lexeme *st_in_quotes(lexer_info *linfo, buffer *buf)
{
#ifdef LEXER_DEBUG
print_state("ST_IN_QUOTES", linfo->c);
#endif
switch (linfo->c) {
case EOF:
linfo->state = ST_ERROR;
return NULL;
case '\\':
deferred_get_char(linfo);
linfo->state = ST_BACKSLASH_IN_QUOTES;
break;
case '\"':
deferred_get_char(linfo);
linfo->state = ST_WORD;
break;
case '\n':
print_prompt2();
/* fallthrough */
default:
add_to_buffer(buf, linfo->c);
deferred_get_char(linfo);
break;
}
return NULL;
}
lexeme *st_backslash(lexer_info *linfo, buffer *buf)
{
#ifdef LEXER_DEBUG
print_state("ST_BACKSLASH", linfo->c);
#endif
switch (linfo->c) {
case EOF:
linfo->state = ST_ERROR;
return NULL;
case '\n':
/* Ignore newline symbol */
print_prompt2();
break;
/* Non bash-like behaviour. In bash substitution
* makes in $'string' costruction. */
case 'a':
add_to_buffer(buf, '\a');
break;
case 'b':
add_to_buffer(buf, '\b');
break;
case 'f':
add_to_buffer(buf, '\f');
break;
case 'n':
add_to_buffer(buf, '\n');
break;
case 'r':
add_to_buffer(buf, '\r');
break;
case 't':
add_to_buffer(buf, '\t');
break;
case 'v':
add_to_buffer(buf, '\v');
break;
default:
add_to_buffer(buf, linfo->c);
break;
}
deferred_get_char(linfo);
linfo->state = ST_WORD;
return NULL;
}
/* snarfheredoc -- read a heredoc until the eof marker */
extern Tree *snarfheredoc(const char *eof, Boolean quoted) {
Tree *tree, **tailp;
Buffer *buf;
unsigned char *s;
assert(quoted || strchr(eof, '$') == NULL); /* can never be typed (whew!) */
if (strchr(eof, '\n') != NULL) {
yyerror("here document eof-marker contains a newline");
return NULL;
}
disablehistory = TRUE;
for (tree = NULL, tailp = &tree, buf = openbuffer(0);;) {
int c;
print_prompt2();
for (s = (unsigned char *) eof; (c = GETC()) == *s; s++)
;
if (*s == '\0' && (c == '\n' || c == EOF)) {
if (buf->current == 0 && tree != NULL)
freebuffer(buf);
else
*tailp = treecons(mk(nQword, sealcountedbuffer(buf)), NULL);
break;
}
if (s != (unsigned char *) eof)
buf = bufncat(buf, eof, s - (unsigned char *) eof);
for (;; c = GETC()) {
if (c == EOF) {
yyerror("incomplete here document");
freebuffer(buf);
disablehistory = FALSE;
return NULL;
}
if (c == '$' && !quoted && (c = GETC()) != '$') {
Tree *var;
UNGETC(c);
if (buf->current == 0)
freebuffer(buf);
else {
*tailp = treecons(mk(nQword, sealcountedbuffer(buf)), NULL);
tailp = &(*tailp)->CDR;
}
var = getherevar();
if (var == NULL) {
freebuffer(buf);
disablehistory = FALSE;
return NULL;
}
*tailp = treecons(var, NULL);
tailp = &(*tailp)->CDR;
buf = openbuffer(0);
continue;
}
buf = bufputc(buf, c);
if (c == '\n')
break;
}
}
disablehistory = FALSE;
return tree->CDR == NULL ? tree->CAR : tree;
}
extern int yylex(void) {
static Boolean dollar = FALSE;
int c;
size_t i; /* The purpose of all these local assignments is to */
const char *meta; /* allow optimizing compilers like gcc to load these */
char *buf = tokenbuf; /* values into registers. On a sparc this is a */
YYSTYPE *y = &yylval; /* win, in code size *and* execution time */
if (goterror) {
