Rename type ts_lr_parser -> TSStateMachine
This commit is contained in:
parent
27f6eb725d
commit
0ec3faba3e
5 changed files with 96 additions and 95 deletions
|
|
@ -11,7 +11,7 @@ extern "C" {
|
|||
#include "tree_sitter/runtime.h"
|
||||
#include "tree_sitter/parser/lexer.h"
|
||||
#include "tree_sitter/parser/stack.h"
|
||||
#include "tree_sitter/parser/lr_parser.h"
|
||||
#include "tree_sitter/parser/state_machine.h"
|
||||
|
||||
#define SYMBOL_NAMES \
|
||||
static const char *ts_symbol_names[]
|
||||
|
|
@ -73,10 +73,10 @@ ts_lexer_start_token(lexer);
|
|||
SYMBOL_NAMES;
|
||||
|
||||
static const TSTree * ts_parse(void *data, TSInput input, TSInputEdit *edit) {
|
||||
ts_lr_parser *parser = (ts_lr_parser *)data;
|
||||
ts_lr_parser_initialize(parser, input, edit);
|
||||
TSStateMachine *parser = (TSStateMachine *)data;
|
||||
ts_state_machine_initialize(parser, input, edit);
|
||||
for (;;) {
|
||||
const TSTree *tree = ts_lr_parser_parse(parser, ts_symbol_names);
|
||||
const TSTree *tree = ts_state_machine_parse(parser, ts_symbol_names);
|
||||
if (tree) return tree;
|
||||
}
|
||||
}
|
||||
|
|
@ -85,9 +85,9 @@ static const TSTree * ts_parse(void *data, TSInput input, TSInputEdit *edit) {
|
|||
TSParser constructor_name() { \
|
||||
return (TSParser) { \
|
||||
.parse_fn = ts_parse, \
|
||||
.free_fn = ts_lr_parser_free, \
|
||||
.free_fn = ts_state_machine_free, \
|
||||
.symbol_names = ts_symbol_names, \
|
||||
.data = ts_lr_parser_make( \
|
||||
.data = ts_state_machine_make( \
|
||||
SYMBOL_COUNT, \
|
||||
(const TSParseAction *)ts_parse_actions, \
|
||||
ts_lex_states, \
|
||||
|
|
|
|||
|
|
@ -51,16 +51,17 @@ typedef struct {
|
|||
const TSStateId *lex_states;
|
||||
TSTree * (* lex_fn)(TSLexer *, TSStateId);
|
||||
} config;
|
||||
} ts_lr_parser;
|
||||
} TSStateMachine;
|
||||
|
||||
ts_lr_parser * ts_lr_parser_make(size_t symbol_count,
|
||||
const TSParseAction *parse_table,
|
||||
const TSStateId *lex_states,
|
||||
TSTree * (* lex_fn)(TSLexer *, TSStateId),
|
||||
const int *hidden_symbol_flags);
|
||||
void ts_lr_parser_free(void *data);
|
||||
void ts_lr_parser_initialize(ts_lr_parser *parser, TSInput input, TSInputEdit *edit);
|
||||
TSTree * ts_lr_parser_parse(ts_lr_parser *parser, const char **symbol_names);
|
||||
TSStateMachine * ts_state_machine_make(
|
||||
size_t symbol_count,
|
||||
const TSParseAction *parse_table,
|
||||
const TSStateId *lex_states,
|
||||
TSTree * (* lex_fn)(TSLexer *, TSStateId),
|
||||
const int *hidden_symbol_flags);
|
||||
void ts_state_machine_free(void *data);
|
||||
void ts_state_machine_initialize(TSStateMachine *, TSInput, TSInputEdit *);
|
||||
TSTree * ts_state_machine_parse(TSStateMachine *, const char **symbol_names);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
|
|
@ -5,7 +5,7 @@
|
|||
extern "C" {
|
||||
#endif
|
||||
|
||||
#include "tree_sitter/parser/lr_parser.h"
|
||||
#include "tree_sitter/parser/state_machine.h"
|
||||
|
||||
enum {
|
||||
dummy_sym1 = 2,
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
#include "runtime/runtime_spec_helper.h"
|
||||
#include "runtime/helpers/spy_reader.h"
|
||||
#include "runtime/helpers/dummy_parser.h"
|
||||
#include "tree_sitter/parser/lr_parser.h"
|
||||
#include "tree_sitter/parser/state_machine.h"
|
||||
|
||||
TSTree *lex_fn_node_to_return;
|
||||
TSStateId lex_fn_state_received;
|
||||
|
|
@ -16,12 +16,12 @@ TSTree * fake_lex(TSLexer *lexer, TSStateId state_id) {
|
|||
START_TEST
|
||||
|
||||
