sadge to have to do a step backwards but it has to work
This commit is contained in:
parent
d4ea200c1c
commit
92823aff80
16 changed files with 1215 additions and 897 deletions
|
|
@ -28,8 +28,7 @@
|
|||
static const t_u32 MAX_VERSION_COUNT = 4;
|
||||
static const t_u32 MAX_VERSION_COUNT_OVERFLOW = 6;
|
||||
static const t_u32 MAX_SUMMARY_DEPTH = 1;
|
||||
static const t_u32 MAX_COST_DIFFERENCE = 16
|
||||
* ERROR_COST_PER_SKIPPED_TREE;
|
||||
static const t_u32 MAX_COST_DIFFERENCE = 16 * ERROR_COST_PER_SKIPPED_TREE;
|
||||
|
||||
typedef struct s_error_status t_error_status;
|
||||
typedef enum e_error_comparison t_error_comparison;
|
||||
|
|
@ -269,16 +268,16 @@ static bool ts_parser__better_version_exists(TSParser *self,
|
|||
return false;
|
||||
}
|
||||
|
||||
static bool ts_parser__call_main_lex_fn(TSParser *self, TSLexMode lex_mode)
|
||||
{
|
||||
return self->language->lex_fn(&self->lexer.data, lex_mode.lex_state);
|
||||
}
|
||||
|
||||
static bool ts_parser__call_keyword_lex_fn(TSParser *self, TSLexMode lex_mode)
|
||||
{
|
||||
(void)(lex_mode);
|
||||
return self->language->keyword_lex_fn(&self->lexer.data, 0);
|
||||
}
|
||||
// static bool ts_parser__call_main_lex_fn(TSParser *self, TSLexMode lex_mode)
|
||||
// {
|
||||
// return self->language->lex_fn(&self->lexer.data, lex_mode.lex_state);
|
||||
// }
|
||||
//
|
||||
// static bool ts_parser__call_keyword_lex_fn(TSParser *self, TSLexMode lex_mode)
|
||||
// {
|
||||
// (void)(lex_mode);
|
||||
// return self->language->keyword_lex_fn(&self->lexer.data, 0);
|
||||
// }
|
||||
|
||||
static void ts_parser__external_scanner_create(TSParser *self)
|
||||
{
|
||||
|
|
@ -299,6 +298,7 @@ static t_u32 ts_parser__external_scanner_serialize(TSParser *self)
|
|||
{
|
||||
t_u32 length;
|
||||
|
||||
|
||||
length = self->language->external_scanner.serialize(self->external_scanner_payload,
|
||||
self->lexer.debug_buffer);
|
||||
if (length > TREE_SITTER_SERIALIZATION_BUFFER_SIZE)
|
||||
|
|
@ -319,6 +319,7 @@ static void ts_parser__external_scanner_deserialize(TSParser *self,
|
|||
{
|
||||
data = ts_external_scanner_state_data(&external_token->external_scanner_state);
|
||||
length = external_token->external_scanner_state.length;
|
||||
printf("HERE\n");
|
||||
}
|
||||
self->language->external_scanner.deserialize(self->external_scanner_payload,
|
||||
data, length);
|
||||
|
|
@ -430,7 +431,7 @@ static t_subtree ts_parser__lex(TSParser *self, t_stack_version version,
|
|||
ts_lexer_reset(&self->lexer, current_position);
|
||||
}
|
||||
ts_lexer_start(&self->lexer);
|
||||
found_token = ts_parser__call_main_lex_fn(self, lex_mode);
|
||||
found_token = self->language->lex_fn(&self->lexer.data, lex_mode.lex_state);
|
||||
ts_lexer_finish(&self->lexer, &lookahead_end_byte);
|
||||
if (found_token)
|
||||
break ;
|
||||
|
|
@ -485,7 +486,7 @@ static t_subtree ts_parser__lex(TSParser *self, t_stack_version version,
|
|||
end_byte = self->lexer.token_end_position.bytes;
|
||||
ts_lexer_reset(&self->lexer, self->lexer.token_start_position);
|
||||
ts_lexer_start(&self->lexer);
|
||||
is_keyword = ts_parser__call_keyword_lex_fn(self, lex_mode);
|
||||
is_keyword = self->language->keyword_lex_fn(&self->lexer.data, 0);
|
||||
if (is_keyword && self->lexer.token_end_position.bytes == end_byte
|
||||
&& ts_language_has_actions(self->language, parse_state,
|
||||
self->lexer.data.result_symbol))
|
||||
|
|
|
|||
1198
parser/src/scanner.c
Normal file
1198
parser/src/scanner.c
Normal file
File diff suppressed because it is too large
Load diff
|
|
@ -1,52 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* advance_words.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/01 19:28:19 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/01 20:10:14 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "me/string/string.h"
