bshell/bshell.runtime/parse/syntax/command.c

#include "../syntax.h"

#include <fx/encoding.h>

static bool parse_fword(struct bshell_parse_ctx *ctx, struct bshell_ast_node **out)
{
	if (peek_token_type(ctx) != BSHELL_TOK_WORD_START) {
		return false;
	}

	discard_token(ctx);

	struct bshell_fstring_ast_node *fstring
		= (struct bshell_fstring_ast_node *)bshell_ast_node_create(
			BSHELL_AST_FSTRING);
	if (!fstring) {
		ctx->p_status = BSHELL_ERR_NO_MEMORY;
		return false;
	}

	bool ok = true;
	while (ok) {
		if (peek_token_type(ctx) == BSHELL_TOK_WORD_END) {
			discard_token(ctx);
			break;
		}

		struct bshell_ast_node *item = NULL;
		if (peek_token_type(ctx) == BSHELL_TOK_WORD) {
			struct bshell_word_ast_node *n
				= (struct bshell_word_ast_node *)
					bshell_ast_node_create(BSHELL_AST_WORD);
			if (!n) {
				ctx->p_status = BSHELL_ERR_NO_MEMORY;
				ok = false;
				break;
			}

			n->n_value = claim_token(ctx);
			item = (struct bshell_ast_node *)n;
		} else {
			if (!parse_arith_value(ctx, &item)) {
				ok = false;
				break;
			}
		}

		fx_queue_push_back(&fstring->n_elements, &item->n_entry);
	}

	if (!ok) {
		bshell_ast_node_destroy((struct bshell_ast_node *)fstring);
		fstring = NULL;
	}

	*out = (struct bshell_ast_node *)fstring;
	return ok;
	return false;
}

static bool parse_cmdcall_arg(
	struct bshell_parse_ctx *ctx,
	struct bshell_ast_node **out)
{
	if (ctx->p_status != BSHELL_SUCCESS) {
		return false;
	}

	struct bshell_lex_token *tok = peek_token(ctx);
	if (!tok) {
		return false;
	}

	struct bshell_ast_node *arg = NULL;

	switch (tok->tok_type) {
	case BSHELL_TOK_WORD_START:
		return parse_fword(ctx, out);
	case BSHELL_TOK_STR_START:
		return parse_fstring(ctx, out);
	case BSHELL_TOK_WORD: {
		struct bshell_word_ast_node *n
			= (struct bshell_word_ast_node *)bshell_ast_node_create(
				BSHELL_AST_WORD);
		if (!n) {
			ctx->p_status = BSHELL_ERR_NO_MEMORY;
			return false;
		}

		n->n_value = claim_token(ctx);
		*out = (struct bshell_ast_node *)n;
		return true;
	}

	case BSHELL_TOK_VAR: {
		struct bshell_var_ast_node *n
			= (struct bshell_var_ast_node *)bshell_ast_node_create(
				BSHELL_AST_VAR);
		if (!n) {
			ctx->p_status = BSHELL_ERR_NO_MEMORY;
			return false;
		}

		n->n_ident = claim_token(ctx);
		*out = (struct bshell_ast_node *)n;
		return true;
	}

	case BSHELL_TOK_VAR_SPLAT: {
		struct bshell_var_splat_ast_node *n
			= (struct bshell_var_splat_ast_node *)
				bshell_ast_node_create(BSHELL_AST_VAR_SPLAT);
		if (!n) {
			ctx->p_status = BSHELL_ERR_NO_MEMORY;
			return false;
		}

		n->n_ident = claim_token(ctx);
		*out = (struct bshell_ast_node *)n;
		return true;
	}

	case BSHELL_TOK_STRING: {
		struct bshell_string_ast_node *n
			= (struct bshell_string_ast_node *)
				bshell_ast_node_create(BSHELL_AST_STRING);
		if (!n) {
			ctx->p_status = BSHELL_ERR_NO_MEMORY;
			return false;
		}

		n->n_value = claim_token(ctx);
		*out = (struct bshell_ast_node *)n;
		return true;
	}

	case BSHELL_TOK_SYMBOL:
		switch (tok->tok_symbol) {
		case BSHELL_SYM_LEFT_PAREN:
		case BSHELL_SYM_LEFT_BRACE:
		case BSHELL_SYM_DOLLAR_LEFT_PAREN:
		case BSHELL_SYM_AT_LEFT_BRACE:
		case BSHELL_SYM_AT_LEFT_PAREN:
			return parse_arith_value(ctx, out);
		default:
			report_error(
				ctx,
				"encountered unsupported command arg");
			return false;
		}
		break;
	default:
		report_error(ctx, "encountered unsupported command arg");
		return false;
	}

	return true;
}

static bool parse_redirect_to_fd(
	struct bshell_parse_ctx *ctx,
	unsigned int in_fd,
	bool append,
	struct bshell_ast_node **out)
{
	if (ctx->p_status != BSHELL_SUCCESS) {
		return false;
	}

	struct bshell_redirection_ast_node *redirect
		= (struct bshell_redirection_ast_node *)bshell_ast_node_create(
			BSHELL_AST_REDIRECTION);

	redirect->n_in = in_fd;
	redirect->n_append = append;

	if (!parse_symbol(ctx, BSHELL_SYM_AMPERSAND)) {
		bshell_ast_node_destroy((struct bshell_ast_node *)redirect);
		return false;
	}

	struct bshell_lex_token *out_tok = NULL;
	struct bshell_ast_node *out_expr = NULL;
	long long out_fd = -1;

	if (peek_word(ctx, &out_tok)) {
		const char *s = out_tok->tok_str;
		char *ep;
		out_fd = strtoll(s, &ep, 10);
		if (*ep == '\0') {
			discard_token(ctx);
			out_tok = NULL;
		} else {
			out_fd = -1;
		}
	} else if (!parse_cmdcall_arg(ctx, &out_expr)) {
		return false;
	}

	redirect->n_out_is_fd = (out_fd >= 0) || out_expr;
	redirect->n_out_is_expr = out_expr != NULL;
	redirect->n_out = (unsigned int)out_fd;
	redirect->n_out_path_expr = out_expr;
	if (out_tok) {
		redirect->n_out_tok = claim_token(ctx);
		redirect->n_out_path = out_tok->tok_str;
	}

	*out = (struct bshell_ast_node *)redirect;
	return true;
}

static bool parse_redirect_to_file_squashed(
	struct bshell_parse_ctx *ctx,
	unsigned int in_fd,
	bool append,
	const char *str,
	struct bshell_ast_node **out)
{
	if (ctx->p_status != BSHELL_SUCCESS) {
		return false;
	}

	struct bshell_lex_token *tok = peek_token(ctx);
	if (*str == '\0') {
		return false;
	}

	struct bshell_redirection_ast_node *redirect
		= (struct bshell_redirection_ast_node *)bshell_ast_node_create(
			BSHELL_AST_REDIRECTION);

	redirect->n_in = in_fd;
	redirect->n_append = append;
	redirect->n_out_is_fd = false;
	redirect->n_out_is_expr = false;
	redirect->n_out_path = str;

	redirect->n_out_tok = claim_token(ctx);

	*out = (struct bshell_ast_node *)redirect;
	return true;
}

static bool parse_redirect_to_file_separate(
	struct bshell_parse_ctx *ctx,
	unsigned int in_fd,
	bool append,
	struct bshell_ast_node **out)
{
	if (ctx->p_status != BSHELL_SUCCESS) {
		return false;
	}

	struct bshell_ast_node *out_path = NULL;
	if (!parse_cmdcall_arg(ctx, &out_path)) {
		ctx->p_status = BSHELL_ERR_BAD_SYNTAX;
		return false;
	}

	struct bshell_redirection_ast_node *redirect
		= (struct bshell_redirection_ast_node *)bshell_ast_node_create(
			BSHELL_AST_REDIRECTION);

	redirect->n_in = in_fd;
	redirect->n_append = append;
	redirect->n_out_is_fd = false;
	redirect->n_out_is_expr = true;
	redirect->n_out_path_expr = out_path;

