fixup! buffer: add base64url encoding option

Backport parts of dae283d
nodejs · targos · Jan 15, 2021 · Aug 8, 2021 · Aug 13, 2021 · 84500373e53cc176fa61cdb559c401f45d93110c
commit 84500373e53cc176fa61cdb559c401f45d93110c
diff --git a/src/base64-inl.h b/src/base64-inl.h
@@ -123,12 +123,13 @@ size_t base64_decode(char* const dst, const size_t dstlen,
 inline size_t base64_encode(const char* src,
                             size_t slen,
                             char* dst,
-                            size_t dlen) {
+                            size_t dlen,
+                            Base64Mode mode) {
   // We know how much we'll write, just make sure that there's space.
-  CHECK(dlen >= base64_encoded_size(slen) &&
+  CHECK(dlen >= base64_encoded_size(slen, mode) &&
         "not enough space provided for base64 encode");
 
-  dlen = base64_encoded_size(slen);
+  dlen = base64_encoded_size(slen, mode);
 
   unsigned a;
   unsigned b;
@@ -137,9 +138,7 @@ inline size_t base64_encode(const char* src,
   unsigned k;
   unsigned n;
 
-  static const char table[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-                              "abcdefghijklmnopqrstuvwxyz"
-                              "0123456789+/";
+  const char* table = base64_select_table(mode);
 
   i = 0;
   k = 0;
@@ -164,16 +163,19 @@ inline size_t base64_encode(const char* src,
       a = src[i + 0] & 0xff;
       dst[k + 0] = table[a >> 2];
       dst[k + 1] = table[(a & 3) << 4];
-      dst[k + 2] = '=';
-      dst[k + 3] = '=';
+      if (mode == Base64Mode::NORMAL) {
+        dst[k + 2] = '=';
+        dst[k + 3] = '=';
+      }
       break;
     case 2:
       a = src[i + 0] & 0xff;
       b = src[i + 1] & 0xff;
       dst[k + 0] = table[a >> 2];
       dst[k + 1] = table[((a & 3) << 4) | (b >> 4)];
       dst[k + 2] = table[(b & 0x0f) << 2];
-      dst[k + 3] = '=';
+      if (mode == Base64Mode::NORMAL)
+        dst[k + 3] = '=';
       break;
   }
 

diff --git a/src/base64.h b/src/base64.h
@@ -5,13 +5,40 @@
 
 #include "util.h"
 
+#include <cmath>
 #include <cstddef>
 #include <cstdint>
 
 namespace node {
 //// Base 64 ////
-static inline constexpr size_t base64_encoded_size(size_t size) {
-  return ((size + 2) / 3 * 4);
+
+enum class Base64Mode {
+  NORMAL,
+  URL
+};
+
+static constexpr char base64_table[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+                                       "abcdefghijklmnopqrstuvwxyz"
+                                       "0123456789+/";
+
+static constexpr char base64_table_url[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+                                           "abcdefghijklmnopqrstuvwxyz"
+                                           "0123456789-_";
+
+static inline const char* base64_select_table(Base64Mode mode) {
+  switch (mode) {
+    case Base64Mode::NORMAL: return base64_table;
+    case Base64Mode::URL: return base64_table_url;
+    default: UNREACHABLE();
+  }
+}
+
+static inline constexpr size_t base64_encoded_size(
+    size_t size,
+    Base64Mode mode = Base64Mode::NORMAL) {
+  return mode == Base64Mode::NORMAL
+      ? ((size + 2) / 3 * 4)
+      : std::ceil(static_cast<double>(size * 4) / 3);
 }
 
 // Doesn't check for padding at the end.  Can be 1-2 bytes over.
@@ -32,7 +59,8 @@ size_t base64_decode(char* const dst, const size_t dstlen,
 inline size_t base64_encode(const char* src,
                             size_t slen,
                             char* dst,
-                            size_t dlen);
+                            size_t dlen,
+                            Base64Mode mode = Base64Mode::NORMAL);
 }  // namespace node
 
 

diff --git a/src/node.h b/src/node.h
@@ -664,7 +664,18 @@ inline void NODE_SET_PROTOTYPE_METHOD(v8::Local<v8::FunctionTemplate> recv,
 #define NODE_SET_PROTOTYPE_METHOD node::NODE_SET_PROTOTYPE_METHOD
 
 // BINARY is a deprecated alias of LATIN1.
-enum encoding {ASCII, UTF8, BASE64, UCS2, BINARY, HEX, BUFFER, LATIN1 = BINARY};
+// BASE64URL is not currently exposed to the JavaScript side.
+enum encoding {
+  ASCII,
+  UTF8,
+  BASE64,
+  UCS2,
+  BINARY,
+  HEX,
+  BUFFER,
+  BASE64URL,
+  LATIN1 = BINARY
+};
 
 NODE_EXTERN enum encoding ParseEncoding(
     v8::Isolate* isolate,

diff --git a/src/string_bytes.cc b/src/string_bytes.cc
@@ -358,6 +358,8 @@ size_t StringBytes::Write(Isolate* isolate,
       break;
     }
 
+    case BASE64URL:
+      // Fall through
     case BASE64:
       if (str->IsExternalOneByte()) {
         auto ext = str->GetExternalOneByteStringResource();
@@ -425,6 +427,8 @@ Maybe<size_t> StringBytes::StorageSize(Isolate* isolate,
       data_size = str->Length() * sizeof(uint16_t);
       break;
 
+    case BASE64URL:
+      // Fall through
     case BASE64:
       data_size = base64_decoded_size_fast(str->Length());
       break;
@@ -466,6 +470,8 @@ Maybe<size_t> StringBytes::Size(Isolate* isolate,
     case UCS2:
       return Just(str->Length() * sizeof(uint16_t));
 
+    case BASE64URL:
+      // Fall through
     case BASE64: {
       String::Value value(isolate, str);
       return Just(base64_decoded_size(*value, value.length()));
@@ -691,6 +697,20 @@ MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
       return ExternOneByteString::New(isolate, dst, dlen, error);
     }
 
+    case BASE64URL: {
+      size_t dlen = base64_encoded_size(buflen, Base64Mode::URL);
+      char* dst = node::UncheckedMalloc(dlen);
+      if (dst == nullptr) {
+        *error = node::ERR_MEMORY_ALLOCATION_FAILED(isolate);
+        return MaybeLocal<Value>();
+      }
+
+      size_t written = base64_encode(buf, buflen, dst, dlen, Base64Mode::URL);
+      CHECK_EQ(written, dlen);
+
+      return ExternOneByteString::New(isolate, dst, dlen, error);
+    }
+
     case HEX: {
       size_t dlen = buflen * 2;
       char* dst = node::UncheckedMalloc(dlen);

diff --git a/test/cctest/test_base64.cc b/test/cctest/test_base64.cc
@@ -44,6 +44,20 @@ TEST(Base64Test, Encode) {
        "IGRlc2VydW50IG1vbGxpdCBhbmltIGlkIGVzdCBsYWJvcnVtLg==");
 }
 
+TEST(Base64Test, EncodeURL) {
+  auto test = [](const char* string, const char* base64_string) {
+    const size_t len = strlen(base64_string);
+    char* const buffer = new char[len + 1];
+    buffer[len] = 0;
+    base64_encode(string, strlen(string), buffer, len, node::Base64Mode::URL);
+    EXPECT_STREQ(base64_string, buffer);
+    delete[] buffer;
+  };
+
+  test("\x68\xd9\x16\x25\x5c\x1e\x40\x92\x2d\xfb", "aNkWJVweQJIt-w");
+  test("\xac\xc7\x93\xaa\x83\x6f\xc3\xe3\x3f\x75", "rMeTqoNvw-M_dQ");
+}
+
 TEST(Base64Test, Decode) {
   auto test = [](const char* base64_string, const char* string) {
     const size_t len = strlen(string);
@@ -75,6 +89,7 @@ TEST(Base64Test, Decode) {
   test("YWJj ZGVm", "abcdef");
   test("Y W J j Z G V m", "abcdef");
   test("Y   W\n JjZ \nG Vm", "abcdef");
+  test("rMeTqoNvw-M_dQ", "\xac\xc7\x93\xaa\x83\x6f\xc3\xe3\x3f\x75");
 
   const char* text =
       "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do "

diff --git a/test/parallel/test-buffer-alloc.js b/test/parallel/test-buffer-alloc.js
@@ -345,7 +345,7 @@ const base64flavors = ['base64', 'base64url'];
   assert.strictEqual(Buffer.from(quote).toString('base64'), expected);
   assert.strictEqual(
     Buffer.from(quote).toString('base64url'),
-    expected.replaceAll('+', '-').replaceAll('/', '_').replaceAll('=', '')
+    expected.replace(/\+/g, '-').replace(/\//g, '_').replace(/=/g, '')
   );
 
   base64flavors.forEach((encoding) => {