|
||||
#include "me/types.h"
|
||||
#include "parser/parser.h"
|
||||
|
||||
void advance_word_inner(TSLexer *lexer, bool *empty, t_i32 *quote)
|
||||
{
|
||||
*empty = true;
|
||||
*quote = 0;
|
||||
if (lexer->lookahead == '\'' || lexer->lookahead == '"')
|
||||
{
|
||||
*quote = lexer->lookahead;
|
||||
lexer->advance(lexer, false);
|
||||
}
|
||||
}
|
||||
|
||||
bool advance_word(TSLexer *lexer, t_string *unquoted_word)
|
||||
{
|
||||
bool empty;
|
||||
t_i32 quote;
|
||||
|
||||
advance_word_inner(lexer, &empty, "e);
|
||||
while (lexer->lookahead && !((quote && (lexer->lookahead == quote
|
||||
|| lexer->lookahead == '\r' || lexer->lookahead == '\n'))
|
||||
|| (!quote && (me_isspace(lexer->lookahead)))))
|
||||
{
|
||||
if (lexer->lookahead == '\\')
|
||||
{
|
||||
lexer->advance(lexer, false);
|
||||
if (!lexer->lookahead)
|
||||
return (false);
|
||||
}
|
||||
empty = false;
|
||||
string_push_char(unquoted_word, lexer->lookahead);
|
||||
lexer->advance(lexer, false);
|
||||
}
|
||||
if (quote && lexer->lookahead == quote)
|
||||
lexer->advance(lexer, false);
|
||||
return (!empty);
|
||||
}
|
||||
|
|
@ -1,68 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* deserialize.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/01 15:06:56 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/01 20:08:37 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/mem/mem.h"
|
||||
#include "me/types.h"
|
||||
#include "me/vec/vec_heredoc.h"
|
||||
#include "parser/inner/heredoc.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
void reset(t_scanner *scanner);
|
||||
|
||||
void actual_reset(\
|
||||
t_scanner *scanner, const t_u8 *buffer, t_usize i, t_u32 *size)
|
||||
{
|
||||
t_heredoc *heredoc;
|
||||
t_usize delim_size;
|
||||
|
||||
heredoc = NULL;
|
||||
if (i < scanner->heredocs.len)
|
||||
heredoc = vec_heredoc_get(&scanner->heredocs, i);
|
||||
else
|
||||
{
|
||||
vec_heredoc_push(&scanner->heredocs, heredoc_new());
|
||||
heredoc = vec_heredoc_last(&scanner->heredocs);
|
||||
}
|
||||
heredoc->is_raw = buffer[(*size)++];
|
||||
heredoc->started = buffer[(*size)++];
|
||||
heredoc->allows_indent = buffer[(*size)++];
|
||||
mem_copy(&delim_size, &buffer[(*size)], sizeof(t_usize));
|
||||
(*size) += sizeof(t_usize);
|
||||
string_reserve(&heredoc->delimiter, delim_size + 1);
|
||||
heredoc->delimiter.len = delim_size - 1;
|
||||
if (delim_size > 0)
|
||||
{
|
||||
mem_copy(heredoc->delimiter.buf, &buffer[(*size)], delim_size);
|
||||
(*size) += delim_size;
|
||||
}
|
||||
}
|
||||
|
||||
void tree_sitter_sh_external_scanner_deserialize(t_scanner *scanner,
|
||||
const t_u8 *buffer, t_u32 length)
|
||||
{
|
||||
t_u32 size;
|
||||
t_u32 heredoc_count;
|
||||
t_usize i;
|
||||
|
||||
if (length == 0)
|
||||
return (reset(scanner));
|
||||
size = 0;
|
||||
scanner->last_glob_paren_depth = buffer[size++];
|
||||
scanner->ext_was_in_double_quote = buffer[size++];
|
||||
scanner->ext_saw_outside_quote = buffer[size++];
|
||||
heredoc_count = (t_u8)buffer[size++];
|
||||
i = 0;
|
||||
while (i < heredoc_count)
|
||||
actual_reset(scanner, buffer, i++, &size);
|
||||
if (size != length)
|
||||
me_abort("size != length");
|
||||
}
|
||||
|
|
@ -1,103 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* heredoc.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/01 19:33:04 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:31:02 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "me/str/str.h"
|
||||
#include "me/types.h"
|
||||
#include "me/vec/vec_heredoc.h"
|
||||
#include "parser/inner/heredoc.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
#include "parser/parser.h"
|
||||
|
||||
bool scan_heredoc_start(t_heredoc *heredoc, TSLexer *lexer)
|
||||
{
|
||||
bool found_delimiter;
|
||||
|
||||
while (me_isspace(lexer->lookahead))
|
||||
lexer->advance(lexer, true);
|
||||
lexer->result_symbol = HEREDOC_START;
|
||||
heredoc->is_raw = lexer->lookahead == '\'' || lexer->lookahead == '"'
|
||||
|| lexer->lookahead == '\\';
|
||||
found_delimiter = advance_word(lexer, &heredoc->delimiter);
|
||||
if (!found_delimiter)
|
||||
return (string_clear(&heredoc->delimiter), false);
|
||||
return (found_delimiter);
|
||||
}
|
||||
|
||||
bool scan_heredoc_end_identifier(t_heredoc *heredoc, TSLexer *lexer)
|
||||
{
|
||||
t_i32 size;
|
||||
|
||||
size = 0;
|
||||
string_clear(&heredoc->current_leading_word);
|
||||
if (heredoc->delimiter.len > 0)
|
||||
{
|
||||
while (lexer->lookahead != '\0' && lexer->lookahead != '\n'
|
||||
&& (t_i32)heredoc->delimiter.buf[size] == lexer->lookahead
|
||||
&& heredoc->current_leading_word.len < heredoc->delimiter.len)
|
||||
{
|
||||
string_push_char(&heredoc->current_leading_word, lexer->lookahead);
|
||||
lexer->advance(lexer, false);
|
||||
size++;
|
||||
}
|
||||
}
|
||||
if (heredoc->delimiter.len == 0)
|
||||
return (false);
|
||||
return (str_compare(heredoc->current_leading_word.buf,
|
||||
heredoc->delimiter.buf));
|
||||
}
|
||||
|
||||
bool scan_heredoc_content_nullbyte(struct s_heredoc_scan_state *state);
|
||||
bool scan_heredoc_content_backslash(struct s_heredoc_scan_state *state);
|
||||
bool scan_heredoc_content_dollar(struct s_heredoc_scan_state *state);
|
||||
bool scan_heredoc_content_newline(struct s_heredoc_scan_state *state);
|
||||
bool scan_heredoc_content_other(struct s_heredoc_scan_state *state);
|
||||
|
||||
bool scan_heredoc_content(t_scanner *scanner, TSLexer *lexer,
|
||||
enum e_token_type middle_type, enum e_token_type end_type)
|
||||
{
|
||||
struct s_heredoc_scan_state state;
|
||||
t_heredoc_content_func func;
|
||||
|
||||
state = (struct s_heredoc_scan_state){.did_advance = false, .lexer = lexer,
|
||||
.heredoc = vec_heredoc_last(&scanner->heredocs), .scanner = scanner,
|
||||
.middle_type = middle_type, .end_type = end_type,
|
||||
.return_value = false};
|
||||
while (true)
|
||||
{
|
||||
if (lexer->lookahead == '\0')
|
||||
func = scan_heredoc_content_nullbyte;
|
||||
else if (lexer->lookahead == '\\')
|
||||
func = scan_heredoc_content_backslash;
|
||||
else if (lexer->lookahead == '$')
|
||||
func = scan_heredoc_content_dollar;
|
||||
else if (lexer->lookahead == '\n')
|
||||
func = scan_heredoc_content_newline;
|
||||
else
|
||||
func = scan_heredoc_content_other;
|
||||
if (func(&state))
|
||||
return (state.return_value);
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
|
||||
bool scan_heredoc_end(t_scanner *scanner, TSLexer *lexer)
|
||||
{
|
||||
t_heredoc *heredoc;
|
||||
|
||||
heredoc = vec_heredoc_last(&scanner->heredocs);
|
||||
string_free(heredoc->current_leading_word);
|
||||
string_free(heredoc->delimiter);
|
||||
vec_heredoc_pop(&scanner->heredocs, NULL);
|
||||
lexer->result_symbol = HEREDOC_END;
|
||||
return (true);
|
||||
}
|
||||
|
|
@ -1,117 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* heredoc_functions.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/01 19:36:53 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 18:04:32 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "me/str/str.h"
|
||||
#include "me/types.h"
|
||||
#include "me/vec/vec_heredoc.h"
|
||||
#include "parser/inner/heredoc.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
#include "parser/parser.h"
|
||||
|
||||
bool scan_heredoc_end_identifier(t_heredoc *heredoc, TSLexer *lexer);
|
||||
|
||||
bool scan_heredoc_content_nullbyte(struct s_heredoc_scan_state *state)
|
||||
{
|
||||
if (state->lexer->eof(state->lexer) && state->did_advance)
|
||||
{
|
||||
reset_heredoc(state->heredoc);
|
||||
state->lexer->result_symbol = state->end_type;
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
return (state->return_value = false, true);
|
||||
}
|
||||
|
||||
bool scan_heredoc_content_backslash(struct s_heredoc_scan_state *state)
|
||||
{
|
||||
state->did_advance = true;
|
||||
state->lexer->advance(state->lexer, false);
|
||||
state->lexer->advance(state->lexer, false);
|
||||
return (false);
|
||||
}
|
||||
|
||||
bool scan_heredoc_content_dollar(struct s_heredoc_scan_state *state)
|
||||
{
|
||||
if (state->heredoc->is_raw)
|
||||
{
|
||||
state->did_advance = true;
|
||||
state->lexer->advance(state->lexer, false);
|
||||
}
|
||||
if (state->did_advance)
|
||||
{
|
||||
state->lexer->mark_end(state->lexer);
|
||||
state->lexer->result_symbol = state->middle_type;
|
||||
state->heredoc->started = true;
|
||||
state->lexer->advance(state->lexer, false);
|
||||
if (me_isalpha(state->lexer->lookahead)
|
||||
|| state->lexer->lookahead == '{' || state->lexer->lookahead == '(')
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
if (state->middle_type == HEREDOC_BODY_BEGINNING
|
||||
&& state->lexer->get_column(state->lexer) == 0)
|
||||
{
|
||||
state->lexer->result_symbol = state->middle_type;
|
||||
state->heredoc->started = true;
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
return (state->return_value = false, true);
|
||||
}
|
||||
|
||||
bool scan_heredoc_content_newline(struct s_heredoc_scan_state *state)
|
||||
{
|
||||
if (!state->did_advance)
|
||||
state->lexer->advance(state->lexer, true);
|
||||
else
|
||||
state->lexer->advance(state->lexer, false);
|
||||
state->did_advance = true;
|
||||
if (state->heredoc->allows_indent)
|
||||
{
|
||||
while (me_isspace(state->lexer->lookahead))
|
||||
state->lexer->advance(state->lexer, false);
|
||||
}
|
||||
state->lexer->result_symbol = state->end_type;
|
||||
if (state->heredoc->started)
|
||||
state->lexer->result_symbol = state->middle_type;
|
||||
state->lexer->mark_end(state->lexer);
|
||||
if (scan_heredoc_end_identifier(state->heredoc, state->lexer))
|
||||
{
|
||||
if (state->lexer->result_symbol == HEREDOC_END)
|
||||
vec_heredoc_pop(&state->scanner->heredocs, NULL);
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
|
||||
bool scan_heredoc_content_other(struct s_heredoc_scan_state *state)
|
||||
{
|
||||
if (state->lexer->get_column(state->lexer) == 0)
|
||||
{
|
||||
while (me_isspace(state->lexer->lookahead))
|
||||
state->lexer->advance(state->lexer, !state->did_advance);
|
||||
if (state->end_type != SIMPLE_HEREDOC_BODY)
|
||||
{
|
||||
state->lexer->result_symbol = state->middle_type;
|
||||
if (scan_heredoc_end_identifier(state->heredoc, state->lexer))
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
if (state->end_type == SIMPLE_HEREDOC_BODY)
|
||||
{
|
||||
state->lexer->result_symbol = state->end_type;
|
||||
state->lexer->mark_end(state->lexer);
|
||||
if (scan_heredoc_end_identifier(state->heredoc, state->lexer))
|
||||
return (state->return_value = true, true);
|
||||
}
|
||||
}
|
||||
state->did_advance = true;
|
||||
state->lexer->advance(state->lexer, false);
|
||||
return (false);
|
||||
}
|
||||
|
|
@ -1,47 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* lifetime.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 13:17:17 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:45:47 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/mem/mem.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
void *tree_sitter_sh_external_scanner_create(void)
|
||||
{
|
||||
t_scanner *scanner;
|
||||
|
||||
scanner = mem_alloc(sizeof(*scanner));
|
||||
scanner->heredocs = vec_heredoc_new(16, heredoc_free);
|
||||
return (scanner);
|
||||
}
|
||||
|
||||