goterror = FALSE;
return NL;
}
/* rc variable-names may contain only alnum, '*' and '_', so use dnw if we are scanning one. */
meta = (dollar ? dnw : nw);
dollar = FALSE;
if (newline) {
--input->lineno; /* slight space optimization; print_prompt2() always increments lineno */
print_prompt2();
newline = FALSE;
}
top: while ((c = GETC()) == ' ' || c == '\t')
w = NW;
if (c == EOF)
return ENDFILE;
if (!meta[(unsigned char) c]) { /* it's a word or keyword. */
InsertFreeCaret();
w = RW;
i = 0;
do {
buf[i++] = c;
if (i >= bufsize)
buf = tokenbuf = erealloc(buf, bufsize *= 2);
} while ((c = GETC()) != EOF && !meta[(unsigned char) c]);
UNGETC(c);
buf[i] = '\0';
w = KW;
if (buf[1] == '\0') {
int k = *buf;
if (k == '@' || k == '~')
return k;
} else if (*buf == 'f') {
if (streq(buf + 1, "n")) return FN;
if (streq(buf + 1, "or")) return FOR;
} else if (*buf == 'l') {
if (streq(buf + 1, "ocal")) return LOCAL;
if (streq(buf + 1, "et")) return LET;
} else if (streq(buf, "~~"))
return EXTRACT;
else if (streq(buf, "%closure"))
return CLOSURE;
w = RW;
y->str = gcdup(buf);
return WORD;
}
if (c == '`' || c == '!' || c == '$' || c == '\'') {
InsertFreeCaret();
if (c == '!')
w = KW;
}
switch (c) {
case '!':
return '!';
case '`':
c = GETC();
if (c == '`')
return BACKBACK;
UNGETC(c);
return '`';
case '$':
dollar = TRUE;
switch (c = GETC()) {
case '#': return COUNT;
case '^': return FLAT;
case '&': return PRIM;
default: UNGETC(c); return '$';
}
case '\'':
w = RW;
i = 0;
while ((c = GETC()) != '\'' || (c = GETC()) == '\'') {
buf[i++] = c;
if (c == '\n')
print_prompt2();
if (c == EOF) {
w = NW;
scanerror("eof in quoted string");
return ERROR;
}
if (i >= bufsize)
buf = tokenbuf = erealloc(buf, bufsize *= 2);
}
UNGETC(c);
buf[i] = '\0';
y->str = gcdup(buf);
return QWORD;
case '\\':
if ((c = GETC()) == '\n') {
print_prompt2();
UNGETC(' ');
goto top; /* Pretend it was just another space. */
}
if (c == EOF) {
UNGETC(EOF);
goto badescape;
}
UNGETC(c);
c = '\\';
InsertFreeCaret();
w = RW;
c = GETC();
switch (c) {
case 'a': *buf = '\a'; break;
case 'b': *buf = '\b'; break;
case 'e': *buf = '\033'; break;
case 'f': *buf = '\f'; break;
case 'n': *buf = '\n'; break;
case 'r': *buf = '\r'; break;
case 't': *buf = '\t'; break;
case 'x': case 'X': {
int n = 0;
for (;;) {
c = GETC();
if (!isxdigit(c))
break;
n = (n << 4)
| (c - (isdigit(c) ? '0' : ((islower(c) ? 'a' : 'A') - 0xA)));
}
if (n == 0)
goto badescape;
UNGETC(c);
*buf = n;
break;
}
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
int n = 0;
do {
n = (n << 3) | (c - '0');
c = GETC();
} while (isodigit(c));
if (n == 0)
goto badescape;
UNGETC(c);
*buf = n;
break;
}
default:
if (isalnum(c)) {
badescape:
scanerror("bad backslash escape");
return ERROR;
}
*buf = c;
break;
}
buf[1] = 0;
y->str = gcdup(buf);
return QWORD;
case '#':
while ((c = GETC()) != '\n') /* skip comment until newline */
if (c == EOF)
return ENDFILE;
/* FALLTHROUGH */
case '\n':
input->lineno++;
newline = TRUE;
w = NW;
return NL;
case '(':
