Move logic for lexical error handling outside of lexer functions
This way, less logic needs to be exposed in parser.h
This commit is contained in:
parent
1c52c30111
commit
4f0c83ba01
5 changed files with 47 additions and 85 deletions
|
|
@ -28,19 +28,11 @@ typedef struct {
|
|||
bool structural : 1;
|
||||
} TSSymbolMetadata;
|
||||
|
||||
typedef enum {
|
||||
TSTransitionTypeMain,
|
||||
TSTransitionTypeSeparator,
|
||||
TSTransitionTypeError,
|
||||
} TSTransitionType;
|
||||
|
||||
typedef struct TSLexer {
|
||||
void (*advance)(struct TSLexer *, TSStateId, TSTransitionType);
|
||||
void (*advance)(struct TSLexer *, TSStateId, bool);
|
||||
|
||||
TSLength current_position;
|
||||
TSLength token_end_position;
|
||||
TSLength token_start_position;
|
||||
TSLength error_end_position;
|
||||
|
||||
const char *chunk;
|
||||
size_t chunk_start;
|
||||
|
|
@ -48,10 +40,7 @@ typedef struct TSLexer {
|
|||
|
||||
size_t lookahead_size;
|
||||
int32_t lookahead;
|
||||
TSStateId starting_state;
|
||||
TSSymbol result_symbol;
|
||||
bool result_follows_error;
|
||||
int32_t first_unexpected_character;
|
||||
|
||||
TSInput input;
|
||||
TSDebugger debugger;
|
||||
|
|
@ -108,14 +97,14 @@ struct TSLanguage {
|
|||
|
||||
#define ADVANCE(state_value) \
|
||||
{ \
|
||||
lexer->advance(lexer, state_value, TSTransitionTypeMain); \
|
||||
lexer->advance(lexer, state_value, false); \
|
||||
state = state_value; \
|
||||
goto next_state; \
|
||||
}
|
||||
|
||||
#define SKIP(state_value) \
|
||||
{ \
|
||||
lexer->advance(lexer, state_value, TSTransitionTypeSeparator); \
|
||||
lexer->advance(lexer, state_value, true); \
|
||||
state = state_value; \
|
||||
goto next_state; \
|
||||
}
|
||||
|
|
|
|||
|
|
@ -14,6 +14,10 @@ static inline void ts_length_set_unknown(TSLength *self) {
|
|||
self->columns = 0;
|
||||
}
|
||||
|
||||
static inline TSLength ts_length_min(TSLength len1, TSLength len2) {
|
||||
return (len1.chars < len2.chars) ? len1 : len2;
|
||||
}
|
||||
|
||||
static inline TSLength ts_length_add(TSLength len1, TSLength len2) {
|
||||
TSLength result;
|
||||
result.chars = len1.chars + len2.chars;
|
||||
|
|
|
|||
|
|
@ -47,8 +47,7 @@ static void ts_lexer__get_lookahead(TSLexer *self) {
|
|||
LOG_LOOKAHEAD();
|
||||
}
|
||||
|
||||
static void ts_lexer__advance(TSLexer *self, TSStateId state,
|
||||
TSTransitionType transition_type) {
|
||||
static void ts_lexer__advance(TSLexer *self, TSStateId state, bool skip) {
|
||||
|
||||
if (self->chunk == empty_chunk)
|
||||
return;
|
||||
|
|
@ -64,25 +63,11 @@ static void ts_lexer__advance(TSLexer *self, TSStateId state,
|
|||
}
|
||||
}
|
||||
|
||||
switch (transition_type) {
|
||||
case TSTransitionTypeSeparator:
|
||||
if (self->result_follows_error) {
|
||||
LOG("skip_error state:%d", state);
|
||||
} else {
|
||||
LOG("skip_separator state:%d", state);
|
||||
self->token_start_position = self->current_position;
|
||||
}
|
||||
break;
|
||||
case TSTransitionTypeError:
|
||||
LOG("skip_error state:%d", state);
|
||||
self->result_follows_error = true;
|
||||
self->error_end_position = self->current_position;
|
||||
if (!self->first_unexpected_character)
|
||||
self->first_unexpected_character = self->lookahead;
|
||||
break;
|
||||
default:
|
||||
LOG("advance state:%d", state);
|
||||
break;
|
||||
if (skip) {
|
||||
LOG("skip_separator state:%d", state);
|
||||
self->token_start_position = self->current_position;
|
||||
} else {
|
||||
LOG("advance state:%d", state);
|
||||
}
|
||||
|
||||
if (self->current_position.bytes >= self->chunk_start + self->chunk_size)
|
||||
|
|
@ -108,7 +93,6 @@ void ts_lexer_init(TSLexer *self) {
|
|||
|
||||
static inline void ts_lexer__reset(TSLexer *self, TSLength position) {
|
||||
self->token_start_position = position;
|
||||
self->token_end_position = position;
|
||||
self->current_position = position;
|
||||
|
||||
self->chunk = 0;
|
||||
|
|
@ -132,32 +116,11 @@ void ts_lexer_reset(TSLexer *self, TSLength position) {
|
|||
void ts_lexer_start(TSLexer *self, TSStateId lex_state) {
|
||||
LOG("start_lex state:%d, pos:%lu", lex_state, self->current_position.chars);
|
||||
|
||||
self->starting_state = lex_state;
|
||||
self->token_start_position = self->current_position;
|
||||
self->result_follows_error = false;
|
||||
self->result_symbol = 0;
|
||||
self->first_unexpected_character = 0;
|
||||
|
||||
if (!self->chunk)
|
||||
ts_lexer__get_chunk(self);
|
||||
if (!self->lookahead_size)
|
||||
ts_lexer__get_lookahead(self);
|
||||
}
|
||||
|
||||
void ts_lexer_finish(TSLexer *self, TSLexerResult *result) {
|
||||
result->padding =
|
||||
ts_length_sub(self->token_start_position, self->token_end_position);
