NetBSD/usr.bin/vi/common/search.c

515 lines
12 KiB
C
Raw Normal View History

1998-01-09 11:03:16 +03:00
/* $NetBSD: search.c,v 1.3 1998/01/09 08:07:08 perry Exp $ */
1994-08-17 20:17:19 +04:00
/*-
* Copyright (c) 1992, 1993, 1994
* The Regents of the University of California. All rights reserved.
1996-05-20 06:01:36 +04:00
* Copyright (c) 1992, 1993, 1994, 1995, 1996
* Keith Bostic. All rights reserved.
1994-08-17 20:17:19 +04:00
*
1996-05-20 06:01:36 +04:00
* See the LICENSE file for redistribution information.
1994-08-17 20:17:19 +04:00
*/
1996-05-20 06:01:36 +04:00
#include "config.h"
1994-08-17 20:17:19 +04:00
#ifndef lint
static const char sccsid[] = "@(#)search.c 10.24 (Berkeley) 5/26/96";
1994-08-17 20:17:19 +04:00
#endif /* not lint */
#include <sys/types.h>
#include <sys/queue.h>
#include <bitstring.h>
#include <ctype.h>
#include <errno.h>
#include <limits.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
1996-05-20 06:01:36 +04:00
#include "common.h"
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
typedef enum { S_EMPTY, S_EOF, S_NOPREV, S_NOTFOUND, S_SOF, S_WRAP } smsg_t;
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
static void search_msg __P((SCR *, smsg_t));
static int search_setup __P((SCR *, dir_t, char *, char **, u_int));
1994-08-17 20:17:19 +04:00
/*
1996-05-20 06:01:36 +04:00
* search_setup --
* Set up a search.
1994-08-17 20:17:19 +04:00
*/
static int
1996-05-20 06:01:36 +04:00
search_setup(sp, dir, ptrn, epp, flags)
1994-08-17 20:17:19 +04:00
SCR *sp;
1996-05-20 06:01:36 +04:00
dir_t dir;
1994-08-17 20:17:19 +04:00
char *ptrn, **epp;
u_int flags;
1996-05-20 06:01:36 +04:00
{
recno_t lno;
int delim;
1994-08-17 20:17:19 +04:00
char *p, *t;
1996-05-20 06:01:36 +04:00
/* If the file is empty, it's a fast search. */
if (sp->lno <= 1) {
if (db_last(sp, &lno))
return (1);
if (lno == 0) {
if (LF_ISSET(SEARCH_MSG))
search_msg(sp, S_EMPTY);
1994-08-17 20:17:19 +04:00
return (1);
}
1996-05-20 06:01:36 +04:00
}
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
if (LF_ISSET(SEARCH_PARSE)) { /* Parse the string. */
/*
* Use the saved pattern if no pattern supplied, or if only
* the delimiter character supplied.
*
* !!!
* Only the pattern itself was saved, historically vi didn't
* reuse addressing or delta information.
*/
if (ptrn == NULL)
goto prev;
if (ptrn[1] == '\0') {
if (epp != NULL)
*epp = ptrn + 1;
goto prev;
1994-08-17 20:17:19 +04:00
}
1996-05-20 06:01:36 +04:00
if (ptrn[0] == ptrn[1]) {
if (epp != NULL)
*epp = ptrn + 2;
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
/* Complain if we don't have a previous pattern. */
prev: if (sp->re == NULL) {
search_msg(sp, S_NOPREV);
return (1);
}
/* Compile the search pattern if necessary. */
if (!F_ISSET(sp, SC_RE_SEARCH) &&
re_compile(sp, sp->re, NULL, NULL, &sp->re_c,
RE_C_SEARCH |
(LF_ISSET(SEARCH_MSG) ? 0 : RE_C_SILENT)))
return (1);
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
/* Set the search direction. */
if (LF_ISSET(SEARCH_SET))
sp->searchdir = dir;
return (0);
}
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
/*
* Set the delimiter, and move forward to the terminating
* delimiter, handling escaped delimiters.
*
* QUOTING NOTE:
* Only discard an escape character if it escapes a delimiter.
