lexer.h File Reference

#include "./length.h"
#include "./subtree.h"
#include "tree_sitter/api.h"
#include "tree_sitter/parser.h"

Go to the source code of this file.

Classes
struct	Lexer

Functions
void	ts_lexer_init (Lexer *)
void	ts_lexer_delete (Lexer *)
void	ts_lexer_set_input (Lexer *, TSInput)
void	ts_lexer_reset (Lexer *, Length)
void	ts_lexer_start (Lexer *)
void	ts_lexer_finish (Lexer *, uint32_t *)
void	ts_lexer_advance_to_end (Lexer *)
void	ts_lexer_mark_end (Lexer *)
bool	ts_lexer_set_included_ranges (Lexer *self, const TSRange *ranges, uint32_t count)
TSRange *	ts_lexer_included_ranges (const Lexer *self, uint32_t *count)

Function Documentation

ts_lexer_advance_to_end()

void ts_lexer_advance_to_end

(

Lexer *

self

)

Definition at line 357 of file lexer.c.

                                          {
  while (self->chunk) {
    ts_lexer__advance(&self->data, false);
  }
}

References ts_lexer__advance().

Referenced by parser__halt_parse().

ts_lexer_delete()

void ts_lexer_delete

(

Lexer *

self

)

Definition at line 304 of file lexer.c.

                                  {
  ts_free(self->included_ranges);
}

References ts_free.

Referenced by ts_parser_delete().

ts_lexer_finish()

void ts_lexer_finish	(	Lexer *	self,
		uint32_t *	lookahead_end_byte
	)

Definition at line 337 of file lexer.c.

                                                                {
  if (length_is_undefined(self->token_end_position)) {
    ts_lexer__mark_end(&self->data);
  }
 
  uint32_t current_lookahead_end_byte = self->current_position.bytes + 1;
 
  // In order to determine that a byte sequence is invalid UTF8 or UTF16,
  // the character decoding algorithm may have looked at the following byte.
  // Therefore, the next byte *after* the current (invalid) character
  // affects the interpretation of the current character.
  if (self->data.lookahead == TS_DECODE_ERROR) {
    current_lookahead_end_byte++;
  }
 
  if (current_lookahead_end_byte > *lookahead_end_byte) {
    *lookahead_end_byte = current_lookahead_end_byte;
  }
}

References length_is_undefined(), TS_DECODE_ERROR, and ts_lexer__mark_end().

Referenced by ts_parser__lex().

ts_lexer_included_ranges()

TSRange* ts_lexer_included_ranges	(	const Lexer *	self,
		uint32_t *	count
	)

Definition at line 395 of file lexer.c.

                                                                      {
  *count = self->included_range_count;
  return self->included_ranges;
}

References count.

Referenced by ts_parser_included_ranges().

ts_lexer_init()

void ts_lexer_init

(

Lexer *

self

)

Definition at line 275 of file lexer.c.

                                {
  *self = (Lexer) {
    .data = {
      // The lexer's methods are stored as struct fields so that generated
      // parsers can call them without needing to be linked against this
      // library.
      .advance = ts_lexer__advance,
      .mark_end = ts_lexer__mark_end,
      .get_column = ts_lexer__get_column,
      .is_at_included_range_start = ts_lexer__is_at_included_range_start,
      .eof = ts_lexer__eof,
      .lookahead = 0,
      .result_symbol = 0,
    },
    .chunk = NULL,
    .chunk_size = 0,
    .chunk_start = 0,
    .current_position = {0, {0, 0}},
    .logger = {
      .payload = NULL,
      .log = NULL
    },
    .included_ranges = NULL,
    .included_range_count = 0,
    .current_included_range_index = 0,
  };
  ts_lexer_set_included_ranges(self, NULL, 0);
}

References NULL, ts_lexer__advance(), ts_lexer__eof(), ts_lexer__get_column(), ts_lexer__is_at_included_range_start(), ts_lexer__mark_end(), and ts_lexer_set_included_ranges().

Referenced by parser_init(), and ts_parser_new().

ts_lexer_mark_end()

void ts_lexer_mark_end

(

Lexer *

self

)

Definition at line 363 of file lexer.c.

                                    {
  ts_lexer__mark_end(&self->data);
}

References ts_lexer__mark_end().

Referenced by ts_parser__handle_error().

ts_lexer_reset()

void ts_lexer_reset	(	Lexer *	self,
		Length	position
	)

Definition at line 316 of file lexer.c.

                                                  {
  if (position.bytes != self->current_position.bytes) {
    ts_lexer_goto(self, position);
  }
}

References Length::bytes, and ts_lexer_goto().

Referenced by parser__lex(), ts_parser__handle_error(), ts_parser__lex(), and ts_parser_reset().

ts_lexer_set_included_ranges()

bool ts_lexer_set_included_ranges	(	Lexer *	self,
		const TSRange *	ranges,
		uint32_t	count
	)

Definition at line 367 of file lexer.c.

  {
  if (count == 0 || !ranges) {
    ranges = &DEFAULT_RANGE;
    count = 1;
  } else {
    uint32_t previous_byte = 0;
    for (unsigned i = 0; i < count; i++) {
      const TSRange *range = &ranges[i];
      if (
        range->start_byte < previous_byte ||
        range->end_byte < range->start_byte
      ) return false;
      previous_byte = range->end_byte;
    }
  }
 
  size_t size = count * sizeof(TSRange);
  self->included_ranges = ts_realloc(self->included_ranges, size);
  memcpy(self->included_ranges, ranges, size);
  self->included_range_count = count;
  ts_lexer_goto(self, self->current_position);
  return true;
}

References count, DEFAULT_RANGE, i, memcpy(), capstone::range, ts_lexer_goto(), and ts_realloc.

Referenced by ts_lexer_init(), and ts_parser_set_included_ranges().

ts_lexer_set_input()

void ts_lexer_set_input	(	Lexer *	self,
		TSInput	input
	)

Definition at line 308 of file lexer.c.

                                                    {
  self->input = input;
  ts_lexer__clear_chunk(self);
  ts_lexer_goto(self, self->current_position);
}

References input(), ts_lexer__clear_chunk(), and ts_lexer_goto().

Referenced by parser__start(), and ts_parser_parse().

ts_lexer_start()

void ts_lexer_start

(

Lexer *

self

)

Definition at line 322 of file lexer.c.

                                 {
  self->token_start_position = self->current_position;
  self->token_end_position = LENGTH_UNDEFINED;
  self->data.result_symbol = 0;
  self->did_get_column = false;
  if (!ts_lexer__eof(&self->data)) {
    if (!self->chunk_size) ts_lexer__get_chunk(self);
    if (!self->lookahead_size) ts_lexer__get_lookahead(self);
    if (
      self->current_position.bytes == 0 &&
      self->data.lookahead == BYTE_ORDER_MARK
    ) ts_lexer__advance(&self->data, true);
  }
}

References BYTE_ORDER_MARK, LENGTH_UNDEFINED, ts_lexer__advance(), ts_lexer__eof(), ts_lexer__get_chunk(), and ts_lexer__get_lookahead().

Referenced by parser__lex(), and ts_parser__lex().