2015-09-19 13:19:49 -07:00
|
|
|
#include "runtime/parser.h"
|
2015-11-20 12:55:01 -08:00
|
|
|
#include <assert.h>
|
2014-07-10 13:14:52 -07:00
|
|
|
#include <stdio.h>
|
2016-06-02 14:04:48 -07:00
|
|
|
#include <limits.h>
|
2014-10-09 14:02:03 -07:00
|
|
|
#include <stdbool.h>
|
2014-07-10 13:14:52 -07:00
|
|
|
#include "tree_sitter/runtime.h"
|
|
|
|
|
#include "runtime/tree.h"
|
2014-07-30 23:40:02 -07:00
|
|
|
#include "runtime/lexer.h"
|
2014-09-26 16:15:07 -07:00
|
|
|
#include "runtime/length.h"
|
2016-02-17 20:41:29 -08:00
|
|
|
#include "runtime/array.h"
|
2015-11-20 12:00:49 -08:00
|
|
|
#include "runtime/language.h"
|
2016-01-15 15:08:42 -08:00
|
|
|
#include "runtime/alloc.h"
|
2016-05-09 14:31:44 -07:00
|
|
|
#include "runtime/reduce_action.h"
|
2016-08-31 10:51:59 -07:00
|
|
|
#include "runtime/error_costs.h"
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2016-10-05 14:02:49 -07:00
|
|
|
#define LOG(...) \
|
|
|
|
|
if (self->lexer.logger.log) { \
|
|
|
|
|
snprintf(self->lexer.debug_buffer, TS_DEBUG_BUFFER_SIZE, __VA_ARGS__); \
|
|
|
|
|
self->lexer.logger.log(self->lexer.logger.payload, TSLogTypeParse, \
|
|
|
|
|
self->lexer.debug_buffer); \
|
|
|
|
|
} \
|
|
|
|
|
if (self->print_debugging_graphs) { \
|
|
|
|
|
fprintf(stderr, "graph {\nlabel=\""); \
|
|
|
|
|
fprintf(stderr, __VA_ARGS__); \
|
|
|
|
|
fprintf(stderr, "\"\n}\n\n"); \
|
2016-02-23 09:45:27 -08:00
|
|
|
}
|
|
|
|
|
|
2016-04-04 12:25:57 -07:00
|
|
|
#define LOG_STACK() \
|
|
|
|
|
if (self->print_debugging_graphs) { \
|
|
|
|
|
ts_stack_print_dot_graph(self->stack, self->language->symbol_names, \
|
|
|
|
|
stderr); \
|
|
|
|
|
fputs("\n\n", stderr); \
|
2016-02-23 09:45:27 -08:00
|
|
|
}
|
|
|
|
|
|
2016-06-22 21:04:35 -07:00
|
|
|
#define LOG_TREE() \
|
|
|
|
|
if (self->print_debugging_graphs) { \
|
|
|
|
|
ts_tree_print_dot_graph(self->finished_tree, self->language, stderr); \
|
2016-06-22 22:03:27 -07:00
|
|
|
fputs("\n", stderr); \
|
2016-06-22 21:04:35 -07:00
|
|
|
}
|
|
|
|
|
|
2016-03-02 09:55:25 -08:00
|
|
|
#define SYM_NAME(symbol) ts_language_symbol_name(self->language, symbol)
|
2014-10-13 21:20:08 -07:00
|
|
|
|
2016-03-07 20:06:46 -08:00
|
|
|
typedef struct {
|
2016-08-29 12:08:58 -07:00
|
|
|
Parser *parser;
|
2016-03-07 20:06:46 -08:00
|
|
|
TSSymbol lookahead_symbol;
|
2016-05-09 14:31:44 -07:00
|
|
|
TreeArray *trees_above_error;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t tree_count_above_error;
|
2016-04-15 21:28:00 -07:00
|
|
|
bool found_repair;
|
2016-05-09 14:31:44 -07:00
|
|
|
ReduceAction best_repair;
|
2016-04-15 21:28:00 -07:00
|
|
|
TSStateId best_repair_next_state;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t best_repair_skip_count;
|
2016-03-07 20:06:46 -08:00
|
|
|
} ErrorRepairSession;
|
2016-03-02 21:03:55 -08:00
|
|
|
|
2016-03-07 20:06:46 -08:00
|
|
|
typedef struct {
|
2016-11-14 17:25:55 -08:00
|
|
|
Parser *parser;
|
|
|
|
|
TSSymbol lookahead_symbol;
|
|
|
|
|
} SkipPrecedingTreesSession;
|
2016-03-03 10:21:57 -08:00
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static void parser__push(Parser *self, StackVersion version, Tree *tree,
|
2016-08-29 12:08:58 -07:00
|
|
|
TSStateId state) {
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, version, tree, false, state);
|
2016-06-22 22:32:38 -07:00
|
|
|
ts_tree_release(tree);
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static bool parser__breakdown_top_of_stack(Parser *self, StackVersion version) {
|
2016-03-31 12:03:07 -07:00
|
|
|
bool did_break_down = false;
|
2016-06-22 22:32:38 -07:00
|
|
|
bool pending = false;
|
2015-12-06 21:10:47 -08:00
|
|
|
|
2015-12-02 11:24:13 -08:00
|
|
|
do {
|
2016-04-04 12:25:57 -07:00
|
|
|
StackPopResult pop = ts_stack_pop_pending(self->stack, version);
|
2016-03-03 11:05:37 -08:00
|
|
|
if (!pop.slices.size)
|
2016-03-31 12:03:07 -07:00
|
|
|
break;
|
2015-12-06 21:10:47 -08:00
|
|
|
|
2016-03-31 12:03:07 -07:00
|
|
|
did_break_down = true;
|
2016-06-22 22:32:38 -07:00
|
|
|
pending = false;
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < pop.slices.size; i++) {
|
2016-03-03 11:05:37 -08:00
|
|
|
StackSlice slice = pop.slices.contents[i];
|
2016-04-04 12:25:57 -07:00
|
|
|
TSStateId state = ts_stack_top_state(self->stack, slice.version);
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *parent = *array_front(&slice.trees);
|
2016-06-22 22:32:38 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = 0; j < parent->child_count; j++) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *child = parent->children[j];
|
2016-06-22 22:32:38 -07:00
|
|
|
pending = child->child_count > 0;
|
2016-03-31 12:03:07 -07:00
|
|
|
|
2016-06-22 22:32:38 -07:00
|
|
|
if (child->symbol == ts_builtin_sym_error) {
|
2016-10-05 14:02:49 -07:00
|
|
|
state = ERROR_STATE;
|
2016-06-22 22:32:38 -07:00
|
|
|
} else if (!child->extra) {
|
2016-11-14 08:36:06 -08:00
|
|
|
state = ts_language_next_state(self->language, state, child->symbol);
|
2015-12-06 21:10:47 -08:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, slice.version, child, pending, state);
|
2016-01-19 18:07:24 -08:00
|
|
|
}
|
2015-12-02 11:24:13 -08:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = 1; j < slice.trees.size; j++) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *tree = slice.trees.contents[j];
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__push(self, slice.version, tree, state);
|
2016-01-19 18:07:24 -08:00
|
|
|
}
|
2016-01-21 14:07:38 -07:00
|
|
|
|
2016-07-01 15:08:19 -07:00
|
|
|
LOG("breakdown_top_of_stack tree:%s", SYM_NAME(parent->symbol));
|
|
|
|
|
LOG_STACK();
|
|
|
|
|
|
2016-10-05 14:02:49 -07:00
|
|
|
ts_stack_decrease_push_count(self->stack, slice.version,
|
|
|
|
|
parent->child_count + 1);
|
2016-06-22 22:32:38 -07:00
|
|
|
ts_tree_release(parent);
|
|
|
|
|
array_delete(&slice.trees);
|
2016-02-23 17:35:50 -08:00
|
|
|
}
|
2016-06-22 22:32:38 -07:00
|
|
|
} while (pending);
|
2016-01-19 18:07:24 -08:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
return did_break_down;
|
2015-12-02 11:24:13 -08:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static bool parser__breakdown_lookahead(Parser *self, Tree **lookahead,
|
2016-08-29 12:08:58 -07:00
|
|
|
TSStateId state,
|
|
|
|
|
ReusableNode *reusable_node) {
|
2016-07-17 13:35:43 -07:00
|
|
|
bool result = false;
|
|
|
|
|
while (reusable_node->tree->child_count > 0 &&
|
2016-10-05 14:02:49 -07:00
|
|
|
(self->is_split || reusable_node->tree->parse_state != state ||
|
2016-07-17 13:35:43 -07:00
|
|
|
reusable_node->tree->fragile_left ||
|
|
|
|
|
reusable_node->tree->fragile_right)) {
|
|
|
|
|
LOG("state_mismatch sym:%s", SYM_NAME(reusable_node->tree->symbol));
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_breakdown(reusable_node);
|
2016-07-17 13:35:43 -07:00
|
|
|
result = true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (result) {
|
|
|
|
|
ts_tree_release(*lookahead);
|
|
|
|
|
ts_tree_retain(*lookahead = reusable_node->tree);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return result;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static bool parser__can_reuse(Parser *self, TSStateId state, Tree *tree,
|
2016-08-29 12:08:58 -07:00
|
|
|
TableEntry *table_entry) {
|
2016-12-20 17:06:20 -08:00
|
|
|
TSLexMode current_lex_mode = self->language->lex_modes[state];
|
|
|
|
|
if (tree->first_leaf.lex_mode.lex_state == current_lex_mode.lex_state &&
|
2016-12-21 11:24:41 -08:00
|
|
|
tree->first_leaf.lex_mode.external_lex_state == current_lex_mode.external_lex_state)
|
2016-07-17 13:35:43 -07:00
|
|
|
return true;
|
2017-01-05 10:06:43 -08:00
|
|
|
if (tree->size.bytes == 0) return false;
|
2016-07-17 13:35:43 -07:00
|
|
|
if (!table_entry->is_reusable)
|
|
|
|
|
return false;
|
|
|
|
|
if (!table_entry->depends_on_lookahead)
|
|
|
|
|
return true;
|
2016-08-30 10:58:25 -07:00
|
|
|
return tree->child_count > 1 && tree->error_cost == 0;
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
static bool parser__condense_stack(Parser *self) {
|
2016-06-02 14:04:48 -07:00
|
|
|
bool result = false;
|
|
|
|
|
for (StackVersion i = 0; i < ts_stack_version_count(self->stack); i++) {
|
|
|
|
|
if (ts_stack_is_halted(self->stack, i)) {
|
|
|
|
|
ts_stack_remove_version(self->stack, i);
|
2016-08-31 10:51:59 -07:00
|
|
|
result = true;
|
2016-06-02 14:04:48 -07:00
|
|
|
i--;
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2016-08-31 10:51:59 -07:00
|
|
|
ErrorStatus error_status = ts_stack_error_status(self->stack, i);
|
2016-06-02 14:04:48 -07:00
|
|
|
|
2016-08-31 10:51:59 -07:00
|
|
|
for (StackVersion j = 0; j < i; j++) {
|
2016-06-02 14:04:48 -07:00
|
|
|
if (ts_stack_merge(self->stack, j, i)) {
|
2016-08-31 10:51:59 -07:00
|
|
|
result = true;
|
|
|
|
|
i--;
|
2016-06-02 14:04:48 -07:00