*/
for (delim = *ptrn, p = t = ++ptrn;; *t++ = *p++) {
1994-08-17 20:17:19 +04:00
if (p[0] == '\0' || p[0] == delim) {
if (p[0] == delim)
++p;
*t = '\0';
break;
}
if (p[1] == delim && p[0] == '\\')
++p;
}
if (epp != NULL)
*epp = p;
}
/* Compile the RE. */
1996-05-20 06:01:36 +04:00
if (re_compile(sp, ptrn, &sp->re, &sp->re_len, &sp->re_c,
RE_C_SEARCH |
(LF_ISSET(SEARCH_MSG) ? 0 : RE_C_SILENT) |
(LF_ISSET(SEARCH_TAG) ? RE_C_TAG : 0) |
(LF_ISSET(SEARCH_CSCOPE) ? RE_C_CSCOPE : 0)))
return (1);
/* Set the search direction. */
if (LF_ISSET(SEARCH_SET))
1994-08-17 20:17:19 +04:00
sp->searchdir = dir;
1996-05-20 06:01:36 +04:00
return (0);
1994-08-17 20:17:19 +04:00
}
/*
1996-05-20 06:01:36 +04:00
* f_search --
* Do a forward search.
*
* PUBLIC: int f_search __P((SCR *, MARK *, MARK *, char *, char **, u_int));
1994-08-17 20:17:19 +04:00
*/
int
1996-05-20 06:01:36 +04:00
f_search(sp, fm, rm, ptrn, eptrn, flags)
1994-08-17 20:17:19 +04:00
SCR *sp;
MARK *fm, *rm;
char *ptrn, **eptrn;
1996-05-20 06:01:36 +04:00
u_int flags;
1994-08-17 20:17:19 +04:00
{
1996-05-20 06:01:36 +04:00
busy_t btype;
1994-08-17 20:17:19 +04:00
recno_t lno;
1996-05-20 06:01:36 +04:00
regmatch_t match[1];
1994-08-17 20:17:19 +04:00
size_t coff, len;
1996-05-20 06:01:36 +04:00
int cnt, eval, rval, wrapped;
1994-08-17 20:17:19 +04:00
char *l;
1996-05-20 06:01:36 +04:00
if (search_setup(sp, FORWARD, ptrn, eptrn, flags))
1994-08-17 20:17:19 +04:00
return (1);
if (LF_ISSET(SEARCH_FILE)) {
lno = 1;
coff = 0;
} else {
1996-05-20 06:01:36 +04:00
if (db_get(sp, fm->lno, DBG_FATAL, &l, &len))
1994-08-17 20:17:19 +04:00
return (1);
1996-05-20 06:01:36 +04:00
lno = fm->lno;
/*
* If doing incremental search, start searching at the previous
* column, so that we search a minimal distance and still match
* special patterns, e.g., \< for beginning of a word.
*
* Otherwise, start searching immediately after the cursor. If
* at the end of the line, start searching on the next line.
* This is incompatible (read bug fix) with the historic vi --
* searches for the '$' pattern never moved forward, and the
* "-t foo" didn't work if the 'f' was the first character in
* the file.
*/
if (LF_ISSET(SEARCH_INCR)) {
if ((coff = fm->cno) != 0)
--coff;
} else if (fm->cno + 1 >= len) {
coff = 0;
lno = fm->lno + 1;
if (db_get(sp, lno, 0, &l, &len)) {
1994-08-17 20:17:19 +04:00
if (!O_ISSET(sp, O_WRAPSCAN)) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_EOF);
1994-08-17 20:17:19 +04:00
return (1);
}
lno = 1;
1996-05-20 06:01:36 +04:00
}
} else
1994-08-17 20:17:19 +04:00
coff = fm->cno + 1;
}
1996-05-20 06:01:36 +04:00
btype = BUSY_ON;
for (cnt = INTERRUPT_CHECK, rval = 1, wrapped = 0;; ++lno, coff = 0) {
if (cnt-- == 0) {
if (INTERRUPTED(sp))
break;
if (LF_ISSET(SEARCH_MSG)) {
search_busy(sp, btype);
btype = BUSY_UPDATE;
}
cnt = INTERRUPT_CHECK;
1994-08-17 20:17:19 +04:00
}
1996-05-20 06:01:36 +04:00
if (wrapped && lno > fm->lno || db_get(sp, lno, 0, &l, &len)) {
1994-08-17 20:17:19 +04:00
if (wrapped) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_NOTFOUND);
1994-08-17 20:17:19 +04:00
break;
}
if (!O_ISSET(sp, O_WRAPSCAN)) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_EOF);
1994-08-17 20:17:19 +04:00
break;
}
lno = 0;
wrapped = 1;
continue;
}
/* If already at EOL, just keep going. */
1996-05-20 06:01:36 +04:00
if (len != 0 && coff == len)
1994-08-17 20:17:19 +04:00
continue;
/* Set the termination. */
match[0].rm_so = coff;
match[0].rm_eo = len;
#if defined(DEBUG) && 0
TRACE(sp, "F search: %lu from %u to %u\n",
1996-05-20 06:01:36 +04:00
lno, coff, len != 0 ? len - 1 : len);
1994-08-17 20:17:19 +04:00
#endif
/* Search the line. */
1996-05-20 06:01:36 +04:00
eval = regexec(&sp->re_c, l, 1, match,
1994-08-17 20:17:19 +04:00
(match[0].rm_so == 0 ? 0 : REG_NOTBOL) | REG_STARTEND);
if (eval == REG_NOMATCH)
continue;
if (eval != 0) {
1996-05-20 06:01:36 +04:00
if (LF_ISSET(SEARCH_MSG))
re_error(sp, eval, &sp->re_c);
else
(void)sp->gp->scr_bell(sp);
1994-08-17 20:17:19 +04:00
break;
}
/*
1996-05-20 06:01:36 +04:00
* XXX
* Warn if the search wrapped. This message is only displayed
* if we're in interactive mode and there are no keys in the
* queue. The problem is that the command is going to succeed,
* and the message isn't an error, it's merely informational.
* So, if we're reading the .exrc file, displaying it causes
* the screen to pause before entering the file, which is bad.
* Or, if a macro causes it to be repeatedly displayed, e.g.,
* the pattern only occurs once in the file and wrapscan is set,
* you lose, particularly if the macro does something like:
1996-05-20 06:01:36 +04:00
* :map K /pattern/^MjK
* Each new search will display the message and the /pattern/
* will immediately overwrite it, with strange results. The
* System V vi displays the "wrapped" message multiple times,
* but it's overwritten each time, so it's not as noticeable.
* Since we don't discard messages, it's a real problem.
*
* XXX
* The test for SC_SCR_EX or SC_SCR_VI feels wrong to me. I
* didn't originally intend for that bit to reflect that the
* user has "gone interactive", but that's the only solution
* I see.
1994-08-17 20:17:19 +04:00
*/
if (wrapped && LF_ISSET(SEARCH_MSG) &&
F_ISSET(sp, SC_SCR_EX | SC_SCR_VI) && !KEYS_WAITING(sp))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_WRAP);
1994-08-17 20:17:19 +04:00
#if defined(DEBUG) && 0
1996-05-20 06:01:36 +04:00
TRACE(sp, "F search: %qu to %qu\n",
match[0].rm_so, match[0].rm_eo);
1994-08-17 20:17:19 +04:00
#endif
1996-05-20 06:01:36 +04:00
rm->lno = lno;
rm->cno = match[0].rm_so;
/*
* If a change command, it's possible to move beyond the end
* of a line. Historic vi generally got this wrong (e.g. try
* "c?$<cr>"). Not all that sure this gets it right, there
* are lots of strange cases.
*/
if (!LF_ISSET(SEARCH_EOL) && rm->cno >= len)
rm->cno = len != 0 ? len - 1 : 0;
1994-08-17 20:17:19 +04:00
rval = 0;
break;
}
1996-05-20 06:01:36 +04:00
if (LF_ISSET(SEARCH_MSG))
search_busy(sp, BUSY_OFF);
1994-08-17 20:17:19 +04:00
return (rval);
}
1996-05-20 06:01:36 +04:00
/*
* b_search --
* Do a backward search.
*
* PUBLIC: int b_search __P((SCR *, MARK *, MARK *, char *, char **, u_int));
*/
1994-08-17 20:17:19 +04:00
int
1996-05-20 06:01:36 +04:00
b_search(sp, fm, rm, ptrn, eptrn, flags)
1994-08-17 20:17:19 +04:00
SCR *sp;
MARK *fm, *rm;
char *ptrn, **eptrn;
1996-05-20 06:01:36 +04:00
u_int flags;
1994-08-17 20:17:19 +04:00
{
1996-05-20 06:01:36 +04:00
busy_t btype;
1994-08-17 20:17:19 +04:00
recno_t lno;
1996-05-20 06:01:36 +04:00
regmatch_t match[1];
size_t coff, last, len;
int cnt, eval, rval, wrapped;
1994-08-17 20:17:19 +04:00
char *l;
1996-05-20 06:01:36 +04:00
if (search_setup(sp, BACKWARD, ptrn, eptrn, flags))
1994-08-17 20:17:19 +04:00
return (1);
1996-05-20 06:01:36 +04:00
/*
* If doing incremental search, set the "starting" position past the
* current column, so that we search a minimal distance and still
* match special patterns, e.g., \> for the end of a word. This is
* safe when the cursor is at the end of a line because we only use
* it for comparison with the location of the match.
*
* Otherwise, start searching immediately before the cursor. If in
* the first column, start search on the previous line.
*/
if (LF_ISSET(SEARCH_INCR)) {
lno = fm->lno;
coff = fm->cno + 1;
} else {
if (fm->cno == 0) {
if (fm->lno == 1 && !O_ISSET(sp, O_WRAPSCAN)) {
1994-08-17 20:17:19 +04:00
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_SOF);
1994-08-17 20:17:19 +04:00
return (1);
}
lno = fm->lno - 1;
1996-05-20 06:01:36 +04:00
} else
lno = fm->lno;
coff = fm->cno;
}
1994-08-17 20:17:19 +04:00
1996-05-20 06:01:36 +04:00
btype = BUSY_ON;
for (cnt = INTERRUPT_CHECK, rval = 1, wrapped = 0;; --lno, coff = 0) {
if (cnt-- == 0) {
if (INTERRUPTED(sp))
break;
if (LF_ISSET(SEARCH_MSG)) {
search_busy(sp, btype);
btype = BUSY_UPDATE;
}
cnt = INTERRUPT_CHECK;
1994-08-17 20:17:19 +04:00
}
if (wrapped && lno < fm->lno || lno == 0) {
if (wrapped) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_NOTFOUND);
1994-08-17 20:17:19 +04:00
break;
}
if (!O_ISSET(sp, O_WRAPSCAN)) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_SOF);
1994-08-17 20:17:19 +04:00
break;
}
1996-05-20 06:01:36 +04:00
if (db_last(sp, &lno))
break;
1994-08-17 20:17:19 +04:00
if (lno == 0) {
if (LF_ISSET(SEARCH_MSG))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_EMPTY);
1994-08-17 20:17:19 +04:00
break;
}
++lno;
wrapped = 1;
continue;
}
1996-05-20 06:01:36 +04:00
if (db_get(sp, lno, 0, &l, &len))
break;
1994-08-17 20:17:19 +04:00
/* Set the termination. */
match[0].rm_so = 0;
match[0].rm_eo = len;
#if defined(DEBUG) && 0
TRACE(sp, "B search: %lu from 0 to %qu\n", lno, match[0].rm_eo);
#endif
/* Search the line. */
1996-05-20 06:01:36 +04:00
eval = regexec(&sp->re_c, l, 1, match,
1994-08-17 20:17:19 +04:00
(match[0].rm_eo == len ? 0 : REG_NOTEOL) | REG_STARTEND);
if (eval == REG_NOMATCH)
continue;
if (eval != 0) {
1996-05-20 06:01:36 +04:00
if (LF_ISSET(SEARCH_MSG))
re_error(sp, eval, &sp->re_c);
else
(void)sp->gp->scr_bell(sp);
1994-08-17 20:17:19 +04:00
break;
}
/* Check for a match starting past the cursor. */
if (coff != 0 && match[0].rm_so >= coff)
continue;
1996-05-20 06:01:36 +04:00
/*
* XXX
* See the comment in f_search() for more information.
*/
if (wrapped && LF_ISSET(SEARCH_MSG) &&
F_ISSET(sp, SC_SCR_EX | SC_SCR_VI) && !KEYS_WAITING(sp))
1996-05-20 06:01:36 +04:00
search_msg(sp, S_WRAP);
1994-08-17 20:17:19 +04:00
#if defined(DEBUG) && 0
1996-05-20 06:01:36 +04:00
TRACE(sp, "B found: %qu to %qu\n",
match[0].rm_so, match[0].rm_eo);
1994-08-17 20:17:19 +04:00
#endif
1996-05-20 06:01:36 +04:00
/*
* We now have the first match on the line. Step through the
* line character by character until find the last acceptable
* match. This is painful, we need a better interface to regex
* to make this work.
*/
for (;;) {
last = match[0].rm_so++;
if (match[0].rm_so >= len)
break;
match[0].rm_eo = len;
eval = regexec(&sp->re_c, l, 1, match,
(match[0].rm_so == 0 ? 0 : REG_NOTBOL) |
REG_STARTEND);
if (eval == REG_NOMATCH)
break;
if (eval != 0) {
if (LF_ISSET(SEARCH_MSG))
re_error(sp, eval, &sp->re_c);
else
(void)sp->gp->scr_bell(sp);
goto err;
1994-08-17 20:17:19 +04:00
}
1996-05-20 06:01:36 +04:00
if (coff && match[0].rm_so >= coff)
break;
1994-08-17 20:17:19 +04:00
}
1996-05-20 06:01:36 +04:00
rm->lno = lno;
/* See comment in f_search(). */
if (!LF_ISSET(SEARCH_EOL) && last >= len)
rm->cno = len != 0 ? len - 1 : 0;
else
rm->cno = last;
1994-08-17 20:17:19 +04:00
rval = 0;
break;
}
1996-05-20 06:01:36 +04:00
err: if (LF_ISSET(SEARCH_MSG))
search_busy(sp, BUSY_OFF);
1994-08-17 20:17:19 +04:00
return (rval);
}
/*
1996-05-20 06:01:36 +04:00
* search_msg --
* Display one of the search messages.
1994-08-17 20:17:19 +04:00
*/
1996-05-20 06:01:36 +04:00
static void
search_msg(sp, msg)
1994-08-17 20:17:19 +04:00
SCR *sp;
1996-05-20 06:01:36 +04:00
smsg_t msg;
1994-08-17 20:17:19 +04:00
{
1996-05-20 06:01:36 +04:00
switch (msg) {
case S_EMPTY:
msgq(sp, M_ERR, "072|File empty; nothing to search");
break;
case S_EOF:
msgq(sp, M_ERR,
"073|Reached end-of-file without finding the pattern");
break;
case S_NOPREV:
msgq(sp, M_ERR, "074|No previous search pattern");
break;
case S_NOTFOUND:
msgq(sp, M_ERR, "075|Pattern not found");
break;
case S_SOF:
msgq(sp, M_ERR,
"076|Reached top-of-file without finding the pattern");
break;
case S_WRAP:
msgq(sp, M_ERR, "077|Search wrapped");
break;
default:
abort();
1994-08-17 20:17:19 +04:00
}
}
/*
1996-05-20 06:01:36 +04:00
* search_busy --
* Put up the busy searching message.
1994-08-17 20:17:19 +04:00
*
1996-05-20 06:01:36 +04:00
* PUBLIC: void search_busy __P((SCR *, busy_t));
1994-08-17 20:17:19 +04:00
*/
void
1996-05-20 06:01:36 +04:00
search_busy(sp, btype)
1994-08-17 20:17:19 +04:00
SCR *sp;
1996-05-20 06:01:36 +04:00
busy_t btype;
1994-08-17 20:17:19 +04:00
{
1996-05-20 06:01:36 +04:00
sp->gp->scr_busy(sp, "078|Searching...", btype);
1994-08-17 20:17:19 +04:00
}