ex.c

早期freebsd实现

		/*
		 * QUOTING NOTE:
		 *
		 * The historic implementation ignored all escape characters
		 * so there was no way to put a space or newline into the +cmd
		 * field.  We do a simplistic job of fixing it by moving to the
		 * first whitespace character that isn't escaped by a literal
		 * next character.  The literal next characters are stripped
		 * as they're no longer useful.
		 */
		if (cmdlen > 0 && ch == '+') {
			++cmd;
			--cmdlen;
			for (arg1 = p = cmd; cmdlen > 0; --cmdlen, ++cmd) {
				ch = *cmd;
				if (IS_ESCAPE(sp, ch) && cmdlen > 1) {
					--cmdlen;
					ch = *++cmd;
				} else if (isblank(ch))
					break;
				*p++ = ch;
			}
			arg1_len = cmd - arg1;

			/* Reset, so the first argument isn't reparsed. */
			save_cmd = cmd;
		}
	} else if (cp == &cmds[C_BANG] ||
	    cp == &cmds[C_GLOBAL] || cp == &cmds[C_VGLOBAL]) {
		cmd += cmdlen;
		cmdlen = 0;
	} else if (cp == &cmds[C_READ] || cp == &cmds[C_WRITE]) {
		/*
		 * Move to the next character.  If it's a '!', it's a filter
		 * command and we want to eat it all, otherwise, we're done.
		 */
		for (; cmdlen > 0; --cmdlen, ++cmd) {
			ch = *cmd;
			if (!isblank(ch))
				break;
		}
		if (cmdlen > 0 && ch == '!') {
			cmd += cmdlen;
			cmdlen = 0;
		}
	} else if (cp == &cmds[C_SUBSTITUTE]) {
		/*
		 * Move to the next non-whitespace character, we'll use it as
		 * the delimiter.  If the character isn't an alphanumeric or
		 * a '|', it's the delimiter, so parse it.  Otherwise, we're
		 * into something like ":s g", so use the special substitute
		 * command.
		 */
		for (; cmdlen > 0; --cmdlen, ++cmd)
			if (!isblank(cmd[0]))
				break;

		if (isalnum(cmd[0]) || cmd[0] == '|')
			cp = &cmd_subagain;
		else if (cmdlen > 0) {
			/*
			 * QUOTING NOTE:
			 *
			 * Backslashes quote delimiter characters for RE's.
			 * The backslashes are NOT removed since they'll be
			 * used by the RE code.  Move to the third delimiter
			 * that's not escaped (or the end of the command).
			 */
			delim = *cmd;
			++cmd;
			--cmdlen;
			for (cnt = 2; cmdlen > 0 && cnt; --cmdlen, ++cmd)
				if (cmd[0] == '\\' && cmdlen > 1) {
					++cmd;
					--cmdlen;
				} else if (cmd[0] == delim)
					--cnt;
		}
	}
	/*
	 * Use normal quoting and termination rules to find the end
	 * of this command.
	 *
	 * QUOTING NOTE:
	 *
	 * Historically, vi permitted ^V's to escape <newline>'s in the .exrc
	 * file.  It was almost certainly a bug, but that's what bug-for-bug
	 * compatibility means, Grasshopper.  Also, ^V's escape the command
	 * delimiters.  Literal next quote characters in front of the newlines,
	 * '|' characters or literal next characters are stripped as as they're
	 * no longer useful.
	 */
	for (p = cmd, cnt = 0; cmdlen > 0; --cmdlen, ++cmd) {
		ch = cmd[0];
		if (IS_ESCAPE(sp, ch) && cmdlen > 1) {
			tmp = cmd[1];
			if (tmp == '\n' || tmp == '|') {
				if (tmp == '\n')
					++sp->if_lno;
				--cmdlen;
				++cmd;
				++cnt;
				ch = tmp;
			}
		} else if (ch == '\n' || ch == '|') {
			if (ch == '\n')
				nl = 1;
			--cmdlen;
			break;
		}
		*p++ = ch;
	}

	/*
	 * Save off the next command information, go back to the
	 * original start of the command.
	 */
	p = cmd + 1;
	cmd = save_cmd;
	save_cmd = p;
	save_cmdlen = cmdlen;
	cmdlen = ((save_cmd - cmd) - 1) - cnt;

	/*
	 * !!!
	 * The "set tags" command historically used a backslash, not the
	 * user's literal next character, to escape whitespace.  Handle
	 * it here instead of complicating the argv_exp3() code.  Note,
	 * this isn't a particularly complex trap, and if backslashes were
	 * legal in set commands, this would have to be much more complicated.
	 */
	if (cp == &cmds[C_SET])
		for (p = cmd, len = cmdlen; len > 0; --len, ++p)
			if (*p == '\\')
				*p = CH_LITERAL;

	/*
	 * Set the default addresses.  It's an error to specify an address for
	 * a command that doesn't take them.  If two addresses are specified
	 * for a command that only takes one, lose the first one.  Two special
	 * cases here, some commands take 0 or 2 addresses.  For most of them
	 * (the E_ADDR2_ALL flag), 0 defaults to the entire file.  For one
	 * (the `!' command, the E_ADDR2_NONE flag), 0 defaults to no lines.
	 *
	 * Also, if the file is empty, some commands want to use an address of
	 * 0, i.e. the entire file is 0 to 0, and the default first address is
	 * 0.  Otherwise, an entire file is 1 to N and the default line is 1.
	 * Note, we also add the E_ZERO flag to the command flags, for the case
	 * where the 0 address is only valid if it's a default address.
	 *
	 * Also, set a flag if we set the default addresses.  Some commands
	 * (ex: z) care if the user specified an address of if we just used
	 * the current cursor.
	 */
	switch (LF_ISSET(E_ADDR1|E_ADDR2|E_ADDR2_ALL|E_ADDR2_NONE)) {
	case E_ADDR1:				/* One address: */
		switch (exc.addrcnt) {
		case 0:				/* Default cursor/empty file. */
			exc.addrcnt = 1;
			F_SET(&exc, E_ADDRDEF);
			if (LF_ISSET(E_ZERODEF)) {
				if (file_lline(sp, ep, &lno))
					goto err;
				if (lno == 0) {
					exc.addr1.lno = 0;
					LF_SET(E_ZERO);
				} else
					exc.addr1.lno = sp->lno;
			} else
				exc.addr1.lno = sp->lno;
			exc.addr1.cno = sp->cno;
			break;
		case 1:
			break;
		case 2:				/* Lose the first address. */
			exc.addrcnt = 1;
			exc.addr1 = exc.addr2;
		}
		break;
	case E_ADDR2_NONE:			/* Zero/two addresses: */
		if (exc.addrcnt == 0)		/* Default to nothing. */
			break;
		goto two;
	case E_ADDR2_ALL:			/* Zero/two addresses: */
		if (exc.addrcnt == 0) {		/* Default entire/empty file. */
			exc.addrcnt = 2;
			F_SET(&exc, E_ADDRDEF);
			if (file_lline(sp, ep, &exc.addr2.lno))
				goto err;
			if (LF_ISSET(E_ZERODEF) && exc.addr2.lno == 0) {
				exc.addr1.lno = 0;
				LF_SET(E_ZERO);
			} else
				exc.addr1.lno = 1;
			exc.addr1.cno = exc.addr2.cno = 0;
			F_SET(&exc, E_ADDR2_ALL);
			break;
		}
		/* FALLTHROUGH */
	case E_ADDR2:				/* Two addresses: */
two:		switch (exc.addrcnt) {
		case 0:				/* Default cursor/empty file. */
			exc.addrcnt = 2;
			F_SET(&exc, E_ADDRDEF);
			if (LF_ISSET(E_ZERODEF) && sp->lno == 1) {
				if (file_lline(sp, ep, &lno))
					goto err;
				if (lno == 0) {
					exc.addr1.lno = exc.addr2.lno = 0;
					LF_SET(E_ZERO);
				} else
					exc.addr1.lno = exc.addr2.lno = sp->lno;
			} else
				exc.addr1.lno = exc.addr2.lno = sp->lno;
			exc.addr1.cno = exc.addr2.cno = sp->cno;
			break;
		case 1:				/* Default to first address. */
			exc.addrcnt = 2;
			exc.addr2 = exc.addr1;
			break;
		case 2:
			break;
		}
		break;
	default:
		if (exc.addrcnt)		/* Error. */
			goto usage;
	}

	/*
	 * !!!
	 * The ^D scroll command historically scrolled half the screen size
	 * rows down, rounded down, or to EOF.  It was an error if the cursor
	 * was already at EOF.  (Leading addresses were permitted, but were
	 * then ignored.)
	 */
	if (cp == &cmds[C_SCROLL]) {
		exc.addrcnt = 2;
		exc.addr1.lno = sp->lno + 1;
		exc.addr2.lno = sp->lno + 1 + (O_VAL(sp, O_LINES) + 1) / 2;
		exc.addr1.cno = exc.addr2.cno = sp->cno;
		if (file_lline(sp, ep, &lno))
			goto err;
		if (lno != 0 && lno > sp->lno && exc.addr2.lno > lno)
			exc.addr2.lno = lno;
	}

	flagoff = 0;
	for (p = cp->syntax; *p != '\0'; ++p) {
		/*
		 * The write command is sensitive to leading whitespace, e.g.
		 * "write !" is different from "write!".  If not the write
		 * command, skip leading whitespace.
		 */
		if (cp != &cmds[C_WRITE])
			for (; cmdlen > 0; --cmdlen, ++cmd) {
				ch = *cmd;
				if (!isblank(ch))
					break;
			}