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2016-08-31 10:51:59 -07:00
|
|
|
switch (error_status_compare(error_status,
|
|
|
|
|
ts_stack_error_status(self->stack, j))) {
|
|
|
|
|
case -1:
|
|
|
|
|
ts_stack_remove_version(self->stack, j);
|
|
|
|
|
result = true;
|
|
|
|
|
i--;
|
|
|
|
|
j--;
|
|
|
|
|
break;
|
|
|
|
|
case 1:
|
|
|
|
|
ts_stack_remove_version(self->stack, i);
|
|
|
|
|
result = true;
|
|
|
|
|
i--;
|
|
|
|
|
break;
|
|
|
|
|
}
|
2016-06-02 14:04:48 -07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return result;
|
|
|
|
|
}
|
|
|
|
|
|
2016-12-20 17:06:20 -08:00
|
|
|
static StackIterateAction parser__restore_external_scanner_callback(
|
|
|
|
|
void *payload, TSStateId state, TreeArray *trees, uint32_t tree_count,
|
|
|
|
|
bool is_done, bool is_pending) {
|
|
|
|
|
Parser *self = payload;
|
|
|
|
|
if (tree_count > 0) {
|
|
|
|
|
Tree *tree = *array_back(trees);
|
2016-12-21 13:59:56 -08:00
|
|
|
if (tree->has_external_token_state) {
|
2017-01-05 10:06:43 -08:00
|
|
|
const TSExternalTokenState *state = ts_tree_last_external_token_state(tree);
|
2016-12-21 13:59:56 -08:00
|
|
|
self->language->external_scanner.deserialize(
|
|
|
|
|
self->external_scanner_payload,
|
2017-01-05 10:06:43 -08:00
|
|
|
*state
|
2016-12-21 13:59:56 -08:00
|
|
|
);
|
2017-01-05 10:06:43 -08:00
|
|
|
LOG("deserialized_external_scanner");
|
2016-12-20 17:06:20 -08:00
|
|
|
return StackIterateStop;
|
2016-12-02 22:03:48 -08:00
|
|
|
}
|
2017-01-05 10:06:43 -08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (is_done) {
|
|
|
|
|
LOG("no_previous_external_token");
|
2016-12-20 17:06:20 -08:00
|
|
|
self->language->external_scanner.reset(self->external_scanner_payload);
|
|
|
|
|
return StackIterateStop;
|
2016-12-02 22:03:48 -08:00
|
|
|
}
|
|
|
|
|
|
2016-12-20 17:06:20 -08:00
|
|
|
return StackIterateNone;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void parser__restore_external_scanner(Parser *self, StackVersion version) {
|
2016-12-21 13:59:56 -08:00
|
|
|
if (!self->lexer.needs_to_restore_external_scanner) return;
|
2017-01-05 10:06:43 -08:00
|
|
|
LOG("restore_external_scanner");
|
2016-12-20 17:06:20 -08:00
|
|
|
StackPopResult pop = ts_stack_iterate(self->stack, version, parser__restore_external_scanner_callback, self);
|
|
|
|
|
if (pop.slices.size > 0) {
|
|
|
|
|
StackSlice slice = pop.slices.contents[0];
|
|
|
|
|
for (size_t i = 1; i < slice.trees.size; i++) {
|
|
|
|
|
Tree *tree = slice.trees.contents[i];
|
|
|
|
|
if (tree->has_external_tokens) {
|
|
|
|
|
printf("RE-SCANNING TREE: %s\n", ts_tree_string(tree, self->language, true));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
ts_tree_array_delete(&slice.trees);
|
|
|
|
|
}
|
2016-12-05 11:50:24 -08:00
|
|
|
}
|
2016-06-21 07:28:04 -07:00
|
|
|
|
2016-12-20 17:06:20 -08:00
|
|
|
static Tree *parser__lex(Parser *self, StackVersion version) {
|
|
|
|
|
TSStateId parse_state = ts_stack_top_state(self->stack, version);
|
|
|
|
|
Length start_position = ts_stack_top_position(self->stack, version);
|
2016-12-05 11:50:24 -08:00
|
|
|
TSLexMode lex_mode = self->language->lex_modes[parse_state];
|
2016-12-20 17:06:20 -08:00
|
|
|
const bool *external_tokens = ts_language_enabled_external_tokens(
|
|
|
|
|
self->language,
|
2016-12-21 11:24:41 -08:00
|
|
|
lex_mode.external_lex_state
|
2016-12-20 17:06:20 -08:00
|
|
|
);
|
2016-12-05 11:50:24 -08:00
|
|
|
|
2016-12-20 17:06:20 -08:00
|
|
|
bool found_external_token = false;
|
2016-12-05 11:50:24 -08:00
|
|
|
bool found_error = false;
|
2016-09-03 23:40:57 -07:00
|
|
|
bool skipped_error = false;
|
|
|
|
|
int32_t first_error_character = 0;
|
2016-11-09 20:59:05 -08:00
|
|
|
Length error_start_position, error_end_position;
|
2016-12-20 17:06:20 -08:00
|
|
|
ts_lexer_reset(&self->lexer, start_position);
|
|
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
|
Length current_position = self->lexer.current_position;
|
|
|
|
|
|
|
|
|
|
if (external_tokens) {
|
2016-12-21 11:24:41 -08:00
|
|
|
LOG("lex_external state:%d, row:%u, column:%u", lex_mode.external_lex_state,
|
2016-12-20 17:06:20 -08:00
|
|
|
current_position.extent.row, current_position.extent.column);
|
|
|
|
|
parser__restore_external_scanner(self, version);
|
|
|
|
|
ts_lexer_start(&self->lexer);
|
|
|
|
|
if (self->language->external_scanner.scan(self->external_scanner_payload,
|
|
|
|
|
&self->lexer.data, external_tokens)) {
|
2016-12-21 13:59:56 -08:00
|
|
|
self->lexer.last_external_token_end_byte = self->lexer.current_position.bytes;
|
|
|
|
|
self->lexer.needs_to_restore_external_scanner = false;
|
2016-12-20 17:06:20 -08:00
|
|
|
found_external_token = true;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
ts_lexer_reset(&self->lexer, current_position);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LOG("lex_internal state:%d, row:%u, column:%u", lex_mode.lex_state,
|
|
|
|
|
current_position.extent.row, current_position.extent.column);
|
|
|
|
|
ts_lexer_start(&self->lexer);
|
|
|
|
|
if (self->language->lex_fn(&self->lexer.data, lex_mode.lex_state)) {
|
|
|
|
|
break;
|
|
|
|
|
}
|
2016-09-03 23:40:57 -07:00
|
|
|
|
2016-12-05 11:50:24 -08:00
|
|
|
if (!found_error) {
|
2016-09-03 22:46:14 -07:00
|
|
|
LOG("retry_in_error_mode");
|
2016-12-05 11:50:24 -08:00
|
|
|
found_error = true;
|
|
|
|
|
lex_mode = self->language->lex_modes[ERROR_STATE];
|
2016-12-20 17:06:20 -08:00
|
|
|
external_tokens = ts_language_enabled_external_tokens(
|
|
|
|
|
self->language,
|
2016-12-21 11:24:41 -08:00
|
|
|
lex_mode.external_lex_state
|
2016-12-20 17:06:20 -08:00
|
|
|
);
|
2016-09-19 13:35:08 -07:00
|
|
|
ts_lexer_reset(&self->lexer, start_position);
|
2016-09-03 22:46:14 -07:00
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2016-09-19 13:35:08 -07:00
|
|
|
if (!skipped_error) {
|
2016-12-20 17:06:20 -08:00
|
|
|
LOG("skip_unrecognized_character");
|
|
|
|
|
skipped_error = true;
|
2016-09-19 13:35:08 -07:00
|
|
|
error_start_position = self->lexer.token_start_position;
|
2016-10-16 21:10:25 -07:00
|
|
|
first_error_character = self->lexer.data.lookahead;
|
2016-09-03 22:46:14 -07:00
|
|
|
}
|
|
|
|
|
|
2016-09-19 13:35:08 -07:00
|
|
|
if (self->lexer.current_position.bytes == error_end_position.bytes) {
|
2016-10-16 21:10:25 -07:00
|
|
|
if (self->lexer.data.lookahead == 0) {
|
|
|
|
|
self->lexer.data.result_symbol = ts_builtin_sym_error;
|
2016-09-19 13:35:08 -07:00
|
|
|
break;
|
2016-09-03 23:40:57 -07:00
|
|
|
}
|
2016-12-05 16:36:34 -08:00
|
|
|
self->lexer.data.advance(&self->lexer, false);
|
2016-09-03 22:46:14 -07:00
|
|
|
}
|
|
|
|
|
|
2016-09-19 13:35:08 -07:00
|
|
|
error_end_position = self->lexer.current_position;
|
2016-06-17 21:26:03 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *result;
|
2016-09-03 23:40:57 -07:00
|
|
|
if (skipped_error) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Length padding = length_sub(error_start_position, start_position);
|
|
|
|
|
Length size = length_sub(error_end_position, error_start_position);
|
2016-09-03 23:40:57 -07:00
|
|
|
ts_lexer_reset(&self->lexer, error_end_position);
|
|
|
|
|
result = ts_tree_make_error(size, padding, first_error_character);
|
2016-05-20 20:26:03 -07:00
|
|
|
} else {
|
2016-10-05 14:02:49 -07:00
|
|
|
TSSymbol symbol = self->lexer.data.result_symbol;
|
2016-12-21 11:24:41 -08:00
|
|
|
if (found_external_token) symbol = self->language->external_scanner.symbol_map[symbol];
|
2016-12-20 17:06:20 -08:00
|
|
|
|
2016-12-02 22:03:48 -08:00
|
|
|
Length padding = length_sub(self->lexer.token_start_position, start_position);
|
|
|
|
|
Length size = length_sub(self->lexer.current_position, self->lexer.token_start_position);
|
|
|
|
|
TSSymbolMetadata metadata = ts_language_symbol_metadata(self->language, symbol);
|
|
|
|
|
result = ts_tree_make_leaf(symbol, padding, size, metadata);
|
2016-12-20 17:06:20 -08:00
|
|
|
|
|
|
|
|
if (found_external_token) {
|
|
|
|
|
result->has_external_tokens = true;
|
|
|
|
|
if (self->language->external_scanner.serialize(self->external_scanner_payload, result->external_token_state)) {
|
|
|
|
|
result->has_external_token_state = true;
|
|
|
|
|
self->last_external_token = result;
|
|
|
|
|
}
|
|
|
|
|
}
|
2016-05-20 20:26:03 -07:00
|
|
|
}
|
|
|
|
|
|
2016-09-03 23:40:57 -07:00
|
|
|
result->parse_state = parse_state;
|
2016-12-20 17:06:20 -08:00
|
|
|
result->first_leaf.lex_mode = lex_mode;
|
|
|
|
|
|
|
|
|
|
LOG("lexed_lookahead sym:%s, size:%u", SYM_NAME(result->symbol), result->size.bytes);
|
2016-05-20 20:26:03 -07:00
|
|
|
return result;
|
|
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
static void parser__clear_cached_token(Parser *self) {
|
2016-07-01 15:08:19 -07:00
|
|
|
ts_tree_release(self->cached_token);
|
|
|
|
|
self->cached_token = NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static Tree *parser__get_lookahead(Parser *self, StackVersion version,
|
2016-12-20 17:06:20 -08:00
|
|
|
ReusableNode *reusable_node,
|
|
|
|
|
bool *is_fresh) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Length position = ts_stack_top_position(self->stack, version);
|
2014-10-14 09:32:11 -07:00
|
|
|
|
2016-04-01 21:17:23 -07:00
|
|
|
while (reusable_node->tree) {
|
2016-09-13 13:08:52 -07:00
|
|
|
if (reusable_node->byte_index > position.bytes) {
|
2016-12-20 17:06:20 -08:00
|
|
|
LOG("before_reusable_node sym:%s", SYM_NAME(reusable_node->tree->symbol));
|
2015-11-18 08:47:15 -08:00
|
|
|
break;
|
2015-09-13 19:47:45 -07:00
|
|
|
}
|
2014-10-14 09:32:11 -07:00
|
|
|
|
2016-09-13 13:08:52 -07:00
|
|
|
if (reusable_node->byte_index < position.bytes) {
|
2016-12-20 17:06:20 -08:00
|
|
|
LOG("past_reusable sym:%s", SYM_NAME(reusable_node->tree->symbol));
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_pop(reusable_node);
|
2015-09-13 19:47:45 -07:00
|
|
|
continue;
|
|
|
|
|
}
|
2014-10-14 09:32:11 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
if (reusable_node->tree->has_changes) {
|
2016-11-14 12:15:24 -08:00
|
|
|
LOG("cant_reuse_changed tree:%s, size:%u",
|
2016-07-01 15:08:19 -07:00
|
|
|
SYM_NAME(reusable_node->tree->symbol),
|
2016-09-13 13:08:52 -07:00
|
|
|
reusable_node->tree->size.bytes);
|
2017-01-05 10:06:43 -08:00
|
|
|
if (!reusable_node_breakdown(reusable_node)) {
|
|
|
|
|
reusable_node_pop(reusable_node);
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__breakdown_top_of_stack(self, version);
|
2016-07-01 15:08:19 -07:00
|
|
|
}
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
if (reusable_node->tree->symbol == ts_builtin_sym_error) {
|
2016-11-14 12:15:24 -08:00
|
|
|
LOG("cant_reuse_error tree:%s, size:%u",
|
2016-07-01 15:08:19 -07:00
|
|
|
SYM_NAME(reusable_node->tree->symbol),
|
2016-09-13 13:08:52 -07:00
|
|
|
reusable_node->tree->size.bytes);
|
2017-01-05 10:06:43 -08:00
|
|
|
if (!reusable_node_breakdown(reusable_node)) {
|
|
|
|
|
reusable_node_pop(reusable_node);
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__breakdown_top_of_stack(self, version);
|
2016-06-22 22:03:27 -07:00
|
|
|
}
|
2015-09-13 19:47:45 -07:00
|
|
|
continue;
|
2014-10-14 09:32:11 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *result = reusable_node->tree;
|
2016-01-29 17:31:43 -08:00
|
|
|
ts_tree_retain(result);
|
2015-11-18 08:47:15 -08:00
|
|
|
return result;
|
2014-10-14 09:32:11 -07:00
|
|
|
}
|
2014-10-22 12:54:46 -07:00
|
|
|
|
2016-09-13 13:08:52 -07:00
|
|
|
if (self->cached_token && position.bytes == self->cached_token_byte_index) {
|
2016-07-01 15:08:19 -07:00
|
|
|
ts_tree_retain(self->cached_token);
|
|
|
|
|
return self->cached_token;
|
|
|
|
|
}
|
|
|
|
|
|
2016-12-20 17:06:20 -08:00
|
|
|
*is_fresh = true;
|
|
|
|
|
return parser__lex(self, version);
|
2015-11-18 08:47:15 -08:00
|
|
|
}
|
2015-10-06 16:22:58 -07:00
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static bool parser__select_tree(Parser *self, Tree *left, Tree *right) {
|
2016-04-24 00:54:20 -07:00
|
|
|
if (!left)
|
|
|
|
|
return true;
|
|
|
|
|
if (!right)
|
|
|
|
|
return false;
|
2016-08-30 10:58:25 -07:00
|
|
|
if (right->error_cost < left->error_cost) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("select_smaller_error symbol:%s, over_symbol:%s",
|
2016-06-23 11:42:43 -07:00
|
|
|
SYM_NAME(right->symbol), SYM_NAME(left->symbol));
|
2016-04-24 00:54:20 -07:00
|
|
|
return true;
|
2015-12-08 12:25:41 -08:00
|
|
|
}
|
2016-08-30 10:58:25 -07:00
|
|
|
if (left->error_cost < right->error_cost) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("select_smaller_error symbol:%s, over_symbol:%s",
|
2016-06-23 11:42:43 -07:00
|
|
|
SYM_NAME(left->symbol), SYM_NAME(right->symbol));
|
2016-04-24 00:54:20 -07:00
|
|
|
return false;
|
|
|
|
|
}
|
2014-10-09 14:02:03 -07:00
|
|
|
|
2016-05-09 14:31:44 -07:00
|
|
|
int comparison = ts_tree_compare(left, right);
|
|
|
|
|
switch (comparison) {
|
|
|
|
|
case -1:
|
2016-06-23 11:42:43 -07:00
|
|
|
LOG("select_earlier symbol:%s, over_symbol:%s", SYM_NAME(left->symbol),
|
|
|
|
|
SYM_NAME(right->symbol));
|
2016-05-09 14:31:44 -07:00
|
|
|
return false;
|
|
|
|
|
break;
|
|
|
|
|
case 1:
|
2016-06-23 11:42:43 -07:00
|
|
|
LOG("select_earlier symbol:%s, over_symbol:%s", SYM_NAME(right->symbol),
|
|
|
|
|
SYM_NAME(left->symbol));
|
2016-05-09 14:31:44 -07:00
|
|
|
return true;
|
|
|
|
|
default:
|
2016-06-23 11:42:43 -07:00
|
|
|
LOG("select_existing symbol:%s, over_symbol:%s", SYM_NAME(left->symbol),
|
|
|
|
|
SYM_NAME(right->symbol));
|
2016-05-09 14:31:44 -07:00
|
|
|
return false;
|
|
|
|
|
}
|
2016-04-18 11:16:56 -07:00
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
static bool parser__better_version_exists(Parser *self, StackVersion version,
|
2016-08-31 10:51:59 -07:00
|
|
|
ErrorStatus my_error_status) {
|
|
|
|
|
if (self->finished_tree &&
|
|
|
|
|
self->finished_tree->error_cost <= my_error_status.cost)
|
2016-08-06 21:22:44 -07:00
|
|
|
return true;
|
|
|
|
|
|
2016-06-26 22:51:48 -07:00
|
|
|
for (StackVersion i = 0, n = ts_stack_version_count(self->stack); i < n; i++) {
|
|
|
|
|
if (i == version || ts_stack_is_halted(self->stack, i))
|
|
|
|
|
continue;
|
|
|
|
|
|
2016-08-31 10:51:59 -07:00
|
|
|
switch (error_status_compare(my_error_status,
|
|
|
|
|
ts_stack_error_status(self->stack, i))) {
|
|
|
|
|
case -1:
|
|
|
|
|
LOG("halt_other version:%u", i);
|
|
|
|
|
ts_stack_halt(self->stack, i);
|
|
|
|
|
break;
|
|
|
|
|
case 1:
|
|
|
|
|
return true;
|
2016-06-26 22:51:48 -07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static void parser__shift(Parser *self, StackVersion version, TSStateId state,
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *lookahead, bool extra) {
|
2016-06-12 17:26:15 -07:00
|
|
|
if (extra != lookahead->extra) {
|
2016-04-18 11:16:56 -07:00
|
|
|
TSSymbolMetadata metadata =
|
|
|
|
|
ts_language_symbol_metadata(self->language, lookahead->symbol);
|
|
|
|
|
if (metadata.structural && ts_stack_version_count(self->stack) > 1) {
|
2016-11-04 09:18:38 -07:00
|
|
|
lookahead = ts_tree_make_copy(lookahead);
|
2016-04-18 11:16:56 -07:00
|
|
|
} else {
|
|
|
|
|
ts_tree_retain(lookahead);
|
|
|
|
|
}
|
2016-06-12 17:26:15 -07:00
|
|
|
lookahead->extra = extra;
|
2016-04-18 11:16:56 -07:00
|
|
|
} else {
|
|
|
|
|
ts_tree_retain(lookahead);
|
|
|
|
|
}
|
|
|
|
|
|
2016-04-02 20:58:19 -07:00
|
|
|
bool is_pending = lookahead->child_count > 0;
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, version, lookahead, is_pending, state);
|
2016-04-18 11:16:56 -07:00
|
|
|
ts_tree_release(lookahead);
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static bool parser__switch_children(Parser *self, Tree *tree,
|
2016-11-14 12:15:24 -08:00
|
|
|
Tree **children, uint32_t count) {
|
2016-04-24 00:54:20 -07:00
|
|
|
self->scratch_tree.symbol = tree->symbol;
|
|
|
|
|
self->scratch_tree.child_count = 0;
|
|
|
|
|
ts_tree_set_children(&self->scratch_tree, count, children);
|
2016-08-29 12:08:58 -07:00
|
|
|
if (parser__select_tree(self, tree, &self->scratch_tree)) {
|
2016-04-24 00:54:20 -07:00
|
|
|
tree->size = self->scratch_tree.size;
|
|
|
|
|
tree->padding = self->scratch_tree.padding;
|
2016-08-30 10:58:25 -07:00
|
|
|
tree->error_cost = self->scratch_tree.error_cost;
|
2016-04-24 00:54:20 -07:00
|
|
|
tree->children = self->scratch_tree.children;
|
|
|
|
|
tree->child_count = self->scratch_tree.child_count;
|
|
|
|
|
tree->named_child_count = self->scratch_tree.named_child_count;
|
|
|
|
|
tree->visible_child_count = self->scratch_tree.visible_child_count;
|
|
|
|
|
return true;
|
|
|
|
|
} else {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
static StackPopResult parser__reduce(Parser *self, StackVersion version,
|
|
|
|
|
TSSymbol symbol, unsigned count,
|
|
|
|
|
bool fragile, bool allow_skipping) {
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t initial_version_count = ts_stack_version_count(self->stack);
|
2016-11-14 17:25:55 -08:00
|
|
|
|
2016-04-18 11:16:56 -07:00
|
|
|
StackPopResult pop = ts_stack_pop_count(self->stack, version, count);
|
2016-11-14 17:25:55 -08:00
|
|
|
if (pop.stopped_at_error)
|
|
|
|
|
return pop;
|
2016-04-07 10:40:33 -07:00
|
|
|
|
2016-04-18 11:16:56 -07:00
|
|
|
const TSLanguage *language = self->language;
|
|
|
|
|
TSSymbolMetadata metadata = ts_language_symbol_metadata(language, symbol);
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < pop.slices.size; i++) {
|
2016-03-03 11:05:37 -08:00
|
|
|
StackSlice slice = pop.slices.contents[i];
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// Extra tokens on top of the stack should not be included in this new parent
|
|
|
|
|
// node. They will be re-pushed onto the stack after the parent node is
|
|
|
|
|
// created and pushed.
