Refactor document to allow other input data structures
This commit is contained in:
Max Brunsfeld
parent
72a0c0e09c
commit
0f4b6332cb
3 changed files with 76 additions and 33 deletions
|
|
@ -37,7 +37,8 @@ typedef struct {
|
|||
} ts_stack_entry;
|
||||
|
||||
typedef struct {
|
||||
const char *input;
|
||||
ts_input input;
|
||||
const char *current_chunk;
|
||||
size_t position;
|
||||
size_t token_end_position;
|
||||
size_t token_start_position;
|
||||
|
|
@ -51,12 +52,13 @@ typedef struct {
|
|||
static void ts_lex(ts_parser *parser);
|
||||
static const ts_symbol * ts_recover(ts_state state, ts_state *to_state, size_t *count);
|
||||
|
||||
static ts_parser ts_parser_make(const char *input) {
|
||||
static ts_parser ts_parser_make(ts_input input) {
|
||||
ts_parser result = {
|
||||
.input = input,
|
||||
.token_start_position = 0,
|
||||
.token_end_position = 0,
|
||||
.position = 0,
|
||||
.current_chunk = input.read_fn(input.data),
|
||||
.lookahead_node = NULL,
|
||||
.prev_lookahead_node = NULL,
|
||||
.lex_state = 0,
|
||||
|
|
@ -67,7 +69,7 @@ static ts_parser ts_parser_make(const char *input) {
|
|||
}
|
||||
|
||||
static char ts_parser_lookahead_char(const ts_parser *parser) {
|
||||
return parser->input[parser->position];
|
||||
return parser->current_chunk[parser->position];
|
||||
}
|
||||
|
||||
static ts_symbol ts_parser_lookahead_sym(const ts_parser *parser) {
|
||||
|
|
@ -136,14 +138,22 @@ static void ts_parser_reduce(ts_parser *parser, ts_symbol symbol, int immediate_
|
|||
ts_parser_shrink_stack(parser, new_stack_size);
|
||||
DEBUG_PARSE("reduce: %s, state: %u \n", ts_symbol_names[symbol], ts_parser_parse_state(parser));
|
||||
}
|
||||
|
||||
static void ts_parser_advance(ts_parser *parser, ts_state lex_state) {
|
||||
DEBUG_LEX("character: '%c' \n", ts_parser_lookahead_char(parser));
|
||||
if (ts_parser_lookahead_char(parser))
|
||||
|
||||
static void ts_parser_advance(ts_parser *parser) {
|
||||
if (parser->current_chunk && parser->current_chunk[parser->position]) {
|
||||
parser->position++;
|
||||
parser->lex_state = lex_state;
|
||||
} else {
|
||||
parser->current_chunk = parser->input.read_fn(parser->input.data);
|
||||
parser->position = 0;
|
||||
}
|
||||
}
|
||||
|
||||
static void ts_parser_advance_to_state(ts_parser *parser, ts_state lex_state) {
|
||||
DEBUG_LEX("character: '%c' \n", ts_parser_lookahead_char(parser));
|
||||
ts_parser_advance(parser);
|
||||
parser->lex_state = lex_state;
|
||||
}
|
||||
|
||||
static void ts_parser_set_lookahead_sym(ts_parser *parser, ts_symbol symbol) {
|
||||
DEBUG_LEX("token: %s \n", ts_symbol_names[symbol]);
|
||||
size_t size = parser->position - parser->token_start_position;
|
||||
|
|
@ -159,7 +169,7 @@ static ts_tree * ts_parser_tree(ts_parser *parser) {
|
|||
|
||||
static void ts_parser_skip_whitespace(ts_parser *parser) {
|
||||
while (isspace(ts_parser_lookahead_char(parser)))
|
||||
parser->position++;
|
||||
ts_parser_advance(parser);
|
||||
parser->token_start_position = parser->position;
|
||||
}
|
||||
|
||||
|
|
@ -198,7 +208,7 @@ static int ts_parser_handle_error(ts_parser *parser, size_t count, const ts_symb
|
|||
static void ts_lex(ts_parser *parser)
|
||||
|
||||
#define PARSE_FN() \
|
||||
static const ts_tree * ts_parse(const char *input)
|
||||
static const ts_tree * ts_parse(ts_input input)
|
||||
|
||||
#define SYMBOL_NAMES \
|
||||
static const char *ts_symbol_names[] =
|
||||
|
|
@ -243,7 +253,7 @@ parser->lex_state
|
|||
{ ts_parser_shift(parser, state); goto next_state; }
|
||||
|
||||
#define ADVANCE(state_index) \
|
||||
{ ts_parser_advance(parser, state_index); goto next_state; }
|
||||
{ ts_parser_advance_to_state(parser, state_index); goto next_state; }
|
||||
|
||||
#define REDUCE(symbol, child_count, collapse_flags) \
|
||||
{ \
|
||||
|
|
|
|||
|
|
@ -40,10 +40,15 @@ char * ts_tree_error_string(const ts_tree *tree, const char **names);
|
|||
size_t ts_tree_child_count(const ts_tree *tree);
|
||||
ts_tree ** ts_tree_children(const ts_tree *tree);
|
||||
|
||||