		/*
		 * Quit when reach the end of the command, unless it's a
		 * command that does its own parsing, in which case we want
		 * to build a reasonable argv for it.  This code guarantees
		 * that there will be an argv when the function gets called,
		 * so the correct test is for a length of 0, not for the
		 * argc > 0.
		 */
		if (cmdlen == 0 && *p != '!' && *p != 'S' && *p != 's')
			break;

		switch (*p) {
		case '!':				/* ! */
			if (*cmd == '!') {
				++cmd;
				--cmdlen;
				F_SET(&exc, E_FORCE);
			}
			break;
		case '1':				/* +, -, #, l, p */
			/*
			 * !!!
			 * Historically, some flags were ignored depending
			 * on where they occurred in the command line.  For
			 * example, in the command, ":3+++p--#", historic vi
			 * acted on the '#' flag, but ignored the '-' flags.
			 * It's unambiguous what the flags mean, so we just
			 * handle them regardless of the stupidity of their
			 * location.
			 */
			for (; cmdlen; --cmdlen, ++cmd)
				switch (*cmd) {
				case '+':
					++flagoff;
					break;
				case '-':
					--flagoff;
					break;
				case '#':
					F_SET(&exc, E_F_HASH);
					break;
				case 'l':
					F_SET(&exc, E_F_LIST);
					break;
				case 'p':
					F_SET(&exc, E_F_PRINT);
					break;
				default:
					goto end1;
				}
end1:			break;
		case '2':				/* -, ., +, ^ */
		case '3':				/* -, ., +, ^, = */
			for (; cmdlen; --cmdlen, ++cmd)
				switch (*cmd) {
				case '-':
					F_SET(&exc, E_F_DASH);
					break;
				case '.':
					F_SET(&exc, E_F_DOT);
					break;
				case '+':
					F_SET(&exc, E_F_PLUS);
					break;
				case '^':
					F_SET(&exc, E_F_CARAT);
					break;
				case '=':
					if (*p == '3') {
						F_SET(&exc, E_F_EQUAL);
						break;
					}
					/* FALLTHROUGH */
				default:
					goto end2;
				}
end2:			break;
		case 'b':				/* buffer */
			/*
			 * Digits can't be buffer names in ex commands, or the
			 * command "d2" would be a delete into buffer '2', and
			 * not a two-line deletion.
			 */
			if (!isdigit(cmd[0])) {
				exc.buffer = *cmd;
				++cmd;
				--cmdlen;
				F_SET(&exc, E_BUFFER);
			}
			break;
		case 'c':				/* count [01+a] */
			++p;
			/* Validate any signed value. */
			if (!isdigit(*cmd) &&
			    (*p != '+' || (*cmd != '+' && *cmd != '-')))
				break;
			/* If a signed value, set appropriate flags. */
			if (*cmd == '-')
				F_SET(&exc, E_COUNT_NEG);
			else if (*cmd == '+')
				F_SET(&exc, E_COUNT_POS);
/* 8-bit XXX */		if ((lno = strtol(cmd, &t, 10)) == 0 && *p != '0') {
				msgq(sp, M_ERR, "Count may not be zero.");
				goto err;
			}
			cmdlen -= (t - cmd);
			cmd = t;
			/*
			 * Count as address offsets occur in commands taking
			 * two addresses.  Historic vi practice was to use
			 * the count as an offset from the *second* address.
			 *
			 * Set a count flag; some underlying commands (see
			 * join) do different things with counts than with
			 * line addresses.
			 */
			if (*p == 'a') {
				exc.addr1 = exc.addr2;
				exc.addr2.lno = exc.addr1.lno + lno - 1;
			} else
				exc.count = lno;
			F_SET(&exc, E_COUNT);
			break;
		case 'f':				/* file */
			if (argv_exp2(sp, ep,
			    &exc, cmd, cmdlen, cp == &cmds[C_BANG]))
				goto err;
			goto countchk;
		case 'l':				/* line */
			if (ep_line(sp, ep, &cur, &cmd, &cmdlen, &tmp))
				goto err;
			/* Line specifications are always required. */
			if (!tmp) {
				msgq(sp, M_ERR,
				     "%s: bad line specification", cmd);
				goto err;
			}
			exc.lineno = cur.lno;
			break;
		case 'S':				/* string, file exp. */
			if (argv_exp1(sp, ep,
			    &exc, cmd, cmdlen, cp == &cmds[C_BANG]))
				goto err;
			goto addr2;
		case 's':				/* string */
			if (argv_exp0(sp, ep, &exc, cmd, cmdlen))
				goto err;
			goto addr2;
		case 'W':				/* word string */
			/*
			 * QUOTING NOTE:
			 *
			 * Literal next characters escape the following
			 * character.  Quoting characters are stripped
			 * here since they are no longer useful.
			 *
			 * First there was the word.
			 */
			for (p = t = cmd; cmdlen > 0; --cmdlen, ++cmd) {
				ch = *cmd;
				if (IS_ESCAPE(sp, ch) && cmdlen > 1) {
					--cmdlen;
					*p++ = *++cmd;
				} else if (isblank(ch)) {
					++cmd;
					--cmdlen;
					break;
				} else
					*p++ = ch;
			}
			if (argv_exp0(sp, ep, &exc, t, p - t))
				goto err;

			/* Delete intervening whitespace. */
			for (; cmdlen > 0; --cmdlen, ++cmd) {
				ch = *cmd;
				if (!isblank(ch))
					break;
			}
			if (cmdlen == 0)
				goto usage;

			/* Followed by the string. */
			for (p = t = cmd; cmdlen > 0; --cmdlen, ++cmd, ++p) {
				ch = *cmd;
				if (IS_ESCAPE(sp, ch) && cmdlen > 1) {
					--cmdlen;
					*p = *++cmd;
				} else
					*p = ch;
			}
			if (argv_exp0(sp, ep, &exc, t, p - t))
				goto err;
			goto addr2;
		case 'w':				/* word */
			if (argv_exp3(sp, ep, &exc, cmd, cmdlen))
				goto err;
countchk:		if (*++p != 'N') {		/* N */
				/*
				 * If a number is specified, must either be
				 * 0 or that number, if optional, and that
				 * number, if required.
				 */
				num = *p - '0';
				if ((*++p != 'o' || exp->argsoff != 0) &&
				    exp->argsoff != num)
					goto usage;
			}
			goto addr2;
		default:
			msgq(sp, M_ERR,
			    "Internal syntax table error (%s: %c).",
			    cp->name, *p);
		}
	}

	/* Skip trailing whitespace. */
	for (; cmdlen; --cmdlen) {
		ch = *cmd++;
		if (!isblank(ch))
			break;
	}

	/*
	 * There shouldn't be anything left, and no more required
	 * fields, i.e neither 'l' or 'r' in the syntax string.
	 */
	if (cmdlen || strpbrk(p, "lr")) {
usage:		msgq(sp, M_ERR, "Usage: %s.", cp->usage);
		goto err;
	}

	/* Verify that the addresses are legal. */
addr2:	switch (exc.addrcnt) {
	case 2:
		if (file_lline(sp, ep, &lno))
			goto err;
		/*
		 * Historic ex/vi permitted commands with counts to go past
		 * EOF.  So, for example, if the file only had 5 lines, the
		 * ex command "1,6>" would fail, but the command ">300"
		 * would succeed.  Since we don't want to have to make all
		 * of the underlying commands handle random line numbers,
		 * fix it here.
		 */
		if (exc.addr2.lno > lno)
			if (F_ISSET(&exc, E_COUNT))
				exc.addr2.lno = lno;
			else {
				if (lno == 0)
					msgq(sp, M_ERR, "The file is empty.");
				else
					msgq(sp, M_ERR,
					    "Only %lu line%s in the file.",
					    lno, lno > 1 ? "s" : "");
				goto err;
			}
		/* FALLTHROUGH */
	case 1:
		num = exc.addr1.lno;
		/*
		 * If it's a "default vi command", zero is okay.  Historic
		 * vi allowed this, note, it's also the hack that allows
		 * "vi + nonexistent_file" to work.
		 */
		if (num == 0 && (!IN_VI_MODE(sp) || uselastcmd != 1) &&
		    !LF_ISSET(E_ZERO)) {
			msgq(sp, M_ERR,
			    "The %s command doesn't permit an address of 0.",
			    cp->name);
			goto err;
		}
		if (file_lline(sp, ep, &lno))
			goto err;
		if (num > lno) {
			if (lno == 0)
				msgq(sp, M_ERR, "The file is empty.");
			else
				msgq(sp, M_ERR, "Only %lu line%s in the file.",
				    lno, lno > 1 ? "s" : "");
			goto err;
		}
		break;