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t child_count = slice.trees.size;
|
2016-04-07 10:40:33 -07:00
|
|
|
while (child_count > 0 && slice.trees.contents[child_count - 1]->extra)
|
|
|
|
|
child_count--;
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
Tree *parent = ts_tree_make_node(symbol, child_count, slice.trees.contents, metadata);
|
2016-02-08 12:08:15 -08:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// This pop operation may have caused multiple stack versions to collapse
|
|
|
|
|
// into one, because they all diverged from a common state. In that case,
|
|
|
|
|
// choose one of the arrays of trees to be the parent node's children, and
|
|
|
|
|
// delete the rest of the tree arrays.
|
2016-04-24 00:54:20 -07:00
|
|
|
while (i + 1 < pop.slices.size) {
|
|
|
|
|
StackSlice next_slice = pop.slices.contents[i + 1];
|
|
|
|
|
if (next_slice.version != slice.version)
|
|
|
|
|
break;
|
|
|
|
|
i++;
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t child_count = next_slice.trees.size;
|
2016-04-24 00:54:20 -07:00
|
|
|
while (child_count > 0 && next_slice.trees.contents[child_count - 1]->extra)
|
|
|
|
|
child_count--;
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
if (parser__switch_children(self, parent, next_slice.trees.contents, child_count)) {
|
2016-04-24 00:54:20 -07:00
|
|
|
ts_tree_array_delete(&slice.trees);
|
|
|
|
|
slice = next_slice;
|
|
|
|
|
} else {
|
|
|
|
|
ts_tree_array_delete(&next_slice.trees);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-04-18 11:16:56 -07:00
|
|
|
TSStateId state = ts_stack_top_state(self->stack, slice.version);
|
2016-11-14 17:25:55 -08:00
|
|
|
TSStateId next_state = ts_language_next_state(language, state, symbol);
|
|
|
|
|
if (fragile || self->is_split || pop.slices.size > 1 || initial_version_count > 1) {
|
2016-03-07 20:06:46 -08:00
|
|
|
parent->fragile_left = true;
|
|
|
|
|
parent->fragile_right = true;
|
2016-06-27 14:39:12 -07:00
|
|
|
parent->parse_state = TS_TREE_STATE_NONE;
|
2016-04-07 10:40:33 -07:00
|
|
|
} else {
|
2016-04-18 11:16:56 -07:00
|
|
|
parent->parse_state = state;
|
2016-03-07 20:06:46 -08:00
|
|
|
}
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// If this pop operation terminated at the end of an error region, then
|
|
|
|
|
// create two stack versions: one in which the parent node is interpreted
|
|
|
|
|
// normally, and one in which the parent node is skipped.
|
2016-11-14 08:36:06 -08:00
|
|
|
if (state == ERROR_STATE && allow_skipping) {
|
2016-11-14 17:25:55 -08:00
|
|
|
StackVersion other_version = ts_stack_copy_version(self->stack, slice.version);
|
2016-09-06 10:22:16 -07:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, other_version, parent, false, ERROR_STATE);
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = parent->child_count; j < slice.trees.size; j++) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *tree = slice.trees.contents[j];
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, other_version, tree, false, ERROR_STATE);
|
2016-06-24 10:28:05 -07:00
|
|
|
}
|
2016-09-06 10:22:16 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
ErrorStatus error_status = ts_stack_error_status(self->stack, other_version);
|
2016-09-06 10:22:16 -07:00
|
|
|
if (parser__better_version_exists(self, version, error_status))
|
|
|
|
|
ts_stack_remove_version(self->stack, other_version);
|
2015-11-20 00:01:53 -08:00
|
|
|
}
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// Push the parent node onto the stack, along with any extra tokens that
|
|
|
|
|
// were previously on top of the stack.
|
2016-11-14 08:36:06 -08:00
|
|
|
parser__push(self, slice.version, parent, next_state);
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = parent->child_count; j < slice.trees.size; j++) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *tree = slice.trees.contents[j];
|
2016-11-14 08:36:06 -08:00
|
|
|
parser__push(self, slice.version, tree, next_state);
|
2015-12-02 07:53:15 -08:00
|
|
|
}
|
2015-07-08 17:34:21 -07:00
|
|
|
}
|
2015-05-27 10:53:02 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
for (StackVersion i = initial_version_count; i < ts_stack_version_count(self->stack); i++) {
|
2016-06-02 14:04:48 -07:00
|
|
|
for (StackVersion j = initial_version_count; j < i; j++) {
|
|
|
|
|
if (ts_stack_merge(self->stack, j, i)) {
|
|
|
|
|
i--;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2016-04-24 00:55:19 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
return pop;
|
2015-06-15 15:24:15 -07:00
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
static inline const TSParseAction *parser__reductions_after_sequence(
|
|
|
|
|
Parser *self, TSStateId start_state, const TreeArray *trees_below,
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t tree_count_below, const TreeArray *trees_above,
|
|
|
|
|
TSSymbol lookahead_symbol, uint32_t *count) {
|
2016-05-09 14:31:44 -07:00
|
|
|
TSStateId state = start_state;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t child_count = 0;
|
2016-06-10 13:15:45 -07:00
|
|
|
*count = 0;
|
2016-05-09 14:31:44 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < trees_below->size; i++) {
|
2016-08-29 09:32:57 -07:00
|
|
|
if (child_count == tree_count_below)
|
|
|
|
|
break;
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *tree = trees_below->contents[trees_below->size - 1 - i];
|
2016-11-14 08:36:06 -08:00
|
|
|
TSStateId next_state = ts_language_next_state(self->language, state, tree->symbol);
|
|
|
|
|
if (next_state == ERROR_STATE)
|
2016-06-10 13:15:45 -07:00
|
|
|
return NULL;
|
2016-11-14 08:36:06 -08:00
|
|
|
if (next_state != state) {
|
|
|
|
|
child_count++;
|
|
|
|
|
state = next_state;
|
|
|
|
|
}
|
2016-06-10 13:15:45 -07:00
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < trees_above->size; i++) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *tree = trees_above->contents[i];
|
2016-11-14 08:36:06 -08:00
|
|
|
TSStateId next_state = ts_language_next_state(self->language, state, tree->symbol);
|
|
|
|
|
if (next_state == ERROR_STATE)
|
2016-06-10 13:15:45 -07:00
|
|
|
return NULL;
|
2016-11-14 08:36:06 -08:00
|
|
|
if (next_state != state) {
|
|
|
|
|
child_count++;
|
|
|
|
|
state = next_state;
|
|
|
|
|
}
|
2016-06-10 13:15:45 -07:00
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
|
2016-06-12 17:41:25 -07:00
|
|
|
const TSParseAction *actions =
|
|
|
|
|
ts_language_actions(self->language, state, lookahead_symbol, count);
|
2016-06-10 13:15:45 -07:00
|
|
|
|
2016-06-21 22:53:48 -07:00
|
|
|
if (*count > 0 && actions[*count - 1].type != TSParseActionTypeReduce) {
|
2016-06-10 13:15:45 -07:00
|
|
|
(*count)--;
|
2016-06-21 22:53:48 -07:00
|
|
|
}
|
2016-06-10 13:15:45 -07:00
|
|
|
|
2016-10-26 11:10:56 -07:00
|
|
|
while (*count > 0 && actions[0].params.child_count < child_count) {
|
2016-06-10 13:15:45 -07:00
|
|
|
actions++;
|
|
|
|
|
(*count)--;
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
|
|
|
|
|
2016-10-26 11:10:56 -07:00
|
|
|
while (*count > 0 && actions[*count - 1].params.child_count > child_count) {
|
2016-06-10 13:15:45 -07:00
|
|
|
(*count)--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return actions;
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
static StackIterateAction parser__repair_error_callback(
|
2016-11-14 12:15:24 -08:00
|
|
|
void *payload, TSStateId state, TreeArray *trees, uint32_t tree_count,
|
2016-05-09 14:31:44 -07:00
|
|
|
bool is_done, bool is_pending) {
|
2016-06-10 13:15:45 -07:00
|
|
|
|
2016-03-07 20:06:46 -08:00
|
|
|
ErrorRepairSession *session = (ErrorRepairSession *)payload;
|
2016-08-29 12:08:58 -07:00
|
|
|
Parser *self = session->parser;
|
2016-03-07 20:06:46 -08:00
|
|
|
TSSymbol lookahead_symbol = session->lookahead_symbol;
|
2016-05-09 14:31:44 -07:00
|
|
|
ReduceActionSet *repairs = &self->reduce_actions;
|
|
|
|
|
TreeArray *trees_above_error = session->trees_above_error;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t tree_count_above_error = session->tree_count_above_error;
|
2016-06-10 13:15:45 -07:00
|
|
|
|
2016-04-15 21:28:00 -07:00
|
|
|
StackIterateAction result = StackIterateNone;
|
2016-03-07 20:06:46 -08:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t last_repair_count = -1;
|
|
|
|
|
uint32_t repair_reduction_count = -1;
|
2016-06-10 13:15:45 -07:00
|
|
|
const TSParseAction *repair_reductions = NULL;
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < repairs->size; i++) {
|