if (w == RW) /* not keywords, so let & friends work */
c = SUB;
/* FALLTHROUGH */
case ';':
case '^':
case ')':
case '=':
case '{': case '}':
w = NW;
return c;
case '&':
w = NW;
c = GETC();
if (c == '&')
return ANDAND;
UNGETC(c);
return '&';
case '|': {
int p[2];
w = NW;
c = GETC();
if (c == '|')
return OROR;
if (!getfds(p, c, 1, 0))
return ERROR;
if (p[1] == CLOSED) {
scanerror("expected digit after '='"); /* can't close a pipe */
return ERROR;
}
y->tree = mk(nPipe, p[0], p[1]);
return PIPE;
}
{
char *cmd;
int fd[2];
case '<':
fd[0] = 0;
if ((c = GETC()) == '>')
if ((c = GETC()) == '>') {
c = GETC();
cmd = "%open-append";
} else
cmd = "%open-write";
else if (c == '<')
if ((c = GETC()) == '<') {
c = GETC();
cmd = "%here";
} else
cmd = "%heredoc";
else if (c == '=')
return CALL;
else
cmd = "%open";
goto redirection;
case '>':
fd[0] = 1;
if ((c = GETC()) == '>')
if ((c = GETC()) == '<') {
c = GETC();
cmd = "%open-append";
} else
cmd = "%append";
else if (c == '<') {
c = GETC();
cmd = "%open-create";
} else
cmd = "%create";
goto redirection;
redirection:
w = NW;
if (!getfds(fd, c, fd[0], DEFAULT))
return ERROR;
if (fd[1] != DEFAULT) {
y->tree = (fd[1] == CLOSED)
? mkclose(fd[0])
: mkdup(fd[0], fd[1]);
return DUP;
}
y->tree = mkredircmd(cmd, fd[0]);
return REDIR;
}
default:
assert(c != '\0');
w = NW;
return c; /* don't know what it is, let yacc barf on it */
}
}
extern int yylex() {
static bool dollar = FALSE;
bool saw_meta = FALSE;
int c;
SIZE_T i; /* The purpose of all these local assignments is to */
char *meta; /* allow optimizing compilers like gcc to load these */
char *buf = realbuf; /* values into registers. On a sparc this is a */
YYSTYPE *y = &yylval; /* win, in code size *and* execution time */
if (errset) {
errset = FALSE;
return '\n';
}
/* rc variable-names may contain only alnum, '*' and '_', so use dnw if we are scanning one. */
meta = (dollar ? dnw : nw);
dollar = FALSE;
if (newline) {
--lineno; /* slight space optimization; print_prompt2() always increments lineno */
print_prompt2();
newline = FALSE;
}
top: while ((c = gchar()) == ' ' || c == '\t')
w = NW;
if (c == EOF)
return END;
if (!meta[(unsigned char) c]) { /* it's a word or keyword. */
checkfreecaret;
w = RW;
i = 0;
read: do {
buf[i++] = c;
if (c == '?' || c == '[' || c == '*')
saw_meta = TRUE;
if (i >= bufsize)
buf = realbuf = erealloc(buf, bufsize *= 2);
} while ((c = gchar()) != EOF && !meta[(unsigned char) c]);
while (c == '\\') {
if ((c = gchar()) == '\n') {
print_prompt2();
c = ' '; /* Pretend a space was read */
break;
} else {
bs: if (meta != dnw) { /* all words but varnames may have a bslash */
buf[i++] = '\\';
if (i >= bufsize)
buf = realbuf = erealloc(buf, bufsize *= 2);
if (!meta[(unsigned char) c])
goto read;
} else {
ugchar(c);
c = '\\';
break;
}
}
}
ugchar(c);
buf[i] = '\0';
w = KW;
if (i == 2) {
if (*buf == 'i' && buf[1] == 'f') return IF;
if (*buf == 'f' && buf[1] == 'n') return FN;
if (*buf == 'i' && buf[1] == 'n') return IN;
}
if (streq(buf, "for")) return FOR;
