/* UReplaceableCallbacks callback */
static void Xcopy(UReplaceable* rep, int32_t start, int32_t limit, int32_t dest) {
XReplaceable* x = (XReplaceable*)rep;
int32_t newLen = Xlength(rep) + limit - start;
UChar* newText = (UChar*) malloc(sizeof(UChar) * (newLen+1));
u_strncpy(newText, x->text, dest);
u_strncpy(newText + dest, x->text + start, limit - start);
u_strcpy(newText + dest + limit - start, x->text + dest);
free(x->text);
x->text = newText;
}
/* UReplaceableCallbacks callback */
static void Xreplace(UReplaceable* rep, int32_t start, int32_t limit,
const UChar* text, int32_t textLength) {
XReplaceable* x = (XReplaceable*)rep;
int32_t newLen = Xlength(rep) + limit - start + textLength;
UChar* newText = (UChar*) malloc(sizeof(UChar) * (newLen+1));
u_strncpy(newText, x->text, start);
u_strncpy(newText + start, text, textLength);
u_strcpy(newText + start + textLength, x->text + limit);
free(x->text);
x->text = newText;
}
/* Unget a char: handles case when we are already at the start of the buffer */
static void
ungetsc(int c)
{
struct sretrace_info *rp = retrace_info;
if (backslash_skip)
backslash_skip--;
/* Don't unget EOF... */
if (source->str == null && c == '\0')
return;
while (rp) {
if (Xlength(rp->xs, rp->xp))
rp->xp--;
rp = rp->next;
}
ungetsc_i(c);
}
char *
get_phys_path(const char *path)
{
XString xs;
char *xp;
Xinit(xs, xp, strlen(path) + 1, ATEMP);
xp = do_phys_path(&xs, xp, path);
if (!xp)
return ((char *) 0);
if (Xlength(xs, xp) == 0)
Xput(xs, xp, '/');
Xput(xs, xp, '\0');
return (Xclose(xs, xp));
}
int
c_read(char **wp)
{
int c = 0;
int expand = 1, history = 0;
int expanding;
int ecode = 0;
char *cp;
int fd = 0;
struct shf *shf;
int optc;
const char *emsg;
XString cs, xs;
struct tbl *vp;
char *xp = NULL;
while ((optc = ksh_getopt(wp, &builtin_opt, "prsu,")) != -1)
switch (optc) {
case 'p':
if ((fd = coproc_getfd(R_OK, &emsg)) < 0) {
bi_errorf("-p: %s", emsg);
return 1;
}
break;
case 'r':
expand = 0;
break;
case 's':
history = 1;
break;
case 'u':
if (!*(cp = builtin_opt.optarg))
fd = 0;
else if ((fd = check_fd(cp, R_OK, &emsg)) < 0) {
bi_errorf("-u: %s: %s", cp, emsg);
return 1;
}
break;
case '?':
return 1;
}
wp += builtin_opt.optind;
if (*wp == NULL)
*--wp = "REPLY";
/* Since we can't necessarily seek backwards on non-regular files,
* don't buffer them so we can't read too much.
*/
shf = shf_reopen(fd, SHF_RD | SHF_INTERRUPT | can_seek(fd), shl_spare);
if ((cp = strchr(*wp, '?')) != NULL) {
*cp = 0;
if (isatty(fd)) {
/* at&t ksh says it prints prompt on fd if it's open
* for writing and is a tty, but it doesn't do it
* (it also doesn't check the interactive flag,
* as is indicated in the Kornshell book).
*/
shellf("%s", cp+1);
}
}
/* If we are reading from the co-process for the first time,
* make sure the other side of the pipe is closed first. This allows
* the detection of eof.
*
* This is not compatible with at&t ksh... the fd is kept so another
* coproc can be started with same output, however, this means eof
* can't be detected... This is why it is closed here.
* If this call is removed, remove the eof check below, too.
* coproc_readw_close(fd);
*/
if (history)
Xinit(xs, xp, 128, ATEMP);
expanding = 0;
Xinit(cs, cp, 128, ATEMP);
for (; *wp != NULL; wp++) {
for (cp = Xstring(cs, cp); ; ) {
if (c == '\n' || c == EOF)
break;
while (1) {
c = shf_getc(shf);
if (c == '\0')
continue;
if (c == EOF && shf_error(shf) &&
shf_errno(shf) == EINTR) {
/* Was the offending signal one that
* would normally kill a process?
* If so, pretend the read was killed.
*/
ecode = fatal_trap_check();
/* non fatal (eg, CHLD), carry on */
if (!ecode) {
shf_clearerr(shf);
continue;
}
}
break;
}
if (history) {
Xcheck(xs, xp);
Xput(xs, xp, c);
}
Xcheck(cs, cp);
if (expanding) {
expanding = 0;
if (c == '\n') {
c = 0;
if (Flag(FTALKING_I) && isatty(fd)) {
/* set prompt in case this is
* called from .profile or $ENV
*/
set_prompt(PS2, NULL);
pprompt(prompt, 0);
}
} else if (c != EOF)
Xput(cs, cp, c);
continue;
}
if (expand && c == '\\') {
expanding = 1;
continue;
}
if (c == '\n' || c == EOF)
break;
if (ctype(c, C_IFS)) {
if (Xlength(cs, cp) == 0 && ctype(c, C_IFSWS))
continue;
if (wp[1])
break;
}
Xput(cs, cp, c);
}
/* strip trailing IFS white space from last variable */
if (!wp[1])
while (Xlength(cs, cp) && ctype(cp[-1], C_IFS) &&
ctype(cp[-1], C_IFSWS))
cp--;
Xput(cs, cp, '\0');
vp = global(*wp);
/* Must be done before setting export. */
if (vp->flag & RDONLY) {
shf_flush(shf);
bi_errorf("%s is read only", *wp);
return 1;
}
if (Flag(FEXPORT))
typeset(*wp, EXPORT, 0, 0, 0);
if (!setstr(vp, Xstring(cs, cp), KSH_RETURN_ERROR)) {
shf_flush(shf);
return 1;
}
}
shf_flush(shf);
if (history) {
Xput(xs, xp, '\0');
source->line++;
histsave(source->line, Xstring(xs, xp), 1);
Xfree(xs, xp);
}
/* if this is the co-process fd, close the file descriptor
* (can get eof if and only if all processes are have died, ie,
* coproc.njobs is 0 and the pipe is closed).
*/
if (c == EOF && !ecode)
coproc_read_close(fd);
return ecode ? ecode : c == EOF;
}
/*
* Makes a filename into result using the following algorithm.
* - make result NULL
* - if file starts with '/', append file to result & set cdpathp to NULL
* - if file starts with ./ or ../ append cwd and file to result
* and set cdpathp to NULL
* - if the first element of cdpathp doesnt start with a '/' xx or '.' xx
* then cwd is appended to result.
* - the first element of cdpathp is appended to result
* - file is appended to result
* - cdpathp is set to the start of the next element in cdpathp (or NULL
* if there are no more elements.
* The return value indicates whether a non-null element from cdpathp
* was appended to result.
*/
int
make_path(const char *cwd, const char *file,
char **cdpathp, /* & of : separated list */
XString *xsp,
int *phys_pathp)
{
int rval = 0;
int use_cdpath = 1;
char *plist;
int len;
int plen = 0;
char *xp = Xstring(*xsp, xp);
if (!file)
file = null;
if (file[0] == '/') {
*phys_pathp = 0;
use_cdpath = 0;
} else {
if (file[0] == '.') {
char c = file[1];
if (c == '.')
c = file[2];
if (c == '/' || c == '\0')
use_cdpath = 0;
}
plist = *cdpathp;
if (!plist)
use_cdpath = 0;
else if (use_cdpath) {
char *pend;
for (pend = plist; *pend && *pend != ':'; pend++)
;
plen = pend - plist;
*cdpathp = *pend ? ++pend : (char *) 0;
}
if ((use_cdpath == 0 || !plen || plist[0] != '/') &&
(cwd && *cwd)) {
len = strlen(cwd);
XcheckN(*xsp, xp, len);
memcpy(xp, cwd, len);
xp += len;
if (cwd[len - 1] != '/')
Xput(*xsp, xp, '/');
}
*phys_pathp = Xlength(*xsp, xp);
if (use_cdpath && plen) {
XcheckN(*xsp, xp, plen);
memcpy(xp, plist, plen);
xp += plen;
if (plist[plen - 1] != '/')
Xput(*xsp, xp, '/');
rval = 1;
}
}
len = strlen(file) + 1;
XcheckN(*xsp, xp, len);
memcpy(xp, file, len);
if (!use_cdpath)
*cdpathp = (char *) 0;
return (rval);
}
/**
* Makes a filename into result using the following algorithm.