typedef const ts_tree * ts_parse_fn(const char *);
|
||||
typedef struct {
|
||||
void *data;
|
||||
const char * (* read_fn)(void *data);
|
||||
int (* seek_fn)(void *data, size_t position);
|
||||
void (* release_fn)(void *data);
|
||||
} ts_input;
|
||||
|
||||
typedef struct {
|
||||
ts_parse_fn *parse_fn;
|
||||
const ts_tree * (* parse_fn)(ts_input);
|
||||
const char **symbol_names;
|
||||
} ts_parse_config;
|
||||
|
||||
|
|
@ -51,10 +56,11 @@ typedef struct ts_document ts_document;
|
|||
|
||||
ts_document * ts_document_make();
|
||||
void ts_document_free(ts_document *);
|
||||
void ts_document_set_parser(ts_document *document, ts_parse_config config);
|
||||
void ts_document_set_input_string(ts_document *document, const char *text);
|
||||
const ts_tree * ts_document_tree(const ts_document *document);
|
||||
const char * ts_document_string(const ts_document *document);
|
||||
void ts_document_set_parser(ts_document *, ts_parse_config);
|
||||
void ts_document_set_input(ts_document *, ts_input input);
|
||||
void ts_document_set_input_string(ts_document *, const char *text);
|
||||
const ts_tree * ts_document_tree(const ts_document *);
|
||||
const char * ts_document_string(const ts_document *);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,11 +1,10 @@
|
|||
#include "tree_sitter/runtime.h"
|
||||
#include <string.h>
|
||||
|
||||
struct ts_document {
|
||||
ts_parse_fn *parse_fn;
|
||||
const char **symbol_names;
|
||||
ts_parse_config parse_config;
|
||||
const ts_tree *tree;
|
||||
size_t error_count;
|
||||
ts_tree **errors;
|
||||
};
|
||||
|
||||
ts_document * ts_document_make() {
|
||||
|
|
@ -17,14 +16,7 @@ void ts_document_free(ts_document *document) {
|
|||
}
|
||||
|
||||
void ts_document_set_parser(ts_document *document, ts_parse_config config) {
|
||||
document->parse_fn = config.parse_fn;
|
||||
document->symbol_names = config.symbol_names;
|
||||
}
|
||||
|
||||
void ts_document_set_input_string(ts_document *document, const char *text) {
|
||||
const ts_tree * result = document->parse_fn(text);
|
||||
document->tree = result;
|
||||
document->errors = NULL;
|
||||
document->parse_config = config;
|
||||
}
|
||||
|
||||
const ts_tree * ts_document_tree(const ts_document *document) {
|
||||
|
|
@ -32,9 +24,44 @@ const ts_tree * ts_document_tree(const ts_document *document) {
|
|||
}
|
||||
|
||||
const char * ts_document_string(const ts_document *document) {
|
||||
if (document->error_count > 0) {
|
||||
return ts_tree_error_string(document->errors[0], document->symbol_names);
|
||||
} else {
|
||||
return ts_tree_string(document->tree, document->symbol_names);
|
||||
}
|
||||
return ts_tree_string(document->tree, document->parse_config.symbol_names);
|
||||
}
|
||||
|
||||
void ts_document_set_input(ts_document *document, ts_input input) {
|
||||
document->tree = document->parse_config.parse_fn(input);
|
||||
}
|
||||
|
||||
typedef struct {
|
||||
const char *string;
|
||||
size_t position;
|
||||
size_t length;
|
||||
} ts_string_input_data;
|
||||
|
||||
const char * ts_string_input_read(void *d) {
|
||||
ts_string_input_data *data = (ts_string_input_data *)d;
|
||||
return data->string + data->position;
|
||||
}
|
||||
|
||||
int ts_string_input_seek(void *d, size_t position) {
|
||||
ts_string_input_data *data = (ts_string_input_data *)d;
|
||||
data->position = position;
|
||||
return (position < data->length);
|
||||
}
|
||||
|
||||
ts_input ts_string_input_make(const char *string) {
|
||||
ts_string_input_data *data = new ts_string_input_data();
|
||||
data->string = string;
|
||||
data->position = 0;
|
||||
data->length = strlen(string);
|
||||
ts_input input = {
|
||||
.data = (void *)data,
|
||||
.read_fn = ts_string_input_read,
|
||||
.seek_fn = ts_string_input_seek,
|
||||
.release_fn = free,
|
||||
};
|
||||
return input;
|
||||
}
|
||||
|
||||
void ts_document_set_input_string(ts_document *document, const char *text) {
|
||||
ts_document_set_input(document, ts_string_input_make(text));
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue