bshell/bshell/parse/lex/command.c

#include "lex-internal.h"

static bool char_can_continue_word(struct lex_ctx *ctx, fx_wchar c)
{
	if (fx_wchar_is_alnum(c)) {
		return true;
	}

	if (fx_wchar_is_space(c)) {
		return false;
	}

	if (c == '$') {
		return true;
	}

	if (char_can_begin_symbol_in_state(ctx, c, LEX_STATE_WORD)) {
		return false;
	}

	return true;
}

static enum bshell_status command_symbol(struct lex_ctx *ctx)
{
	const struct lex_token_def *sym = NULL;
	enum bshell_status status = read_symbol(ctx, &sym);

	if (status != BSHELL_SUCCESS) {
		return status;
	}

	struct lex_token *tok = NULL;
	switch (sym->id) {
	case SYM_SQUOTE:
		status = read_literal_string(ctx, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}
		enqueue_token(ctx, tok);
		return BSHELL_SUCCESS;

	case SYM_HASH:
		return read_line_comment(ctx);
	case SYM_DQUOTE:
		if (!lex_state_push(ctx, LEX_STATE_STRING, 0)) {
			return BSHELL_ERR_NO_MEMORY;
		}

		return BSHELL_SUCCESS;
	case SYM_DOLLAR:
		status = read_var(ctx, TOK_VAR, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}

		if (char_can_continue_word(ctx, peek_char(ctx))) {
			lex_state_push(ctx, LEX_STATE_WORD, 0);
		}

		enqueue_token(ctx, tok);
		return status;
	case SYM_AT:
		status = read_var(ctx, TOK_VAR_SPLAT, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}

		enqueue_token(ctx, tok);
		return status;
	case SYM_DOLLAR_LEFT_BRACE:
		status = read_braced_var(ctx, TOK_VAR, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}

		if (char_can_continue_word(ctx, peek_char(ctx))) {
			lex_state_push(ctx, LEX_STATE_WORD, 0);
		}

		enqueue_token(ctx, tok);
		return status;
	case SYM_AT_LEFT_BRACE:
		status = read_braced_var(ctx, TOK_VAR_SPLAT, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}

		enqueue_token(ctx, tok);
		return status;
	default:
		break;
	}

	push_symbol(ctx, sym->id);

	switch (sym->id) {
	case SYM_LEFT_PAREN:
		lex_state_push(
			ctx,
			LEX_STATE_STATEMENT,
			STATEMENT_F_DISABLE_KEYWORDS);
		return BSHELL_SUCCESS;
	case SYM_DOLLAR_LEFT_PAREN:
		lex_state_push(ctx, LEX_STATE_STATEMENT, 0);
		return BSHELL_SUCCESS;
	case SYM_RIGHT_PAREN:
		lex_state_pop(ctx);
		return BSHELL_SUCCESS;
	case SYM_SEMICOLON:
		lex_state_change(ctx, LEX_STATE_STATEMENT);
		return BSHELL_SUCCESS;
	default:
		break;
	}

	return BSHELL_SUCCESS;
}

static enum bshell_status command_word(struct lex_ctx *ctx)
{
	struct lex_token *word = NULL;
	enum bshell_status status
		= read_word(ctx, READ_NO_NUMBER_RECOGNITION, &word);
	if (status != BSHELL_SUCCESS) {
		return status;
	}

	bool continue_word = false;

	fx_wchar c = peek_char(ctx);
	if (char_can_begin_symbol_in_state(ctx, c, LEX_STATE_WORD)) {
		continue_word = true;
	}

	if (char_has_flags(ctx, c, LEX_TOKEN_TERMINATES_WORD)) {
		continue_word = false;
	}

	if (continue_word) {
		lex_state_push(ctx, LEX_STATE_WORD, 0);
	}

	enqueue_token(ctx, word);
	return BSHELL_SUCCESS;
}

enum bshell_status command_pump_token(struct lex_ctx *ctx)
{
	fx_wchar c = peek_char(ctx);
	bool newline = false;

	set_token_start(ctx);
	while (fx_wchar_is_space(c)) {
		if (c == '\n') {
			newline = true;
		}

		set_token_end(ctx);
		advance_char_noread(ctx);
		c = peek_char_noread(ctx);
	}

	if (newline) {
		struct lex_token *tok = lex_token_create(TOK_LINEFEED);
		enqueue_token(ctx, tok);
		lex_state_change(ctx, LEX_STATE_STATEMENT);
		return BSHELL_SUCCESS;
	}

	if (char_can_begin_symbol(ctx, c)) {
		return command_symbol(ctx);
	}

	return command_word(ctx);
}

const struct lex_state_type lex_command_state = {
	.s_id = LEX_STATE_COMMAND,
	.s_pump_token = command_pump_token,
};