2016-05-09 14:31:44 -07:00
|
|
|
ReduceAction *repair = &repairs->contents[i];
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t count_needed_below_error = repair->count - tree_count_above_error;
|
2016-06-10 13:15:45 -07:00
|
|
|
if (count_needed_below_error > tree_count)
|
|
|
|
|
break;
|
2016-04-15 21:28:00 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t skip_count = tree_count - count_needed_below_error;
|
2016-05-09 14:31:44 -07:00
|
|
|
if (session->found_repair && skip_count >= session->best_repair_skip_count) {
|
|
|
|
|
array_erase(repairs, i--);
|
|
|
|
|
continue;
|
|
|
|
|
}
|
2016-04-15 21:28:00 -07:00
|
|
|
|
2016-11-14 08:36:06 -08:00
|
|
|
TSStateId state_after_repair = ts_language_next_state(self->language, state, repair->symbol);
|
2016-11-14 10:35:33 -08:00
|
|
|
if (state == ERROR_STATE || state_after_repair == ERROR_STATE)
|
|
|
|
|
continue;
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t action_count;
|
2016-11-14 10:35:33 -08:00
|
|
|
ts_language_actions(self->language, state_after_repair, lookahead_symbol, &action_count);
|
|
|
|
|
if (action_count == 0)
|
2016-05-09 14:31:44 -07:00
|
|
|
continue;
|
2016-03-02 21:03:55 -08:00
|
|
|
|
2016-06-10 13:15:45 -07:00
|
|
|
if (count_needed_below_error != last_repair_count) {
|
|
|
|
|
last_repair_count = count_needed_below_error;
|
2016-08-29 12:08:58 -07:00
|
|
|
repair_reductions = parser__reductions_after_sequence(
|
2016-06-12 17:41:25 -07:00
|
|
|
self, state, trees, count_needed_below_error, trees_above_error,
|
|
|
|
|
lookahead_symbol, &repair_reduction_count);
|
2016-06-10 13:15:45 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = 0; j < repair_reduction_count; j++) {
|
2016-10-26 11:10:56 -07:00
|
|
|
if (repair_reductions[j].params.symbol == repair->symbol) {
|
2016-06-10 13:15:45 -07:00
|
|
|
result |= StackIteratePop;
|
|
|
|
|
session->found_repair = true;
|
|
|
|
|
session->best_repair = *repair;
|
|
|
|
|
session->best_repair_skip_count = skip_count;
|
|
|
|
|
session->best_repair_next_state = state_after_repair;
|
|
|
|
|
array_erase(repairs, i--);
|
|
|
|
|
break;
|
|
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
2015-12-02 07:53:15 -08:00
|
|
|
}
|
2016-03-02 21:03:55 -08:00
|
|
|
|
2016-05-09 14:31:44 -07:00
|
|
|
if (repairs->size == 0)
|
2016-04-15 21:28:00 -07:00
|
|
|
result |= StackIterateStop;
|
|
|
|
|
|
|
|
|
|
return result;
|
2015-02-21 10:41:23 -08:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static bool parser__repair_error(Parser *self, StackSlice slice,
|
2016-11-14 17:25:55 -08:00
|
|
|
TSSymbol lookahead_symbol, TableEntry entry) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("repair_error");
|
2016-03-07 20:06:46 -08:00
|
|
|
ErrorRepairSession session = {
|
|
|
|
|
.parser = self,
|
2016-07-17 13:35:43 -07:00
|
|
|
.lookahead_symbol = lookahead_symbol,
|
2016-04-15 21:28:00 -07:00
|
|
|
.found_repair = false,
|
2016-05-09 14:31:44 -07:00
|
|
|
.trees_above_error = &slice.trees,
|
2016-06-10 13:15:45 -07:00
|
|
|
.tree_count_above_error = ts_tree_array_essential_count(&slice.trees),
|
2016-03-07 20:06:46 -08:00
|
|
|
};
|
2014-07-20 20:27:33 -07:00
|
|
|
|
2016-05-09 14:31:44 -07:00
|
|
|
array_clear(&self->reduce_actions);
|
2016-11-14 17:25:55 -08:00
|
|
|
for (uint32_t i = 0; i < entry.action_count; i++) {
|
|
|
|
|
if (entry.actions[i].type == TSParseActionTypeReduce) {
|
|
|
|
|
TSSymbol symbol = entry.actions[i].params.symbol;
|
|
|
|
|
uint32_t child_count = entry.actions[i].params.child_count;
|
2016-06-14 14:48:31 -07:00
|
|
|
if ((child_count > session.tree_count_above_error) ||
|
|
|
|
|
(child_count == session.tree_count_above_error &&
|
|
|
|
|
!ts_language_symbol_metadata(self->language, symbol).visible))
|
2016-11-14 12:15:24 -08:00
|
|
|
array_push(&self->reduce_actions, ((ReduceAction){
|
|
|
|
|
.symbol = symbol,
|
|
|
|
|
.count = child_count
|
|
|
|
|
}));
|
2016-06-14 14:48:31 -07:00
|
|
|
}
|
|
|
|
|
}
|
2016-04-15 21:28:00 -07:00
|
|
|
|
|
|
|
|
StackPopResult pop = ts_stack_iterate(
|
2016-11-14 17:25:55 -08:00
|
|
|
self->stack, slice.version, parser__repair_error_callback, &session);
|
2016-04-18 11:16:56 -07:00
|
|
|
|
2016-04-15 21:28:00 -07:00
|
|
|
if (!session.found_repair) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("no_repair_found");
|
2016-04-18 11:16:56 -07:00
|
|
|
ts_stack_remove_version(self->stack, slice.version);
|
2016-04-10 14:12:24 -07:00
|
|
|
ts_tree_array_delete(&slice.trees);
|
2016-11-04 09:18:38 -07:00
|
|
|
return false;
|
2016-03-03 11:05:37 -08:00
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
|
2016-05-09 14:31:44 -07:00
|
|
|
ReduceAction repair = session.best_repair;
|
2016-04-15 21:28:00 -07:00
|
|
|
TSStateId next_state = session.best_repair_next_state;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t skip_count = session.best_repair_skip_count;
|
|
|
|
|
uint32_t count_below = repair.count - session.tree_count_above_error;
|
2016-04-15 21:28:00 -07:00
|
|
|
TSSymbol symbol = repair.symbol;
|
|
|
|
|
|
|
|
|
|
StackSlice new_slice = array_pop(&pop.slices);
|
2016-05-29 22:36:47 -07:00
|
|
|
TreeArray children = new_slice.trees;
|
2016-04-15 21:28:00 -07:00
|
|
|
ts_stack_renumber_version(self->stack, new_slice.version, slice.version);
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = pop.slices.size - 1; i + 1 > 0; i--) {
|
2016-04-24 00:54:20 -07:00
|
|
|
StackSlice other_slice = pop.slices.contents[i];
|
2016-04-15 21:28:00 -07:00
|
|
|
ts_tree_array_delete(&other_slice.trees);
|
2016-04-24 00:54:20 -07:00
|
|
|
if (other_slice.version != pop.slices.contents[i + 1].version)
|
|
|
|
|
ts_stack_remove_version(self->stack, other_slice.version);
|
2016-04-15 21:28:00 -07:00
|
|
|
}
|
2015-11-11 16:56:58 -08:00
|
|
|
|
2016-05-29 22:36:47 -07:00
|
|
|
TreeArray skipped_children = array_new();
|
2016-11-04 09:18:38 -07:00
|
|
|
array_grow(&skipped_children, skip_count);
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = count_below; i < children.size; i++)
|
2016-05-29 22:36:47 -07:00
|
|
|
array_push(&skipped_children, children.contents[i]);
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *error = ts_tree_make_error_node(&skipped_children);
|
2016-06-10 13:15:45 -07:00
|
|
|
children.size = count_below;
|
2016-05-29 22:36:47 -07:00
|
|
|
array_push(&children, error);
|
2014-09-29 10:33:56 -07:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < slice.trees.size; i++)
|
2016-05-29 22:36:47 -07:00
|
|
|
array_push(&children, slice.trees.contents[i]);
|
2016-04-15 21:28:00 -07:00
|
|
|
array_delete(&slice.trees);
|
2016-04-10 14:12:24 -07:00
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *parent =
|
2016-05-29 22:36:47 -07:00
|
|
|
ts_tree_make_node(symbol, children.size, children.contents,
|
2016-03-07 20:06:46 -08:00
|
|
|
ts_language_symbol_metadata(self->language, symbol));
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__push(self, slice.version, parent, next_state);
|
2016-09-01 10:04:20 -07:00
|
|
|
ts_stack_decrease_push_count(self->stack, slice.version, error->child_count);
|
2016-05-10 15:24:06 -07:00
|
|
|
|
2016-08-31 10:51:59 -07:00
|
|
|
ErrorStatus error_status = ts_stack_error_status(self->stack, slice.version);
|
|
|
|
|
if (parser__better_version_exists(self, slice.version, error_status)) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("no_better_repair_found");
|
2016-06-14 14:48:31 -07:00
|
|
|
ts_stack_halt(self->stack, slice.version);
|
2016-11-04 09:18:38 -07:00
|
|
|
return false;
|
2016-06-14 14:48:31 -07:00
|
|
|
} else {
|
2016-11-14 12:15:24 -08:00
|
|
|
LOG("repair_found sym:%s, child_count:%u, cost:%u", SYM_NAME(symbol),
|
2016-08-30 10:58:25 -07:00
|
|
|
repair.count, parent->error_cost);
|
2016-11-04 09:18:38 -07:00
|
|
|
return true;
|
2016-06-02 14:04:48 -07:00
|
|
|
}
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
static void parser__start(Parser *self, TSInput input, Tree *previous_tree) {
|
2015-09-18 23:20:06 -07:00
|
|
|
if (previous_tree) {
|
2015-11-20 11:49:04 -08:00
|
|
|
LOG("parse_after_edit");
|
2015-09-13 19:47:45 -07:00
|
|
|
} else {
|
2015-11-20 11:49:04 -08:00
|
|
|
LOG("new_parse");
|
2015-09-13 19:47:45 -07:00
|
|
|
}
|
2015-09-18 23:20:06 -07:00
|
|
|
|
2016-12-20 13:12:01 -08:00
|
|
|
if (self->language->external_scanner.reset) {
|
|
|
|
|
self->language->external_scanner.reset(self->external_scanner_payload);
|
|
|
|
|
}
|
2016-11-30 09:34:47 -08:00
|
|
|
|
2015-12-03 10:00:39 -08:00
|
|
|
ts_lexer_set_input(&self->lexer, input);
