mothra: ignore http content-type and encoding and just do content sniffing

1 files changed, 65 insertions, 0 deletions

diff --git a/sys/src/cmd/mothra/snoop.c b/sys/src/cmd/mothra/snoop.c
new file mode 100644
index 000000000..3ab8075f7
--- /dev/null
+++ b/sys/src/cmd/mothra/snoop.c
@@ -0,0 +1,65 @@
+#include <u.h>
+#include <libc.h>
+#include <draw.h>
+#include <event.h>
+#include <panel.h>
+#include <ctype.h>
+#include "mothra.h"
+
+int
+snooptype(int fd)
+{
+	int pfd[2], typ, n;
+	char buf[1024];
+
+	typ = PLAIN;
+	if((n = readn(fd, buf, sizeof(buf)-1)) < 0)
+		return typ;
+	buf[n] = 0;
+	if(cistrstr(buf, "<?xml") ||
+		cistrstr(buf, "<!DOCTYPE") ||
+		cistrstr(buf, "<HTML"))
+		typ = HTML;
+	else if(memcmp(buf, "\x1F\x8B", 2) == 0)
+		typ = GUNZIP;
+	else if(memcmp(buf, "\377\330\377", 3) == 0)
+		typ = JPEG;
+	else if(memcmp(buf, "\211PNG\r\n\032\n", 3) == 0)
+		typ = PNG;
+	else if(memcmp(buf, "GIF", 3) == 0)
+		typ = GIF;
+	else if(memcmp(buf, "BM", 2) == 0)
+		typ = BMP;
+	else if(memcmp(buf, "PK\x03\x04", 4) == 0)
+		typ = PAGE;
+	else if(memcmp(buf, "%PDF-", 5) == 0 || strstr(buf, "%!"))
+		typ = PAGE;
+	else if(memcmp(buf, "x T ", 4) == 0)
+		typ = PAGE;
+	else if(memcmp(buf, "\xF7\x02\x01\x83\x92\xC0\x1C;", 8) == 0)
+		typ = PAGE;
+	else if(memcmp(buf, "\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1", 8) == 0)
+		typ = PAGE;
+	else if(memcmp(buf, "\111\111\052\000", 4) == 0) 
+		typ = PAGE;
+	else if(memcmp(buf, "\115\115\000\052", 4) == 0)
+		typ = PAGE;
+	if(pipe(pfd) >= 0){
+		switch(rfork(RFFDG|RFPROC|RFNOWAIT)){
+		case -1:
+			break;
+		case 0:
+			close(pfd[0]);
+			do {
+				if(write(pfd[1], buf, n) != n)
+					break;
+			} while((n = read(fd, buf, sizeof(buf))) > 0);
+			exits(nil);
+		default:
+			dup(pfd[0], fd);
+		}
+		close(pfd[1]);
+		close(pfd[0]);
+	}
+	return typ;
+}