describe("LR Parsers", [&]() {
|
||||
ts_lr_parser *parser;
|
||||
TSStateMachine *parser;
|
||||
SpyReader *reader;
|
||||
|
||||
before_each([&]() {
|
||||
reader = new SpyReader("some structured text", 5);
|
||||
parser = ts_lr_parser_make(dummy_parser.symbol_count,
|
||||
parser = ts_state_machine_make(dummy_parser.symbol_count,
|
||||
(const TSParseAction *)dummy_parser.parse_table,
|
||||
dummy_parser.lex_states,
|
||||
fake_lex,
|
||||
|
|
@ -34,12 +34,12 @@ describe("LR Parsers", [&]() {
|
|||
|
||||
describe("when starting at the beginning of the input (edit is NULL)", [&]() {
|
||||
before_each([&]() {
|
||||
ts_lr_parser_initialize(parser, reader->input, nullptr);
|
||||
ts_state_machine_initialize(parser, reader->input, nullptr);
|
||||
});
|
||||
|
||||
it("runs the lexer with the lex state corresponding to the initial state", [&]() {
|
||||
lex_fn_node_to_return = ts_tree_make_leaf(dummy_sym2, 5, 1);
|
||||
ts_lr_parser_parse(parser, nullptr);
|
||||
ts_state_machine_parse(parser, nullptr);
|
||||
AssertThat(lex_fn_state_received, Equals(100));
|
||||
});
|
||||
|
||||
|
|
@ -49,12 +49,12 @@ describe("LR Parsers", [&]() {
|
|||
});
|
||||
|
||||
it("advances to the state specified in the action", [&]() {
|
||||
ts_lr_parser_parse(parser, nullptr);
|
||||
ts_state_machine_parse(parser, nullptr);
|
||||
AssertThat(ts_stack_top_state(&parser->stack), Equals(12));
|
||||
});
|
||||
|
||||
it("continues parsing (returns NULL)", [&]() {
|
||||
auto result = ts_lr_parser_parse(parser, nullptr);
|
||||
auto result = ts_state_machine_parse(parser, nullptr);
|
||||
AssertThat(result, Equals((TSTree *)nullptr));
|
||||
});
|
||||
});
|
||||
|
|
@ -65,7 +65,7 @@ describe("LR Parsers", [&]() {
|
|||
});
|
||||
|
||||
it("ends the parse, returning an error tree", [&]() {
|
||||
auto result = ts_lr_parser_parse(parser, nullptr);
|
||||
auto result = ts_state_machine_parse(parser, nullptr);
|
||||
AssertThat(ts_tree_symbol(result), Equals(ts_builtin_sym_error));
|
||||
});
|
||||
});
|
||||
|
|
@ -1,34 +1,34 @@
|
|||
#include "tree_sitter/parser/lr_parser.h"
|
||||
#include "tree_sitter/parser/state_machine.h"
|
||||
#include "runtime/tree.h"
|
||||
|
||||
/*
|
||||
* Private
|
||||
*/
|
||||
|
||||
static const TSParseAction * actions_for_state(ts_lr_parser *parser, TSStateId state) {
|
||||
return parser->config.parse_table + (state * parser->config.symbol_count);
|
||||
static const TSParseAction * actions_for_state(TSStateMachine *machine, TSStateId state) {
|
||||
return machine->config.parse_table + (state * machine->config.symbol_count);
|
||||
}
|
||||
|
||||
void shift(ts_lr_parser *parser, TSStateId 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;
|
||||
void shift(TSStateMachine *machine, TSStateId parse_state, int is_extra) {
|
||||
machine->lookahead->is_extra = is_extra;
|
||||
ts_stack_push(&machine->stack, parse_state, machine->lookahead);
|
||||
machine->lookahead = machine->next_lookahead;
|
||||
machine->next_lookahead = NULL;
|
||||
}
|
||||
|
||||
void reduce(ts_lr_parser *parser, TSSymbol symbol, size_t child_count) {
|
||||
parser->next_lookahead = parser->lookahead;
|
||||
parser->lookahead = ts_stack_reduce(&parser->stack,
|
||||
void reduce(TSStateMachine *machine, TSSymbol symbol, size_t child_count) {
|
||||
machine->next_lookahead = machine->lookahead;
|
||||
machine->lookahead = ts_stack_reduce(&machine->stack,
|
||||
symbol,
|
||||
child_count,
|
||||
parser->config.hidden_symbol_flags,
|
||||