|
2015-10-14 21:52:13 -07:00
|
|
|
ts_stack_clear(self->stack);
|
2017-01-05 10:06:43 -08:00
|
|
|
self->reusable_node = reusable_node_new(previous_tree);
|
2016-07-01 15:08:19 -07:00
|
|
|
self->cached_token = NULL;
|
2015-12-24 22:04:20 -08:00
|
|
|
self->finished_tree = NULL;
|
2015-09-13 19:47:45 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static void parser__accept(Parser *self, StackVersion version,
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *lookahead) {
|
2016-09-19 13:35:08 -07:00
|
|
|
lookahead->extra = true;
|
|
|
|
|
assert(lookahead->symbol == ts_builtin_sym_end);
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, version, lookahead, false, 1);
|
2016-04-24 00:54:20 -07:00
|
|
|
StackPopResult pop = ts_stack_pop_all(self->stack, version);
|
2016-03-10 11:57:33 -08:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < pop.slices.size; i++) {
|
2016-04-24 00:54:20 -07:00
|
|
|
StackSlice slice = pop.slices.contents[i];
|
|
|
|
|
TreeArray trees = slice.trees;
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *root = NULL;
|
2016-06-26 11:57:42 -07:00
|
|
|
if (trees.size == 1) {
|
|
|
|
|
root = trees.contents[0];
|
|
|
|
|
array_delete(&trees);
|
|
|
|
|
} else {
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t j = trees.size - 1; j + 1 > 0; j--) {
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *child = trees.contents[j];
|
2016-06-26 11:57:42 -07:00
|
|
|
if (!child->extra) {
|
2016-11-04 09:18:38 -07:00
|
|
|
root = ts_tree_make_copy(child);
|
2016-06-26 11:57:42 -07:00
|
|
|
root->child_count = 0;
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t k = 0; k < child->child_count; k++)
|
2016-06-26 11:57:42 -07:00
|
|
|
ts_tree_retain(child->children[k]);
|
2016-11-04 09:18:38 -07:00
|
|
|
array_splice(&trees, j, 1, child->child_count, child->children);
|
2016-06-26 11:57:42 -07:00
|
|
|
ts_tree_set_children(root, trees.size, trees.contents);
|
|
|
|
|
ts_tree_release(child);
|
|
|
|
|
break;
|
2016-04-24 00:54:20 -07:00
|
|
|
}
|
|
|
|
|
}
|
2015-12-02 07:53:15 -08:00
|
|
|
}
|
2016-06-26 11:57:42 -07:00
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
if (parser__select_tree(self, self->finished_tree, root)) {
|
2016-06-26 11:57:42 -07:00
|
|
|
ts_tree_release(self->finished_tree);
|
2016-07-17 13:35:43 -07:00
|
|
|
assert(root->ref_count > 0);
|
2016-06-26 11:57:42 -07:00
|
|
|
self->finished_tree = root;
|
|
|
|
|
} else {
|
|
|
|
|
ts_tree_release(root);
|
|
|
|
|
}
|
2015-12-02 07:53:15 -08:00
|
|
|
}
|
2016-01-19 18:07:24 -08:00
|
|
|
|
2016-04-24 00:54:20 -07:00
|
|
|
ts_stack_remove_version(self->stack, pop.slices.contents[0].version);
|
2016-06-02 14:04:48 -07:00
|
|
|
ts_stack_halt(self->stack, version);
|
2014-08-09 01:03:55 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
static bool parser__do_potential_reductions(Parser *self, StackVersion version) {
|
2016-05-29 22:36:47 -07:00
|
|
|
bool has_shift_action = false;
|
2016-08-29 09:34:08 -07:00
|
|
|
TSStateId state = ts_stack_top_state(self->stack, version);
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t previous_version_count = ts_stack_version_count(self->stack);
|
2016-08-29 09:34:08 -07:00
|
|
|
|
2016-05-09 14:31:44 -07:00
|
|
|
array_clear(&self->reduce_actions);
|
2016-11-14 08:36:06 -08:00
|
|
|
for (TSSymbol symbol = 0; symbol < self->language->token_count; symbol++) {
|
2016-08-29 09:34:08 -07:00
|
|
|
TableEntry entry;
|
|
|
|
|
ts_language_table_entry(self->language, state, symbol, &entry);
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < entry.action_count; i++) {
|
2016-08-29 09:34:08 -07:00
|
|
|
TSParseAction action = entry.actions[i];
|
2016-05-29 22:36:47 -07:00
|
|
|
if (action.extra)
|
|
|
|
|
continue;
|
|
|
|
|
switch (action.type) {
|
|
|
|
|
case TSParseActionTypeShift:
|
|
|
|
|
case TSParseActionTypeRecover:
|
|
|
|
|
has_shift_action = true;
|
|
|
|
|
break;
|
|
|
|
|
case TSParseActionTypeReduce:
|
2016-10-26 11:10:56 -07:00
|
|
|
if (action.params.child_count > 0)
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_reduce_action_set_add(&self->reduce_actions, (ReduceAction){
|
|
|
|
|
.symbol = action.params.symbol,
|
|
|
|
|
.count = action.params.child_count,
|
|
|
|
|
});
|
2016-05-29 22:36:47 -07:00
|
|
|
default:
|
|
|
|
|
break;
|
|
|
|
|
}
|
2016-03-10 11:57:33 -08:00
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
2016-03-10 11:57:33 -08:00
|
|
|
|
2016-05-29 22:36:47 -07:00
|
|
|
bool did_reduce = false;
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < self->reduce_actions.size; i++) {
|
2016-05-28 21:22:10 -07:00
|
|
|
ReduceAction action = self->reduce_actions.contents[i];
|
2016-11-14 17:25:55 -08:00
|
|
|
StackPopResult reduction =
|
2016-10-05 14:02:49 -07:00
|
|
|
parser__reduce(self, version, action.symbol, action.count, true, false);
|
2016-11-04 09:18:38 -07:00
|
|
|
if (reduction.stopped_at_error) {
|
2016-11-14 17:25:55 -08:00
|
|
|
ts_tree_array_delete(&reduction.slices.contents[0].trees);
|
|
|
|
|
ts_stack_remove_version(self->stack, reduction.slices.contents[0].version);
|
2016-11-04 09:18:38 -07:00
|
|
|
continue;
|
|
|
|
|
} else {
|
|
|
|
|
did_reduce = true;
|
2016-05-28 21:22:10 -07:00
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
|
|
|
|
|
2016-08-29 09:34:08 -07:00
|
|
|
if (did_reduce) {
|
|
|
|
|
if (has_shift_action) {
|
2016-11-04 09:18:38 -07:00
|
|
|
return true;
|
2016-08-29 09:34:08 -07:00
|
|
|
} else {
|
|
|
|
|
ts_stack_renumber_version(self->stack, previous_version_count, version);
|
2016-11-04 09:18:38 -07:00
|
|
|
return false;
|
2016-08-29 09:34:08 -07:00
|
|
|
}
|
|
|
|
|
} else {
|
2016-11-04 09:18:38 -07:00
|
|
|
return true;
|
2016-08-29 09:34:08 -07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
static StackIterateAction parser__skip_preceding_trees_callback(
|
2016-11-14 12:15:24 -08:00
|
|
|
void *payload, TSStateId state, TreeArray *trees, uint32_t tree_count,
|
2016-08-29 17:34:51 -07:00
|
|
|
bool is_done, bool is_pending) {
|
2016-10-05 14:02:49 -07:00
|
|
|
if (tree_count > 0 && state != ERROR_STATE) {
|
2016-11-14 17:25:55 -08:00
|
|
|
SkipPrecedingTreesSession *session = payload;
|
2016-08-29 17:34:51 -07:00
|
|
|
Parser *self = session->parser;
|
|
|
|
|
TSSymbol lookahead_symbol = session->lookahead_symbol;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t action_count;
|
2016-11-14 10:25:26 -08:00
|
|
|
const TSParseAction *actions =
|
|
|
|
|
ts_language_actions(self->language, state, lookahead_symbol, &action_count);
|
|
|
|
|
if (action_count > 0 && actions[0].type == TSParseActionTypeReduce) {
|
2016-08-29 17:34:51 -07:00
|
|
|
return StackIteratePop | StackIterateStop;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return StackIterateNone;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
static bool parser__skip_preceding_trees(Parser *self, StackVersion version,
|
2016-08-29 17:34:51 -07:00
|
|
|
TSSymbol lookahead_symbol) {
|
2016-11-14 17:25:55 -08:00
|
|
|
SkipPrecedingTreesSession session = { self, lookahead_symbol };
|
2016-08-29 17:34:51 -07:00
|
|
|
StackPopResult pop = ts_stack_iterate(
|
2016-11-14 17:25:55 -08:00
|
|
|
self->stack, version, parser__skip_preceding_trees_callback, &session);
|
2016-08-29 17:34:51 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
StackVersion previous_version = STACK_VERSION_NONE;
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < pop.slices.size; i++) {
|
2016-08-29 17:34:51 -07:00
|
|
|
StackSlice slice = pop.slices.contents[i];
|
2016-11-14 17:25:55 -08:00
|
|
|
if (slice.version == previous_version) {
|
2016-08-29 17:34:51 -07:00
|
|
|
ts_tree_array_delete(&slice.trees);
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
previous_version = slice.version;
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *error = ts_tree_make_error_node(&slice.trees);
|
2016-08-29 17:34:51 -07:00
|
|
|
error->extra = true;
|
|
|
|
|
TSStateId state = ts_stack_top_state(self->stack, slice.version);
|
|
|
|
|
parser__push(self, slice.version, error, state);
|
|
|
|
|
}
|
2016-11-14 17:25:55 -08:00
|
|
|
|
|
|
|
|
return pop.slices.size > 0;
|
2016-08-29 17:34:51 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static void parser__handle_error(Parser *self, StackVersion version,
|
2016-08-29 17:34:51 -07:00
|
|
|
TSSymbol lookahead_symbol) {
|
2016-11-14 17:25:55 -08:00
|
|
|
// If there are other stack versions that are clearly better than this one,
|
|
|
|
|
// just halt this version.
