ref: 76f5318cbd58fbe2bfc8cb51856bfa217c673bb7
dir: /utils/rcsh/lex.c/
#include "rc.h"
#include "y.tab.h"
#define NTOK 8192
int getnext(void);
int future=EOF;
int doprompt=1;
int inquote;
int nerror;
char *promptstr;
char tok[NTOK];
int lastdol; /* was the last token read '$' or '$#' or '"'? */
int lastword; /* was the last token read a word or compound word terminator? */
int lastc;
void
kinit(void)
{
kenter(FOR, "for");
kenter(IN, "in");
kenter(WHILE, "while");
kenter(IF, "if");
kenter(NOT, "not");
kenter(TWIDDLE, "~");
kenter(BANG, "!");
kenter(SUBSHELL, "@");
kenter(SWITCH, "switch");
kenter(FN, "fn");
}
int
wordchr(int c)
{
return !strchr("\n \t\r#;&|^$=`'{}()<>", c) && c!=EOF;
}
int
idchr(int c)
{
/*
* Formerly:
* return 'a'<=c && c<='z' || 'A'<=c && c<='Z' || '0'<=c && c<='9'
* || c=='_' || c=='*';
*/
return c>' ' && !strchr("!\"#$%&'()+,-./:;<=>?@[\\]^`{|}~", c);
}
/*
* Look ahead in the input stream
*/
int
nextc(void)
{
if(future==EOF)
future=getnext();
return future;
}
/*
* Consume the lookahead character.
*/
int
advance(void)
{
int c=nextc();
lastc=future;
future=EOF;
return c;
}
/*
* read a character from the input stream
*/
int
getnext(void)
{
register int c;
static peekc=EOF;
if(peekc!=EOF){
c=peekc;
peekc=EOF;
return c;
}
if(runq->eof) return EOF;
if(doprompt)
pprompt();
c=rchr(runq->cmdfd);
if(!inquote && c=='\\'){
c=rchr(runq->cmdfd);
if(c=='\n'){
doprompt=1;
c=' ';
}
else{
peekc=c;
c='\\';
}
}
doprompt=doprompt || c=='\n' || c==EOF;
if(c==EOF) runq->eof++;
else if(flag['V'] || ndot>=2 && flag['v'])
pchr(err, c);
return c;
}
void
pprompt(void)
{
Var *prompt;
if(runq->iflag){
pstr(err, promptstr);
flush(err);
prompt=vlook("prompt");
if(prompt->val && prompt->val->next)
promptstr=prompt->val->next->word;
else
promptstr="\t";
}
runq->lineno++;
doprompt=0;
}
void
skipwhite(void)
{
int c;
for(;;){
c=nextc();
if(c=='#'){ /* Why did this used to be if(!inquote && c=='#') ?? */
for(;;){
c=nextc();
if(c=='\n' || c==EOF) break;
advance();
}
}
if(c==' ' || c=='\t' || c=='\r') advance();
else return;
}
}
void
skipnl(void)
{
int c;
for(;;){
skipwhite();
c=nextc();
if(c!='\n') return;
advance();
}
}
int
nextis(int c)
{
if(nextc()==c){
advance();
return 1;
}
return 0;
}
char *
addtok(char *p, int val)
{
if(p==0) return 0;
if(p==&tok[NTOK]){
*p=0;
yyerror("token buffer too short");
return 0;
}
*p++=val;
return p;
}
char *
addutf(char *p, int c)
{
p=addtok(p, c);
if(twobyte(c)) /* 2-byte escape */
return addtok(p, advance());
if(threebyte(c)){ /* 3-byte escape */
p=addtok(p, advance());
return addtok(p, advance());
}
return p;
}
int
yylex(void)
{
int c, d=nextc();
char *w=tok;
Tree *t;
yylval.tree=0;
/*
* Embarassing sneakiness: if the last token read was a quoted or unquoted
* WORD then we alter the meaning of what follows. If the next character
* is `(', we return SUB (a subscript paren) and consume the `('. Otherwise,
* if the next character is the first character of a simple or compound word,
* we insert a `^' before it.
*/
if(lastword){
lastword=0;
if(d=='('){
advance();
strcpy(tok, "( [SUB]");
return SUB;
}
if(wordchr(d) || d=='\'' || d=='`' || d=='$' || d=='"'){
strcpy(tok, "^");
return '^';
}
}
inquote=0;
skipwhite();
switch(c=advance()){
case EOF:
lastdol=0;
strcpy(tok, "EOF");
return EOF;
case '$':
lastdol=1;
if(nextis('#')){
strcpy(tok, "$#");
return COUNT;
}
if(nextis('"')){
strcpy(tok, "$\"");
return '"';
}
strcpy(tok, "$");
return '$';
case '&':
lastdol=0;
if(nextis('&')){
skipnl();
strcpy(tok, "&&");
return ANDAND;
}
strcpy(tok, "&");
return '&';
case '|':
lastdol=0;
if(nextis(c)){
skipnl();
strcpy(tok, "||");
return OROR;
}
case '<':
case '>':
lastdol=0;
/*
* funny redirection tokens:
* redir: arrow | arrow '[' fd ']'
* arrow: '<' | '<<' | '>' | '>>' | '|'
* fd: digit | digit '=' | digit '=' digit
* digit: '0'|'1'|'2'|'3'|'4'|'5'|'6'|'7'|'8'|'9'
* some possibilities are nonsensical and get a message.
*/
*w++=c;
t=newtree();
switch(c){
case '|':
t->type=PIPE;
t->fd0=1;
t->fd1=0;
break;
case '>':
t->type=REDIR;
if(nextis(c)){
t->rtype=APPEND;
*w++=c;
}
else t->rtype=WRITE;
t->fd0=1;
break;
case '<':
t->type=REDIR;
if(nextis(c)){
t->rtype=HERE;
*w++=c;
}
else t->rtype=READ;
t->fd0=0;
break;
}
if(nextis('[')){
*w++='[';
c=advance();
*w++=c;
if(c<'0' || '9'<c){
RedirErr:
*w=0;
yyerror(t->type==PIPE?"pipe syntax"
:"redirection syntax");
return EOF;
}
t->fd0=0;
do{
t->fd0=t->fd0*10+c-'0';
*w++=c;
c=advance();
}while('0'<=c && c<='9');
if(c=='='){
*w++='=';
if(t->type==REDIR)
t->type=DUP;
c=advance();
if('0'<=c && c<='9'){
t->rtype=DUPFD;
t->fd1=t->fd0;
t->fd0=0;
do{
t->fd0=t->fd0*10+c-'0';
*w++=c;
c=advance();
}while('0'<=c && c<='9');
}
else{
if(t->type==PIPE) goto RedirErr;
t->rtype=CLOSE;
}
}
if(c!=']' || t->type==DUP && (t->rtype==HERE || t->rtype==APPEND))
goto RedirErr;
*w++=']';
}
*w='\0';
yylval.tree=t;
if(t->type==PIPE) skipnl();
return t->type;
case '\'':
lastdol=0;
lastword=1;
inquote=1;
for(;;){
c=advance();
if(c==EOF) break;
if(c=='\''){
if(nextc()!='\'')
break;
advance();
}
w=addutf(w, c);
}
if(w!=0) *w='\0';
t=token(tok, WORD);
t->quoted=1;
yylval.tree=t;
return t->type;
}
if(!wordchr(c)){
lastdol=0;
tok[0]=c;
tok[1]='\0';
return c;
}
for(;;){
/* next line should have (char)c==GLOB, but ken's compiler is broken */
if(c=='*' || c=='[' || c=='?' || c==(unsigned char)GLOB)
w=addtok(w, GLOB);
w=addutf(w, c);
c=nextc();
if(lastdol?!idchr(c):!wordchr(c)) break;
advance();
}
Out:
lastword=1;
lastdol=0;
if(w!=0) *w='\0';
t=klook(tok);
if(t->type!=WORD) lastword=0;
t->quoted=0;
yylval.tree=t;
return t->type;
}
void
yyerror(char *m)
{
pfmt(err, "rc: ");
if(runq->cmdfile) pfmt(err, "file %s: ", runq->cmdfile);
if(!runq->iflag) pfmt(err, "line %d: ", runq->lineno);
if(tok[0] && tok[0]!='\n') pfmt(err, "token %q: ", tok);
pfmt(err, "%s\n", m);
flush(err);
lastword=0;
lastdol=0;
while(lastc!='\n' && lastc!=EOF) advance();
nerror++;
}