bool tree_sitter_sh_external_scanner_scan(\
|
||||
void *payload, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
t_scanner *scanner;
|
||||
|
||||
scanner = (t_scanner *)payload;
|
||||
return (scan(scanner, lexer, valid_symbols));
|
||||
}
|
||||
|
||||
void tree_sitter_sh_external_scanner_destroy(void *payload)
|
||||
{
|
||||
vec_heredoc_free(((t_scanner *)payload)->heredocs);
|
||||
mem_free((t_scanner *)payload);
|
||||
}
|
||||
|
||||
void reset(t_scanner *scanner)
|
||||
{
|
||||
t_usize i;
|
||||
|
||||
i = 0;
|
||||
while (i < scanner->heredocs.len)
|
||||
reset_heredoc(&scanner->heredocs.buffer[i++]);
|
||||
}
|
||||
|
|
@ -1,59 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 17:40:46 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:41:52 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
static bool scan2(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
if (valid_symbols[HEREDOC_CONTENT] && scanner->heredocs.len > 0
|
||||
&& vec_heredoc_last(&scanner->heredocs)->started
|
||||
&& !valid_symbols[ERROR_RECOVERY])
|
||||
return (scan_heredoc_content(scanner, lexer, HEREDOC_CONTENT,
|
||||
HEREDOC_END));
|
||||
if (valid_symbols[HEREDOC_START] && !valid_symbols[ERROR_RECOVERY]
|
||||
&& scanner->heredocs.len > 0)
|
||||
return (scan_heredoc_start(vec_heredoc_last(&scanner->heredocs),
|
||||
lexer));
|
||||
if ((valid_symbols[VARIABLE_NAME] || valid_symbols[FILE_DESCRIPTOR]
|
||||
|| valid_symbols[HEREDOC_ARROW]) && !valid_symbols[ERROR_RECOVERY])
|
||||
return (scan_varname(scanner, lexer, valid_symbols));
|
||||
if (valid_symbols[BARE_DOLLAR] && !valid_symbols[ERROR_RECOVERY]
|
||||
&& scan_bare_dollar(lexer))
|
||||
return (true);
|
||||
if (valid_symbols[EXPANSION_WORD])
|
||||
return (scan_expansion_word(scanner, lexer, valid_symbols));
|
||||
return (false);
|
||||
}
|
||||
|
||||
bool scan(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
if (check_scan_concat(scanner, lexer, valid_symbols))
|
||||
return (scan_concat(scanner, lexer, valid_symbols));
|
||||
if (scan_double_hash(scanner, lexer, valid_symbols))
|
||||
return (true);
|
||||
if (valid_symbols[EMPTY_VALUE] && (me_isspace(lexer->lookahead)
|
||||
|| lexer->eof(lexer) || lexer->lookahead == ';'
|
||||
|| lexer->lookahead == '&'))
|
||||
return (lexer->result_symbol = EMPTY_VALUE, true);
|
||||
if ((valid_symbols[HEREDOC_BODY_BEGINNING]
|
||||
|| valid_symbols[SIMPLE_HEREDOC_BODY]) && scanner->heredocs.len > 0
|
||||
&& !vec_heredoc_last(&scanner->heredocs)->started
|
||||
&& !valid_symbols[ERROR_RECOVERY])
|
||||
return (scan_heredoc_content(scanner, lexer, HEREDOC_BODY_BEGINNING,
|
||||
SIMPLE_HEREDOC_BODY));
|
||||
if (valid_symbols[HEREDOC_END] && scanner->heredocs.len > 0
|
||||
&& scan_heredoc_end_identifier(vec_heredoc_last(&scanner->heredocs),
|
||||
lexer))
|
||||
return (scan_heredoc_end(scanner, lexer));
|
||||
return (scan2(scanner, lexer, valid_symbols));
|
||||
}
|
||||
|
|
@ -1,57 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan_concat.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 17:37:05 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 18:03:17 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
bool scan_concat(\
|
||||
t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
((void)(scanner), (void)(valid_symbols));
|
||||
lexer->result_symbol = CONCAT;
|
||||
if (lexer->lookahead == '`')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
while (lexer->lookahead != '`' && !lexer->eof(lexer))
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->eof(lexer))
|
||||
return (false);
|
||||
if (lexer->lookahead == '`')
|
||||
lexer->advance(lexer, false);
|
||||
return (me_isspace(lexer->lookahead) || lexer->eof(lexer));
|
||||
}
|
||||