machine->config.hidden_symbol_flags,
|
||||
1);
|
||||
}
|
||||
|
||||
static size_t breakdown_stack(ts_lr_parser *parser, TSInputEdit *edit) {
|
||||
static size_t breakdown_stack(TSStateMachine *machine, TSInputEdit *edit) {
|
||||
if (!edit) return 0;
|
||||
|
||||
TSStack *stack = &parser->stack;
|
||||
TSStack *stack = &machine->stack;
|
||||
size_t position = 0;
|
||||
|
||||
for (;;) {
|
||||
|
|
@ -46,7 +46,7 @@ static size_t breakdown_stack(ts_lr_parser *parser, TSInputEdit *edit) {
|
|||
for (size_t i = 0; i < child_count && position < edit->position; i++) {
|
||||
TSTree *child = children[i];
|
||||
TSStateId state = ts_stack_top_state(stack);
|
||||
TSStateId next_state = actions_for_state(parser, state)[ts_tree_symbol(child)].data.to_state;
|
||||
TSStateId next_state = actions_for_state(machine, state)[ts_tree_symbol(child)].data.to_state;
|
||||
ts_stack_push(stack, next_state, child);
|
||||
ts_tree_retain(child);
|
||||
position += ts_tree_total_size(child);
|
||||
|
|
@ -58,42 +58,42 @@ static size_t breakdown_stack(ts_lr_parser *parser, TSInputEdit *edit) {
|
|||
return position;
|
||||
}
|
||||
|
||||
TSSymbol * expected_symbols(ts_lr_parser *parser, size_t *count) {
|
||||
TSSymbol * expected_symbols(TSStateMachine *machine, size_t *count) {
|
||||
*count = 0;
|
||||
const TSParseAction *actions = actions_for_state(parser, ts_stack_top_state(&parser->stack));
|
||||
for (size_t i = 0; i < parser->config.symbol_count; i++)
|
||||
const TSParseAction *actions = actions_for_state(machine, ts_stack_top_state(&machine->stack));
|
||||
for (size_t i = 0; i < machine->config.symbol_count; i++)
|
||||
if (actions[i].type != ts_parse_action_type_error)
|
||||
++(*count);
|
||||
|
||||
size_t n = 0;
|
||||
TSSymbol *result = malloc(*count * sizeof(*result));
|
||||
for (TSSymbol i = 0; i < parser->config.symbol_count; i++)
|
||||
for (TSSymbol i = 0; i < machine->config.symbol_count; i++)
|
||||
if (actions[i].type != ts_parse_action_type_error)
|
||||
result[n++] = i;
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
int handle_error(ts_lr_parser *parser) {
|
||||
int handle_error(TSStateMachine *machine) {
|
||||
size_t count = 0;
|
||||
const TSSymbol *inputs = expected_symbols(parser, &count);
|
||||
TSTree *error = ts_tree_make_error(ts_lexer_lookahead_char(&parser->lexer),
|
||||
const TSSymbol *inputs = expected_symbols(machine, &count);
|
||||
TSTree *error = ts_tree_make_error(ts_lexer_lookahead_char(&machine->lexer),
|
||||
count,
|
||||
inputs,
|
||||
0,
|
||||
0);
|
||||
|
||||
for (;;) {
|
||||
ts_tree_release(parser->lookahead);
|
||||
size_t position = ts_lexer_position(&parser->lexer);
|
||||
parser->lookahead = parser->config.lex_fn(&parser->lexer, ts_lex_state_error);
|
||||
ts_tree_release(machine->lookahead);
|
||||
size_t position = ts_lexer_position(&machine->lexer);
|
||||
machine->lookahead = machine->config.lex_fn(&machine->lexer, ts_lex_state_error);
|
||||
|
||||
int at_end = 0;
|
||||
if (ts_lexer_position(&parser->lexer) == position)
|
||||
at_end = !ts_lexer_advance(&parser->lexer);
|
||||
if (ts_lexer_position(&machine->lexer) == position)
|
||||
at_end = !ts_lexer_advance(&machine->lexer);
|
||||
|
||||
if (at_end || ts_tree_symbol(parser->lookahead) == ts_builtin_sym_end) {
|
||||
ts_stack_push(&parser->stack, 0, error);
|
||||
if (at_end || ts_tree_symbol(machine->lookahead) == ts_builtin_sym_end) {
|
||||
ts_stack_push(&machine->stack, 0, error);
|
||||
return 0;
|
||||
}
|
||||
|
