/*
* Called by iosetup() (deals with 2>&4, etc.), c_read, c_print to turn
* a string (the X in 2>&X, read -uX, print -uX) into a file descriptor.
*/
int
check_fd(const char *name, int mode, const char **emsgp)
{
int fd = 0, fl;
if (name[0] == 'p' && !name[1])
return (coproc_getfd(mode, emsgp));
while (ksh_isdigit(*name)) {
fd = fd * 10 + ksh_numdig(*name);
if (fd >= FDBASE) {
if (emsgp)
*emsgp = "file descriptor too large";
return (-1);
}
++name;
}
if (*name) {
if (emsgp)
*emsgp = "illegal file descriptor name";
return (-1);
}
if ((fl = fcntl(fd, F_GETFL, 0)) < 0) {
if (emsgp)
*emsgp = "bad file descriptor";
return (-1);
}
fl &= O_ACCMODE;
/*
* X_OK is a kludge to disable this check for dups (x<&1):
* historical shells never did this check (XXX don't know what
* POSIX has to say).
*/
if (!(mode & X_OK) && fl != O_RDWR && (
((mode & R_OK) && fl != O_RDONLY) ||
((mode & W_OK) && fl != O_WRONLY))) {
if (emsgp)
*emsgp = (fl == O_WRONLY) ?
"fd not open for reading" :
"fd not open for writing";
return (-1);
}
return (fd);
}
/* parse a decimal number: returns 0 if string isn't a number, 1 otherwise */
int
getn(const char *s, int *ai)
{
char c;
mksh_ari_u num;
bool neg = false;
num.u = 0;
do {
c = *s++;
} while (ksh_isspace(c));
switch (c) {
case '-':
neg = true;
/* FALLTHROUGH */
case '+':
c = *s++;
break;
}
do {
if (!ksh_isdigit(c))
/* not numeric */
return (0);
if (num.u > 214748364U)
/* overflow on multiplication */
return (0);
num.u = num.u * 10U + (unsigned int)ksh_numdig(c);
/* now: num.u <= 2147483649U */
} while ((c = *s++));
if (num.u > (neg ? 2147483648U : 2147483647U))
/* overflow for signed 32-bit int */
return (0);
if (neg)
num.u = -num.u;
*ai = num.i;
return (1);
}
/*
* Called by iosetup() (deals with 2>&4, etc.), c_read, c_print to turn
* a string (the X in 2>&X, read -uX, print -uX) into a file descriptor.
*/
int
check_fd(const char *name, int mode, const char **emsgp)
{
int fd, fl;
if (!name[0] || name[1])
goto illegal_fd_name;
if (name[0] == 'p')
return (coproc_getfd(mode, emsgp));
if (!ctype(name[0], C_DIGIT)) {
illegal_fd_name:
if (emsgp)
*emsgp = "illegal file descriptor name";
return (-1);
}
if ((fl = fcntl((fd = ksh_numdig(name[0])), F_GETFL, 0)) < 0) {
if (emsgp)
*emsgp = "bad file descriptor";
return (-1);
}
fl &= O_ACCMODE;
/*
* X_OK is a kludge to disable this check for dups (x<&1):
* historical shells never did this check (XXX don't know what
* POSIX has to say).
*/
if (!(mode & X_OK) && fl != O_RDWR && (
((mode & R_OK) && fl != O_RDONLY) ||
((mode & W_OK) && fl != O_WRONLY))) {
if (emsgp)
*emsgp = (fl == O_WRONLY) ?
