an attempt at better file parsing

src/http.c

@@ -300,12 +300,12 @@
     clt->src->cancel(clt->src);
 }
 
 static void req_write_cb(uv_link_t *source, int status, void *arg) {
     UM_LOG(VERB, "request write completed: %d", status);
     free(arg);
 }
 
 static void req_write_body_cb(uv_link_t *source, int status, void *arg) {
     UM_LOG(VERB, "request body write completed: %d", status);
     struct body_chunk_s *chunk = arg;
     if (chunk->cb) {
@@ -314,7 +314,7 @@
     free(chunk);
 }
 
 static void chunk_hdr_wcb(uv_link_t *l, int status, void *arg) {
     if (arg != NULL) {
         free(arg);
     }
@@ -340,7 +340,7 @@
                 uv_link_write((uv_link_t *) &clt->http_link, &buf, 1, NULL, chunk_hdr_wcb, buf.base);
 
                 buf.base = (char*)b->chunk;
                 buf.len = b->len;
                 uv_link_write((uv_link_t *) &clt->http_link, &buf, 1, NULL, req_write_body_cb, b);
 
                 buf.base = "\r\n";
@@ -357,7 +357,7 @@
         else {
             buf = uv_buf_init((char*)b->chunk, (unsigned int)b->len);
             uv_link_write((uv_link_t *) &clt->http_link, &buf, 1, NULL, req_write_body_cb, b);
             if (req->body_sent_size > req->req_body_size) {
                 UM_LOG(WARN, "Supplied data[%ld] is larger than provided Content-Length[%ld]",
                         req->body_sent_size, req->req_body_size);
             }
@@ -779,15 +779,64 @@
     tlsuv_tls_link_free(&clt->tls_link);
 }
 
+
 int tlsuv_parse_url(struct tlsuv_url_s *url, const char *urlstr) {
     memset(url, 0, sizeof(struct tlsuv_url_s));
 
     const char *p = urlstr;
+
+    int file_prefix_len = strlen("file:/");
+    // special handling for file:/, file://, file://host/, file:///
+    if (strncmp(urlstr, "file:/", file_prefix_len) == 0) {
+        url->scheme = p;
+        url->scheme_len = 4; // strlen("file")
+        p += file_prefix_len;
+
+        if (p[0] == '/') {
+            p++;
+            if (p[0] == '/') {
+                // file:/// means empty hostname
+                p++;
+            } else {
+                // file://path means there must be a hostname. find the next slash
+                char *pos = strchr(p, '/');
+                if (pos != NULL) {
+                    size_t index = pos - p;
+                    url->hostname = p;
+                    url->hostname_len = index;
+                    p += index + 1;
+                } else {
+                    // poorly formatted entry. this would be just `file://` or `file://hostnameonly`
+                    url->hostname = p;
+                    url->hostname_len = strlen(p);
+                    return -1;
+                }
+            }
+        } else {
+            //one slash - else empty on purpose to indicate this is expected to be no-op
+        }
+
+#ifdef _WIN32
+        if (strlen(p) > 0 && p[1] == ':') {
+            // expect a windows path to have a drive letter c:, d:, etc.
+        } else {
+            // if no ':' in position 2, back up to pickup the leading slash
+            p--;
+        }
+#else
+        p--; //on non-windows, always backup to pickup the leading slash
+#endif
+        url->path = p;
+        url->path_len = strlen(p);
+        return 0;
+    }
+
+
     int count = 0;
     int rc = sscanf(p, "%*[^:]%n://", &count);
     if (rc == 0 &&
         (p + count)[0] == ':' && (p + count)[1] == '/' && (p + count)[2] == '/'
-        ) {
+            ) {
         url->scheme = p;
         url->scheme_len = count;
         p += (count + 3);
@@ -826,7 +875,7 @@
             return -1;
 
         if (lport > 0 && lport <= UINT16_MAX) {
-            url->port = (uint16_t)lport;
+            url->port = (uint16_t) lport;
             p = pend;
         } else {
             return -1;

tests/Readme.md

@@ -15,7 +15,7 @@ The test server opens the following endpoints:
 Start test server:
 ```console
 cd ./tests/test_server
-go run ./test-server.go -keyfile ../certs/server.key -certfile ../certs/server.crt
+go run ./test-server.go -ca-key ../certs/server.key -ca ../certs/server.crt
 ```
 
 now you're ready to run tests

tests/http_tests.cpp

@@ -1121,13 +1121,28 @@ static void URL_TEST(const char *s, int RES, const char *SCHEME, const char *HOS
         CHECK(url.port == (PORT));
     }
 }
-
+#ifdef _WIN32
+#define LEADING_SLASH ""
+#else
+#define LEADING_SLASH "/"
+#endif
 TEST_CASE("url parse", "[http]") {
     URL_TEST("wss://websocket.org/echo?foo=bar", 0, "wss", "websocket.org", 0, "/echo", "foo=bar");
     URL_TEST("wss://websocket.org:443/echo?foo=bar", 0, "wss", "websocket.org", 443, "/echo", "foo=bar");
     URL_TEST("wss://websocket.org/echo", 0, "wss", "websocket.org", 0, "/echo", nullptr);
     URL_TEST("websocket.org:443/echo", 0, nullptr, "websocket.org", 443, "/echo", nullptr);
+
     URL_TEST("file:///echo.txt", 0, "file", nullptr, 0, "/echo.txt", nullptr);
+    URL_TEST("file:/echo.txt", 0, "file", nullptr, 0, "/echo.txt", nullptr);
+    URL_TEST("file://host/echo.txt", 0, "file", "host", 0, "/echo.txt", nullptr);
+    URL_TEST("file:///some/path/to/echo.txt", 0, "file", nullptr, 0, "/some/path/to/echo.txt", nullptr);
+    URL_TEST("file:///c:/echo.txt", 0, "file", nullptr, 0, LEADING_SLASH "c:/echo.txt", nullptr);
+    URL_TEST("file:///c:/some/other/path/to/echo.txt", 0, "file", nullptr, 0, LEADING_SLASH "c:/some/other/path/to/echo.txt", nullptr);
+
+    URL_TEST("file://", -1, "file", nullptr, 0, "/echo.txt", nullptr);
+    URL_TEST("file://hostnameonly", -1, "file", "hostnameonly", 0, nullptr, nullptr);
+    URL_TEST("file://hostnameonly/", 0, "file", "hostnameonly", 0, "/", nullptr);
+
     URL_TEST("/path/only/echo.txt", 0, nullptr, nullptr, 0, "/path/only/echo.txt", nullptr);
     URL_TEST(":443/echo", -1, nullptr, "websocket.org", 443, "/echo", nullptr);
     URL_TEST("websocket.org:443echo", -1, nullptr, "websocket.org", 443, "/echo", nullptr);