save commit

minishell-codex/src/parser/expand.c

@@ -0,0 +1,113 @@
+#include "minishell.h"
+
+static void buf_append(char **buf, size_t *len, size_t *cap, const char *s)
+{
+	size_t slen;
+	char *newbuf;
+
+	if (!s)
+		return;
+	slen = strlen(s);
+	if (*len + slen + 1 > *cap)
+	{
+		*cap = (*len + slen + 1) * 2;
+		newbuf = (char *)realloc(*buf, *cap);
+		if (!newbuf)
+			return;
+		*buf = newbuf;
+	}
+	memcpy(*buf + *len, s, slen);
+	*len += slen;
+	(*buf)[*len] = '\0';
+}
+
+static void buf_append_char(char **buf, size_t *len, size_t *cap, char c)
+{
+	char tmp[2];
+	tmp[0] = c;
+	tmp[1] = '\0';
+	buf_append(buf, len, cap, tmp);
+}
+
+static char *expand_word(const char *word, t_shell *sh)
+{
+	int in_single = 0;
+	int in_double = 0;
+	size_t i = 0;
+	char *out = NULL;
+	size_t len = 0, cap = 0;
+
+	while (word && word[i])
+	{
+		if (word[i] == '\'' && !in_double)
+		{
+			in_single = !in_single;
+			i++;
+			continue;
+		}
+		if (word[i] == '"' && !in_single)
+		{
+			in_double = !in_double;
+			i++;
+			continue;
+		}
+		if (word[i] == '$' && !in_single)
+		{
+			if (word[i + 1] == '?')
+			{
+				char *v = ms_itoa(sh->last_status);
+				buf_append(&out, &len, &cap, v);
+				free(v);
+				i += 2;
+				continue;
+			}
+			if (ms_is_alpha((unsigned char)word[i + 1]))
+			{
+				size_t j = i + 1;
+				while (word[j] && ms_is_alnum((unsigned char)word[j]))
+					j++;
+				char *name = ms_strndup(word + i + 1, j - (i + 1));
+				char *val = env_get(sh, name);
+				buf_append(&out, &len, &cap, val ? val : "");
+				free(name);
+				i = j;
+				continue;
+			}
+		}
+		buf_append_char(&out, &len, &cap, word[i]);
+		i++;
+	}
+	if (!out)
+		out = ms_strdup("");
+	return out;
+}
+
+int expand_pipeline(t_pipeline *p, t_shell *sh)
+{
+	size_t i;
+	int j;
+	for (i = 0; i < p->count; i++)
+	{
+		for (j = 0; p->cmds[i]->argv && p->cmds[i]->argv[j]; j++)
+		{
+			char *neww = expand_word(p->cmds[i]->argv[j], sh);
+			free(p->cmds[i]->argv[j]);
+			p->cmds[i]->argv[j] = neww;
+		}
+		if (p->cmds[i]->redirs)
+		{
+			t_redir *r = p->cmds[i]->redirs;
+			while (r)
+			{
+				if (r->target)
+				{
+					char *nt = expand_word(r->target, sh);
+					free(r->target);
+					r->target = nt;
+				}
+				r = r->next;
+			}
+		}
+	}
+	return 0;
+}

minishell-codex/src/parser/lexer.c

@@ -0,0 +1,131 @@
+#include "minishell.h"
+
+static t_token *token_new(t_tokentype type, const char *text, size_t len)
+{
+	t_token *t = (t_token *)calloc(1, sizeof(t_token));
+	if (!t)
+		return NULL;
+	t->type = type;
+	if (text)
+		t->text = ms_strndup(text, len);
+	return t;
+}
+
+static void token_add(t_token **head, t_token *new)
+{
+	t_token *cur;
+	if (!new)
+		return;
+	if (!*head)
+	{
+		*head = new;
+		return;
+	}
+	cur = *head;
+	while (cur->next)
+		cur = cur->next;
+	cur->next = new;
+}
+
+void free_tokens(t_token *toks)
+{
+	t_token *n;
+	while (toks)
+	{
+		n = toks->next;
+		free(toks->text);
+		free(toks);
+		toks = n;
+	}
+}
+
+static int is_meta(char c)
+{
+	return (c == '|' || c == '<' || c == '>');
+}
+
+static int read_word(const char *line, size_t *i, t_token **out)
+{
+	size_t start = *i;
+	int in_single = 0;
+	int in_double = 0;
+
+	while (line[*i])
+	{
+		if (line[*i] == '\'' && !in_double)
+			in_single = !in_single;
+		else if (line[*i] == '"' && !in_single)
+			in_double = !in_double;
+		else if (!in_single && !in_double)
+		{
+			if (ms_is_space(line[*i]) || is_meta(line[*i]))
+				break;
+		}
+		(*i)++;
+	}
+	if (in_single || in_double)
+		return 1;
+	*out = token_new(TOK_WORD, line + start, *i - start);
+	return 0;
+}
+
+t_token *lex_line(const char *line, int *error)
+{
+	t_token *toks = NULL;
+	size_t i = 0;
+
+	*error = 0;
+	while (line[i])
+	{
+		if (ms_is_space(line[i]))
+		{
+			i++;
+			continue;
+		}
+		if (line[i] == '|')
+		{
+			token_add(&toks, token_new(TOK_PIPE, "|", 1));
+			i++;
+			continue;
+		}
+		if (line[i] == '<')
+		{
+			if (line[i + 1] == '<')
+			{
+				token_add(&toks, token_new(TOK_HEREDOC, "<<", 2));
+				i += 2;
+			}
+			else
+			{
+				token_add(&toks, token_new(TOK_REDIR_IN, "<", 1));
+				i++;
+			}
+			continue;
+		}
+		if (line[i] == '>')
+		{
+			if (line[i + 1] == '>')
+			{
+				token_add(&toks, token_new(TOK_REDIR_APPEND, ">>", 2));
+				i += 2;
+			}
+			else
+			{
+				token_add(&toks, token_new(TOK_REDIR_OUT, ">", 1));
+				i++;
+			}
+			continue;
+		}
+		{
+			t_token *w = NULL;
+			if (read_word(line, &i, &w))
+			{
+				*error = 1;
+				free_tokens(toks);
+				return NULL;
+			}
+			token_add(&toks, w);
+		}
+	}
+	return toks;
+}

