From bf03074e346b004659196b6c17eee04dbffd3ac2 Mon Sep 17 00:00:00 2001
From: Nicholas Noll <nbnoll@eml.cc>
Date: Fri, 15 Oct 2021 16:18:02 -0700
Subject: feat(rc): working prototype of input->compile->print loop

---
 sys/cmd/rc/lex.c | 207 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 207 insertions(+)
 create mode 100644 sys/cmd/rc/lex.c

(limited to 'sys/cmd/rc/lex.c')

diff --git a/sys/cmd/rc/lex.c b/sys/cmd/rc/lex.c
new file mode 100644
index 0000000..ec9e94d
--- /dev/null
+++ b/sys/cmd/rc/lex.c
@@ -0,0 +1,207 @@
+#include "rc.h"
+#include "parse.h"
+
+static int advance(void);
+// -----------------------------------------------------------------------
+// lexer
+
+struct Lexer
+{
+    int c[2];
+    ushort doprompt;
+    char buf[BUFSIZ];
+};
+
+static struct Lexer lexer = { .c={0, EOF}, .doprompt=1 };
+
+void
+yyerror(char *msg)
+{
+    print(errio, "\nrc: ");
+
+    if(lexer.buf[0] && lexer.buf[0]!='\n')
+        print(errio, "@ %q: ", lexer.buf);
+
+    print(errio, "%s\n", msg);
+    flush(errio);
+
+    while(lexer.c[0] !='\n' && lexer.c[0] != EOF)
+        advance();
+}
+
+int
+readc(void)
+{
+    int c;
+    static int peek = EOF;
+
+    if(peek!=EOF){
+        c = peek;
+        peek = EOF;
+        return c;
+    }
+    if(shell->flag.eof)
+        return EOF;
+
+    if(!prompt(&lexer.doprompt))
+        exit(1); // XXX: hack for signal handling right now...
+
+    c = get(shell->cmd.io);
+
+    lexer.doprompt = lexer.doprompt || c=='\n' || c==EOF;
+
+    if(c==EOF)
+        shell->flag.eof = 1;
+
+    return c;
+}
+
+static
+int
+peekc(void)
+{
+    if(lexer.c[1] == EOF)
+        lexer.c[1] = readc();
+
+    return lexer.c[1];
+}
+
+static
+int
+advance(void)
+{
+    int c = peekc();
+    lexer.c[0] = lexer.c[1], lexer.c[1] = EOF;
+
+    return c;
+}
+
+static
+void
+skipws(void)
+{
+    int c;
+    for(;;){
+        c = peekc();
+        if(c== ' ' || c == '\t')
+            advance();
+        else
+            return;
+    }
+}
+
+static
+void
+skipnl(void)
+{
+    int c;
+    for(;;){
+        c = peekc();
+        if(c== ' ' || c == '\t' || c == '\n')
+            advance();
+        else
+            return;
+    }
+}
+
+static
+int
+nextis(int c)
+{
+    if(peekc()==c){
+        advance();
+        return 1;
+    }
+    return 0;
+}
+
+static
+char *
+putbyte(char *buf, int c)
+{
+    if(!buf)
+        return buf;
+
+    if(buf == arrend(lexer.buf)){
+        fatal("lexer: out of buffer space");
+        return nil;
+    }
+    *buf++ = c;
+    return buf;
+}
+
+#define	onebyte(c)	     ((c&0x80)==0x00)
+#define	twobyte(c)	     ((c&0xe0)==0xc0)
+#define	threebyte(c)	 ((c&0xf0)==0xe0)
+#define	fourbyte(c)	     ((c&0xf8)==0xf0)
+
+static
+char *
+putrune(char *buf, int c)
+{
+    buf = putbyte(buf, c);
+    if(onebyte(c))
+        return buf;
+    if(twobyte(c))
+        return putbyte(buf,c);
+    if(threebyte(c)){
+        buf = putbyte(buf,c);
+        return putbyte(buf,c);
+    }
+    if(fourbyte(c)){
+        buf = putbyte(buf,c);
+        buf = putbyte(buf,c);
+        return putbyte(buf,c);
+    }
+    fatal("malformed utf8 stream");
+
+    return nil;
+}
+
+// -----------------------------------------------------------------------
+// exported functions
+
+int
+iswordchar(int c)
+{
+    return !strchr("\n \t#;&|^$=`'{}()<>", c) && c!=EOF;
+}
+
+int
+yylex(void)
+{
+    int  c;
+    Tree *node;
+    char *w = lexer.buf;
+
+    yylval.tree = nil;
+
+    skipws();
+    switch(c=advance()){
+    case EOF:
+        return EOF;
+    case '&':
+        lexer.buf[0] = '&';
+        lexer.buf[1] = 0;
+        return '&';
+
+    default:
+        ;
+    }
+    if(!iswordchar(c))
+        return c;
+
+    for(;;){
+        w = putrune(w, c);
+        c = peekc();
+        if(!iswordchar(c))
+            break;
+        advance();
+    }
+    w[0] = 0;
+
+    node = token(Tword, lexer.buf);
+
+    yylval.tree = node;
+    return node->type;
+}
-- 
cgit v1.2.1