mshow: print plain text safely

Makefile

@@ -24,7 +24,7 @@ maddr magrep mexport mflag mgenmid mhdr mlist mpick mscan msed mseq mshow msort
   mthread : seq.o slurp.o
 maddr magrep mhdr mpick mscan mshow : rfc2047.o
 magrep mshow : rfc2045.o
-mshow : filter.o
+mshow : filter.o safe_u8putstr.c
 msort : mystrverscmp.o
 mmime : slurp.o
 

blaze822.h

@@ -82,3 +82,8 @@ time_t tm_to_secs(const struct tm *tm);
 // slurp.c
 
 int slurp(char *filename, char **bufo, off_t *leno);
+
+// safe_u8putstr.c
+
+#include <stdio.h>
+void safe_u8putstr(char *s0, size_t l, FILE *stream);

mshow.c

@@ -30,43 +30,43 @@ static char *Oflag;
 struct message *filters;
 
 static int mimecount;
+static int safe_output;
 
 static char defaultAflags[] = "text/plain:text/html";
 static char *Aflag = defaultAflags;
 
+static int
+printable(int c)
+{
+	return (unsigned)c-0x20 < 0x5f;
+}
+
+int
+print_ascii(char *body, size_t bodylen)
+{
+	if (safe_output) {
+		safe_u8putstr(body, bodylen, stdout);
+		return bodylen;
+	} else {
+		return fwrite(body, 1, bodylen, stdout);
+	}
+}
+
 void
 printhdr(char *hdr)
 {
 	int uc = 1;
 
-	while (*hdr && *hdr != ':') {
+	while (*hdr && *hdr != ':' && printable(*hdr)) {
 		putc(uc ? toupper(*hdr) : *hdr, stdout);
 		uc = (*hdr == '-');
 		hdr++;
 	}
 
-	if (*hdr)
-		printf("%s\n", hdr);
-}
-
-int
-print_ascii(char *body, size_t bodylen)
-{
-	if (!memchr(body, '\r', bodylen))
-		return fwrite(body, 1, bodylen, stdout);
-
-	// crlf normalization required
-	size_t i;
-	for (i = 0; i < bodylen; i++) {
-		if (body[i] == '\r') {
-			if (!(i+1 < bodylen && body[i+1] == '\n'))
-				putc_unlocked('\n', stdout);
-			continue;
-		}
-		putc_unlocked(body[i], stdout);
+	if (*hdr) {
+		print_ascii(hdr, strlen(hdr));
+		fputc('\n', stdout);
 	}
-
-	return bodylen;
 }
 
 void
@@ -529,7 +529,8 @@ print_date_header(char *v)
 		now = time(0);
 	}
 
-	printf("Date: %s", v);
+	printf("Date: ");
+	print_ascii(v, strlen(v));
 	
 	time_t t = blaze822_date(v);
 	if (t == -1) {
@@ -608,7 +609,10 @@ print_decode_header(char *h, char *v)
 	char d[4096];
 	blaze822_decode_rfc2047(d, v, sizeof d, "UTF-8");
 	printhdr(h);
-	printf(": %s\n", d);
+	fputc(':', stdout);
+	fputc(' ', stdout);
+	print_ascii(d, strlen(d));
+	fputc('\n', stdout);
 }
 
 void
@@ -674,7 +678,7 @@ show(char *file)
 	printf("\n");
 
 	if (rflag || !blaze822_check_mime(msg)) {  // raw body
-		fwrite(blaze822_body(msg), 1, blaze822_bodylen(msg), stdout);
+		print_ascii(blaze822_body(msg), blaze822_bodylen(msg));
 		goto done;
 	}
 
@@ -713,6 +717,9 @@ main(int argc, char *argv[])
 			exit(1);
 		}
 
+	if (!rflag && !Oflag)
+		safe_output = 1;
+
 	if (xflag) { // extract
 		extract(xflag, argc-optind, argv+optind, 0);
 	} else if (Oflag) { // extract to stdout

safe_u8putstr.c

@@ -0,0 +1,68 @@
+#include <stdio.h>
+#include <stdint.h>
+
+void
+safe_u8putstr(char *s0, size_t l, FILE *stream)
+{
+	// tty-safe output of s, with relaxed utf-8 semantics:
+	// - C0 and C1 are displayed as escape sequences
+	// - valid utf8 is printed as is
+	// - rest is printed bytewise as is (probably latin1)
+	// - translate CRLF to CR
+
+	unsigned char *s = (unsigned char* )s0;
+	unsigned char *e = s + l;
+
+	while (s < e) {
+		if ((*s & 0x80) == 0) {
+			if (*s < 32 &&
+			    *s != ' ' && *s != '\t' && *s != '\n' && *s != '\r') {
+				// C0
+				fputc(0xe2, stream);
+				fputc(0x90, stream);
+				fputc(0x80+*s, stream);
+			} else if (*s == 127) {
+				// DEL
+				fputc(0xe2, stream);
+				fputc(0x90, stream);
+				fputc(0xa1, stream);
+			} else if (*s == '\r') {
+				if (e - s > 1 && s[1] == '\n')
+					s++;
+				fputc(*s, stream);
+			} else {
+				// safe ASCII
+				fputc(*s, stream);
+			}
+		} else if ((*s & 0xc0) == 0x80) {
+			// C1
+			fputc(0xe2, stream);
+			fputc(0x90, stream);
+			fputc(0x80+0x1b, stream);
+
+			fputc(0xe2, stream);
+			fputc(0x90, stream);
+			fputc(*s, stream);
+		} else {
+			uint32_t f = 0;
+			if (e - s >= 4)
+				f = (s[0]<<24) | (s[1]<<16) | (s[2]<<8) | s[3];
+			else if (e - s == 3)
+				f = (s[0]<<24) | (s[1]<<16) | (s[2]<<8);
+			else if (e - s == 2)
+				f = (s[0]<<24) | (s[1]<<16);
+			else if (e - s == 1)
+				f = (s[0]<<24);
+			
+			if      ((f & 0xe0c00000) == 0xc0800000) goto u2;
+			else if ((f & 0xf0c0c000) == 0xe0808000) goto u3;
+			else if ((f & 0xf8c0c0c0) == 0xf0808080) goto u4;
+			else    /* invalid utf8 */               goto u1;
+u4:				fputc(*s++, stream);
+u3:				fputc(*s++, stream);
+u2:				fputc(*s++, stream);
+u1:				fputc(*s, stream);
+		}
+		s++;
+	}
+}