|
2016-08-31 10:51:59 -07:00
|
|
|
ErrorStatus error_status = ts_stack_error_status(self->stack, version);
|
|
|
|
|
error_status.count++;
|
|
|
|
|
if (parser__better_version_exists(self, version, error_status)) {
|
2016-08-29 09:34:08 -07:00
|
|
|
ts_stack_halt(self->stack, version);
|
|
|
|
|
LOG("bail_on_error");
|
2016-11-04 09:18:38 -07:00
|
|
|
return;
|
2016-08-29 09:34:08 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LOG("handle_error");
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// If the current lookahead symbol would have been valid in some previous
|
|
|
|
|
// state on the stack, create one stack version that repairs the error
|
|
|
|
|
// immediately by simply skipping all of the trees that came after that state.
|
|
|
|
|
if (parser__skip_preceding_trees(self, version, lookahead_symbol)) {
|
|
|
|
|
LOG("skip_preceding_trees");
|
|
|
|
|
LOG_STACK();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Perform any reductions that could have happened in this state, regardless
|
|
|
|
|
// of the lookahead.
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t previous_version_count = ts_stack_version_count(self->stack);
|
2016-08-29 09:34:08 -07:00
|
|
|
for (StackVersion v = version; v < ts_stack_version_count(self->stack);) {
|
2016-11-04 09:18:38 -07:00
|
|
|
if (parser__do_potential_reductions(self, v)) {
|
|
|
|
|
if (v == version) {
|
|
|
|
|
v = previous_version_count;
|
|
|
|
|
} else {
|
|
|
|
|
v++;
|
|
|
|
|
}
|
2016-08-29 09:34:08 -07:00
|
|
|
}
|
|
|
|
|
}
|
2016-05-29 22:36:47 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
// Push a discontinuity onto the stack. Merge all of the stack versions that
|
|
|
|
|
// were created in the previous step.
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, version, NULL, false, ERROR_STATE);
|
2016-05-29 22:36:47 -07:00
|
|
|
while (ts_stack_version_count(self->stack) > previous_version_count) {
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_stack_push(self->stack, previous_version_count, NULL, false, ERROR_STATE);
|
2016-05-29 22:36:47 -07:00
|
|
|
assert(ts_stack_merge(self->stack, version, previous_version_count));
|
|
|
|
|
}
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static void parser__recover(Parser *self, StackVersion version, TSStateId state,
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *lookahead) {
|
2016-06-02 14:04:48 -07:00
|
|
|
if (lookahead->symbol == ts_builtin_sym_end) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("recover_eof");
|
2016-06-02 14:04:48 -07:00
|
|
|
TreeArray children = array_new();
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *parent = ts_tree_make_error_node(&children);
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__push(self, version, parent, 1);
|
|
|
|
|
parser__accept(self, version, lookahead);
|
2016-06-02 14:04:48 -07:00
|
|
|
}
|
|
|
|
|
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("recover state:%u", state);
|
2016-05-09 14:31:44 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
StackVersion new_version = ts_stack_copy_version(self->stack, version);
|
2016-08-31 14:29:34 -07:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__shift(
|
2016-10-05 14:02:49 -07:00
|
|
|
self, new_version, ERROR_STATE, lookahead,
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_language_symbol_metadata(self->language, lookahead->symbol).extra);
|
2016-08-31 14:29:34 -07:00
|
|
|
ErrorStatus error_status = ts_stack_error_status(self->stack, new_version);
|
|
|
|
|
if (parser__better_version_exists(self, version, error_status)) {
|
|
|
|
|
ts_stack_remove_version(self->stack, new_version);
|
|
|
|
|
LOG("bail_on_recovery");
|
|
|
|
|
}
|
2016-03-07 20:06:46 -08:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__shift(self, version, state, lookahead, false);
|
2016-03-07 20:06:46 -08:00
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
static void parser__advance(Parser *self, StackVersion version,
|
2016-08-29 12:08:58 -07:00
|
|
|
ReusableNode *reusable_node) {
|
2016-07-17 13:35:43 -07:00
|
|
|
bool validated_lookahead = false;
|
2016-12-20 17:06:20 -08:00
|
|
|
Tree *lookahead = parser__get_lookahead(self, version, reusable_node, &validated_lookahead);
|
2016-07-17 13:35:43 -07:00
|
|
|
|
2015-10-07 12:58:32 -07:00
|
|
|
for (;;) {
|
2016-04-04 12:25:57 -07:00
|
|
|
TSStateId state = ts_stack_top_state(self->stack, version);
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
TableEntry table_entry;
|
2016-12-20 17:06:20 -08:00
|
|
|
ts_language_table_entry(self->language, state, lookahead->first_leaf.symbol, &table_entry);
|
2016-07-17 13:35:43 -07:00
|
|
|
|
|
|
|
|
if (!validated_lookahead) {
|
2016-08-29 12:08:58 -07:00
|
|
|
if (!parser__can_reuse(self, state, lookahead, &table_entry)) {
|
2016-12-20 17:06:20 -08:00
|
|
|
if (lookahead == reusable_node->tree) {
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_pop_leaf(reusable_node);
|
2016-12-20 17:06:20 -08:00
|
|
|
} else {
|
2016-08-29 12:08:58 -07:00
|
|
|
parser__clear_cached_token(self);
|
2016-12-20 17:06:20 -08:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
ts_tree_release(lookahead);
|
2016-12-20 17:06:20 -08:00
|
|
|
lookahead = parser__get_lookahead(self, version, reusable_node, &validated_lookahead);
|
2016-07-17 13:35:43 -07:00
|
|
|
continue;
|
|
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
validated_lookahead = true;
|
2016-12-20 17:06:20 -08:00
|
|
|
LOG("reused_lookahead sym:%s, size:%u", SYM_NAME(lookahead->symbol), lookahead->size.bytes);
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-06-21 22:58:06 -07:00
|
|
|
bool reduction_stopped_at_error = false;
|
2016-04-10 14:12:24 -07:00
|
|
|
StackVersion last_reduction_version = STACK_VERSION_NONE;
|
|
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
for (uint32_t i = 0; i < table_entry.action_count; i++) {
|
2016-07-17 13:35:43 -07:00
|
|
|
TSParseAction action = table_entry.actions[i];
|
2016-04-10 14:12:24 -07:00
|
|
|
|
2015-10-07 12:58:32 -07:00
|
|
|
switch (action.type) {
|
2016-04-10 14:12:24 -07:00
|
|
|
case TSParseActionTypeShift: {
|
2016-11-14 08:36:06 -08:00
|
|
|
bool extra = action.extra;
|
|
|
|
|
TSStateId next_state;
|
|
|
|
|
|
|
|
|
|
if (action.extra) {
|
|
|
|
|
next_state = state;
|
|
|
|
|
LOG("shift_extra");
|
|
|
|
|
} else {
|
|
|
|
|
next_state = action.params.to_state;
|
|
|
|
|
LOG("shift state:%u", next_state);
|
|
|
|
|
}
|
|
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
if (lookahead->child_count > 0) {
|
2016-12-20 17:06:20 -08:00
|
|
|
if (parser__breakdown_lookahead(self, &lookahead, state, reusable_node)) {
|
2016-08-29 12:08:58 -07:00
|
|
|
if (!parser__can_reuse(self, state, lookahead, &table_entry)) {
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_pop(reusable_node);
|
2016-07-17 13:35:43 -07:00
|
|
|
ts_tree_release(lookahead);
|
2016-12-20 17:06:20 -08:00
|
|
|
lookahead = parser__get_lookahead(self, version, reusable_node, &validated_lookahead);
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-14 08:36:06 -08:00
|
|
|
next_state = ts_language_next_state(self->language, state, lookahead->symbol);
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-14 08:36:06 -08:00
|
|
|
parser__shift(self, version, next_state, lookahead, extra);
|
2016-07-17 13:35:43 -07:00
|
|
|
|
|
|
|
|
if (lookahead == reusable_node->tree)
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_pop(reusable_node);
|
2016-07-17 13:35:43 -07:00
|
|
|
|
|
|
|
|
ts_tree_release(lookahead);
|
2016-11-04 09:18:38 -07:00
|
|
|
return;
|
2016-04-10 14:12:24 -07:00
|
|
|
}
|
2015-10-07 12:58:32 -07:00
|
|
|
|
2016-04-10 14:12:24 -07:00
|
|
|
case TSParseActionTypeReduce: {
|
2016-06-21 22:58:06 -07:00
|
|
|
if (reduction_stopped_at_error)
|
2016-06-21 22:53:48 -07:00
|
|
|
continue;
|
|
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
unsigned child_count = action.params.child_count;
|
|
|
|
|
TSSymbol symbol = action.params.symbol;
|
|
|
|
|
