
2014-07-30 23:40:02 -07:00
|
|
|
parser->language->symbol_names[action.data.symbol],
|
2014-07-20 21:43:27 -07:00
|
|
|
action.data.child_count);
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_parser_reduce(parser, action.data.symbol, action.data.child_count);
|
|
|
|
|
return NULL;
|
|
|
|
|
case TSParseActionTypeReduceExtra:
|
|
|
|
|
if (!ts_parser_reduce_extra(parser, action.data.symbol))
|
|
|
|
|
goto error;
|
|
|
|
|
DEBUG_PARSE("REDUCE EXTRA");
|
|
|
|
|
return NULL;
|
|
|
|
|
case TSParseActionTypeAccept:
|
|
|
|
|
DEBUG_PARSE("ACCEPT");
|
|
|
|
|
return ts_parser_tree_root(parser);
|
|
|
|
|
case TSParseActionTypeError:
|
|
|
|
|
goto error;
|
|
|
|
|
default:
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2014-07-10 13:14:52 -07:00
|
|
|
error:
|
2014-07-20 20:27:33 -07:00
|
|
|
DEBUG_PARSE("ERROR");
|
|
|
|
|
if (!ts_parser_handle_error(parser))
|
|
|
|
|
return ts_parser_tree_root(parser);
|
|
|
|
|
else
|
|
|
|
|
return NULL;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2014-07-20 21:43:27 -07:00
|
|
|
const TSTree *ts_parser_parse(TSParser *parser, TSInput input,
|
|
|
|
|
TSInputEdit *edit) {
|
2014-07-20 20:27:33 -07:00
|
|
|
ts_parser_start(parser, input, edit);
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2014-07-20 20:27:33 -07:00
|
|
|
for (;;) {
|
|
|
|
|
const TSTree *tree = ts_parser_step(parser);
|
2014-07-20 21:43:27 -07:00
|
|
|
if (tree)
|
|
|
|
|
return tree;
|
2014-07-20 20:27:33 -07:00
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|