raphael/minishell
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Updated WIP

Browse source
This commit is contained in:
Maieul BOYER 2024-09-02 16:28:13 +02:00
parent 7540096f6d
commit 993efe4287
No known key found for this signature in database
11 changed files with 170 additions and 426 deletions
Download patch file Download diff file Expand all files Collapse all files

2
Makefile
View file

@ -6,7 +6,7 @@
# By: rparodi <rparodi@student.42.fr> +#+ +:+ +#+ #
# +#+#+#+#+#+ +#+ #
# Created: 2023/11/12 11:05:05 by rparodi #+# #+# #
# Updated: 2024/08/31 17:01:44 by maiboyer ### ########.fr #
# Updated: 2024/09/02 16:17:16 by maiboyer ### ########.fr #
# #
# **************************************************************************** #

1
exec/Filelist.exec.mk
View file

@ -12,5 +12,6 @@ run_arithmetic/_run_arith \
run_arithmetic/_to_ast_node \
run_arithmetic/arithmetic \
run_arithmetic/arithmetic_operation \
run_arithmetic/operator_bis \
run_ast \

3
exec/src/run_arithmetic/_get_op.c
View file

@ -6,13 +6,12 @@
/* By: rparodi <rparodi@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/08/30 17:56:29 by rparodi #+# #+# */
/* Updated: 2024/08/30 17:58:54 by rparodi ### ########.fr */
/* Updated: 2024/09/02 13:45:52 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
#include "exec/_run_arith.h"
#include "me/types.h"
#include "me/convert/str_to_numbers.h"
/// ADD OPERATOR STUFF
t_error _binary_get_op(t_ast_arithmetic_operator op, t_arith_op_func *out)

50
exec/src/run_arithmetic/operator_bis.c Normal file
View file

@ -0,0 +1,50 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* operator_bis.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/09/02 13:45:46 by maiboyer #+# #+# */
/* Updated: 2024/09/02 13:50:59 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
#include "exec/_run_arith.h"
#include "me/types.h"
t_error _postfix_op_dec(t_ast_node self, t_state *state, t_i64 *out)
{
(void)(self);
(void)(state);
(void)(out);
return (ERROR);
}
t_error _postfix_op_inc(t_ast_node self, t_state *state, t_i64 *out)
{
(void)(self);
(void)(state);
(void)(out);
return (ERROR);
}
t_error _unary_op_minus(t_ast_node self, t_state *state, t_i64 *out)
{
t_i64 val;
(void)(self);
(void)(state);
(void)(out);
if (_get_node_number(self, state, &val))
return (ERROR);
val = -val;
*out = val;
return (NO_ERROR);
}
t_error _unary_op_plus(t_ast_node self, t_state *state, t_i64 *out)
{
return (_get_node_number(self, state, out));
}

13
exec/src/run_ast.c
View file

@ -6,7 +6,7 @@
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/07/11 17:22:29 by maiboyer #+# #+# */
/* Updated: 2024/08/14 18:30:23 by maiboyer ### ########.fr */
/* Updated: 2024/09/02 13:44:31 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
@ -136,7 +136,7 @@ t_error _handle_no_operator(t_ast_expansion *self, t_state *state, t_expansion_r
if (val == NULL)
return (value->exists = false, NO_ERROR);
value->exists = true;
value->value = *val;
value->value = str_clone(*val);
return (NO_ERROR);
};
@ -306,10 +306,17 @@ t_error _ast_get_str__expansion(t_ast_node elem, t_word_iterator *state, t_vec_e
t_error _ast_get_str__arimethic_expansion(t_ast_node elem, t_word_iterator *state, t_vec_estr *out)
{
t_str out_str;
t_i64 out_num;
if (elem == NULL || state == NULL || out == NULL || elem->kind != AST_ARITHMETIC_EXPANSION)
return (ERROR);
// vec_estr_push(out, (t_expandable_str){.do_expand = state->res.kind == AST_WORD_NO_QUOTE, .value = str_clone(exp_out.value)});
if (run_arithmetic_expansion(&elem->data.arithmetic_expansion, state->state, &out_num))
return (ERROR);
if (i64_to_str(out_num, &out_str))
return (ERROR);
vec_estr_push(out, (t_expandable_str){.do_expand = state->res.kind == AST_WORD_NO_QUOTE, .value = out_str});
return (NO_ERROR);
}
t_error _ast_get_str__command_substitution(t_ast_node elem, t_word_iterator *state, t_vec_estr *out)

2
parser/Filelist.parser.mk
View file

@ -37,6 +37,8 @@ scanner/advance_words \
scanner/deserialize \
scanner/heredoc \
scanner/heredoc_functions \
scanner/lifetime \
scanner/scan_dollar \
scanner/serialize \
stack/stack_add_link \
stack/stack_funcs1 \

