Add base64url_encode()

2 files changed, 86 insertions, 9 deletions

diff --git a/libbutl/base64.cxx b/libbutl/base64.cxx
index 4466f24..282f7c2 100644
--- a/libbutl/base64.cxx
+++ b/libbutl/base64.cxx
@@ -16,19 +16,20 @@ namespace butl
   static const char codes[] =
     "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
+  static const char codes_url[] =
+    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_";
+
   // base64-encode the data in the iterator range [i, e). Write the encoded
-  // data starting at the iterator position o.
+  // data starting at the iterator position o. If url is true, encode using
+  // base64url.
   //
   template <typename I, typename O>
   static void
-  base64_encode (I& i, const I& e, O& o)
+  base64_encode (I& i, const I& e, O& o, bool url = false)
   {
     const size_t un (65); // Non-existing index of the codes string.
     for (size_t n (0); i != e; ++n)
     {
-      if (n && n % 19 == 0)
-        *o++ = '\n'; // Split into lines, like the base64 utility does.
-
       auto next = [&i] () {return static_cast<unsigned char> (*i++);};
 
       unsigned char c (next ());
@@ -51,10 +52,26 @@ namespace butl
         i4 = c & 0x3F;
       }
 
-      *o++ = codes[i1];
-      *o++ = codes[i2];
-      *o++ = i3 == un ? '=' : codes[i3];
-      *o++ = i4 == un ? '=' : codes[i4];
+      if (!url)
+      {
+        if (n && n % 19 == 0)
+          *o++ = '\n'; // Split into lines, like the base64 utility does.
+
+        *o++ = codes[i1];
+        *o++ = codes[i2];
+        *o++ = i3 == un ? '=' : codes[i3];
+        *o++ = i4 == un ? '=' : codes[i4];
+      }
+      // base64url: different 63rd and 64th characters and no padding or
+      // newlines.
+      //
+      else
+      {
+        *o++ = codes_url[i1];
+        *o++ = codes_url[i2];
+        if (i3 != un) *o++ = codes_url[i3];
+        if (i4 != un) *o++ = codes_url[i4];
+      }
     }
   }
 
@@ -170,6 +187,47 @@ namespace butl
     return r;
   }
 
+  string
+  base64url_encode (istream& is)
+  {
+    if (!is.good ())
+      throw invalid_argument ("bad stream");
+
+    string r;
+    istreambuf_iterator<char> i (is);
+    back_insert_iterator<string> o (r);
+
+    base64_encode (i, istreambuf_iterator<char> (), o, true /* url */);
+    is.setstate (istream::eofbit);
+    return r;
+  }
+
+  void
+  base64url_encode (ostream& os, istream& is)
+  {
+    if (!os.good () || !is.good ())
+      throw invalid_argument ("bad stream");
+
+    istreambuf_iterator<char> i (is);
+    ostreambuf_iterator<char> o (os);
+    base64_encode (i, istreambuf_iterator<char> (), o, true /* url */);
+
+    if (o.failed ())
+      os.setstate (istream::badbit);
+
+    is.setstate (istream::eofbit);
+  }
+
+  string
+  base64url_encode (const std::vector<char>& v)
+  {
+    string r;
+    back_insert_iterator<string> o (r);
+    auto i (v.begin ());
+    base64_encode (i, v.end (), o, true /* url */);
+    return r;
+  }
+
   void
   base64_decode (ostream& os, istream& is)
   {
diff --git a/libbutl/base64.hxx b/libbutl/base64.hxx
index f38e62f..a0d1450 100644
--- a/libbutl/base64.hxx
+++ b/libbutl/base64.hxx
@@ -27,6 +27,25 @@ namespace butl
   LIBBUTL_SYMEXPORT std::string
   base64_encode (const std::vector<char>&);
 
+  // Encode a stream or a buffer using base64url (RFC4648), a base64 variant
+  // with different 62nd and 63rd alphabet characters (- and _ instead of ~
+  // and .; to make it filesystem safe) and optional padding because the
+  // padding character `=` would have to be percent-encoded to be safe in
+  // URLs. This implementation does not output any padding, newlines or any
+  // other whitespace (which is required, for example, by RFC7519: JSON Web
+  // Token (JWT) and RFC7515: JSON Web Signature (JWS)).
+  //
+  // Note that base64url decoding has not yet been implemented.
+  //
+  LIBBUTL_SYMEXPORT void
+  base64url_encode (std::ostream&, std::istream&);
+
+  LIBBUTL_SYMEXPORT std::string
+  base64url_encode (std::istream&);
+
+  LIBBUTL_SYMEXPORT std::string
+  base64url_encode (const std::vector<char>&);
+
   // Base64-decode a stream or a string. Throw invalid_argument if the input
   // is not a valid base64 representation. If reading from a stream, check if
   // it has badbit, failbit, or eofbit set and throw invalid_argument if