Merge pull request #1679 from lightpanda-io/escape_data_uri

Escape DataURIs

src/browser/ScriptManager.zig

@@ -1051,23 +1051,35 @@ fn parseDataURI(allocator: Allocator, src: []const u8) !?[]const u8 {
 
     const uri = src[5..];
     const data_starts = std.mem.indexOfScalar(u8, uri, ',') orelse return null;
+    const data = uri[data_starts + 1 ..];
 
-    var data = uri[data_starts + 1 ..];
+    const unescaped = try URL.unescape(allocator, data);
 
-    // Extract the encoding.
     const metadata = uri[0..data_starts];
-    if (std.mem.endsWith(u8, metadata, ";base64")) {
+    if (std.mem.endsWith(u8, metadata, ";base64") == false) {
-        const decoder = std.base64.standard.Decoder;
+        return unescaped;
-        const decoded_size = try decoder.calcSizeForSlice(data);
-
-        const buffer = try allocator.alloc(u8, decoded_size);
-        errdefer allocator.free(buffer);
-
-        try decoder.decode(buffer, data);
-        data = buffer;
     }
 
-    return data;
+    // Forgiving base64 decode per WHATWG spec:
+    // https://infra.spec.whatwg.org/#forgiving-base64-decode
+    // Step 1: Remove all ASCII whitespace
+    var stripped = try std.ArrayList(u8).initCapacity(allocator, unescaped.len);
+    for (unescaped) |c| {
+        if (!std.ascii.isWhitespace(c)) {
+            stripped.appendAssumeCapacity(c);
+        }
+    }
+    const trimmed = std.mem.trimRight(u8, stripped.items, "=");
+
+    // Length % 4 == 1 is invalid
+    if (trimmed.len % 4 == 1) {
+        return error.InvalidCharacterError;
+    }
+
+    const decoded_size = std.base64.standard_no_pad.Decoder.calcSizeForSlice(trimmed) catch return error.InvalidCharacterError;
+    const buffer = try allocator.alloc(u8, decoded_size);
+    std.base64.standard_no_pad.Decoder.decode(buffer, trimmed) catch return error.InvalidCharacterError;
+    return buffer;
 }
 
 const testing = @import("../testing.zig");

src/browser/URL.zig

@@ -642,6 +642,33 @@ pub fn getRobotsUrl(arena: Allocator, url: [:0]const u8) ![:0]const u8 {
     );
 }
 
+pub fn unescape(arena: Allocator, input: []const u8) ![]const u8 {
+    if (std.mem.indexOfScalar(u8, input, '%') == null) {
+        return input;
+    }
+
+    var result = try std.ArrayList(u8).initCapacity(arena, input.len);
+
+    var i: usize = 0;
+    while (i < input.len) {
+        if (input[i] == '%' and i + 2 < input.len) {
+            const hex = input[i + 1 .. i + 3];
+            const byte = std.fmt.parseInt(u8, hex, 16) catch {
+                result.appendAssumeCapacity(input[i]);
+                i += 1;
+                continue;
+            };
+            result.appendAssumeCapacity(byte);
+            i += 3;
+        } else {
+            result.appendAssumeCapacity(input[i]);
+            i += 1;
+        }
+    }
+
+    return result.items;
+}
+
 const testing = @import("../testing.zig");
 test "URL: isCompleteHTTPUrl" {
     try testing.expectEqual(true, isCompleteHTTPUrl("http://example.com/about"));
@@ -1233,3 +1260,68 @@ test "URL: getRobotsUrl" {
         try testing.expectString("https://example.com/robots.txt", url);
     }
 }
+
+test "URL: unescape" {
+    defer testing.reset();
+    const arena = testing.arena_allocator;
+
+    {
+        const result = try unescape(arena, "hello world");
+        try testing.expectEqual("hello world", result);
+    }
+
+    {
+        const result = try unescape(arena, "hello%20world");
+        try testing.expectEqual("hello world", result);
+    }
+
+    {
+        const result = try unescape(arena, "%48%65%6c%6c%6f");
+        try testing.expectEqual("Hello", result);
+    }
+
+    {
+        const result = try unescape(arena, "%48%65%6C%6C%6F");
+        try testing.expectEqual("Hello", result);
+    }
+
+    {
+        const result = try unescape(arena, "a%3Db");
+        try testing.expectEqual("a=b", result);
+    }
+
+    {
+        const result = try unescape(arena, "a%3DB");
+        try testing.expectEqual("a=B", result);
+    }
+
+    {
+        const result = try unescape(arena, "ZDIgPSAndHdvJzs%3D");
+        try testing.expectEqual("ZDIgPSAndHdvJzs=", result);
+    }
+
+    {
+        const result = try unescape(arena, "%5a%44%4d%67%50%53%41%6e%64%47%68%79%5a%57%55%6e%4f%77%3D%3D");
+        try testing.expectEqual("ZDMgPSAndGhyZWUnOw==", result);
+    }
+
+    {
+        const result = try unescape(arena, "hello%2world");
+        try testing.expectEqual("hello%2world", result);
+    }
+
+    {
+        const result = try unescape(arena, "hello%ZZworld");
+        try testing.expectEqual("hello%ZZworld", result);
+    }
+
+    {
+        const result = try unescape(arena, "hello%");
+        try testing.expectEqual("hello%", result);
+    }
+
+    {
+        const result = try unescape(arena, "hello%2");
+        try testing.expectEqual("hello%2", result);
+    }
+}

src/browser/tests/legacy/html/script/script.html

@@ -19,3 +19,13 @@
 </script>
 
 <script id=datauri src="data:text/plain;charset=utf-8;base64,dGVzdGluZy5leHBlY3RFcXVhbCh0cnVlLCB0cnVlKTs="></script>
+
+<script id=datauri_url_encoded_text src="data:text/javascript,testing.expectEqual(3, 3);"></script>
+
+<script id=datauri_encoded_padding src="data:text/javascript;base64,dGVzdGluZy5leHBlY3RFcXVhbCgxLCAxKTs%3D"></script>
+
+<script id=datauri_fully_encoded src="data:text/javascript;base64,%64%47%56%7a%64%47%6c%75%5a%79%35%6c%65%48%42%6c%59%33%52%46%63%58%56%68%62%43%67%79%4c%43%41%79%4b%54%73%3d"></script>
+
+<script id=datauri_with_whitespace src="data:text/javascript;base64,%20ZD%20Qg%0D%0APS%20An%20Zm91cic%0D%0A%207%20"></script>
+
+<script id=datauri_url_encoded_unicode src="data:text/javascript,testing.expectEqual(4%2C%204)%3B"></script>