* - make result NULL
* - if file starts with '/', append file to result & set cdpathp to NULL
* - if file starts with ./ or ../ append cwd and file to result
* and set cdpathp to NULL
* - if the first element of cdpathp doesnt start with a '/' xx or '.' xx
* then cwd is appended to result.
* - the first element of cdpathp is appended to result
* - file is appended to result
* - cdpathp is set to the start of the next element in cdpathp (or NULL
* if there are no more elements.
* The return value indicates whether a non-null element from cdpathp
* was appended to result.
*/
static int
make_path(const char *cwd, const char *file,
/* pointer to colon-separated list */
char **cdpathp,
XString *xsp,
int *phys_pathp)
{
int rval = 0;
bool use_cdpath = true;
char *plist;
size_t len, plen = 0;
char *xp = Xstring(*xsp, xp);
if (!file)
file = null;
if (mksh_abspath(file)) {
*phys_pathp = 0;
use_cdpath = false;
} else {
if (file[0] == '.') {
char c = file[1];
if (c == '.')
c = file[2];
if (c == '/' || c == '\0')
use_cdpath = false;
}
plist = *cdpathp;
if (!plist)
use_cdpath = false;
else if (use_cdpath) {
char *pend = plist;
while (*pend && *pend != MKSH_PATHSEPC)
++pend;
plen = pend - plist;
*cdpathp = *pend ? pend + 1 : NULL;
}
if ((!use_cdpath || !plen || !mksh_abspath(plist)) &&
(cwd && *cwd)) {
len = strlen(cwd);
XcheckN(*xsp, xp, len);
memcpy(xp, cwd, len);
xp += len;
if (cwd[len - 1] != '/')
Xput(*xsp, xp, '/');
}
*phys_pathp = Xlength(*xsp, xp);
if (use_cdpath && plen) {
XcheckN(*xsp, xp, plen);
memcpy(xp, plist, plen);
xp += plen;
if (plist[plen - 1] != '/')
Xput(*xsp, xp, '/');
rval = 1;
}
}
len = strlen(file) + 1;
XcheckN(*xsp, xp, len);
memcpy(xp, file, len);
if (!use_cdpath)
*cdpathp = NULL;
return (rval);
}
char *
do_realpath(const char *upath)
{
char *xp, *ip, *tp, *ipath, *ldest = NULL;
XString xs;
size_t pos, len;
int llen;
struct stat sb;
#ifdef MKSH__NO_PATH_MAX
size_t ldestlen = 0;
#define pathlen sb.st_size
#define pathcnd (ldestlen < (pathlen + 1))
#else
#define pathlen PATH_MAX
#define pathcnd (!ldest)
#endif
/* max. recursion depth */
int symlinks = 32;
if (mksh_abspath(upath)) {
/* upath is an absolute pathname */
strdupx(ipath, upath, ATEMP);
} else {
/* upath is a relative pathname, prepend cwd */
if ((tp = ksh_get_wd()) == NULL || !mksh_abspath(tp))
return (NULL);
ipath = shf_smprintf("%s%s%s", tp, "/", upath);
afree(tp, ATEMP);
}
/* ipath and upath are in memory at the same time -> unchecked */
Xinit(xs, xp, strlen(ip = ipath) + 1, ATEMP);
/* now jump into the deep of the loop */
goto beginning_of_a_pathname;
while (*ip) {
/* skip slashes in input */
while (*ip == '/')
++ip;
if (!*ip)
break;
/* get next pathname component from input */
tp = ip;
while (*ip && *ip != '/')
++ip;
len = ip - tp;
/* check input for "." and ".." */
if (tp[0] == '.') {
if (len == 1)
/* just continue with the next one */
continue;
else if (len == 2 && tp[1] == '.') {
/* strip off last pathname component */
while (xp > Xstring(xs, xp))
if (*--xp == '/')
break;
/* then continue with the next one */
continue;
}
}
/* store output position away, then append slash to output */
pos = Xsavepos(xs, xp);
/* 1 for the '/' and len + 1 for tp and the NUL from below */
XcheckN(xs, xp, 1 + len + 1);
Xput(xs, xp, '/');
/* append next pathname component to output */
memcpy(xp, tp, len);
xp += len;
*xp = '\0';
/* lstat the current output, see if it's a symlink */
if (mksh_lstat(Xstring(xs, xp), &sb)) {
/* lstat failed */
if (errno == ENOENT) {
/* because the pathname does not exist */
while (*ip == '/')
/* skip any trailing slashes */
++ip;
/* no more components left? */
if (!*ip)
/* we can still return successfully */
break;
/* more components left? fall through */
}
/* not ENOENT or not at the end of ipath */
goto notfound;
}
/* check if we encountered a symlink? */
if (S_ISLNK(sb.st_mode)) {
#ifndef MKSH__NO_SYMLINK
/* reached maximum recursion depth? */
if (!symlinks--) {
/* yep, prevent infinite loops */
errno = ELOOP;
goto notfound;
}
/* get symlink(7) target */
if (pathcnd) {
#ifdef MKSH__NO_PATH_MAX
if (notoktoadd(pathlen, 1)) {
errno = ENAMETOOLONG;
goto notfound;
}
#endif
ldest = aresize(ldest, pathlen + 1, ATEMP);
}
llen = readlink(Xstring(xs, xp), ldest, pathlen);
if (llen < 0)
/* oops... */
goto notfound;
ldest[llen] = '\0';
/*
* restart if symlink target is an absolute path,
* otherwise continue with currently resolved prefix
*/
/* append rest of current input path to link target */
tp = shf_smprintf("%s%s%s", ldest, *ip ? "/" : "", ip);
afree(ipath, ATEMP);
ip = ipath = tp;
if (!mksh_abspath(ldest)) {
/* symlink target is a relative path */
xp = Xrestpos(xs, xp, pos);
} else
#endif
{
/* symlink target is an absolute path */
xp = Xstring(xs, xp);
beginning_of_a_pathname:
/* assert: (ip == ipath)[0] == '/' */
/* assert: xp == xs.beg => start of path */
/* exactly two leading slashes? (SUSv4 3.266) */
if (ip[1] == '/' && ip[2] != '/') {
/* keep them, e.g. for UNC pathnames */
Xput(xs, xp, '/');
}
}
}
/* otherwise (no symlink) merely go on */
}
/*
* either found the target and successfully resolved it,
* or found its parent directory and may create it
*/
if (Xlength(xs, xp) == 0)
/*
* if the resolved pathname is "", make it "/",
* otherwise do not add a trailing slash
*/
Xput(xs, xp, '/');
Xput(xs, xp, '\0');
/*
* if source path had a trailing slash, check if target path
* is not a non-directory existing file
*/
if (ip > ipath && ip[-1] == '/') {
if (stat(Xstring(xs, xp), &sb)) {
if (errno != ENOENT)
goto notfound;
} else if (!S_ISDIR(sb.st_mode)) {
errno = ENOTDIR;
goto notfound;
}
/* target now either does not exist or is a directory */
}
/* return target path */
if (ldest != NULL)
afree(ldest, ATEMP);
afree(ipath, ATEMP);
return (Xclose(xs, xp));
notfound:
/* save; freeing memory might trash it */
llen = errno;
if (ldest != NULL)
afree(ldest, ATEMP);
afree(ipath, ATEMP);
Xfree(xs, xp);
errno = llen;
return (NULL);
#undef pathlen
#undef pathcnd
}
static void
getsc_line(Source *s)
{
char *xp = Xstring(s->xs, xp);
int interactive = Flag(FTALKING) && s->type == SSTDIN;
int have_tty = interactive && (s->flags & SF_TTY);
/* Done here to ensure nothing odd happens when a timeout occurs */
XcheckN(s->xs, xp, LINE);
*xp = '\0';
s->start = s->str = xp;
if (have_tty && ksh_tmout) {
ksh_tmout_state = TMOUT_READING;
alarm(ksh_tmout);
}
#ifdef EDIT
if (have_tty && (0
# ifdef VI
|| Flag(FVI)
# endif /* VI */
# ifdef EMACS
|| Flag(FEMACS) || Flag(FGMACS)
# endif /* EMACS */
))
{
int nread;
nread = x_read(xp, LINE);
if (nread < 0) /* read error */
nread = 0;
xp[nread] = '\0';
xp += nread;
}
else
#endif /* EDIT */
{
if (interactive) {
pprompt(prompt, 0);
} else
s->line++;
while (1) {
char *p = shf_getse(xp, Xnleft(s->xs, xp), s->u.shf);
if (!p && shf_error(s->u.shf)
&& shf_errno(s->u.shf) == EINTR)
{
shf_clearerr(s->u.shf);
if (trap)
runtraps(0);
continue;
}
if (!p || (xp = p, xp[-1] == '\n'))
break;
/* double buffer size */
xp++; /* move past null so doubling works... */
XcheckN(s->xs, xp, Xlength(s->xs, xp));
xp--; /* ...and move back again */
}
/* flush any unwanted input so other programs/builtins
* can read it. Not very optimal, but less error prone
* than flushing else where, dealing with redirections,
* etc..
