bshell/bshell.runtime/parse/lex/command.c

#include "lex-internal.h"

#include <bshell/parse/token.h>

static bool char_can_continue_word(struct bshell_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, BSHELL_LEX_STATE_WORD)) {
		return false;
	}

	return true;
}

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

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

	handle_lex_state_transition(ctx, sym->id);

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

	case BSHELL_SYM_HASH:
		return read_line_comment(ctx);
	case BSHELL_SYM_DOLLAR:
		status = read_var(ctx, BSHELL_TOK_VAR, &tok);
		if (status != BSHELL_SUCCESS) {
			return status;
		}

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

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

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

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

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

	push_symbol(ctx, sym->id);

	return BSHELL_SUCCESS;
}

static bool string_is_redirection(const char *s)
{
	if (!*s) {
		return false;
	}

	if (!strcmp(s, ">") || !strcmp(s, ">>")) {
		return true;
	}

	long nr_angles = 0;
	for (size_t i = 0; s[i];) {
		fx_wchar c = fx_wchar_utf8_codepoint_decode(s);
		if (fx_wchar_is_number(c)) {
			if (nr_angles) {
				return false;
			}
		} else if (c == '>') {
			nr_angles++;

			if (nr_angles > 2) {
				return false;
			}
		} else {
			return false;
		}

		s += fx_wchar_utf8_codepoint_stride(s);
	}

	return true;
}

static enum bshell_status command_word(struct bshell_lex_ctx *ctx)
{
	struct bshell_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);
	const char *s = word->tok_str;
	if (char_can_begin_symbol_in_state(ctx, c, BSHELL_LEX_STATE_WORD)) {
		continue_word = true;
	}

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

	if (string_is_redirection(s)) {
		continue_word = false;
	}

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

	enqueue_token(ctx, word);
	return BSHELL_SUCCESS;
}

enum bshell_status command_pump_token(struct bshell_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 bshell_lex_token *tok
			= bshell_lex_token_create(BSHELL_TOK_LINEFEED);
		enqueue_token(ctx, tok);
		handle_lex_state_transition(ctx, BSHELL_TOK_LINEFEED);
		return BSHELL_SUCCESS;
	}

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

	return command_word(ctx);
}

const struct bshell_lex_state_link links[] = {
	LINK_PUSH(BSHELL_SYM_DQUOTE, BSHELL_LEX_STATE_STRING, 0),
	LINK_PUSH(
		BSHELL_SYM_LEFT_PAREN,
		BSHELL_LEX_STATE_STATEMENT,
		STATEMENT_F_DISABLE_KEYWORDS),
	LINK_PUSH(BSHELL_SYM_DOLLAR_LEFT_PAREN, BSHELL_LEX_STATE_STATEMENT, 0),
	LINK_POP(BSHELL_SYM_RIGHT_PAREN),
	LINK_POP(BSHELL_SYM_RIGHT_BRACE),
	LINK_CHANGE(BSHELL_SYM_SEMICOLON, BSHELL_LEX_STATE_STATEMENT),
	LINK_PUSH(BSHELL_SYM_AT_LEFT_BRACE, BSHELL_LEX_STATE_HASHTABLE, 0),
	LINK_CHANGE(BSHELL_TOK_LINEFEED, BSHELL_LEX_STATE_STATEMENT),
	LINK_END,
};

static const unsigned int symbols[] = {
	BSHELL_SYM_DQUOTE,
	BSHELL_SYM_SQUOTE,
	BSHELL_SYM_DOLLAR,
	BSHELL_SYM_DOLLAR_LEFT_PAREN,
	BSHELL_SYM_DOLLAR_LEFT_BRACE,
	BSHELL_SYM_AT,
	BSHELL_SYM_AT_LEFT_BRACE,
	BSHELL_SYM_AT_LEFT_PAREN,
	BSHELL_SYM_AMPERSAND,
	BSHELL_SYM_PIPE,
	BSHELL_SYM_SEMICOLON,
	BSHELL_SYM_RIGHT_PAREN,
	BSHELL_SYM_LEFT_PAREN,
	BSHELL_SYM_LEFT_BRACE,
	BSHELL_SYM_RIGHT_BRACE,
	BSHELL_SYM_NONE,
};

const struct bshell_lex_state_type lex_command_state = {
	.s_id = BSHELL_LEX_STATE_COMMAND,
	.s_pump_token = command_pump_token,
	.s_links = links,
	.s_symbols = symbols,
};