18
parser/include/parser/inner/scanner.h
View file

@ -1,3 +1,15 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* scanner.h :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/09/02 13:22:04 by maiboyer #+# #+# */
/* Updated: 2024/09/02 13:27:44 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
#ifndef SCANNER_H
#define SCANNER_H
@ -53,5 +65,11 @@ struct s_heredoc_scan_state
typedef bool (*t_heredoc_content_func)(struct s_heredoc_scan_state *state);
bool advance_word(TSLexer *lexer, t_string *unquoted_word);
bool scan(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols);
bool scan_bare_dollar(TSLexer *lexer);
bool scan_heredoc_content(t_scanner *scanner, TSLexer *lexer, enum e_token_type middle_type, enum e_token_type end_type);
bool scan_heredoc_end_identifier(t_heredoc *heredoc, TSLexer *lexer);
bool scan_heredoc_start(t_heredoc *heredoc, TSLexer *lexer);
void reset(t_scanner *scanner);
#endif

421
parser/src/scanner.c
View file

@ -6,49 +6,20 @@
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/09/01 14:17:17 by maiboyer #+# #+# */
/* Updated: 2024/09/01 19:55:43 by maiboyer ### ########.fr */
/* Updated: 2024/09/02 16:25:19 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
#include "parser/inner/scanner.h"
#include "me/char/char.h"
#include "me/str/str.h"
#include "me/mem/mem.h"
#include "me/string/string.h"
#include "me/types.h"
#include "me/vec/vec_heredoc.h"
#include "parser/array.h"
#include "parser/inner/heredoc.h"
#include "parser/parser.h"
#include <assert.h>
bool scan_heredoc_content(t_scanner *scanner, TSLexer *lexer, enum e_token_type middle_type, enum e_token_type end_type);
bool scan_heredoc_start(t_heredoc *heredoc, TSLexer *lexer);
bool scan_heredoc_end_identifier(t_heredoc *heredoc, TSLexer *lexer);
bool advance_word(TSLexer *lexer, t_string *unquoted_word);
void reset(t_scanner *scanner)
{
t_usize i;
i = 0;
while (i < scanner->heredocs.len)
reset_heredoc(&scanner->heredocs.buffer[i++]);
}
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);
}
bool scan(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
{
if (valid_symbols[CONCAT] && !valid_symbols[ERROR_RECOVERY])
@ -103,30 +74,21 @@ bool scan(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
}
}
if (valid_symbols[EMPTY_VALUE])
{
if (me_isspace(lexer->lookahead) || lexer->eof(lexer) || lexer->lookahead == ';' || lexer->lookahead == '&')
{
lexer->result_symbol = EMPTY_VALUE;
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)
if (valid_symbols[HEREDOC_END] && scanner->heredocs.len > 0 && scan_heredoc_end_identifier(vec_heredoc_last(&scanner->heredocs), lexer))
{
t_heredoc *heredoc = vec_heredoc_last(&scanner->heredocs);
if (scan_heredoc_end_identifier(heredoc, lexer))
{
string_free(heredoc->current_leading_word);
string_free(heredoc->delimiter);
(void)vec_heredoc_pop(&scanner->heredocs, NULL);
lexer->result_symbol = HEREDOC_END;
return true;
}
return (lexer->result_symbol = HEREDOC_END, true);
}
if (valid_symbols[HEREDOC_CONTENT] && scanner->heredocs.len > 0 && vec_heredoc_last(&scanner->heredocs)->started &&
@ -290,354 +252,8 @@ bool scan(t_scanner *scanner, TSLexer *lexer, const bool *valid_symbols)
}
if (valid_symbols[BARE_DOLLAR] && !valid_symbols[ERROR_RECOVERY] && scan_bare_dollar(lexer))
{
return true;
}
if ((valid_symbols[REGEX]) && !valid_symbols[ERROR_RECOVERY])
{
if (valid_symbols[REGEX])
while (me_isspace(lexer->lookahead))
lexer->advance(lexer, true);
if ((lexer->lookahead != '"' && lexer->lookahead != '\'') || ((lexer->lookahead == '$' || lexer->lookahead == '\'')) ||
(lexer->lookahead == '\''))
{
typedef struct
{
bool done;
bool advanced_once;
bool found_non_alnumdollarunderdash;
bool last_was_escape;
bool in_single_quote;
t_u32 paren_depth;
t_u32 bracket_depth;
t_u32 brace_depth;
} State;
if (lexer->lookahead == '$')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (lexer->lookahead == '(')
return false;
}
lexer->mark_end(lexer);
State state = {false, false, false, false, false, 0, 0, 0};
while (!state.done)
{
if (state.in_single_quote)
{
if (lexer->lookahead == '\'')
{
state.in_single_quote = false;
lexer->advance(lexer, false);
lexer->mark_end(lexer);
}
}
switch (lexer->lookahead)
{
case '\\':
state.last_was_escape = true;
break;
case '\0':
return false;
case '(':
state.paren_depth++;
state.last_was_escape = false;
break;
case '[':
state.bracket_depth++;
state.last_was_escape = false;
break;
case '{':
if (!state.last_was_escape)
state.brace_depth++;
state.last_was_escape = false;
break;
case ')':
if (state.paren_depth == 0)
state.done = true;
state.paren_depth--;
state.last_was_escape = false;
break;
case ']':
if (state.bracket_depth == 0)
state.done = true;
state.bracket_depth--;
state.last_was_escape = false;
break;
case '}':
if (state.brace_depth == 0)
state.done = true;
state.brace_depth--;
state.last_was_escape = false;
break;
case '\'':
// Enter or exit a single-quoted string.
state.in_single_quote = !state.in_single_quote;
lexer->advance(lexer, false);
state.advanced_once = true;
state.last_was_escape = false;
continue;
default:
state.last_was_escape = false;
break;
}
if (!state.done)
{
if (valid_symbols[REGEX])
{
bool was_space = !state.in_single_quote && me_isspace(lexer->lookahead);
lexer->advance(lexer, false);
state.advanced_once = true;
if (!was_space || state.paren_depth > 0)
{
lexer->mark_end(lexer);
}
}
}
}
lexer->result_symbol = REGEX;
return (!(valid_symbols[REGEX] && !state.advanced_once));
}
}
extglob_pattern:
if (valid_symbols[EXTGLOB_PATTERN] && !valid_symbols[ERROR_RECOVERY])
{
// first skip ws, then check for ? * + @ !
while (me_isspace(lexer->lookahead))
lexer->advance(lexer, true);
if (lexer->lookahead == '?' || lexer->lookahead == '*' || lexer->lookahead == '+' || lexer->lookahead == '@' ||
lexer->lookahead == '!' || lexer->lookahead == '-' || lexer->lookahead == ')' || lexer->lookahead == '\\' ||
lexer->lookahead == '.' || lexer->lookahead == '[' || (me_isalpha(lexer->lookahead)))
{
if (lexer->lookahead == '\\')
{
lexer->advance(lexer, false);
if ((me_isspace(lexer->lookahead) || lexer->lookahead == '"') && lexer->lookahead != '\r' && lexer->lookahead != '\n')
lexer->advance(lexer, false);
else
return false;
}
if (lexer->lookahead == ')' && scanner->last_glob_paren_depth == 0)
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (me_isspace(lexer->lookahead))
return false;
}
lexer->mark_end(lexer);
bool was_non_alpha = !me_isalpha(lexer->lookahead);
if (lexer->lookahead != '[')
{
// no esac
if (lexer->lookahead == 'e')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (lexer->lookahead == 's')
{
lexer->advance(lexer, false);
if (lexer->lookahead == 'a')
{
lexer->advance(lexer, false);
if (lexer->lookahead == 'c')
{
lexer->advance(lexer, false);
if (me_isspace(lexer->lookahead))
return false;
}
}
}
}
else
lexer->advance(lexer, false);
}
// -\w is just a word, find something else special
if (lexer->lookahead == '-')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
while (me_isalnum(lexer->lookahead))
lexer->advance(lexer, false);
if (lexer->lookahead == ')' || lexer->lookahead == '\\' || lexer->lookahead == '.')
return false;
lexer->mark_end(lexer);
}
// case item -) or *)
if (lexer->lookahead == ')' && scanner->last_glob_paren_depth == 0)
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (me_isspace(lexer->lookahead))
{
lexer->result_symbol = EXTGLOB_PATTERN;
return was_non_alpha;
}
}
if (me_isspace(lexer->lookahead))
{
lexer->mark_end(lexer);
lexer->result_symbol = EXTGLOB_PATTERN;
scanner->last_glob_paren_depth = 0;
return true;
}
if (lexer->lookahead == '$')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (lexer->lookahead == '{' || lexer->lookahead == '(')
{
lexer->result_symbol = EXTGLOB_PATTERN;
return true;
}
}
if (lexer->lookahead == '|')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
lexer->result_symbol = EXTGLOB_PATTERN;
return true;
}
if (!me_isalnum(lexer->lookahead) && lexer->lookahead != '(' && lexer->lookahead != '"' && lexer->lookahead != '[' &&
lexer->lookahead != '?' && lexer->lookahead != '/' && lexer->lookahead != '\\' && lexer->lookahead != '_' &&
lexer->lookahead != '*')
return false;
typedef struct
{
bool done;
bool saw_non_alphadot;
t_u32 paren_depth;
t_u32 bracket_depth;
t_u32 brace_depth;
} State;
State state = {false, was_non_alpha, scanner->last_glob_paren_depth, 0, 0};
while (!state.done)
{
switch (lexer->lookahead)
{
case '\0':
return false;
case '(':
state.paren_depth++;
break;
case '[':
state.bracket_depth++;
break;
case '{':
state.brace_depth++;
break;
case ')':
if (state.paren_depth == 0)
{
state.done = true;
}
state.paren_depth--;
break;
case ']':
if (state.bracket_depth == 0)
{
state.done = true;
}
state.bracket_depth--;
break;
case '}':
if (state.brace_depth == 0)
{
state.done = true;
}
state.brace_depth--;
break;
}
if (lexer->lookahead == '|')
{
lexer->mark_end(lexer);
lexer->advance(lexer, false);
if (state.paren_depth == 0 && state.bracket_depth == 0 && state.brace_depth == 0)
{
lexer->result_symbol = EXTGLOB_PATTERN;
return true;
}
}
if (!state.done)
{
bool was_space = me_isspace(lexer->lookahead);
if (lexer->lookahead == '$')
{
lexer->mark_end(lexer);
if (!me_isalpha(lexer->lookahead) && lexer->lookahead != '.' && lexer->lookahead != '\\')
state.saw_non_alphadot = true;
lexer->advance(lexer, false);
if (lexer->lookahead == '(' || lexer->lookahead == '{')
{
lexer->result_symbol = EXTGLOB_PATTERN;
scanner->last_glob_paren_depth = state.paren_depth;
return state.saw_non_alphadot;
}
}
if (was_space)
{
lexer->mark_end(lexer);
lexer->result_symbol = EXTGLOB_PATTERN;
scanner->last_glob_paren_depth = 0;
return state.saw_non_alphadot;
}
if (lexer->lookahead == '"')
{
lexer->mark_end(lexer);
lexer->result_symbol = EXTGLOB_PATTERN;
scanner->last_glob_paren_depth = 0;
return state.saw_non_alphadot;
}
if (lexer->lookahead == '\\')
{
if (!me_isalpha(lexer->lookahead) && lexer->lookahead != '.' && lexer->lookahead != '\\')
state.saw_non_alphadot = true;
lexer->advance(lexer, false);
if (me_isspace(lexer->lookahead) || lexer->lookahead == '"')
lexer->advance(lexer, false);
}
else
{
if (!me_isalpha(lexer->lookahead) && lexer->lookahead != '.' && lexer->lookahead != '\\')
state.saw_non_alphadot = true;
lexer->advance(lexer, false);
}
if (!was_space)
lexer->mark_end(lexer);
}
}
lexer->result_symbol = EXTGLOB_PATTERN;
scanner->last_glob_paren_depth = 0;
return state.saw_non_alphadot;
}
scanner->last_glob_paren_depth = 0;
return false;
}
expansion_word:
if (valid_symbols[EXPANSION_WORD])
{
@ -716,26 +332,3 @@ expansion_word:
brace_start:
return false;
}
void *tree_sitter_sh_external_scanner_create()
{
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);
}

46
parser/src/scanner/lifetime.c Normal file
View file

@ -0,0 +1,46 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* lifetime.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/09/02 13:17:17 by maiboyer #+# #+# */
/* Updated: 2024/09/02 13:18:49 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */
#include "me/mem/mem.h"
#include "parser/inner/scanner.h"
void *tree_sitter_sh_external_scanner_create()
{
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++]);
}

28
parser/src/scanner/scan_dollar.c Normal file
View file

@ -0,0 +1,28 @@
/* ************************************************************************** */
/* */
/* ::: :::::::: */
/* scan_dollar.c :+: :+: :+: */
/* +:+ +:+ +:+ */
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/09/02 13:26:00 by maiboyer #+# #+# */
/* Updated: 2024/09/02 13:26:32 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);
}

2
stdme/src/mem/allocator.c
View file

@ -6,7 +6,7 @@
/* By: maiboyer <maiboyer@student.42.fr> +#+ +:+ +#+ */
/* +#+#+#+#+#+ +#+ */
/* Created: 2024/05/14 18:26:27 by maiboyer #+# #+# */
/* Updated: 2024/08/22 16:23:44 by maiboyer ### ########.fr */
/* Updated: 2024/09/02 13:40:13 by maiboyer ### ########.fr */
/* */
/* ************************************************************************** */