For functions that write to an output buffer, ignore the buffer pointer if the passed capacity is set to 0

README.md

@@ -1,5 +1,5 @@
 # url.c
-This is a small library to parse and manipulate URLs in conformance to RFC 3986 and WHATWG.
+This is a small library to parse and manipulate URLs in conformance to RFC 3986 and (most of) the WHATWG specification.
 
 It features
 * No allocations
@@ -15,6 +15,4 @@ You just copy paste `url.h` and `url.c` in your source tree and compile them as
 Some example programs are provided in the `examples/` folder. To compile them, run `build.bat` on Windows or `build.sh` on Linux.
 
 # Testing
-The parser and reference resolver is tested against the URL section of the [web platform tests](https://github.com/web-platform-tests/wpt). Note that not all tests pass. This is due to the parser not doing any normalization and not supporting more niche features such as converting `\` characters to `/` or supporting drive letters like `C:` in the URL's path.
-
-To run the test suite, `cd` into `tests/` and compile the test runner using `build.sh` (Linux) or `build.bat` (Windows). Then, run `./test.out` (Linux) or `./test.exe` (Windows).
+The parser and reference resolver is tested against the URL section of the [web platform tests](https://github.com/web-platform-tests/wpt), which validates implementations that adhere to the WHATWG specification (what the major browsers actually do). To run the test suite, `cd` into `tests/` and compile the test runner using `build.sh` (Linux) or `build.bat` (Windows). Then, run `./test.out` (Linux) or `./test.exe` (Windows). Note that not all tests pass, which is to be expected as adherence to the WHATWG specification is a work-in-progress.

tests/test.c

@@ -31,6 +31,10 @@ void test_printf(char *fmt, ...)
         sizeof(test_output_buffer) - test_output_buffer_used,
         fmt, args);
     va_end(args);
+
+    assert(len > -1);
+    if (len > (int) sizeof(test_output_buffer) - test_output_buffer_used)
+        len = sizeof(test_output_buffer) - test_output_buffer_used;
     test_output_buffer_used += len;
 }
 
@@ -198,15 +202,21 @@ static int json_to_url(cJSON *json, URL *url)
         url->fragment.len--;
     }
 
-    char tmp[8];
-    if (port.len >= (int) sizeof(tmp)) {
-        test_printf(" JSON object's port field is longer than expected\n");
-        return -1;
-    }
-    memcpy(tmp, port.ptr, port.len);
-    tmp[port.len] = 0;
+    if (port.len == 0) {
+        url->no_port = 1;
+        url->port = 0;
+    } else {
+        char tmp[8];
+        if (port.len >= (int) sizeof(tmp)) {
+            test_printf(" JSON object's port field is longer than expected\n");
+            return -1;
+        }
+        memcpy(tmp, port.ptr, port.len);
+        tmp[port.len] = 0;
 
-    url->port = atoi(tmp);
+        url->no_port = 0;
+        url->port = atoi(tmp);
+    }
     return 0;
 }
 
@@ -217,73 +227,103 @@ static bool streq(URL_String a, URL_String b)
     return !memcmp(a.ptr, b.ptr, a.len);
 }
 
-static int compare_urls(URL url_1, URL url_2, char *label_1, char *label_2)
+static int compare_urls(URL parsed, URL expected)
 {
     #define UNPACK(s) (s).len, (s).ptr
 
     bool ok = true;
 
-    if (!streq(url_1.scheme, url_2.scheme)) {
+    if (!streq(parsed.scheme, expected.scheme)) {
         if (ok) test_printf("\n");
         test_printf("  scheme mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.scheme));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.scheme));
+        test_printf("    parsed   [%.*s]\n", UNPACK(parsed.scheme));
+        test_printf("    expected [%.*s]\n", UNPACK(expected.scheme));
         ok = false;
     }
 
-    if (!streq(url_1.username, url_2.username)) {
+    char tmp[1<<9];
+    int ret = url_percent_decode(parsed.username, tmp, sizeof(tmp));
+    assert(ret > -1 && ret < (int) sizeof(tmp));
+
+    if (!streq((URL_String) { tmp, ret }, expected.username)) {
         if (ok) test_printf("\n");
         test_printf("  username mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.username));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.username));
+        test_printf("    parsed   [%.*s]\n", ret, tmp);
+        test_printf("    expected [%.*s]\n", UNPACK(expected.username));
         ok = false;
     }
 
-    if (!streq(url_1.password, url_2.password)) {
+    ret = url_percent_decode(parsed.password, tmp, sizeof(tmp));
+    assert(ret > -1 && ret < (int) sizeof(tmp));
+
+    if (!streq((URL_String) { tmp, ret }, expected.password)) {
         if (ok) test_printf("\n");
         test_printf("  password mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.password));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.password));
+        test_printf("    parsed   [%.*s]\n", ret, tmp);
+        test_printf("    expected [%.*s]\n", UNPACK(expected.password));
         ok = false;
     }
 
-    if (!streq(url_1.host_text, url_2.host_text)) {
+    ret = url_percent_decode(parsed.host_text, tmp, sizeof(tmp));
+    assert(ret > -1 && ret < (int) sizeof(tmp));
+
+    if (!streq((URL_String) { tmp, ret }, expected.host_text)) {
         if (ok) test_printf("\n");
         test_printf("  host mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.host_text));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.host_text));
+        test_printf("    parsed   [%.*s]\n", ret, tmp);
+        test_printf("    expected [%.*s]\n", UNPACK(expected.host_text));
         ok = false;
     }
 
-    if (url_1.port != url_2.port) {
+    if (parsed.no_port != expected.no_port) {
+        if (ok) test_printf("\n");
+        if (parsed.no_port) {
+            test_printf("  port mismatch\n");
+            test_printf("    parsed   (empty)\n");
+            test_printf("    expected [%d]\n", expected.port);
+        } else {
+            test_printf("  port mismatch\n");
+            test_printf("    parsed   [%d]\n", parsed.port);
+            test_printf("    expected (empty)\n");
+        }
+        ok = false;
+    }
+
+    if (parsed.port != expected.port) {
         if (ok) test_printf("\n");
         test_printf("  port mismatch\n");
-        test_printf("    %s [%d]\n", label_1, url_1.port);
-        test_printf("    %s [%d]\n", label_2, url_2.port);
+        test_printf("    parsed   [%d]\n", parsed.port);
+        test_printf("    expected [%d]\n", expected.port);
         ok = false;
     }
 
-    if (!streq(url_1.path, url_2.path)) {
+    if (!streq(parsed.path, expected.path)) {
         if (ok) test_printf("\n");
         test_printf("  path mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.path));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.path));
+        test_printf("    parsed   [%.*s]\n", UNPACK(parsed.path));
+        test_printf("    expected [%.*s]\n", UNPACK(expected.path));
         ok = false;
     }
 
-    if (!streq(url_1.query, url_2.query)) {
+    ret = url_percent_decode(parsed.query, tmp, sizeof(tmp));
+    assert(ret > -1 && ret < (int) sizeof(tmp));
+
+    if (!streq((URL_String) { tmp, ret }, expected.query)) {
         if (ok) test_printf("\n");
         test_printf("  query mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.query));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.query));
+        test_printf("    parsed   [%.*s]\n", ret, tmp);
+        test_printf("    expected [%.*s]\n", UNPACK(expected.query));
         ok = false;
     }
 
-    if (!streq(url_1.fragment, url_2.fragment)) {
+    ret = url_percent_decode(parsed.fragment, tmp, sizeof(tmp));
+    assert(ret > -1 && ret < (int) sizeof(tmp));
+
+    if (!streq((URL_String) { tmp, ret }, expected.fragment)) {
         if (ok) test_printf("\n");
         test_printf("  fragment mismatch\n");
-        test_printf("    %s [%.*s]\n", label_1, UNPACK(url_1.fragment));
-        test_printf("    %s [%.*s]\n", label_2, UNPACK(url_2.fragment));
+        test_printf("    parsed   [%.*s]\n", ret, tmp);
+        test_printf("    expected [%.*s]\n", UNPACK(expected.fragment));
         ok = false;
     }
 
@@ -394,7 +434,7 @@ static int run_test(cJSON *json)
         if (json_to_url(json, &expected) < 0)
             return -1;
 
-        if (!compare_urls(parsed, expected, "parsed", "expected")) {
+        if (!compare_urls(parsed, expected)) {
             test_printf("  input: ");
             test_print_str(src);
             test_printf("\n");
@@ -472,6 +512,7 @@ int main(void)
 
     if (!cJSON_IsArray(root)) {
         fprintf(stderr, "Error: %s doesn't contain a JSON array as expected", path);
+        cJSON_Delete(root);
         free(buf);
         return -1;
     }
@@ -524,6 +565,7 @@ int main(void)
     printf("  failed : %d\n", num_failed);
     printf("  aborted: %d\n", num_aborted);
 
+    cJSON_Delete(root);
     free(buf);
     return 0;
 }

url.c

@@ -524,9 +524,20 @@ int url_parse(char *src, int len, int *pcur, URL *out, int flags)
             // First, try the IPv4 rule
             if (cur < len && is_digit(src[cur])) {
                 if (url_parse_ipv4(src, len, &cur, &out->host_ipv4) == 0) {
-                    out->host_type = URL_HOST_IPV4;
-                    is_ipv4 = 1;
+                    // If we managed to parse an IPv4 address but the
+                    // following character si a valid character for a
+                    // registered name, then it wasn't an IPv4 after
+                    // all. For instance this should not be parsed as
+                    // an IPv4:
+                    //     127.0.0.1.com
+                    if (cur == len || !is_reg_name(src[cur])) {
+                        out->host_type = URL_HOST_IPV4;
+                        is_ipv4 = 1;
+                    }
                 }
+
+                if (!is_ipv4)
+                    cur = host_off;
             }
 
             if (!is_ipv4) {
@@ -569,25 +580,26 @@ int url_parse(char *src, int len, int *pcur, URL *out, int flags)
         out->no_port = 1;
         out->port    = 0;
         if (cur < len && src[cur] == ':') {
-
-            // The WHATWG standard forbids port numbers with the
-            // "file" protocol.
-            if ((flags & URL_FLAG_RFC3986) == 0) {
-                if (streqcase(out->scheme, S("file")))
-                    return -1;
-            }
-
             cur++; // Consume the ':'
-            out->no_port = 0;
 
-            while (cur < len && is_digit(src[cur])) {
+            if (cur < len && is_digit(src[cur])) {
 
-                int n = src[cur] - '0';
-                cur++;
+                // The WHATWG standard forbids port numbers with the
+                // "file" protocol.
+                if ((flags & URL_FLAG_RFC3986) == 0) {
+                    if (streqcase(out->scheme, S("file")))
+                        return -1;
+                }
 
-                if (out->port > (UINT16_MAX - n) / 10)
-                    return -1; // Overflow
-                out->port = out->port * 10 + n;
+                out->no_port = 0;
+                do {
+                    int n = src[cur] - '0';
+                    cur++;
+
+                    if (out->port > (UINT16_MAX - n) / 10)
+                        return -1; // Overflow
+                    out->port = out->port * 10 + n;
+                } while (cur < len && is_digit(src[cur]));
             }
         }
 
@@ -845,6 +857,8 @@ static void append_fragment(Builder *b, URL_String fragment)
 
 int url_serialize(URL url, URL *base, char *dst, int cap)
 {
+    ASSERT(cap == 0 || dst != NULL);
+
     if (base != NULL && base->scheme.len == 0)
         return -1; // Base is not an absolute URL
 
@@ -948,6 +962,8 @@ static URL_b32 is_white_space(char c)
 
 int url_remove_white_space(char *src, int len, char *dst, int cap)
 {
+    ASSERT(cap == 0 || dst != NULL);
+
     while (len > 0 && is_white_space(src[0])) {
         src++;
         len--;
@@ -970,6 +986,8 @@ int url_remove_white_space(char *src, int len, char *dst, int cap)
 
 int url_percent_decode(URL_String str, char *dst, int cap)
 {
+    ASSERT(cap == 0 || dst != NULL);
+
     char *src = str.ptr;
     int   len = str.len;
     int   rd  = 0;