	*out = (struct bshell_ast_node *)redirect;
	return true;
}

bool parse_redirect(struct bshell_parse_ctx *ctx, struct bshell_ast_node **out)
{
	struct bshell_lex_token *tok = peek_token(ctx);
	if (!tok || tok->tok_type != BSHELL_TOK_WORD) {
		return false;
	}

	unsigned int in_fd = 1;
	const char *str = tok->tok_str;
	bool append = false;

	if (fx_wchar_is_number(*str)) {
		in_fd = 0;
		while (fx_wchar_is_number(*str)) {
			in_fd *= 10;
			in_fd += *str - '0';
			str++;
		}
	}

	if (*str != '>') {
		return false;
	}

	str++;
	if (*str == '>') {
		append = true;
		str++;
	}

	if (*str != '\0') {
		return parse_redirect_to_file_squashed(
			ctx,
			in_fd,
			append,
			str,
			out);
	}

	discard_token(ctx);

	if (parse_redirect_to_fd(ctx, in_fd, append, out)) {
		return true;
	}

	if (parse_redirect_to_file_separate(ctx, in_fd, append, out)) {
		return true;
	}

	return false;
}

static bool peek_cmdcall_item(struct bshell_parse_ctx *ctx, bool unrestricted)
{
	/* each token type falls into one of three categories:
	 * - cmdcall item: the token can be used as part of a command call. the
	 *   token indicates the start of a command call.
	 * - NOT a cmdcall item: the token cannot be used as part of a command
	 *   call, usually because it as a cmdcall operator like | or &.
	 *   encountering one of these tokens ends the cmdcall currently being
	 *   parsed.
	 * - RESTRICTED cmdcall item: the token can be used as part of a
	 *   command, but will not be considered the start of a cmdcall. to run
	 *   a command with this token as its name, the call operator must be
	 *   used.
	 */
	switch (peek_token_type(ctx)) {
	case BSHELL_TOK_KEYWORD:
	case BSHELL_TOK_INT:
	case BSHELL_TOK_DOUBLE:
	case BSHELL_TOK_VAR:
	case BSHELL_TOK_VAR_SPLAT:
	case BSHELL_TOK_STRING:
	case BSHELL_TOK_WORD_START:
		return unrestricted;
	case BSHELL_TOK_SYMBOL:
		switch (peek_unknown_symbol(ctx)) {
		case BSHELL_SYM_PLUS:
		case BSHELL_SYM_HYPHEN:
			return unrestricted;
		case BSHELL_SYM_PIPE:
		case BSHELL_SYM_AMPERSAND:
		case BSHELL_SYM_SEMICOLON:
		case BSHELL_SYM_RIGHT_PAREN:
		case BSHELL_SYM_RIGHT_BRACE:
		case BSHELL_SYM_RIGHT_BRACKET:
			return false;
		default:
			return true;
		}
	case BSHELL_TOK_NONE:
	case BSHELL_TOK_LINEFEED:
		return false;
	default:
		return true;
	}
}

bool parse_cmdcall(struct bshell_parse_ctx *ctx, struct bshell_ast_node **out)
{
	struct bshell_cmdcall_ast_node *node
		= (struct bshell_cmdcall_ast_node *)bshell_ast_node_create(
			BSHELL_AST_CMDCALL);
	if (!node) {
		ctx->p_status = BSHELL_ERR_NO_MEMORY;
		return false;
	}

	struct bshell_ast_node *child = NULL;
	bool unrestricted = false;
	bool ok = true;
	bool stop = false;