|
||||
|
||||
if (self->result_follows_error) {
|
||||
result->symbol = ts_builtin_sym_error;
|
||||
result->size =
|
||||
ts_length_sub(self->error_end_position, self->token_start_position);
|
||||
result->first_unexpected_character = self->first_unexpected_character;
|
||||
ts_lexer_reset(self, self->error_end_position);
|
||||
} else {
|
||||
result->symbol = self->result_symbol;
|
||||
result->size =
|
||||
ts_length_sub(self->current_position, self->token_start_position);
|
||||
self->token_end_position = self->current_position;
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -7,18 +7,10 @@ extern "C" {
|
|||
|
||||
#include "tree_sitter/parser.h"
|
||||
|
||||
typedef struct {
|
||||
TSSymbol symbol;
|
||||
TSLength padding;
|
||||
TSLength size;
|
||||
int32_t first_unexpected_character;
|
||||
} TSLexerResult;
|
||||
|
||||
void ts_lexer_init(TSLexer *);
|
||||
void ts_lexer_set_input(TSLexer *, TSInput);
|
||||
void ts_lexer_reset(TSLexer *, TSLength);
|
||||
void ts_lexer_start(TSLexer *, TSStateId);
|
||||
void ts_lexer_finish(TSLexer *, TSLexerResult *);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
|
|
|
|||
|
|
@ -243,18 +243,23 @@ static bool parser__condense_stack(Parser *self) {
|
|||
}
|
||||
|
||||
static TSTree *parser__lex(Parser *self, TSStateId parse_state) {
|
||||
TSStateId state = self->language->lex_states[parse_state];
|
||||
LOG("lex state:%d", state);
|
||||
TSStateId start_state = self->language->lex_states[parse_state];
|
||||
TSStateId current_state = start_state;
|
||||
TSLength start_position = self->lexer.current_position;
|
||||
TSLength position = start_position;
|
||||
LOG("lex state:%d", start_state);
|
||||
|
||||
TSStateId current_state = state;
|
||||
TSLength position = self->lexer.current_position;
|
||||
ts_lexer_start(&self->lexer, state);
|
||||
bool skipped_error = false;
|
||||
int32_t first_error_character = 0;
|
||||
TSLength error_start_position, error_end_position;
|
||||
|
||||
ts_lexer_start(&self->lexer, start_state);
|
||||
|
||||
while (!self->language->lex_fn(&self->lexer, current_state)) {
|
||||
if (current_state != TS_STATE_ERROR) {
|
||||
LOG("retry_in_error_mode");
|
||||
ts_lexer_reset(&self->lexer, position);
|
||||
ts_lexer_start(&self->lexer, state);
|
||||
ts_lexer_start(&self->lexer, start_state);
|
||||
current_state = TS_STATE_ERROR;
|
||||
continue;
|
||||
}
|
||||
|
|
@ -265,30 +270,39 @@ static TSTree *parser__lex(Parser *self, TSStateId parse_state) {
|
|||
}
|
||||
|
||||
if (self->lexer.current_position.chars == position.chars) {
|
||||
self->lexer.advance(&self->lexer, TS_STATE_ERROR, TSTransitionTypeError);
|
||||
if (!skipped_error) {
|
||||
error_start_position = self->lexer.current_position;
|
||||
first_error_character = self->lexer.lookahead;
|
||||
}
|
||||
skipped_error = true;
|
||||
self->lexer.advance(&self->lexer, TS_STATE_ERROR, false);
|
||||
error_end_position = self->lexer.current_position;
|
||||
}
|
||||
|
||||
position = self->lexer.current_position;
|
||||
}
|
||||
|
||||
TSLexerResult lex_result;
|
||||
ts_lexer_finish(&self->lexer, &lex_result);
|
||||
|
||||
TSTree *result;
|
||||
if (lex_result.symbol == ts_builtin_sym_error) {
|
||||
result = ts_tree_make_error(lex_result.size, lex_result.padding,
|
||||
lex_result.first_unexpected_character);
|
||||
|
||||
if (skipped_error) {
|
||||
error_start_position = ts_length_min(error_start_position, self->lexer.token_start_position);
|
||||
TSLength padding = ts_length_sub(error_start_position, start_position);
|
||||
TSLength size = ts_length_sub(error_end_position, error_start_position);
|
||||
ts_lexer_reset(&self->lexer, error_end_position);
|
||||
result = ts_tree_make_error(size, padding, first_error_character);
|
||||
} else {
|
||||
result = ts_tree_make_leaf(
|
||||
lex_result.symbol, lex_result.padding, lex_result.size,
|
||||
ts_language_symbol_metadata(self->language, lex_result.symbol));
|
||||
if (!result)
|
||||
return NULL;
|
||||
result->parse_state = parse_state;
|
||||
TSSymbol symbol = self->lexer.result_symbol;
|
||||
TSLength padding = ts_length_sub(self->lexer.token_start_position, start_position);
|
||||
TSLength size = ts_length_sub(self->lexer.current_position, self->lexer.token_start_position);
|
||||
result = ts_tree_make_leaf(symbol, padding, size,
|
||||
ts_language_symbol_metadata(self->language, symbol));
|
||||
}
|
||||
|
||||
result->first_leaf.lex_state = state;
|
||||
if (!result)
|
||||
return NULL;
|
||||
|
||||
result->parse_state = parse_state;
|
||||
result->first_leaf.lex_state = start_state;
|
||||
return result;
|
||||
}
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue