Re: Marshaling with ARGF

Hi,

At Thu, 17 Oct 2002 03:24:27 +0900,
Tom Payne wrote:
> Anyone on the Ruby dev team care to modifiy marshal.c so that objects
> like ARGF can be used (i.e. use rb_funcall(obj, rb_intern("read"),
> ....) rather than hardcoded calls to rb_io_fread, etc.)?

Dirty hack.


Index: marshal.c
===================================================================
RCS file: /cvs/ruby/src/ruby/marshal.c,v
retrieving revision 1.72
diff -u -2 -p -r1.72 marshal.c
--- marshal.c	25 Sep 2002 14:52:37 -0000	1.72
+++ marshal.c	17 Oct 2002 02:31:20 -0000
@@ -77,9 +77,10 @@ shortlen(len, ds)
 static ID s_dump, s_load;
 static ID s_dump_data, s_load_data, s_alloc;
+static ID s_getc, s_read;
 
 struct dump_arg {
     VALUE obj;
     FILE *fp;
-    VALUE str;
+    VALUE str, dest;
     st_table *symbol;
     st_table *data;
@@ -96,10 +97,28 @@ static void w_long _((long, struct dump_
 
 static void
+w_byten(s, n, arg)
+    char *s;
+    int n;
+    struct dump_arg *arg;
+{
+    if (arg->fp) {
+	fwrite(s, 1, n, arg->fp);
+    }
+    else {
+	VALUE buf = arg->str;
+	rb_str_buf_cat(buf, s, n);
+	if (arg->dest && RSTRING(buf)->len >= BUFSIZ) {
+	    rb_io_write(arg->dest, buf);
+	    rb_str_resize(buf, 0);
+	}
+    }
+}
+
+static void
 w_byte(c, arg)
     char c;
     struct dump_arg *arg;
 {
-    if (arg->fp) putc(c, arg->fp);
-    else rb_str_buf_cat(arg->str, &c, 1);
+    w_byten(&c, 1, arg);
 }
 
@@ -111,10 +130,5 @@ w_bytes(s, n, arg)
 {
     w_long(n, arg);
-    if (arg->fp) {
-	fwrite(s, 1, n, arg->fp);
-    }
-    else {
-	rb_str_buf_cat(arg->str, s, n);
-    }
+    w_byten(s, n, arg);
 }
 
@@ -587,4 +601,5 @@ marshal_dump(argc, argv)
 	else port = a1;
     }
+    arg.dest = 0;
     if (port) {
 	if (rb_obj_is_kind_of(port, rb_cIO)) {
@@ -597,5 +612,7 @@ marshal_dump(argc, argv)
 	}
 	else {
-	    rb_raise(rb_eTypeError, "instance of IO needed");
+	    arg.fp = 0;
+	    arg.str = rb_str_buf_new(0);
+	    arg.dest = port;
 	}
     }
@@ -642,4 +659,10 @@ r_byte(arg)
 	if (c == EOF) rb_eof_error();
     }
+    else if (!arg->end) {
+	VALUE src = (VALUE)arg->ptr;
+	VALUE v = rb_funcall2(src, s_getc, 0, 0);
+	if (NIL_P(v)) rb_eof_error();
+	c = (unsigned char)FIX2INT(v);
+    }
     else if (arg->ptr < arg->end) {
 	c = *(unsigned char*)arg->ptr++;
@@ -651,16 +674,4 @@ r_byte(arg)
 }
 
-static unsigned short
-r_short(arg)
-    struct load_arg *arg;
-{
-    unsigned short x;
-
-    x =  r_byte(arg);
-    x |= r_byte(arg)<<8;
-
-    return x;
-}
-
 static void
 long_toobig(size)
@@ -729,4 +740,12 @@ r_bytes0(len, arg)
 	}
     }
+    else if (!arg->end) {
+	VALUE src = (VALUE)arg->ptr;
+	VALUE n = LONG2NUM(len);
+	str = rb_funcall2(src, s_read, 1, &n);
+	if (NIL_P(str)) goto too_short;
+	Check_Type(str, T_STRING);
+	if (RSTRING(str)->len != len) goto too_short;
+    }
     else {
 	if (arg->ptr + len > arg->end) {
@@ -934,4 +953,5 @@ r_object0(arg, proc)
 	    long len;
 	    BDIGIT *digits;
+	    VALUE data;
 
 	    NEWOBJ(big, struct RBignum);
@@ -939,4 +959,5 @@ r_object0(arg, proc)
 	    big->sign = (r_byte(arg) == '+');
 	    len = r_long(arg);
+	    data = r_bytes0(len * 2, arg);
 #if SIZEOF_BDIGITS == SIZEOF_SHORT
 	    big->len = len;
@@ -945,21 +966,26 @@ r_object0(arg, proc)
 #endif
 	    big->digits = digits = ALLOC_N(BDIGIT, big->len);
-	    while (len > 0) {
+	    MEMCPY(digits, RSTRING(data)->ptr, char, len * 2);
 #if SIZEOF_BDIGITS > SIZEOF_SHORT
+	    MEMZERO((char *)digits + len * 2, char,
+		    big->len * sizeof(BDIGIT) - len * 2);
+#endif
+	    len = big->len;
+	    while (len > 0) {
+		unsigned char *p = (unsigned char *)digits;
 		BDIGIT num = 0;
+#if SIZEOF_BDIGITS > SIZEOF_SHORT
 		int shift = 0;
 		int i;
 
-		for (i=0; i<SIZEOF_BDIGITS; i+=2) {
-		    int j = r_short(arg);
-		    num |= j << shift;
-		    shift += BITSPERSHORT;
-		    if (--len == 0) break;
+		for (i=0; i<SIZEOF_BDIGITS; i++) {
+		    num |= (int)p[i] << shift;
+		    shift += 8;
 		}
-		*digits++ = num;
 #else
-		*digits++ = r_short(arg);
-		len--;
+		num = p[0] | (p[1] << 8);
 #endif
+		*digits++ = num;
+		len--;
 	    }
 	    v = rb_big_norm((VALUE)big);
@@ -1191,5 +1217,8 @@ marshal_load(argc, argv)
     }
     else {
-	rb_raise(rb_eTypeError, "instance of IO needed");
+	arg.taint = OBJ_TAINTED(port); /* original taintedness */
+	arg.fp = 0;
+	arg.ptr = (char *)port;
+	arg.end = 0;
     }
 
@@ -1226,4 +1255,7 @@ Init_marshal()
     s_load_data = rb_intern("_load_data");
     s_alloc = rb_intern("_alloc");
+    s_getc = rb_intern("getc");
+    s_read = rb_intern("read");
+
     rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
     rb_define_module_function(rb_mMarshal, "load", marshal_load, -1);


-- 
Nobu Nakada