	if (parse_symbol(ctx, BSHELL_SYM_AMPERSAND)) {
		unrestricted = true;
	}

	if (!peek_cmdcall_item(ctx, unrestricted)) {
		return false;
	}

	struct bshell_lex_token *tok = peek_token(ctx);
	if (!tok) {
		return false;
	}

	if (!parse_cmdcall_arg(ctx, &child)) {
		return false;
	}

	fx_queue_push_back(&node->n_args, &child->n_entry);

	while (ok && !stop) {
		if (!peek_cmdcall_item(ctx, true)) {
			break;
		}

		struct bshell_lex_token *tok = peek_token(ctx);
		if (!tok) {
			break;
		}

		if (parse_redirect(ctx, &child)) {
			fx_queue_push_back(&node->n_redirect, &child->n_entry);
		} else if (parse_cmdcall_arg(ctx, &child)) {
			fx_queue_push_back(&node->n_args, &child->n_entry);
		} else {
			ctx->p_status = BSHELL_ERR_BAD_SYNTAX;
			ok = false;
			break;
		}
	}

	if (!ok) {
		bshell_ast_node_destroy((struct bshell_ast_node *)node);
		node = NULL;
	}

	*out = (struct bshell_ast_node *)node;
	return ok;
}

bool peek_command(struct bshell_parse_ctx *ctx)
{
	if (peek_symbol(ctx, BSHELL_SYM_AMPERSAND)) {
		return true;
	}

	return peek_cmdcall_item(ctx, false);
}

bool parse_command(struct bshell_parse_ctx *ctx, struct bshell_ast_node **out)
{
	struct bshell_ast_node *cmdcall = NULL;
	if (!parse_cmdcall(ctx, &cmdcall)) {
		return false;
	}

	struct bshell_pipeline_ast_node *pipeline = NULL;

	while (1) {
		if (peek_symbol(ctx, BSHELL_SYM_SEMICOLON)
		    || peek_linefeed(ctx)) {
			break;
		}

		if (!parse_symbol(ctx, BSHELL_SYM_PIPE)) {
			break;
		}

		if (!pipeline) {
			pipeline = (struct bshell_pipeline_ast_node *)
				bshell_ast_node_create(BSHELL_AST_PIPELINE);
			if (!pipeline) {
				ctx->p_status = BSHELL_ERR_NO_MEMORY;
				bshell_ast_node_destroy(cmdcall);
				return false;
			}

			fx_queue_push_back(
				&pipeline->n_stages,
				&cmdcall->n_entry);
		}

		if (!parse_cmdcall(ctx, &cmdcall)) {
			ctx->p_status = BSHELL_ERR_BAD_SYNTAX;
			return false;
		}

		fx_queue_push_back(&pipeline->n_stages, &cmdcall->n_entry);
	}

	if (pipeline) {
		*out = (struct bshell_ast_node *)pipeline;
	} else {
		*out = cmdcall;
	}

	return true;
}

bool parse_pipeline(
	struct bshell_parse_ctx *ctx,
	struct bshell_ast_node *first_item,
	struct bshell_ast_node **out)
{
	struct bshell_pipeline_ast_node *pipeline
		= (struct bshell_pipeline_ast_node *)bshell_ast_node_create(
			BSHELL_AST_PIPELINE);

	fx_queue_push_back(&pipeline->n_stages, &first_item->n_entry);

	while (1) {
		if (!parse_symbol(ctx, BSHELL_SYM_PIPE)) {
			break;
		}

		struct bshell_ast_node *cmdcall = NULL;
		if (!parse_cmdcall(ctx, &cmdcall)) {
			ctx->p_status = BSHELL_ERR_BAD_SYNTAX;
			return false;
		}

		fx_queue_push_back(&pipeline->n_stages, &cmdcall->n_entry);
	}

	*out = (struct bshell_ast_node *)pipeline;
	return true;
}