minishell-codex/src/parser/parser.c

@@ -0,0 +1,175 @@
+#include "minishell.h"
+
+static t_command *command_new(void)
+{
+	t_command *cmd = (t_command *)calloc(1, sizeof(t_command));
+	return cmd;
+}
+
+static void command_add_arg(t_command *cmd, const char *text)
+{
+	char **new_argv;
+	int i;
+
+	if (!cmd || !text)
+		return;
+	i = cmd->argc;
+	new_argv = (char **)calloc((size_t)i + 2, sizeof(char *));
+	if (!new_argv)
+		return;
+	for (int j = 0; j < i; j++)
+		new_argv[j] = cmd->argv[j];
+	new_argv[i] = ms_strdup(text);
+	new_argv[i + 1] = NULL;
+	free(cmd->argv);
+	cmd->argv = new_argv;
+	cmd->argc += 1;
+}
+
+static void command_add_redir(t_command *cmd, t_redirtype type, const char *target)
+{
+	t_redir *r = (t_redir *)calloc(1, sizeof(t_redir));
+	t_redir *cur;
+	if (!r)
+		return;
+	r->type = type;
+	r->target = ms_strdup(target);
+	r->fd = -1;
+	r->next = NULL;
+	if (!cmd->redirs)
+	{
+		cmd->redirs = r;
+		return;
+	}
+	cur = cmd->redirs;
+	while (cur->next)
+		cur = cur->next;
+	cur->next = r;
+}
+
+static void free_redirs(t_redir *r)
+{
+	t_redir *n;
+	while (r)
+	{
+		n = r->next;
+		if (r->fd != -1)
+			close(r->fd);
+		free(r->target);
+		free(r);
+		r = n;
+	}
+}
+
+void free_pipeline(t_pipeline *p)
+{
+	size_t i;
+	if (!p)
+		return;
+	for (i = 0; i < p->count; i++)
+	{
+		if (p->cmds[i])
+		{
+			for (int j = 0; p->cmds[i]->argv && p->cmds[i]->argv[j]; j++)
+				free(p->cmds[i]->argv[j]);
+			free(p->cmds[i]->argv);
+			free(p->cmds[i]->path);
+			free_redirs(p->cmds[i]->redirs);
+			free(p->cmds[i]);
+		}
+	}
+	free(p->cmds);
+	free(p);
+}
+
+static int pipeline_add_cmd(t_pipeline *p, t_command *cmd)
+{
+	t_command **new_cmds;
+	size_t i;
+
+	new_cmds = (t_command **)calloc(p->count + 1, sizeof(t_command *));
+	if (!new_cmds)
+		return 1;
+	for (i = 0; i < p->count; i++)
+		new_cmds[i] = p->cmds[i];
+	new_cmds[p->count] = cmd;
+	free(p->cmds);
+	p->cmds = new_cmds;
+	p->count += 1;
+	return 0;
+}
+
+t_pipeline *parse_tokens(t_token *toks, int *error)
+{
+	t_pipeline *p;
+	t_command *cmd;
+	t_token *cur;
+
+	*error = 0;
+	p = (t_pipeline *)calloc(1, sizeof(t_pipeline));
+	if (!p)
+		return NULL;
+	cmd = command_new();
+	if (!cmd)
+	{
+		free(p);
+		return NULL;
+	}
+	cur = toks;
+	while (cur)
+	{
+		if (cur->type == TOK_PIPE)
+		{
+			if (cmd->argc == 0)
+			{
+				*error = 1;
+				free_pipeline(p);
+				free(cmd);
+				return NULL;
+			}
+			pipeline_add_cmd(p, cmd);
+			cmd = command_new();
+			if (!cmd)
+			{
+				*error = 1;
+				free_pipeline(p);
+				return NULL;
+			}
+			cur = cur->next;
+			continue;
+		}
+		if (cur->type == TOK_REDIR_IN || cur->type == TOK_REDIR_OUT
+			|| cur->type == TOK_REDIR_APPEND || cur->type == TOK_HEREDOC)
+		{
+			if (!cur->next || cur->next->type != TOK_WORD)
+			{
+				*error = 1;
+				free_pipeline(p);
+				free(cmd);
+				return NULL;
+			}
+			if (cur->type == TOK_REDIR_IN)
+				command_add_redir(cmd, REDIR_IN, cur->next->text);
+			else if (cur->type == TOK_REDIR_OUT)
+				command_add_redir(cmd, REDIR_OUT, cur->next->text);
+			else if (cur->type == TOK_REDIR_APPEND)
+				command_add_redir(cmd, REDIR_APPEND, cur->next->text);
+			else if (cur->type == TOK_HEREDOC)
+				command_add_redir(cmd, REDIR_HEREDOC, cur->next->text);
+			cur = cur->next->next;
+			continue;
+		}
+		if (cur->type == TOK_WORD)
+			command_add_arg(cmd, cur->text);
+		cur = cur->next;
+	}
+	if (cmd->argc == 0)
+	{
+		*error = 1;
+		free_pipeline(p);
+		free(cmd);
+		return NULL;
+	}
+	pipeline_add_cmd(p, cmd);
+	return p;
+}