"fd not open for reading" :
"fd not open for writing";
return (-1);
}
return (fd);
}
/*
* process the string available via fg (get a char)
* and fp (put back a char) for backslash escapes,
* assuming the first call to *fg gets the char di-
* rectly after the backslash; return the character
* (0..0xFF), Unicode (wc + 0x100), or -1 if no known
* escape sequence was found
*/
int
unbksl(bool cstyle, int (*fg)(void), void (*fp)(int))
{
int wc, i, c, fc;
fc = (*fg)();
switch (fc) {
case 'a':
/*
* according to the comments in pdksh, \007 seems
* to be more portable than \a (due to HP-UX cc,
* Ultrix cc, old pcc, etc.) so we avoid the escape
* sequence altogether in mksh and assume ASCII
*/
wc = 7;
break;
case 'b':
wc = '\b';
break;
case 'c':
if (!cstyle)
goto unknown_escape;
c = (*fg)();
wc = CTRL(c);
break;
case 'E':
case 'e':
wc = 033;
break;
case 'f':
wc = '\f';
break;
case 'n':
wc = '\n';
break;
case 'r':
wc = '\r';
break;
case 't':
wc = '\t';
break;
case 'v':
/* assume ASCII here as well */
wc = 11;
break;
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
if (!cstyle)
goto unknown_escape;
/* FALLTHROUGH */
case '0':
if (cstyle)
(*fp)(fc);
/*
* look for an octal number with up to three
* digits, not counting the leading zero;
* convert it to a raw octet
*/
wc = 0;
i = 3;
while (i--)
if ((c = (*fg)()) >= ord('0') && c <= ord('7'))
wc = (wc << 3) + ksh_numdig(c);
else {
(*fp)(c);
break;
}
break;
case 'U':
i = 8;
if (/* CONSTCOND */ 0)
/* FALLTHROUGH */
case 'u':
i = 4;
if (/* CONSTCOND */ 0)
/* FALLTHROUGH */
case 'x':
i = cstyle ? -1 : 2;
/**
* x: look for a hexadecimal number with up to
* two (C style: arbitrary) digits; convert
* to raw octet (C style: Unicode if >0xFF)
* u/U: look for a hexadecimal number with up to
* four (U: eight) digits; convert to Unicode
*/
wc = 0;
while (i--) {
wc <<= 4;
if ((c = (*fg)()) >= ord('0') && c <= ord('9'))
wc += ksh_numdig(c);
else if (c >= ord('A') && c <= ord('F'))
wc += ksh_numuc(c) + 10;
else if (c >= ord('a') && c <= ord('f'))
wc += ksh_numlc(c) + 10;
else {
wc >>= 4;
(*fp)(c);
break;
}
}
if ((cstyle && wc > 0xFF) || fc != 'x')
/* Unicode marker */
wc += 0x100;
break;
case '\'':
if (!cstyle)
goto unknown_escape;
wc = '\'';
break;
case '\\':
wc = '\\';
break;
default:
unknown_escape:
(*fp)(fc);
return (-1);
}
return (wc);
}
static int
getint(struct tbl *vp, mksh_ari_u *nump, bool arith)
{
mksh_uari_t c, num = 0, base = 10;
const char *s;
bool have_base = false, neg = false;
if (vp->flag & SPECIAL)
getspec(vp);
/* XXX is it possible for ISSET to be set and val.s to be NULL? */
if (!(vp->flag & ISSET) || (!(vp->flag & INTEGER) && vp->val.s == NULL))
return (-1);
if (vp->flag & INTEGER) {
nump->i = vp->val.i;
return (vp->type);
}
s = vp->val.s + vp->type;
do {
c = (unsigned char)*s++;
} while (ksh_isspace(c));
switch (c) {
case '-':
neg = true;
/* FALLTHROUGH */
case '+':
c = (unsigned char)*s++;
break;
}
if (c == '0' && arith) {
if (ksh_eq(s[0], 'X', 'x')) {
/* interpret as hexadecimal */
base = 16;
++s;
goto getint_c_style_base;
} else if (Flag(FPOSIX) && ksh_isdigit(s[0]) &&
!(vp->flag & ZEROFIL)) {
/* interpret as octal (deprecated) */
base = 8;
getint_c_style_base:
have_base = true;
c = (unsigned char)*s++;
}
}
do {
if (c == '#') {
/* ksh-style base determination */
if (have_base || num < 1)
return (-1);
if ((base = num) == 1) {
/* mksh-specific extension */
unsigned int wc;
if (!UTFMODE)
wc = *(const unsigned char *)s;
else if (utf_mbtowc(&wc, s) == (size_t)-1)
/* OPTU-8 -> OPTU-16 */
/*
* (with a twist: 1#\uEF80 converts
* the same as 1#\x80 does, thus is
* not round-tripping correctly XXX)
*/
wc = 0xEF00 + *(const unsigned char *)s;
nump->u = (mksh_uari_t)wc;
return (1);
} else if (base > 36)
base = 10;
num = 0;
have_base = true;
continue;
}
if (ksh_isdigit(c))
c = ksh_numdig(c);
else if (ksh_isupper(c))
c = ksh_numuc(c) + 10;
else if (ksh_islower(c))
c = ksh_numlc(c) + 10;
else
return (-1);
if (c >= base)
return (-1);
/* handle overflow as truncation */
num = num * base + c;
} while ((c = (unsigned char)*s++));
if (neg)
num = -num;
nump->u = num;
return (base);
}
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);
}