bool fragile = action.fragile;
|
2016-05-09 14:31:44 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
LOG("reduce sym:%s, child_count:%u", SYM_NAME(symbol), child_count);
|
2016-04-18 11:16:56 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
StackPopResult reduction =
|
|
|
|
|
parser__reduce(self, version, symbol, child_count, fragile, true);
|
|
|
|
|
StackSlice slice = *array_front(&reduction.slices);
|
2016-11-04 09:18:38 -07:00
|
|
|
if (reduction.stopped_at_error) {
|
|
|
|
|
reduction_stopped_at_error = true;
|
2016-11-14 17:25:55 -08:00
|
|
|
if (!parser__repair_error(self, slice, lookahead->first_leaf.symbol,
|
|
|
|
|
table_entry))
|
|
|
|
|
break;
|
2015-12-17 12:48:55 -08:00
|
|
|
}
|
2016-11-04 09:18:38 -07:00
|
|
|
|
2016-11-14 17:25:55 -08:00
|
|
|
last_reduction_version = slice.version;
|
2015-10-07 12:58:32 -07:00
|
|
|
break;
|
2016-04-10 14:12:24 -07:00
|
|
|
}
|
2015-10-07 12:58:32 -07:00
|
|
|
|
2016-04-10 14:12:24 -07:00
|
|
|
case TSParseActionTypeAccept: {
|
2016-08-31 10:51:59 -07:00
|
|
|
if (ts_stack_error_status(self->stack, version).count > 0)
|
2016-06-22 21:05:05 -07:00
|
|
|
continue;
|
2016-06-26 22:51:48 -07:00
|
|
|
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("accept");
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__accept(self, version, lookahead);
|
2016-07-17 13:35:43 -07:00
|
|
|
ts_tree_release(lookahead);
|
2016-11-04 09:18:38 -07:00
|
|
|
return;
|
2016-05-09 14:31:44 -07:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
case TSParseActionTypeRecover: {
|
|
|
|
|
while (lookahead->child_count > 0) {
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_breakdown(reusable_node);
|
2016-07-17 13:35:43 -07:00
|
|
|
ts_tree_release(lookahead);
|
|
|
|
|
lookahead = reusable_node->tree;
|
|
|
|
|
ts_tree_retain(lookahead);
|
|
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__recover(self, version, action.params.to_state, lookahead);
|
2016-07-17 13:35:43 -07:00
|
|
|
if (lookahead == reusable_node->tree)
|
2017-01-05 10:06:43 -08:00
|
|
|
reusable_node_pop(reusable_node);
|
2016-07-17 13:35:43 -07:00
|
|
|
ts_tree_release(lookahead);
|
2016-11-04 09:18:38 -07:00
|
|
|
return;
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
}
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
if (last_reduction_version != STACK_VERSION_NONE) {
|
|
|
|
|
ts_stack_renumber_version(self->stack, last_reduction_version, version);
|
|
|
|
|
LOG_STACK();
|
2016-06-21 21:27:33 -07:00
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
if (parser__breakdown_top_of_stack(self, version)) {
|
|
|
|
|
continue;
|
2016-06-21 21:27:33 -07:00
|
|
|
}
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-10-05 14:02:49 -07:00
|
|
|
if (state == ERROR_STATE) {
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__push(self, version, lookahead, ERROR_STATE);
|
|
|
|
|
return;
|
2016-08-06 21:25:35 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-10 10:25:32 -08:00
|
|
|
parser__handle_error(self, version, lookahead->first_leaf.symbol);
|
2016-07-01 15:08:19 -07:00
|
|
|
|
2016-07-17 13:35:43 -07:00
|
|
|
if (ts_stack_is_halted(self->stack, version)) {
|
|
|
|
|
ts_tree_release(lookahead);
|
2016-11-04 09:18:38 -07:00
|
|
|
return;
|
2016-07-17 13:35:43 -07:00
|
|
|
}
|
2015-07-08 17:34:21 -07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
bool parser_init(Parser *self) {
|
2016-02-04 11:15:46 -08:00
|
|
|
ts_lexer_init(&self->lexer);
|
2016-05-09 14:31:44 -07:00
|
|
|
array_init(&self->reduce_actions);
|
2016-09-22 18:02:11 -07:00
|
|
|
array_init(&self->tree_path1);
|
|
|
|
|
array_init(&self->tree_path2);
|
2016-11-04 09:18:38 -07:00
|
|
|
array_grow(&self->reduce_actions, 4);
|
2016-01-18 10:44:49 -08:00
|
|
|
self->stack = ts_stack_new();
|
2016-11-04 09:18:38 -07:00
|
|
|
self->finished_tree = NULL;
|
2016-01-18 10:44:49 -08:00
|
|
|
return true;
|
2015-08-16 19:53:34 -07:00
|
|
|
}
|
|
|
|
|
|
2016-12-02 22:03:48 -08:00
|
|
|
void parser_set_language(Parser *self, const TSLanguage *language) {
|
2016-12-06 10:12:49 -08:00
|
|
|
if (self->external_scanner_payload && self->language->external_scanner.destroy)
|
|
|
|
|
self->language->external_scanner.destroy(self->external_scanner_payload);
|
2016-12-04 14:18:30 -08:00
|
|
|
|
2016-12-06 10:12:49 -08:00
|
|
|
if (language && language->external_scanner.create)
|
2016-12-02 22:03:48 -08:00
|
|
|
self->external_scanner_payload = language->external_scanner.create();
|
2016-12-04 14:18:30 -08:00
|
|
|
else
|
2016-12-02 22:03:48 -08:00
|
|
|
self->external_scanner_payload = NULL;
|
2016-12-04 14:18:30 -08:00
|
|
|
|
|
|
|
|
self->language = language;
|
2016-12-02 22:03:48 -08:00
|
|
|
}
|
|
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
void parser_destroy(Parser *self) {
|
2016-02-04 11:15:46 -08:00
|
|
|
if (self->stack)
|
|
|
|
|
ts_stack_delete(self->stack);
|
2016-05-09 14:31:44 -07:00
|
|
|
if (self->reduce_actions.contents)
|
|
|
|
|
array_delete(&self->reduce_actions);
|
2016-09-22 18:02:11 -07:00
|
|
|
if (self->tree_path1.contents)
|
|
|
|
|
array_delete(&self->tree_path1);
|
|
|
|
|
if (self->tree_path2.contents)
|
|
|
|
|
array_delete(&self->tree_path2);
|
2016-12-06 10:12:49 -08:00
|
|
|
parser_set_language(self, NULL);
|
2015-08-16 19:53:34 -07:00
|
|
|
}
|
|
|
|
|
|
2016-11-09 20:59:05 -08:00
|
|
|
Tree *parser_parse(Parser *self, TSInput input, Tree *old_tree) {
|
2016-08-29 12:08:58 -07:00
|
|
|
parser__start(self, input, old_tree);
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-07-01 15:08:19 -07:00
|
|
|
StackVersion version = STACK_VERSION_NONE;
|
2016-11-14 12:15:24 -08:00
|
|
|
uint32_t position = 0, last_position = 0;
|
2016-07-01 15:08:19 -07:00
|
|
|
ReusableNode reusable_node;
|
2015-07-08 17:34:21 -07:00
|
|
|
|
2016-07-01 15:08:19 -07:00
|
|
|
do {
|
2016-06-22 14:10:54 -07:00
|
|
|
for (version = 0; version < ts_stack_version_count(self->stack); version++) {
|
2016-07-01 15:08:19 -07:00
|
|
|
reusable_node = self->reusable_node;
|
2016-06-22 14:10:54 -07:00
|
|
|
last_position = position;
|
2016-04-01 21:17:23 -07:00
|
|
|
|
2016-06-22 14:10:54 -07:00
|
|
|
while (!ts_stack_is_halted(self->stack, version)) {
|
|
|
|
|
position = ts_stack_top_position(self->stack, version).chars;
|
2017-01-05 10:06:43 -08:00
|
|
|
if (position > last_position || (version > 0 && position == last_position))
|
2015-12-26 16:49:23 -08:00
|
|
|
break;
|
2016-02-23 00:09:42 -08:00
|
|
|
|
2016-11-14 12:15:24 -08:00
|
|
|
LOG("process version:%d, version_count:%u, state:%d, row:%u, col:%u",
|
2016-06-23 11:42:43 -07:00
|
|
|
version, ts_stack_version_count(self->stack),
|
|
|
|
|
ts_stack_top_state(self->stack, version),
|
2016-12-20 17:06:20 -08:00
|
|
|
ts_stack_top_position(self->stack, version).extent.row,
|
|
|
|
|
ts_stack_top_position(self->stack, version).extent.column);
|
2015-11-18 08:47:15 -08:00
|
|
|
|
2016-11-04 09:18:38 -07:00
|
|
|
parser__advance(self, version, &reusable_node);
|
2016-06-21 22:53:48 -07:00
|
|
|
LOG_STACK();
|
2015-12-26 16:49:23 -08:00
|
|
|
}
|
2014-08-09 01:03:55 -07:00
|
|
|
}
|
2015-12-08 13:01:33 -08:00
|
|
|
|
2016-07-01 15:08:19 -07:00
|
|
|
self->reusable_node = reusable_node;
|
2016-05-29 22:36:47 -07:00
|
|
|
|
2016-08-29 12:08:58 -07:00
|
|
|
if (parser__condense_stack(self)) {
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("condense");
|
2016-05-29 22:36:47 -07:00
|
|
|
LOG_STACK();
|
|
|
|
|
}
|
|
|
|
|
|
2016-07-01 15:08:19 -07:00
|
|
|
self->is_split = (version > 1);
|
|
|
|
|
} while (version != 0);
|
2016-01-29 17:31:43 -08:00
|
|
|
|
2016-06-22 22:36:11 -07:00
|
|
|
LOG("done");
|
2016-06-22 21:04:35 -07:00
|
|
|
LOG_TREE();
|
2016-06-22 14:10:54 -07:00
|
|
|
ts_stack_clear(self->stack);
|
2016-08-29 12:08:58 -07:00
|
|
|
parser__clear_cached_token(self);
|
2016-11-04 09:18:38 -07:00
|
|
|
ts_tree_assign_parents(self->finished_tree, &self->tree_path1);
|
2016-06-22 14:10:54 -07:00
|
|
|
return self->finished_tree;
|
2014-07-10 13:14:52 -07:00
|
|
|
}
|