if (lexer->lookahead == '\\')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '"' || lexer->lookahead == '\'' \
|
||||
|| lexer->lookahead == '\\')
|
||||
return (true);
|
||||
if (lexer->eof(lexer))
|
||||
return (false);
|
||||
}
|
||||
return (true);
|
||||
}
|
||||
|
||||
bool check_scan_concat(\
|
||||
t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
(void)(scanner);
|
||||
return ((valid_symbols[CONCAT] && !valid_symbols[ERROR_RECOVERY]) \
|
||||
&& (!(lexer->lookahead == 0 || me_isspace(lexer->lookahead) \
|
||||
|| lexer->lookahead == '>' || lexer->lookahead == '<' \
|
||||
|| lexer->lookahead == ')' || lexer->lookahead == '(' \
|
||||
|| lexer->lookahead == ';' || lexer->lookahead == '&' \
|
||||
|| lexer->lookahead == '|' || lexer->lookahead == '{' \
|
||||
|| lexer->lookahead == '}')));
|
||||
}
|
||||
|
|
@ -1,30 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan_dollar.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 13:26:00 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:46:26 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
bool scan_bare_dollar(TSLexer *lexer)
|
||||
{
|
||||
while (me_isspace(lexer->lookahead) \
|
||||
&& lexer->lookahead != '\n' && !lexer->eof(lexer))
|
||||
lexer->advance(lexer, true);
|
||||
if (lexer->lookahead == '$')
|
||||
{
|
||||
lexer->advance(lexer, false);
|
||||
lexer->result_symbol = BARE_DOLLAR;
|
||||
lexer->mark_end(lexer);
|
||||
return (me_isspace(lexer->lookahead) \
|
||||
|| lexer->eof(lexer) || lexer->lookahead == '\"');
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
|
|
@ -1,38 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan_double_hash.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 17:32:35 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 18:00:13 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
bool scan_double_hash(\
|
||||
t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
|
||||
{
|
||||
(void)(scanner);
|
||||
if (valid_symbols[IMMEDIATE_DOUBLE_HASH] && !valid_symbols[ERROR_RECOVERY])
|
||||
{
|
||||
if (lexer->lookahead == '#')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '#')
|
||||
{
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead != '}')
|
||||
{
|
||||
lexer->result_symbol = IMMEDIATE_DOUBLE_HASH;
|
||||
lexer->mark_end(lexer);
|
||||
return (true);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
|
|
@ -1,140 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan_varname.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 17:26:05 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 18:01:41 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
bool scan_varname(t_scanner *scanner, TSLexer *lexer,
|
||||
const bool *valid_symbols)
|
||||
{
|
||||
t_heredoc heredoc;
|
||||
bool is_number;
|
||||
|
||||
while (true)
|
||||
{
|
||||
if ((lexer->lookahead == ' ' || lexer->lookahead == '\t'
|
||||
|| lexer->lookahead == '\r' || (lexer->lookahead == '\n'
|
||||
&& !valid_symbols[NEWLINE]))
|
||||
&& !valid_symbols[EXPANSION_WORD])
|
||||
lexer->advance(lexer, true);
|
||||
else if (lexer->lookahead == '\\')
|
||||
{
|
||||
lexer->advance(lexer, true);
|
||||
if (lexer->eof(lexer))
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->result_symbol = VARIABLE_NAME;
|
||||
return (true);
|
||||
}
|
||||
if (lexer->lookahead == '\r')
|
||||
lexer->advance(lexer, true);
|
||||
if (lexer->lookahead == '\n')
|
||||
lexer->advance(lexer, true);
|
||||
else
|
||||
{
|
||||
if (lexer->lookahead == '\\' && valid_symbols[EXPANSION_WORD])
|
||||
return (scan_expansion_word(scanner, lexer, valid_symbols));
|
||||
return (false);
|
||||
}
|
||||
}
|
||||
else
|
||||
break ;
|
||||
}
|
||||
if (!valid_symbols[EXPANSION_WORD] && (lexer->lookahead == '*'
|
||||
|| lexer->lookahead == '@' || lexer->lookahead == '?'