||||
|
|
@ -101,15 +101,15 @@ int handle_error(ts_lr_parser *parser) {
|
|||
* 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;
|
||||
TSParseAction action_on_error = actions_for_state(parser, stack_state)[ts_builtin_sym_error];
|
||||
for (size_t j = 0; j < machine->stack.size; j++) {
|
||||
size_t i = machine->stack.size - 1 - j;
|
||||
TSStateId stack_state = machine->stack.entries[i].state;
|
||||
TSParseAction action_on_error = actions_for_state(machine, stack_state)[ts_builtin_sym_error];
|
||||
if (action_on_error.type == ts_parse_action_type_shift) {
|
||||
TSStateId state_after_error = action_on_error.data.to_state;
|
||||
if (actions_for_state(parser, state_after_error)[ts_tree_symbol(parser->lookahead)].type != ts_parse_action_type_error) {
|
||||
ts_stack_shrink(&parser->stack, i + 1);
|
||||
ts_stack_push(&parser->stack, state_after_error, error);
|
||||
if (actions_for_state(machine, state_after_error)[ts_tree_symbol(machine->lookahead)].type != ts_parse_action_type_error) {
|
||||
ts_stack_shrink(&machine->stack, i + 1);
|
||||
ts_stack_push(&machine->stack, state_after_error, error);
|
||||
return 1;
|
||||
}
|
||||
}
|
||||
|
|
@ -117,8 +117,8 @@ int handle_error(ts_lr_parser *parser) {
|
|||
}
|
||||
}
|
||||
|
||||
TSTree * get_tree_root(ts_lr_parser *parser) {
|
||||
TSStack *stack = &parser->stack;
|
||||
TSTree * get_tree_root(TSStateMachine *machine) {
|
||||
TSStack *stack = &machine->stack;
|
||||
TSTree *top_node = ts_stack_top_node(stack);
|
||||
if (stack->size <= 1)
|
||||
return top_node;
|
||||
|
|
@ -134,37 +134,37 @@ TSTree * get_tree_root(ts_lr_parser *parser) {
|
|||
child->is_extra = 0;
|
||||
ts_tree_retain(child);
|
||||
TSStateId state = ts_stack_top_state(stack);
|
||||
TSStateId next_state = actions_for_state(parser, state)[ts_tree_symbol(child)].data.to_state;
|
||||
TSStateId next_state = actions_for_state(machine, state)[ts_tree_symbol(child)].data.to_state;
|
||||
ts_stack_push(stack, next_state, child);
|
||||
}
|
||||
|
||||
TSTree *new_node = ts_stack_reduce(stack,
|
||||
top_node->symbol,
|
||||
stack->size,
|
||||
parser->config.hidden_symbol_flags,
|
||||
machine->config.hidden_symbol_flags,
|
||||
0);
|
||||
ts_tree_release(top_node);
|
||||
return new_node;
|
||||
}
|
||||
|
||||
TSParseAction get_next_action(ts_lr_parser *parser) {
|
||||
TSStateId state = ts_stack_top_state(&parser->stack);
|
||||
if (!parser->lookahead)
|
||||
parser->lookahead = parser->config.lex_fn(&parser->lexer, parser->config.lex_states[state]);
|
||||
return actions_for_state(parser, state)[ts_tree_symbol(parser->lookahead)];
|
||||
TSParseAction get_next_action(TSStateMachine *machine) {
|
||||
TSStateId state = ts_stack_top_state(&machine->stack);
|
||||
if (!machine->lookahead)
|
||||
machine->lookahead = machine->config.lex_fn(&machine->lexer, machine->config.lex_states[state]);
|
||||
return actions_for_state(machine, state)[ts_tree_symbol(machine->lookahead)];
|
||||
}
|
||||
|
||||
/*
|
||||
* Public API
|
||||
*/
|
||||
|
||||
ts_lr_parser * ts_lr_parser_make(size_t symbol_count,
|
||||
TSStateMachine * ts_state_machine_make(size_t symbol_count,
|
||||
const TSParseAction *parse_table,
|
||||
const TSStateId *lex_states,
|
||||
TSTree * (* lex_fn)(TSLexer *, TSStateId),
|
||||
const int *hidden_symbol_flags) {
|
||||
ts_lr_parser *result = malloc(sizeof(ts_lr_parser));
|
||||
*result = (ts_lr_parser) {
|
||||
TSStateMachine *result = malloc(sizeof(TSStateMachine));
|
||||
