fix(form-data): Use GetDeflatedUTF8StringLength to calculate encoded length

Author: FreeMasen

builtins/web/form-data/form-data-encoder.cpp

@@ -9,6 +9,7 @@
 #include "../streams/native-stream-source.h"
 
 #include "encode.h"
+#include "jstypes.h"
 #include "mozilla/Assertions.h"
 #include "mozilla/ResultVariant.h"
 
@@ -42,6 +43,35 @@ size_t compute_normalized_len(std::string_view src) {
   return len;
 }
 
+size_t compute_unencoded_normalized_len(JSContext *cx, JS::HandleString value) {
+  auto linear = JS_EnsureLinearString(cx, value);
+  auto len = JS::GetDeflatedUTF8StringLength(linear);
+  size_t js_len = JS_GetStringLength(value);
+
+  for (size_t i = 0; i < js_len; i++) {
+    char16_t ch = 0;
+    if (!JS_GetStringCharAt(cx, value, i, &ch)) {
+      break;
+    }
+    if (ch == CR) {
+      if (i+1 < js_len) {
+        char16_t next = 0;
+        if (JS_GetStringCharAt(cx, value, i + 1, &next)) {
+          if (next == LF) {
+            i += 1;
+            // the character is already accounted for
+            continue;
+          }
+        }
+      }
+      len += 1;
+    } else if (ch == LF) {
+      len += 1; // CRLF
+    }
+  }
+  return len;
+}
+
 // Normalizes newlines in a string by replacing:
 // - CR not followed by LF -> CRLF
 // - LF not preceded by CR -> CRLF
@@ -517,12 +547,11 @@ mozilla::Result<size_t, OutOfMemory> MultipartFormDataImpl::query_length(JSConte
       total += 2 * crlf_len;
 
       RootedValue value_str(cx, entry.value);
-      auto value = core::encode(cx, value_str);
+      JS::RootedString value(cx, JS::ToString(cx, value_str));
       if (!value) {
         return mozilla::Result<size_t, OutOfMemory>(OutOfMemory {});
       }
-
-      total += compute_normalized_len(value);
+      total += compute_unencoded_normalized_len(cx, value);
     } else {
       MOZ_ASSERT(File::is_instance(entry.value));
       RootedObject obj(cx, &entry.value.toObject());

tests/integration/formData/formData.js

@@ -1,8 +1,6 @@
 import { serveTest } from "../test-server.js";
 import { strictEqual, assert } from "../../assert.js";
 
-
-
 export const handler = serveTest(async (t) => {
     await t.test("encode-form-data-simple", async () => {
         let fd = new FormData();
@@ -12,8 +10,6 @@ export const handler = serveTest(async (t) => {
             method: "POST",
             body: fd,
         });
-        let bytes = await req.text();
-        console.log(bytes);
         strictEqual(req.headers.get("content-length"), "250")
         assert(req.headers.get("content-type")?.startsWith("multipart/form-data;"));
     });
@@ -24,9 +20,6 @@ export const handler = serveTest(async (t) => {
             method: "POST",
             body: fd,
         });
-        "".charCodeAt
-        let bytes = await req.text();
-        console.log(bytes);
         strictEqual(req.headers.get("content-length"), "177")
         assert(req.headers.get("content-type")?.startsWith("multipart/form-data;"));
     });
@@ -38,8 +31,6 @@ export const handler = serveTest(async (t) => {
             method: "POST",
             body: fd,
         });
-        let bytes = await req.text();
-        console.log(bytes);
         strictEqual(req.headers.get("content-length"), "277")
         assert(req.headers.get("content-type")?.startsWith("multipart/form-data;"));
     });
@@ -55,9 +46,6 @@ export const handler = serveTest(async (t) => {
             method: "POST",
             body: fd,
         });
-        // ensure the body is encoded
-        await req.text();
-        // console.log(bytes);
         strictEqual(req.headers.get("content-length"), "1231")
         assert(req.headers.get("content-type")?.startsWith("multipart/form-data;"));
     });