hurl

commit 85d6ee233ea16724799840840cb19c955cc0842f
parent a931b2c32a110fc26258afe16cc77d1c7f5d1e44
Author: Hiltjo Posthuma <hiltjo@codemadness.org>
Date:   Mon,  9 Aug 2021 18:54:46 +0200

check received content length if the Content-Length header is set

This prevents corrupted downloads, noticed when my connection was crappy.

Diffstat:
M
hurl.c
 | 
58
++++++++++++++++++++++++++++++++++++++++++++++++++--------

1 file changed, 50 insertions(+), 8 deletions(-)
diff --git a/hurl.c b/hurl.c
@@ -64,6 +64,32 @@ sighandler(int signo)
 }
 
 int
+parse_content_length(const char *s, size_t *length)
+{
+	const char *p;
+	char *end;
+	long long l;
+
+	if (!(p = strcasestr(s, "\r\nContent-Length:")))
+		return -1;
+
+	p += sizeof("\r\nContent-Length:") - 1;
+	p += strspn(p, " \t");
+
+	if (!isdigit(*p))
+		return -1;
+
+	errno = 0;
+	l = strtoll(p, &end, 10);
+	if (errno || p == end || (*end != '\0' && *end != '\r') || l < 0)
+		return -1;
+
+	*length = l;
+
+	return 0;
+}
+
+int
 uri_parse(const char *s, struct uri *u)
 {
 	const char *p = s;
@@ -221,9 +247,9 @@ https_request(void)
 	struct tls *t;
 	char buf[READ_BUF_SIZ], *p;
 	const char *errstr;
-	size_t n, len;
+	size_t bodylen, expectedlen, n, len;
 	ssize_t r;
-	int fd = -1, httpok = 0, ret = 1, stdport;
+	int cs, fd = -1, httpok = 0, ret = 1, stdport;
 
 	if (pledge("stdio dns inet rpath unveil", NULL) == -1)
 		err(1, "pledge");
@@ -308,7 +334,9 @@ https_request(void)
 		goto err;
 	}
 	*p = '\0'; /* NUL terminate header part */
+	cs = parse_content_length(buf, &expectedlen);
 	p += strlen("\r\n\r\n");
+	bodylen = strlen(p); /* (partial) body after header */
 
 	if (httpok) {
 		n = len - (p - buf);
@@ -336,6 +364,7 @@ https_request(void)
 			goto err;
 		}
 		len += r;
+		bodylen += r;
 
 		if (httpok) {
 			r = fwrite(buf, 1, r, stdout);
@@ -349,10 +378,15 @@ https_request(void)
 			break;
 	}
 	if (config_maxresponsesiz && len >= config_maxresponsesiz) {
-		fprintf(stderr, "tls_read: response too big: %zu >= %zu\n",
+		fprintf(stderr, "response too big: %zu >= %zu\n",
 		        len, config_maxresponsesiz);
 		goto err;
 	}
+	if (cs != -1 && expectedlen != bodylen) {
+		fprintf(stderr, "Content-Length mismatch: %zu expected != %zu received\n",
+		        expectedlen, bodylen);
+		goto err;
+	}
 	ret = 0;
 
 err:
@@ -368,9 +402,9 @@ int
 http_request(void)
 {
 	char buf[READ_BUF_SIZ], *p;
-	size_t n, len;
+	size_t bodylen, expectedlen, n, len;
 	ssize_t r;
-	int fd = -1, httpok = 0, ret = 1, stdport;
+	int cs, fd = -1, httpok = 0, ret = 1, stdport;
 
 	if (pledge("stdio dns inet", NULL) == -1)
 		err(1, "pledge");
@@ -426,7 +460,9 @@ http_request(void)
 		goto err;
 	}
 	*p = '\0'; /* NUL terminate header part */
+	cs = parse_content_length(buf, &expectedlen);
 	p += strlen("\r\n\r\n");
+	bodylen = strlen(p); /* (partial) body after header */
 
 	if (httpok) {
 		n = len - (p - buf);
@@ -451,6 +487,7 @@ http_request(void)
 			goto err;
 		}
 		len += r;
+		bodylen += r;
 
 		if (httpok) {
 			r = fwrite(buf, 1, r, stdout);
@@ -464,10 +501,15 @@ http_request(void)
 			break;
 	}
 	if (config_maxresponsesiz && len >= config_maxresponsesiz) {
-		fprintf(stderr, "read: response too big: %zu >= %zu\n",
+		fprintf(stderr, "response too big: %zu >= %zu\n",
 		        len, config_maxresponsesiz);
 		goto err;
 	}
+	if (cs != -1 && expectedlen != bodylen) {
+		fprintf(stderr, "Content-Length mismatch: %zu expected != %zu received\n",
+		        expectedlen, bodylen);
+		goto err;
+	}
 	ret = 0;
 
 err:
@@ -534,7 +576,7 @@ gopher_request(void)
 			break;
 	}
 	if (config_maxresponsesiz && len >= config_maxresponsesiz) {
-		fprintf(stderr, "read: response too big: %zu >= %zu\n",
+		fprintf(stderr, "response too big: %zu >= %zu\n",
 		        len, config_maxresponsesiz);
 		goto err;
 	}
@@ -633,7 +675,7 @@ gophers_request(void)
 			break;
 	}
 	if (config_maxresponsesiz && len >= config_maxresponsesiz) {
-		fprintf(stderr, "read: response too big: %zu >= %zu\n",
+		fprintf(stderr, "response too big: %zu >= %zu\n",
 		        len, config_maxresponsesiz);
 		goto err;
 	}