* todo: reduce size of shf buffer (~128?) if SSTDIN
*/
if (s->type == SSTDIN)
shf_flush(s->u.shf);
}
/* XXX: temporary kludge to restore source after a
* trap may have been executed.
*/
source = s;
if (have_tty && ksh_tmout) {
ksh_tmout_state = TMOUT_EXECUTING;
alarm(0);
}
s->start = s->str = Xstring(s->xs, xp);
strip_nuls(Xstring(s->xs, xp), Xlength(s->xs, xp));
/* Note: if input is all nulls, this is not eof */
if (Xlength(s->xs, xp) == 0) { /* EOF */
if (s->type == SFILE)
shf_fdclose(s->u.shf);
s->str = NULL;
} else if (interactive) {
#ifdef HISTORY
char *p = Xstring(s->xs, xp);
if (cur_prompt == PS1)
while (*p && ctype(*p, C_IFS) && ctype(*p, C_IFSWS))
p++;
if (*p) {
s->line++;
histsave(s->line, s->str, 1);
}
#endif /* HISTORY */
}
if (interactive)
set_prompt(PS2, (Source *) 0);
}
int
yylex(int cf)
{
Lex_state states[STATE_BSIZE], *statep;
State_info state_info;
int c, state;
XString ws; /* expandable output word */
char *wp; /* output word pointer */
char *sp, *dp;
int c2;
Again:
states[0].ls_state = -1;
states[0].ls_info.base = (Lex_state *) 0;
statep = &states[1];
state_info.base = states;
state_info.end = &states[STATE_BSIZE];
Xinit(ws, wp, 64, ATEMP);
backslash_skip = 0;
ignore_backslash_newline = 0;
if (cf&ONEWORD)
state = SWORD;
else if (cf&LETEXPR) {
*wp++ = OQUOTE; /* enclose arguments in (double) quotes */
state = SLETPAREN;
statep->ls_sletparen.nparen = 0;
} else { /* normal lexing */
state = (cf & HEREDELIM) ? SHEREDELIM : SBASE;
while ((c = getsc()) == ' ' || c == '\t')
;
if (c == '#') {
ignore_backslash_newline++;
while ((c = getsc()) != '\0' && c != '\n')
;
ignore_backslash_newline--;
}
ungetsc(c);
}
if (source->flags & SF_ALIAS) { /* trailing ' ' in alias definition */
source->flags &= ~SF_ALIAS;
/* In POSIX mode, a trailing space only counts if we are
* parsing a simple command
*/
if (!Flag(FPOSIX) || (cf & CMDWORD))
cf |= ALIAS;
}
/* Initial state: one of SBASE SHEREDELIM SWORD SASPAREN */
statep->ls_state = state;
/* collect non-special or quoted characters to form word */
while (!((c = getsc()) == 0
|| ((state == SBASE || state == SHEREDELIM)
&& ctype(c, C_LEX1))))
{
Xcheck(ws, wp);
switch (state) {
case SBASE:
if (Flag(FCSHHISTORY) && (source->flags & SF_TTY) && c == '!') {
char **replace = NULL;
c2 = getsc();
if (c2 == '\0')
;
else if (c2 == ' ' || c2 == '\t')
ungetsc(c2);
else if (c2 == '!')
replace = hist_get_newest(0);
else if (isdigit(c2) || c2 == '-' ||
isalpha(c2)) {
int get = !isalpha(c2);
char match[200], *str = match;
*str++ = c2;
do {
if ((c2 = getsc()) == '\0')
break;
if (c2 == '\t' || c2 == ' ' ||
c2 == '\n') {
ungetsc(c2);
break;
}
*str++ = c2;
} while (str < &match[sizeof(match)-1]);
*str = '\0';
if (get) {
int h = findhistrel(match);
if (h >= 0)
replace = &history[h];
} else {
int h = findhist(-1, 0, match, true);
if (h >= 0)
replace = &history[h];
}
}
/*
* XXX ksh history buffer saves un-expanded
* commands. Until the history buffer code is
* changed to contain expanded commands, we
* ignore the bad commands (spinning sucks)
*/
if (replace && **replace == '!')
ungetsc(c2);
else if (replace) {
Source *s;
/* do not strdup replacement via alloc */
s = pushs(SREREAD,
source->areap);
s->start = s->str = *replace;
s->next = source;
source = s;
continue;
} else
ungetsc(c2);
}
if (c == '[' && (cf & (VARASN|ARRAYVAR))) {
*wp = EOS; /* temporary */
if (is_wdvarname(Xstring(ws, wp), false))
{
char *p, *tmp;
if (arraysub(&tmp)) {
*wp++ = CHAR;
*wp++ = c;
for (p = tmp; *p; ) {
Xcheck(ws, wp);
*wp++ = CHAR;
*wp++ = *p++;
}
afree(tmp, ATEMP);
break;
} else {
Source *s;
s = pushs(SREREAD,
source->areap);
s->start = s->str
= s->u.freeme = tmp;
s->next = source;
source = s;
}
}
*wp++ = CHAR;
*wp++ = c;
break;
}
/* fall through.. */
Sbase1: /* includes *(...|...) pattern (*+?@!) */
if (c == '*' || c == '@' || c == '+' || c == '?'
|| c == '!')
{
c2 = getsc();
if (c2 == '(' /*)*/ ) {
*wp++ = OPAT;
*wp++ = c;
PUSH_STATE(SPATTERN);
break;
}
ungetsc(c2);
}
/* fall through.. */
Sbase2: /* doesn't include *(...|...) pattern (*+?@!) */
switch (c) {
case '\\':
c = getsc();
if (c) /* trailing \ is lost */
*wp++ = QCHAR, *wp++ = c;
break;
case '\'':
*wp++ = OQUOTE;
ignore_backslash_newline++;
PUSH_STATE(SSQUOTE);
break;
case '"':
*wp++ = OQUOTE;
PUSH_STATE(SDQUOTE);
break;
default:
goto Subst;
}
break;
Subst:
switch (c) {
case '\\':
c = getsc();
switch (c) {
case '"': case '\\':
case '$': case '`':
*wp++ = QCHAR, *wp++ = c;
break;
default:
Xcheck(ws, wp);
if (c) { /* trailing \ is lost */
*wp++ = CHAR, *wp++ = '\\';
*wp++ = CHAR, *wp++ = c;
}
break;
}
break;
case '$':
c = getsc();
if (c == '(') /*)*/ {
c = getsc();
if (c == '(') /*)*/ {
PUSH_STATE(SASPAREN);
statep->ls_sasparen.nparen = 2;
statep->ls_sasparen.start =
Xsavepos(ws, wp);
*wp++ = EXPRSUB;
} else {
ungetsc(c);
PUSH_STATE(SCSPAREN);
statep->ls_scsparen.nparen = 1;
statep->ls_scsparen.csstate = 0;
*wp++ = COMSUB;
}
} else if (c == '{') /*}*/ {
*wp++ = OSUBST;
*wp++ = '{'; /*}*/
wp = get_brace_var(&ws, wp);
c = getsc();
/* allow :# and :% (ksh88 compat) */
if (c == ':') {
*wp++ = CHAR, *wp++ = c;
c = getsc();
}
/* If this is a trim operation,
* treat (,|,) specially in STBRACE.
*/
if (c == '#' || c == '%') {
ungetsc(c);
PUSH_STATE(STBRACE);
} else {
ungetsc(c);
PUSH_STATE(SBRACE);
}
} else if (ctype(c, C_ALPHA)) {
*wp++ = OSUBST;
*wp++ = 'X';
do {
Xcheck(ws, wp);
*wp++ = c;
c = getsc();
} while (ctype(c, C_ALPHA|C_DIGIT));
*wp++ = '\0';
*wp++ = CSUBST;
*wp++ = 'X';
ungetsc(c);
} else if (ctype(c, C_DIGIT|C_VAR1)) {
Xcheck(ws, wp);
*wp++ = OSUBST;
*wp++ = 'X';
*wp++ = c;
*wp++ = '\0';
*wp++ = CSUBST;
*wp++ = 'X';
} else {
*wp++ = CHAR, *wp++ = '$';
ungetsc(c);
}
break;
case '`':
PUSH_STATE(SBQUOTE);
*wp++ = COMSUB;
/* Need to know if we are inside double quotes
* since sh/at&t-ksh translate the \" to " in
* "`..\"..`".
* This is not done in posix mode (section
* 3.2.3, Double Quotes: "The backquote shall
* retain its special meaning introducing the
* other form of command substitution (see
* 3.6.3). The portion of the quoted string
* from the initial backquote and the
* characters up to the next backquote that
* is not preceded by a backslash (having
* escape characters removed) defines that
* command whose output replaces `...` when
* the word is expanded."
* Section 3.6.3, Command Substitution:
* "Within the backquoted style of command
* substitution, backslash shall retain its
* literal meaning, except when followed by
* $ ` \.").
*/
statep->ls_sbquote.indquotes = 0;
if (!Flag(FPOSIX)) {
Lex_state *s = statep;
Lex_state *base = state_info.base;
while (1) {
for (; s != base; s--) {
if (s->ls_state == SDQUOTE) {
statep->ls_sbquote.indquotes = 1;
break;
}
}
if (s != base)
break;
if (!(s = s->ls_info.base))
break;
base = s-- - STATE_BSIZE;
}
}
break;
default:
*wp++ = CHAR, *wp++ = c;
}
break;
case SSQUOTE:
if (c == '\'') {
POP_STATE();
*wp++ = CQUOTE;
ignore_backslash_newline--;
} else
*wp++ = QCHAR, *wp++ = c;
break;
case SDQUOTE:
if (c == '"') {
POP_STATE();
*wp++ = CQUOTE;
} else
goto Subst;
break;
case SCSPAREN: /* $( .. ) */
/* todo: deal with $(...) quoting properly
* kludge to partly fake quoting inside $(..): doesn't
* really work because nested $(..) or ${..} inside
* double quotes aren't dealt with.
*/
switch (statep->ls_scsparen.csstate) {
case 0: /* normal */
switch (c) {
case '(':
statep->ls_scsparen.nparen++;
break;
case ')':
statep->ls_scsparen.nparen--;
break;
case '\\':
statep->ls_scsparen.csstate = 1;
break;
case '"':
statep->ls_scsparen.csstate = 2;
break;
case '\'':
statep->ls_scsparen.csstate = 4;
ignore_backslash_newline++;
break;
}
break;
case 1: /* backslash in normal mode */
case 3: /* backslash in double quotes */
--statep->ls_scsparen.csstate;
break;
case 2: /* double quotes */
if (c == '"')
statep->ls_scsparen.csstate = 0;
else if (c == '\\')
statep->ls_scsparen.csstate = 3;
break;
case 4: /* single quotes */
if (c == '\'') {
statep->ls_scsparen.csstate = 0;
ignore_backslash_newline--;
}
break;
}
if (statep->ls_scsparen.nparen == 0) {
POP_STATE();
*wp++ = 0; /* end of COMSUB */
} else
*wp++ = c;
break;
case SASPAREN: /* $(( .. )) */
/* todo: deal with $((...); (...)) properly */
/* XXX should nest using existing state machine
* (embed "..", $(...), etc.) */
if (c == '(')
statep->ls_sasparen.nparen++;
else if (c == ')') {
statep->ls_sasparen.nparen--;
if (statep->ls_sasparen.nparen == 1) {
/*(*/
if ((c2 = getsc()) == ')') {
POP_STATE();
*wp++ = 0; /* end of EXPRSUB */
break;
} else {
char *s;
ungetsc(c2);
/* mismatched parenthesis -
* assume we were really
* parsing a $(..) expression
*/
s = Xrestpos(ws, wp,
statep->ls_sasparen.start);
memmove(s + 1, s, wp - s);
*s++ = COMSUB;
*s = '('; /*)*/
wp++;
statep->ls_scsparen.nparen = 1;
statep->ls_scsparen.csstate = 0;
state = statep->ls_state
= SCSPAREN;
}
}
}
*wp++ = c;
break;
case SBRACE:
/*{*/
if (c == '}') {
POP_STATE();
*wp++ = CSUBST;
*wp++ = /*{*/ '}';
} else
goto Sbase1;
break;
case STBRACE:
/* Same as SBRACE, except (,|,) treated specially */
/*{*/
if (c == '}') {
POP_STATE();
*wp++ = CSUBST;
*wp++ = /*{*/ '}';
} else if (c == '|') {
*wp++ = SPAT;
} else if (c == '(') {
*wp++ = OPAT;
*wp++ = ' '; /* simile for @ */
PUSH_STATE(SPATTERN);
} else
goto Sbase1;
break;
case SBQUOTE:
if (c == '`') {
*wp++ = 0;
POP_STATE();
} else if (c == '\\') {
switch (c = getsc()) {
case '\\':
case '$': case '`':
*wp++ = c;
break;
case '"':
if (statep->ls_sbquote.indquotes) {
*wp++ = c;
break;
}
/* fall through.. */
default:
if (c) { /* trailing \ is lost */
*wp++ = '\\';
*wp++ = c;
}
break;
}
} else
*wp++ = c;
break;
case SWORD: /* ONEWORD */
goto Subst;
case SLETPAREN: /* LETEXPR: (( ... )) */
/*(*/
if (c == ')') {
if (statep->ls_sletparen.nparen > 0)
--statep->ls_sletparen.nparen;
/*(*/
else if ((c2 = getsc()) == ')') {
c = 0;
*wp++ = CQUOTE;
goto Done;
} else
ungetsc(c2);
} else if (c == '(')
/* parenthesis inside quotes and backslashes
* are lost, but at&t ksh doesn't count them
* either
*/
++statep->ls_sletparen.nparen;
goto Sbase2;
case SHEREDELIM: /* <<,<<- delimiter */
/* XXX chuck this state (and the next) - use
* the existing states ($ and \`..` should be
* stripped of their specialness after the
* fact).
*/
/* here delimiters need a special case since
* $ and `..` are not to be treated specially
*/
if (c == '\\') {
c = getsc();
if (c) { /* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
} else if (c == '\'') {
PUSH_STATE(SSQUOTE);
*wp++ = OQUOTE;
ignore_backslash_newline++;
} else if (c == '"') {
state = statep->ls_state = SHEREDQUOTE;
*wp++ = OQUOTE;
} else {
*wp++ = CHAR;
*wp++ = c;
}
break;
case SHEREDQUOTE: /* " in <<,<<- delimiter */
if (c == '"') {
*wp++ = CQUOTE;
state = statep->ls_state = SHEREDELIM;
} else {
if (c == '\\') {
switch (c = getsc()) {
case '\\': case '"':
case '$': case '`':
break;
default:
if (c) { /* trailing \ lost */
*wp++ = CHAR;
*wp++ = '\\';
}
break;
}
}
*wp++ = CHAR;
*wp++ = c;
}
break;
case SPATTERN: /* in *(...|...) pattern (*+?@!) */
if ( /*(*/ c == ')') {
*wp++ = CPAT;
POP_STATE();
} else if (c == '|') {
*wp++ = SPAT;
} else if (c == '(') {
*wp++ = OPAT;
*wp++ = ' '; /* simile for @ */
PUSH_STATE(SPATTERN);
} else
goto Sbase1;
break;
}
}
Done:
Xcheck(ws, wp);
if (statep != &states[1])
/* XXX figure out what is missing */
yyerror("no closing quote\n");
/* This done to avoid tests for SHEREDELIM wherever SBASE tested */
if (state == SHEREDELIM)
state = SBASE;
dp = Xstring(ws, wp);
if ((c == '<' || c == '>') && state == SBASE
&& ((c2 = Xlength(ws, wp)) == 0
|| (c2 == 2 && dp[0] == CHAR && digit(dp[1]))))
{
struct ioword *iop =
(struct ioword *) alloc(sizeof(*iop), ATEMP);
if (c2 == 2)
iop->unit = dp[1] - '0';
else
iop->unit = c == '>'; /* 0 for <, 1 for > */
c2 = getsc();
/* <<, >>, <> are ok, >< is not */
if (c == c2 || (c == '<' && c2 == '>')) {
iop->flag = c == c2 ?
(c == '>' ? IOCAT : IOHERE) : IORDWR;
if (iop->flag == IOHERE) {
if ((c2 = getsc()) == '-')
iop->flag |= IOSKIP;
else
ungetsc(c2);
}
} else if (c2 == '&')
iop->flag = IODUP | (c == '<' ? IORDUP : 0);
else {
iop->flag = c == '>' ? IOWRITE : IOREAD;
if (c == '>' && c2 == '|')
iop->flag |= IOCLOB;
else
ungetsc(c2);
}
iop->name = (char *) 0;
iop->delim = (char *) 0;
iop->heredoc = (char *) 0;
Xfree(ws, wp); /* free word */
yylval.iop = iop;
return REDIR;
}
if (wp == dp && state == SBASE) {
Xfree(ws, wp); /* free word */
/* no word, process LEX1 character */
switch (c) {
default:
return c;
case '|':
case '&':
case ';':
if ((c2 = getsc()) == c)
c = (c == ';') ? BREAK :
(c == '|') ? LOGOR :
(c == '&') ? LOGAND :
YYERRCODE;
else if (c == '|' && c2 == '&')
c = COPROC;
else
ungetsc(c2);
return c;
case '\n':
gethere();
if (cf & CONTIN)
goto Again;
return c;
case '(': /*)*/
if (!Flag(FSH)) {
if ((c2 = getsc()) == '(') /*)*/
/* XXX need to handle ((...); (...)) */
c = MDPAREN;
else
ungetsc(c2);
}
return c;
/*(*/
case ')':
return c;
}
}
*wp++ = EOS; /* terminate word */
yylval.cp = Xclose(ws, wp);
if (state == SWORD || state == SLETPAREN) /* ONEWORD? */
return LWORD;
ungetsc(c); /* unget terminator */
/* copy word to unprefixed string ident */
for (sp = yylval.cp, dp = ident; dp < ident+IDENT && (c = *sp++) == CHAR; )
*dp++ = *sp++;
/* Make sure the ident array stays '\0' padded */
memset(dp, 0, (ident+IDENT) - dp + 1);
if (c != EOS)
*ident = '\0'; /* word is not unquoted */
if (*ident != '\0' && (cf&(KEYWORD|ALIAS))) {
struct tbl *p;
int h = hash(ident);
/* { */
if ((cf & KEYWORD) && (p = tsearch(&keywords, ident, h))
&& (!(cf & ESACONLY) || p->val.i == ESAC || p->val.i == '}'))
{
afree(yylval.cp, ATEMP);
return p->val.i;
}
if ((cf & ALIAS) && (p = tsearch(&aliases, ident, h))
&& (p->flag & ISSET))
{
Source *s;
for (s = source; s->type == SALIAS; s = s->next)
if (s->u.tblp == p)
return LWORD;
/* push alias expansion */
s = pushs(SALIAS, source->areap);
s->start = s->str = p->val.s;
s->u.tblp = p;
s->next = source;
source = s;
afree(yylval.cp, ATEMP);
goto Again;
}
}
return LWORD;
}
int
yylex(int cf)
{
Lex_state states[STATE_BSIZE], *statep, *s2, *base;
State_info state_info;
int c, c2, state;
size_t cz;
XString ws; /* expandable output word */
char *wp; /* output word pointer */
char *sp, *dp;
Again:
states[0].type = SINVALID;
states[0].ls_base = NULL;
statep = &states[1];
state_info.base = states;
state_info.end = &state_info.base[STATE_BSIZE];
Xinit(ws, wp, 64, ATEMP);
backslash_skip = 0;
ignore_backslash_newline = 0;
if (cf & ONEWORD)
state = SWORD;
else if (cf & LETEXPR) {
/* enclose arguments in (double) quotes */
*wp++ = OQUOTE;
state = SLETPAREN;
statep->nparen = 0;
} else {
/* normal lexing */
state = (cf & HEREDELIM) ? SHEREDELIM : SBASE;
while ((c = getsc()) == ' ' || c == '\t')
;
if (c == '#') {
ignore_backslash_newline++;
while ((c = getsc()) != '\0' && c != '\n')
;
ignore_backslash_newline--;
}
ungetsc(c);
}
if (source->flags & SF_ALIAS) {
/* trailing ' ' in alias definition */
source->flags &= ~SF_ALIAS;
/* POSIX: trailing space only counts if parsing simple cmd */
if (!Flag(FPOSIX) || (cf & CMDWORD))
cf |= ALIAS;
}
/* Initial state: one of SWORD SLETPAREN SHEREDELIM SBASE */
statep->type = state;
/* collect non-special or quoted characters to form word */
while (!((c = getsc()) == 0 ||
((state == SBASE || state == SHEREDELIM) && ctype(c, C_LEX1)))) {
if (state == SBASE &&
subshell_nesting_type == /*{*/ '}' &&
c == /*{*/ '}')
/* possibly end ${ :;} */
break;
Xcheck(ws, wp);
switch (state) {
case SADELIM:
if (c == '(')
statep->nparen++;
else if (c == ')')
statep->nparen--;
else if (statep->nparen == 0 && (c == /*{*/ '}' ||
c == (int)statep->ls_adelim.delimiter)) {
*wp++ = ADELIM;
*wp++ = c;
if (c == /*{*/ '}' || --statep->ls_adelim.num == 0)
POP_STATE();
if (c == /*{*/ '}')
POP_STATE();
break;
}
/* FALLTHROUGH */
case SBASE:
if (c == '[' && (cf & (VARASN|ARRAYVAR))) {
/* temporary */
*wp = EOS;
if (is_wdvarname(Xstring(ws, wp), false)) {
char *p, *tmp;
if (arraysub(&tmp)) {
*wp++ = CHAR;
*wp++ = c;
for (p = tmp; *p; ) {
Xcheck(ws, wp);
*wp++ = CHAR;
*wp++ = *p++;
}
afree(tmp, ATEMP);
break;
} else {
Source *s;
s = pushs(SREREAD,
source->areap);
s->start = s->str =
s->u.freeme = tmp;
s->next = source;
source = s;
}
}
*wp++ = CHAR;
*wp++ = c;
break;
}
/* FALLTHROUGH */
Sbase1: /* includes *(...|...) pattern (*+?@!) */
if (c == '*' || c == '@' || c == '+' || c == '?' ||
c == '!') {
c2 = getsc();
if (c2 == '(' /*)*/ ) {
*wp++ = OPAT;
*wp++ = c;
PUSH_STATE(SPATTERN);
break;
}
ungetsc(c2);
}
/* FALLTHROUGH */
Sbase2: /* doesn't include *(...|...) pattern (*+?@!) */
switch (c) {
case '\\':
getsc_qchar:
if ((c = getsc())) {
/* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
break;
case '\'':
open_ssquote_unless_heredoc:
if ((cf & HEREDOC))
goto store_char;
*wp++ = OQUOTE;
ignore_backslash_newline++;
PUSH_STATE(SSQUOTE);
break;
case '"':
open_sdquote:
*wp++ = OQUOTE;
PUSH_STATE(SDQUOTE);
break;
case '$':
/*
* processing of dollar sign belongs into
* Subst, except for those which can open
* a string: $'…' and $"…"
*/
subst_dollar_ex:
c = getsc();
switch (c) {
case '"':
goto open_sdquote;
case '\'':
goto open_sequote;
default:
goto SubstS;
}
default:
goto Subst;
}
break;
Subst:
switch (c) {
case '\\':
c = getsc();
switch (c) {
case '"':
if ((cf & HEREDOC))
goto heredocquote;
/* FALLTHROUGH */
case '\\':
case '$': case '`':
store_qchar:
*wp++ = QCHAR;
*wp++ = c;
break;
default:
heredocquote:
Xcheck(ws, wp);
if (c) {
/* trailing \ is lost */
*wp++ = CHAR;
*wp++ = '\\';
*wp++ = CHAR;
*wp++ = c;
}
break;
}
break;
case '$':
c = getsc();
SubstS:
if (c == '(') /*)*/ {
c = getsc();
if (c == '(') /*)*/ {
*wp++ = EXPRSUB;
PUSH_SRETRACE(SASPAREN);
statep->nparen = 2;
*retrace_info->xp++ = '(';
} else {
ungetsc(c);
subst_command:
c = COMSUB;
subst_command2:
sp = yyrecursive(c);
cz = strlen(sp) + 1;
XcheckN(ws, wp, cz);
*wp++ = c;
memcpy(wp, sp, cz);
wp += cz;
}
} else if (c == '{') /*}*/ {
if ((c = getsc()) == '|') {
/*
* non-subenvironment
* value substitution
*/
c = VALSUB;
goto subst_command2;
} else if (ctype(c, C_IFSWS)) {
/*
* non-subenvironment
* "command" substitution
*/
c = FUNSUB;
goto subst_command2;
}
ungetsc(c);
*wp++ = OSUBST;
*wp++ = '{'; /*}*/
wp = get_brace_var(&ws, wp);
c = getsc();
/* allow :# and :% (ksh88 compat) */
if (c == ':') {
*wp++ = CHAR;
*wp++ = c;
c = getsc();
if (c == ':') {
*wp++ = CHAR;
*wp++ = '0';
*wp++ = ADELIM;
*wp++ = ':';
PUSH_STATE(SBRACE);
PUSH_STATE(SADELIM);
statep->ls_adelim.delimiter = ':';
statep->ls_adelim.num = 1;
statep->nparen = 0;
break;
} else if (ksh_isdigit(c) ||
c == '('/*)*/ || c == ' ' ||
/*XXX what else? */
c == '$') {
/* substring subst. */
if (c != ' ') {
*wp++ = CHAR;
*wp++ = ' ';
}
ungetsc(c);
PUSH_STATE(SBRACE);
PUSH_STATE(SADELIM);
statep->ls_adelim.delimiter = ':';
statep->ls_adelim.num = 2;
statep->nparen = 0;
break;
}
} else if (c == '/') {
*wp++ = CHAR;
*wp++ = c;
if ((c = getsc()) == '/') {
*wp++ = ADELIM;
*wp++ = c;
} else
ungetsc(c);
PUSH_STATE(SBRACE);
PUSH_STATE(SADELIM);
statep->ls_adelim.delimiter = '/';
statep->ls_adelim.num = 1;
statep->nparen = 0;
break;
}
/*
* If this is a trim operation,
* treat (,|,) specially in STBRACE.
*/
if (ctype(c, C_SUBOP2)) {
ungetsc(c);
if (Flag(FSH))
PUSH_STATE(STBRACEBOURNE);
else
PUSH_STATE(STBRACEKORN);
} else {
ungetsc(c);
if (state == SDQUOTE ||
state == SQBRACE)
PUSH_STATE(SQBRACE);
else
PUSH_STATE(SBRACE);
}
} else if (ksh_isalphx(c)) {
*wp++ = OSUBST;
*wp++ = 'X';
do {
Xcheck(ws, wp);
*wp++ = c;
c = getsc();
} while (ksh_isalnux(c));
*wp++ = '\0';
*wp++ = CSUBST;
*wp++ = 'X';
ungetsc(c);
} else if (ctype(c, C_VAR1 | C_DIGIT)) {
Xcheck(ws, wp);
*wp++ = OSUBST;
*wp++ = 'X';
*wp++ = c;
*wp++ = '\0';
*wp++ = CSUBST;
*wp++ = 'X';
} else {
*wp++ = CHAR;
*wp++ = '$';
ungetsc(c);
}
break;
case '`':
subst_gravis:
PUSH_STATE(SBQUOTE);
*wp++ = COMSUB;
/*
* We need to know whether we are within double
* quotes, since most shells translate \" to "
* within "…`…\"…`…". This is not done in POSIX
* mode (§2.2.3 Double-Quotes: “The backquote
* shall retain its special meaning introducing
* the other form of command substitution (see
* Command Substitution). The portion of the
* quoted string from the initial backquote and
* the characters up to the next backquote that
* is not preceded by a <backslash>, having
* escape characters removed, defines that
* command whose output replaces "`...`" when
* the word is expanded.”; §2.6.3 Command
* Substitution: “Within the backquoted style
* of command substitution, <backslash> shall
* retain its literal meaning, except when
* followed by: '$', '`', or <backslash>. The
* search for the matching backquote shall be
* satisfied by the first unquoted non-escaped
* backquote; during this search, if a
* non-escaped backquote is encountered[…],
* undefined results occur.”).
*/
statep->ls_bool = false;
if (Flag(FPOSIX))
break;
s2 = statep;
base = state_info.base;
while (/* CONSTCOND */ 1) {
for (; s2 != base; s2--) {
if (s2->type == SDQUOTE) {
statep->ls_bool = true;
break;
}
}
if (s2 != base)
break;
if (!(s2 = s2->ls_base))
break;
base = s2-- - STATE_BSIZE;
}
break;
case QCHAR:
if (cf & LQCHAR) {
*wp++ = QCHAR;
*wp++ = getsc();
break;
}
/* FALLTHROUGH */
default:
store_char:
*wp++ = CHAR;
*wp++ = c;
}
break;
case SEQUOTE:
if (c == '\'') {
POP_STATE();
*wp++ = CQUOTE;
ignore_backslash_newline--;
} else if (c == '\\') {
if ((c2 = unbksl(true, s_get, s_put)) == -1)
c2 = s_get();
if (c2 == 0)
statep->ls_bool = true;
if (!statep->ls_bool) {
char ts[4];
if ((unsigned int)c2 < 0x100) {
*wp++ = QCHAR;
*wp++ = c2;
} else {
cz = utf_wctomb(ts, c2 - 0x100);
ts[cz] = 0;
cz = 0;
do {
*wp++ = QCHAR;
*wp++ = ts[cz];
} while (ts[++cz]);
}
}
} else if (!statep->ls_bool) {
*wp++ = QCHAR;
*wp++ = c;
}
break;
case SSQUOTE:
if (c == '\'') {
POP_STATE();
if ((cf & HEREDOC) || state == SQBRACE)
goto store_char;
*wp++ = CQUOTE;
ignore_backslash_newline--;
} else {
*wp++ = QCHAR;
*wp++ = c;
}
break;
case SDQUOTE:
if (c == '"') {
POP_STATE();
*wp++ = CQUOTE;
} else
goto Subst;
break;
/* $(( ... )) */
case SASPAREN:
if (c == '(')
statep->nparen++;
else if (c == ')') {
statep->nparen--;
if (statep->nparen == 1) {
/* end of EXPRSUB */
POP_SRETRACE();
if ((c2 = getsc()) == /*(*/ ')') {
cz = strlen(sp) - 2;
XcheckN(ws, wp, cz);
memcpy(wp, sp + 1, cz);
wp += cz;
afree(sp, ATEMP);
*wp++ = '\0';
break;
} else {
Source *s;
ungetsc(c2);
/*
* mismatched parenthesis -
* assume we were really
* parsing a $(...) expression
*/
--wp;
s = pushs(SREREAD,
source->areap);
s->start = s->str =
s->u.freeme = sp;
s->next = source;
source = s;
goto subst_command;
}
}
}
/* reuse existing state machine */
goto Sbase2;
case SQBRACE:
if (c == '\\') {
/*
* perform POSIX "quote removal" if the back-
* slash is "special", i.e. same cases as the
* {case '\\':} in Subst: plus closing brace;
* in mksh code "quote removal" on '\c' means
* write QCHAR+c, otherwise CHAR+\+CHAR+c are
* emitted (in heredocquote:)
*/
if ((c = getsc()) == '"' || c == '\\' ||
c == '$' || c == '`' || c == /*{*/'}')
goto store_qchar;
goto heredocquote;
}
goto common_SQBRACE;
case SBRACE:
if (c == '\'')
goto open_ssquote_unless_heredoc;
else if (c == '\\')
goto getsc_qchar;
common_SQBRACE:
if (c == '"')
goto open_sdquote;
else if (c == '$')
goto subst_dollar_ex;
else if (c == '`')
goto subst_gravis;
else if (c != /*{*/ '}')
goto store_char;
POP_STATE();
*wp++ = CSUBST;
*wp++ = /*{*/ '}';
break;
/* Same as SBASE, except (,|,) treated specially */
case STBRACEKORN:
if (c == '|')
*wp++ = SPAT;
else if (c == '(') {
*wp++ = OPAT;
/* simile for @ */
*wp++ = ' ';
PUSH_STATE(SPATTERN);
} else /* FALLTHROUGH */
case STBRACEBOURNE:
if (c == /*{*/ '}') {
POP_STATE();
*wp++ = CSUBST;
*wp++ = /*{*/ '}';
} else
goto Sbase1;
break;
case SBQUOTE:
if (c == '`') {
*wp++ = 0;
POP_STATE();
} else if (c == '\\') {
switch (c = getsc()) {
case 0:
/* trailing \ is lost */
break;
case '$':
case '`':
case '\\':
*wp++ = c;
break;
case '"':
if (statep->ls_bool) {
*wp++ = c;
break;
}
/* FALLTHROUGH */
default:
*wp++ = '\\';
*wp++ = c;
break;
}
} else
*wp++ = c;
break;
/* ONEWORD */
case SWORD:
goto Subst;
/* LETEXPR: (( ... )) */
case SLETPAREN:
if (c == /*(*/ ')') {
if (statep->nparen > 0)
--statep->nparen;
else if ((c2 = getsc()) == /*(*/ ')') {
c = 0;
*wp++ = CQUOTE;
goto Done;
} else {
Source *s;
ungetsc(c2);
/*
* mismatched parenthesis -
* assume we were really
* parsing a (...) expression
*/
*wp = EOS;
sp = Xstring(ws, wp);
dp = wdstrip(sp + 1, WDS_TPUTS);
s = pushs(SREREAD, source->areap);
s->start = s->str = s->u.freeme = dp;
s->next = source;
source = s;
return ('('/*)*/);
}
} else if (c == '(')
/*
* parentheses inside quotes and
* backslashes are lost, but AT&T ksh
* doesn't count them either
*/
++statep->nparen;
goto Sbase2;
/* << or <<- delimiter */
case SHEREDELIM:
/*
* here delimiters need a special case since
* $ and `...` are not to be treated specially
*/
switch (c) {
case '\\':
if ((c = getsc())) {
/* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
break;
case '\'':
goto open_ssquote_unless_heredoc;
case '$':
if ((c2 = getsc()) == '\'') {
open_sequote:
*wp++ = OQUOTE;
ignore_backslash_newline++;
PUSH_STATE(SEQUOTE);
statep->ls_bool = false;
break;
} else if (c2 == '"') {
/* FALLTHROUGH */
case '"':
PUSH_SRETRACE(SHEREDQUOTE);
break;
}
ungetsc(c2);
/* FALLTHROUGH */
default:
*wp++ = CHAR;
*wp++ = c;
}
break;
/* " in << or <<- delimiter */
case SHEREDQUOTE:
if (c != '"')
goto Subst;
POP_SRETRACE();
dp = strnul(sp) - 1;
/* remove the trailing double quote */
*dp = '\0';
/* store the quoted string */
*wp++ = OQUOTE;
XcheckN(ws, wp, (dp - sp) * 2);
dp = sp;
while ((c = *dp++)) {
if (c == '\\') {
switch ((c = *dp++)) {
case '\\':
case '"':
case '$':
case '`':
break;
default:
*wp++ = CHAR;
*wp++ = '\\';
break;
}
}
*wp++ = CHAR;
*wp++ = c;
}
afree(sp, ATEMP);
*wp++ = CQUOTE;
state = statep->type = SHEREDELIM;
break;
/* in *(...|...) pattern (*+?@!) */
case SPATTERN:
if (c == /*(*/ ')') {
*wp++ = CPAT;
POP_STATE();
} else if (c == '|') {
*wp++ = SPAT;
} else if (c == '(') {
*wp++ = OPAT;
/* simile for @ */
*wp++ = ' ';
PUSH_STATE(SPATTERN);
} else
goto Sbase1;
break;
}
}
Done:
Xcheck(ws, wp);
if (statep != &states[1])
/* XXX figure out what is missing */
yyerror("no closing quote\n");
/* This done to avoid tests for SHEREDELIM wherever SBASE tested */
if (state == SHEREDELIM)
state = SBASE;
dp = Xstring(ws, wp);
if (state == SBASE && (
#ifndef MKSH_LEGACY_MODE
(c == '&' && !Flag(FSH) && !Flag(FPOSIX)) ||
#endif
c == '<' || c == '>')) {
struct ioword *iop = alloc(sizeof(struct ioword), ATEMP);
if (Xlength(ws, wp) == 0)
iop->unit = c == '<' ? 0 : 1;
else for (iop->unit = 0, c2 = 0; c2 < Xlength(ws, wp); c2 += 2) {
if (dp[c2] != CHAR)
goto no_iop;
if (!ksh_isdigit(dp[c2 + 1]))
goto no_iop;
iop->unit = iop->unit * 10 + ksh_numdig(dp[c2 + 1]);
if (iop->unit >= FDBASE)
goto no_iop;
}
if (c == '&') {
if ((c2 = getsc()) != '>') {
ungetsc(c2);
goto no_iop;
}
c = c2;
iop->ioflag = IOBASH;
} else
iop->ioflag = 0;
c2 = getsc();
/* <<, >>, <> are ok, >< is not */
if (c == c2 || (c == '<' && c2 == '>')) {
iop->ioflag |= c == c2 ?
(c == '>' ? IOCAT : IOHERE) : IORDWR;
if (iop->ioflag == IOHERE) {
if ((c2 = getsc()) == '-')
iop->ioflag |= IOSKIP;
else if (c2 == '<')
iop->ioflag |= IOHERESTR;
else
ungetsc(c2);
}
} else if (c2 == '&')
iop->ioflag |= IODUP | (c == '<' ? IORDUP : 0);
else {
iop->ioflag |= c == '>' ? IOWRITE : IOREAD;
if (c == '>' && c2 == '|')
iop->ioflag |= IOCLOB;
else
ungetsc(c2);
}
iop->ioname = NULL;
iop->delim = NULL;
iop->heredoc = NULL;
/* free word */
Xfree(ws, wp);
yylval.iop = iop;
return (REDIR);
no_iop:
afree(iop, ATEMP);
}
if (wp == dp && state == SBASE) {
/* free word */
Xfree(ws, wp);
/* no word, process LEX1 character */
if ((c == '|') || (c == '&') || (c == ';') || (c == '('/*)*/)) {
if ((c2 = getsc()) == c)
c = (c == ';') ? BREAK :
(c == '|') ? LOGOR :
(c == '&') ? LOGAND :
/* c == '(' ) */ MDPAREN;
else if (c == '|' && c2 == '&')
c = COPROC;
else if (c == ';' && c2 == '|')
c = BRKEV;
else if (c == ';' && c2 == '&')
c = BRKFT;
else
ungetsc(c2);
#ifndef MKSH_SMALL
if (c == BREAK) {
if ((c2 = getsc()) == '&')
c = BRKEV;
else
ungetsc(c2);
}
#endif
} else if (c == '\n') {
if (cf & HEREDELIM)
ungetsc(c);
else {
gethere();
if (cf & CONTIN)
goto Again;
}
}
return (c);
}
/* terminate word */
*wp++ = EOS;
yylval.cp = Xclose(ws, wp);
if (state == SWORD || state == SLETPAREN
/* XXX ONEWORD? */)
return (LWORD);
/* unget terminator */
ungetsc(c);
/*
* note: the alias-vs-function code below depends on several
* interna: starting from here, source->str is not modified;
* the way getsc() and ungetsc() operate; etc.
*/
/* copy word to unprefixed string ident */
sp = yylval.cp;
dp = ident;
while ((dp - ident) < IDENT && (c = *sp++) == CHAR)
*dp++ = *sp++;
if (c != EOS)
/* word is not unquoted */
dp = ident;
/* make sure the ident array stays NUL padded */
memset(dp, 0, (ident + IDENT) - dp + 1);
if (!(cf & (KEYWORD | ALIAS)))
return (LWORD);
if (*ident != '\0') {
struct tbl *p;
uint32_t h = hash(ident);
if ((cf & KEYWORD) && (p = ktsearch(&keywords, ident, h)) &&
(!(cf & ESACONLY) || p->val.i == ESAC ||
p->val.i == /*{*/ '}')) {
afree(yylval.cp, ATEMP);
return (p->val.i);
}
if ((cf & ALIAS) && (p = ktsearch(&aliases, ident, h)) &&
(p->flag & ISSET)) {
/*
* this still points to the same character as the
* ungetsc'd terminator from above
*/
const char *cp = source->str;
/* prefer POSIX but not Korn functions over aliases */
while (*cp == ' ' || *cp == '\t')
/*
* this is like getsc() without skipping
* over Source boundaries (including not
* parsing ungetsc'd characters that got
* pushed into an SREREAD) which is what
* we want here anyway: find out whether
* the alias name is followed by a POSIX
* function definition
*/
++cp;
/* prefer functions over aliases */
if (cp[0] != '(' || cp[1] != ')') {
Source *s = source;
while (s && (s->flags & SF_HASALIAS))
if (s->u.tblp == p)
return (LWORD);
else
s = s->next;
/* push alias expansion */
s = pushs(SALIAS, source->areap);
s->start = s->str = p->val.s;
s->u.tblp = p;
s->flags |= SF_HASALIAS;
s->next = source;
if (source->type == SEOF) {
/* prevent infinite recursion at EOS */
source->u.tblp = p;
source->flags |= SF_HASALIAS;
}
source = s;
afree(yylval.cp, ATEMP);
goto Again;
}
}
} else if (cf & ALIAS) {
/* retain typeset et al. even when quoted */
if (assign_command((dp = wdstrip(yylval.cp, 0))))
strlcpy(ident, dp, sizeof(ident));
afree(dp, ATEMP);
}
return (LWORD);
}
static void
getsc_line(Source *s)
{
char *xp = Xstring(s->xs, xp), *cp;
bool interactive = Flag(FTALKING) && s->type == SSTDIN;
bool have_tty = tobool(interactive && (s->flags & SF_TTY));
/* Done here to ensure nothing odd happens when a timeout occurs */
XcheckN(s->xs, xp, LINE);
*xp = '\0';
s->start = s->str = xp;
if (have_tty && ksh_tmout) {
ksh_tmout_state = TMOUT_READING;
alarm(ksh_tmout);
}
if (interactive)
change_winsz();
#ifndef MKSH_NO_CMDLINE_EDITING
if (have_tty && (
#if !MKSH_S_NOVI
Flag(FVI) ||
#endif
Flag(FEMACS) || Flag(FGMACS))) {
int nread;
nread = x_read(xp);
if (nread < 0)
/* read error */
nread = 0;
xp[nread] = '\0';
xp += nread;
} else
#endif
{
if (interactive)
pprompt(prompt, 0);
else
s->line++;
while (/* CONSTCOND */ 1) {
char *p = shf_getse(xp, Xnleft(s->xs, xp), s->u.shf);
if (!p && shf_error(s->u.shf) &&
shf_errno(s->u.shf) == EINTR) {
shf_clearerr(s->u.shf);
if (trap)
runtraps(0);
continue;
}
if (!p || (xp = p, xp[-1] == '\n'))
break;
/* double buffer size */
/* move past NUL so doubling works... */
xp++;
XcheckN(s->xs, xp, Xlength(s->xs, xp));
/* ...and move back again */
xp--;
}
/*
* flush any unwanted input so other programs/builtins
* can read it. Not very optimal, but less error prone
* than flushing else where, dealing with redirections,
* etc.
* TODO: reduce size of shf buffer (~128?) if SSTDIN
*/
if (s->type == SSTDIN)
shf_flush(s->u.shf);
}
/*
* XXX: temporary kludge to restore source after a
* trap may have been executed.
*/
source = s;
if (have_tty && ksh_tmout) {
ksh_tmout_state = TMOUT_EXECUTING;
alarm(0);
}
cp = Xstring(s->xs, xp);
rndpush(cp);
s->start = s->str = cp;
strip_nuls(Xstring(s->xs, xp), Xlength(s->xs, xp));
/* Note: if input is all nulls, this is not eof */
if (Xlength(s->xs, xp) == 0) {
/* EOF */
if (s->type == SFILE)
shf_fdclose(s->u.shf);
s->str = NULL;
} else if (interactive && *s->str) {
if (cur_prompt != PS1)
histsave(&s->line, s->str, HIST_APPEND, true);
else if (!ctype(*s->str, C_IFS | C_IFSWS))
histsave(&s->line, s->str, HIST_QUEUE, true);
#if !defined(MKSH_SMALL) && HAVE_PERSISTENT_HISTORY
else
goto check_for_sole_return;
} else if (interactive && cur_prompt == PS1) {
check_for_sole_return:
cp = Xstring(s->xs, xp);
while (*cp && ctype(*cp, C_IFSWS))
++cp;
if (!*cp) {
histsave(&s->line, NULL, HIST_FLUSH, true);
histsync();
}
#endif
}
if (interactive)
set_prompt(PS2, NULL);
}