if (streq(buf, "else")) return ELSE;
if (streq(buf, "switch")) return SWITCH;
if (streq(buf, "while")) return WHILE;
if (streq(buf, "case")) return CASE;
w = RW;
y->word.w = ncpy(buf);
if (saw_meta) {
char *r, *s;
y->word.m = nalloc(strlen(buf) + 1);
for (r = buf, s = y->word.m; *r != '\0'; r++, s++)
*s = (*r == '?' || *r == '[' || *r == '*');
} else {
y->word.m = NULL;
}
return WORD;
}
if (c == '`' || c == '!' || c == '@' || c == '~' || c == '$' || c == '\'') {
checkfreecaret;
if (c == '!' || c == '@' || c == '~')
w = KW;
}
switch (c) {
case '!':
return BANG;
case '@':
return SUBSHELL;
case '~':
return TWIDDLE;
case '`':
c = gchar();
if (c == '`')
return BACKBACK;
ugchar(c);
return '`';
case '$':
dollar = TRUE;
c = gchar();
if (c == '#')
return COUNT;
if (c == '^')
return FLAT;
ugchar(c);
return '$';
case '\'':
w = RW;
i = 0;
do {
buf[i++] = c;
if (c == '\n')
print_prompt2();
if (c == EOF) {
w = NW;
scanerror("eof in quoted string");
return HUH;
}
if (i >= bufsize)
buf = realbuf = erealloc(buf, bufsize *= 2);
} while ((c = gchar()) != '\'' || (c = gchar()) == '\''); /* quote "'" thus: 'how''s it going?' */
ugchar(c);
buf[i] = '\0';
y->word.w = ncpy(buf);
y->word.m = NULL;
return WORD;
case '\\':
if ((c = gchar()) == '\n') {
print_prompt2();
goto top; /* Pretend it was just another space. */
}
ugchar(c);
c = '\\';
checkfreecaret;
c = gchar();
i = 0;
goto bs;
case '(':
if (w == RW) /* SUB's happen only after real words, not keyowrds, so if () and while () work */
c = SUB;
w = NW;
return c;
case '#':
while ((c = gchar()) != '\n') /* skip comment until newline */
if (c == EOF)
return END;
/* FALLTHROUGH */
case '\n':
lineno++;
newline = TRUE;
/* FALLTHROUGH */
case ';':
case '^':
case ')':
case '=':
case '{': case '}':
w = NW;
return c;
case '&':
w = NW;
c = gchar();
if (c == '&')
return ANDAND;
ugchar(c);
return '&';
case '|':
w = NW;
c = gchar();
if (c == '|')
return OROR;
getpair(c);
if (errset)
return HUH;
if ((y->pipe.left = fd_left) == UNSET)
y->pipe.left = 1; /* default to fd 1 */
if ((y->pipe.right = fd_right) == UNSET)
y->pipe.right = 0; /* default to fd 0 */
if (y->pipe.right == CLOSED) {
scanerror("expected digit after '='"); /* can't close a pipe */
return HUH;
}
return PIPE;
case '>':
c = gchar();
if (c == '>') {
c = gchar();
y->redir.type = rAppend;
} else
y->redir.type = rCreate;
y->redir.fd = 1;
goto common;
case '<':
c = gchar();
if (c == '<') {
c = gchar();
if (c == '<') {
c = gchar();
y->redir.type = rHerestring;
} else {
y->redir.type = rHeredoc;
}
} else
y->redir.type = rFrom;
y->redir.fd = 0;
common:
w = NW;
getpair(c);
if (errset)
return HUH;
if (fd_right == UNSET) { /* redirection, not dup */
if (fd_left != UNSET) {
y->redir.fd = fd_left;
return SREDIR;
}
return (y->redir.type == rFrom || y->redir.type == rCreate) ? REDIR : SREDIR;
} else { /* dup; recast yylval */
y->dup.type = y->redir.type;
y->dup.left = fd_left;
y->dup.right = fd_right;
return DUP;
}
default:
w = NW;
return c; /* don't know what it is, let yacc barf on it */
}
}