|
||||
|| lexer->lookahead == '-' || lexer->lookahead == '0'
|
||||
|| lexer->lookahead == '_'))
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '=' || lexer->lookahead == '['
|
||||
|| lexer->lookahead == ':' || lexer->lookahead == '-'
|
||||
|| lexer->lookahead == '%' || lexer->lookahead == '#'
|
||||
|| lexer->lookahead == '/')
|
||||
return (false);
|
||||
if (valid_symbols[EXTGLOB_PATTERN] && me_isspace(lexer->lookahead))
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->result_symbol = EXTGLOB_PATTERN;
|
||||
return (true);
|
||||
}
|
||||
}
|
||||
if (valid_symbols[HEREDOC_ARROW] && lexer->lookahead == '<')
|
||||
{
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '<')
|
||||
{
|
||||
lexer->advance(lexer, false);
|
||||
heredoc = heredoc_new();
|
||||
vec_heredoc_push(&scanner->heredocs, heredoc);
|
||||
lexer->result_symbol = HEREDOC_ARROW;
|
||||
return (true);
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
is_number = true;
|
||||
if (me_isdigit(lexer->lookahead))
|
||||
lexer->advance(lexer, false);
|
||||
else if (me_isalpha(lexer->lookahead) || lexer->lookahead == '_')
|
||||
{
|
||||
is_number = false;
|
||||
lexer->advance(lexer, false);
|
||||
}
|
||||
else
|
||||
{
|
||||
if (lexer->lookahead == '{')
|
||||
return (false);
|
||||
if (valid_symbols[EXPANSION_WORD])
|
||||
return (scan_expansion_word(scanner, lexer, valid_symbols));
|
||||
return (false);
|
||||
}
|
||||
while (true)
|
||||
{
|
||||
if (me_isdigit(lexer->lookahead))
|
||||
lexer->advance(lexer, false);
|
||||
else if (me_isalpha(lexer->lookahead) || lexer->lookahead == '_')
|
||||
is_number = (lexer->advance(lexer, false), false);
|
||||
else
|
||||
break ;
|
||||
}
|
||||
if (is_number && valid_symbols[FILE_DESCRIPTOR] && (lexer->lookahead == '>'
|
||||
|| lexer->lookahead == '<'))
|
||||
return (lexer->result_symbol = FILE_DESCRIPTOR, true);
|
||||
if (valid_symbols[VARIABLE_NAME])
|
||||
{
|
||||
if (lexer->lookahead == '+')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '=' || lexer->lookahead == ':')
|
||||
return (lexer->result_symbol = VARIABLE_NAME, true);
|
||||
return (false);
|
||||
}
|
||||
if (lexer->lookahead == '/')
|
||||
return (false);
|
||||
if (lexer->lookahead == '=' || lexer->lookahead == '['
|
||||
|| (lexer->lookahead == ':' && !valid_symbols[OPENING_PAREN])
|
||||
|| lexer->lookahead == '%' || (lexer->lookahead == '#'
|
||||
&& !is_number) || lexer->lookahead == '@'
|
||||
|| (lexer->lookahead == '-'))
|
||||
return (lexer->mark_end(lexer),
|
||||
lexer->result_symbol = VARIABLE_NAME, true);
|
||||
if (lexer->lookahead == '?')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
lexer->result_symbol = VARIABLE_NAME;
|
||||
return (me_isalpha(lexer->lookahead));
|
||||
}
|
||||
}
|
||||
return (false);
|
||||
}
|
||||
|
|
@ -1,97 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* scan_word.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/02 16:59:16 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:58:43 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/char/char.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
|
||||
bool scan_expansion_word(t_scanner *scanner, TSLexer *lexer,
|
||||
const bool *valid_symbols)
|
||||
{
|
||||
bool advanced_once;
|
||||
bool advance_once_space;
|
||||
|
||||
advanced_once = false;
|
||||
advance_once_space = false;
|
||||
(void)(scanner);
|
||||
(void)(lexer);
|
||||
(void)(valid_symbols);
|
||||
while (true)
|
||||
{
|
||||
if (lexer->lookahead == '\"')
|
||||
return (false);
|
||||
if (lexer->lookahead == '$')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '{' || lexer->lookahead == '('
|
||||
|| lexer->lookahead == '\'' || me_isalnum(lexer->lookahead))
|
||||
{
|
||||
lexer->result_symbol = EXPANSION_WORD;
|
||||
return (advanced_once);
|
||||
}
|
||||
advanced_once = true;
|
||||
}
|
||||
if (lexer->lookahead == '}')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->result_symbol = EXPANSION_WORD;
|
||||
return (advanced_once || advance_once_space);
|
||||
}
|
||||
if (lexer->lookahead == '(' && !(advanced_once || advance_once_space))
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
while (lexer->lookahead != ')' && !lexer->eof(lexer))
|
||||
{
|
||||
if (lexer->lookahead == '$')
|
||||
{
|
||||
lexer->mark_end(lexer);
|
||||
lexer->advance(lexer, false);
|
||||
if (lexer->lookahead == '{' || lexer->lookahead == '('
|
||||
|| lexer->lookahead == '\''
|
||||
|| me_isalnum(lexer->lookahead))
|
||||
{
|
||||
lexer->result_symbol = EXPANSION_WORD;
|
||||
return (advanced_once);
|
||||
}
|
||||
advanced_once = true;
|
||||
}
|
||||
else
|
||||
{
|
||||
advanced_once = advanced_once
|
||||
|| !me_isspace(lexer->lookahead);
|
||||
advance_once_space = advance_once_space
|
||||
|| me_isspace(lexer->lookahead);
|
||||
lexer->advance(lexer, false);
|
||||
}
|
||||
}
|
||||
lexer->mark_end(lexer);
|
||||
if (lexer->lookahead == ')')
|
||||
{
|
||||
advanced_once = true;
|
||||
lexer->advance(lexer, false);
|
||||
lexer->mark_end(lexer);
|
||||
if (lexer->lookahead == '}')
|
||||
return (false);
|
||||
}
|
||||
else
|
||||
return (false);
|
||||
}
|
||||
if (lexer->lookahead == '\'')
|
||||
return (false);
|
||||
if (lexer->eof(lexer))
|
||||
return (false);
|
||||
advanced_once = advanced_once || !me_isspace(lexer->lookahead);
|
||||
advance_once_space = advance_once_space || me_isspace(lexer->lookahead);
|
||||
lexer->advance(lexer, false);
|
||||
}
|
||||
}
|
||||
|
|
@ -1,62 +0,0 @@
|
|||
/* ************************************************************************** */
|
||||
/* */
|
||||
/* ::: :::::::: */
|
||||
/* serialize.c :+: :+: :+: */
|
||||
/* +:+ +:+ +:+ */
|
||||
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
|
||||
/* +#+#+#+#+#+ +#+ */
|
||||
/* Created: 2024/09/01 15:06:56 by maiboyer #+# #+# */
|
||||
/* Updated: 2024/09/02 17:06:26 by maiboyer ### ########.fr */
|
||||
/* */
|
||||
/* ************************************************************************** */
|
||||
|
||||
#include "me/types.h"
|
||||
#include "parser/array.h"
|
||||
#include "parser/inner/heredoc.h"
|
||||
#include "parser/inner/scanner.h"
|
||||
#include "parser/parser.h"
|
||||
|
||||
t_error serialize_heredocs(t_scanner *scanner, t_u8 *buffer, t_u32 *size,
|
||||
t_usize i)
|
||||
{
|
||||
t_heredoc *heredoc;
|
||||
|
||||
heredoc = vec_heredoc_get(&scanner->heredocs, i);
|
||||
if (heredoc == NULL)
|
||||
return (ERROR);
|
||||
if (heredoc->delimiter.len + 1 + sizeof(t_usize)
|
||||
+ (*size) >= TREE_SITTER_SERIALIZATION_BUFFER_SIZE)
|
||||
return (ERROR);
|
||||
buffer[(*size)++] = (char)heredoc->is_raw;
|
||||
buffer[(*size)++] = (char)heredoc->started;
|
||||
buffer[(*size)++] = (char)heredoc->allows_indent;
|
||||
heredoc->delimiter.len++;
|
||||
mem_copy(&buffer[(*size)], &heredoc->delimiter.len, sizeof(t_usize));
|
||||
size += sizeof(t_usize);
|
||||
if (heredoc->delimiter.len > 0)
|
||||
{
|
||||
mem_copy(&buffer[(*size)], heredoc->delimiter.buf,
|
||||
heredoc->delimiter.len);
|
||||
(*size) += heredoc->delimiter.len;
|
||||
}
|
||||
heredoc->delimiter.len--;
|
||||
return (NO_ERROR);
|
||||
}
|
||||
|
||||
t_u32 tree_sitter_sh_external_scanner_serialize(t_scanner *scanner,
|
||||
t_u8 *buffer)
|
||||
{
|
||||
t_u32 size;
|
||||
t_usize i;
|
||||
|
||||
size = 0;
|
||||
buffer[size++] = (char)scanner->last_glob_paren_depth;
|
||||
buffer[size++] = (char)scanner->ext_was_in_double_quote;
|
||||
buffer[size++] = (char)scanner->ext_saw_outside_quote;
|
||||
buffer[size++] = (char)scanner->heredocs.len;
|
||||
i = 0;
|
||||
while (i < scanner->heredocs.len)
|
||||
if (serialize_heredocs(scanner, buffer, &size, i++))
|
||||
return (0);
|
||||
return (size);
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue