fix base64 decode bug (#3337)

crystal-lang · Sep 22, 2016 · 1fff36d · 1fff36d
1 parent d657b9a
commit 1fff36d
Showing 2 changed files with 24 additions and 2 deletions.
diff --git a/spec/std/base64_spec.cr b/spec/std/base64_spec.cr
@@ -3,7 +3,7 @@ require "base64"
 require "crypto/md5"
 
 describe "Base64" do
-  it "simple test" do
+  context "simple test" do
     eqs = {"" => "", "a" => "YQ==\n", "ab" => "YWI=\n", "abc" => "YWJj\n",
       "abcd" => "YWJjZA==\n", "abcde" => "YWJjZGU=\n", "abcdef" => "YWJjZGVm\n",
       "abcdefg" => "YWJjZGVmZw==\n"}
@@ -18,6 +18,18 @@ describe "Base64" do
     end
   end
 
+  context "\n in multiple places" do
+    eqs = {"abcd" => "YWJj\nZA==\n", "abcde" => "YWJj\nZGU=\n", "abcdef" => "YWJj\nZGVm\n",
+      "abcdefg" => "YWJj\nZGVmZw==\n", "abcdefg" => "YWJj\nZGVm\nZw==\n",
+    }
+    eqs.each do |a, b|
+      it "decode from #{b.inspect} to #{a.inspect}" do
+        Base64.decode(b).should eq(a.to_slice)
+        Base64.decode_string(b).should eq(a)
+      end
+    end
+  end
+
   it "encodes byte slice" do
     slice = Slice(UInt8).new(5) { 1_u8 }
     Base64.encode(slice).should eq("AQEBAQE=\n")
@@ -130,6 +142,11 @@ describe "Base64" do
         Base64.decode_string("a")
       end
     end
+
+    it "decode small tail after last \n, was a bug" do
+      s = "Tm93IGlzIHRoZSB0aW1lIGZvciBhbGwgZ29vZCBjb2RlcnMKdG8gbGVhcm4g\nnA==\n"
+      Base64.decode(s).should eq Bytes[78, 111, 119, 32, 105, 115, 32, 116, 104, 101, 32, 116, 105, 109, 101, 32, 102, 111, 114, 32, 97, 108, 108, 32, 103, 111, 111, 100, 32, 99, 111, 100, 101, 114, 115, 10, 116, 111, 32, 108, 101, 97, 114, 110, 32, 156]
+    end
   end
 
   describe "scrict" do

diff --git a/src/base64.cr b/src/base64.cr
@@ -237,6 +237,7 @@ module Base64
     size = data.size
     dt = DECODE_TABLE.to_unsafe
     cstr = data.pointer(size)
+    start_cstr = cstr
     while (size > 0) && (sym = cstr[size - 1]) && (sym == NL || sym == NR || sym == PAD)
       size -= 1
     end
@@ -256,6 +257,10 @@ module Base64
       yield (c << 6 | d).to_u8
     end
 
+    while (cstr < endcstr + 4) && (cstr.value == NL || cstr.value == NR)
+      cstr += 1
+    end
+
     mod = (endcstr - cstr) % 4
     if mod == 2
       a, b = next_decoded_value, next_decoded_value
@@ -275,7 +280,7 @@ module Base64
     res = dt[sym]
     cstr += 1
     if res < 0
-      raise Error.new("Unexpected symbol '#{sym.chr}'")
+      raise Error.new("Unexpected byte 0x#{sym.to_s(16)} at #{cstr - start_cstr - 1}")
     end
     res
   end