*result = (TSStateMachine) {
|
||||
.lexer = ts_lexer_make(),
|
||||
.stack = ts_stack_make(),
|
||||
.config = {
|
||||
|
|
@ -178,25 +178,25 @@ ts_lr_parser * ts_lr_parser_make(size_t symbol_count,
|
|||
return result;
|
||||
}
|
||||
|
||||
void ts_lr_parser_free(void *data) {
|
||||
ts_lr_parser *parser = (ts_lr_parser *)data;
|
||||
if (parser->lookahead) ts_tree_release(parser->lookahead);
|
||||
if (parser->next_lookahead) ts_tree_release(parser->next_lookahead);
|
||||
ts_stack_delete(&parser->stack);
|
||||
free(parser);
|
||||
void ts_state_machine_free(void *data) {
|
||||
TSStateMachine *machine = (TSStateMachine *)data;
|
||||
if (machine->lookahead) ts_tree_release(machine->lookahead);
|
||||
if (machine->next_lookahead) ts_tree_release(machine->next_lookahead);
|
||||
ts_stack_delete(&machine->stack);
|
||||
free(machine);
|
||||
}
|
||||
|
||||
void ts_lr_parser_initialize(ts_lr_parser *parser, TSInput input, TSInputEdit *edit) {
|
||||
if (!edit) ts_stack_shrink(&parser->stack, 0);
|
||||
parser->lookahead = NULL;
|
||||
parser->next_lookahead = NULL;
|
||||
void ts_state_machine_initialize(TSStateMachine *machine, TSInput input, TSInputEdit *edit) {
|
||||
if (!edit) ts_stack_shrink(&machine->stack, 0);
|
||||
machine->lookahead = NULL;
|
||||
machine->next_lookahead = NULL;
|
||||
|
||||
size_t position = breakdown_stack(parser, edit);
|
||||
size_t position = breakdown_stack(machine, edit);
|
||||
input.seek_fn(input.data, position);
|
||||
|
||||
parser->lexer = ts_lexer_make();
|
||||
parser->lexer.input = input;
|
||||
ts_lexer_advance(&parser->lexer);
|
||||
machine->lexer = ts_lexer_make();
|
||||
machine->lexer.input = input;
|
||||
ts_lexer_advance(&machine->lexer);
|
||||
}
|
||||
|
||||
/* #define TS_DEBUG_PARSE */
|
||||
|
|
@ -208,31 +208,31 @@ void ts_lr_parser_initialize(ts_lr_parser *parser, TSInput input, TSInputEdit *e
|
|||
#define DEBUG_PARSE(...)
|
||||
#endif
|
||||
|
||||
TSTree * ts_lr_parser_parse(ts_lr_parser *parser, const char **symbol_names) {
|
||||
TSParseAction action = get_next_action(parser);
|
||||
DEBUG_PARSE("LOOKAHEAD %s", symbol_names[ts_tree_symbol(parser->lookahead)]);
|
||||
TSTree * ts_state_machine_parse(TSStateMachine *machine, const char **symbol_names) {
|
||||
TSParseAction action = get_next_action(machine);
|
||||
DEBUG_PARSE("LOOKAHEAD %s", symbol_names[ts_tree_symbol(machine->lookahead)]);
|
||||
switch (action.type) {
|
||||
case ts_parse_action_type_shift:
|
||||
DEBUG_PARSE("SHIFT %d", action.data.to_state);
|
||||
shift(parser, action.data.to_state, 0);
|
||||
shift(machine, 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);
|
||||
shift(machine, ts_stack_top_state(&machine->stack), 1);
|
||||
return NULL;
|
||||
case ts_parse_action_type_reduce:
|
||||
DEBUG_PARSE("REDUCE %s %d", symbol_names[action.data.symbol], action.data.child_count);
|
||||
reduce(parser, action.data.symbol, action.data.child_count);
|
||||
reduce(machine, action.data.symbol, action.data.child_count);
|
||||
return NULL;
|
||||
case ts_parse_action_type_accept:
|
||||
DEBUG_PARSE("ACCEPT");
|
||||
return get_tree_root(parser);
|
||||
return get_tree_root(machine);
|
||||
case ts_parse_action_type_error:
|
||||
DEBUG_PARSE("ERROR");
|
||||
if (handle_error(parser))
|
||||
if (handle_error(machine))
|
||||
return NULL;
|
||||
else
|
||||
return get_tree_root(parser);
|
||||
return get_tree_root(machine);
|
||||
default:
|
||||
return NULL;
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue