Handle utf8 XML better, don't alias everything

* Don't make everything >126 into aliases; we still keep the
  same (slightly retarded) behaviour of making non-printables
  into aliases though, because xmerl can't tell us if they were
  in the document as an alias or a direct character

* Add some tests to cover this, as well as the 
 in the spec

src/xmerl_c14n.erl

@@ -115,7 +115,7 @@ xml_safe_string(Str, Quotes) when is_list(Str) ->
    [Next | Rest] = Str,
    if
       (not Quotes andalso ([Next] =:= "\n")) -> [Next | xml_safe_string(Rest, Quotes)];
-      (Next < 32) orelse (Next > 126) ->
+      (Next < 32) ->
          lists:flatten(["&#x" ++ integer_to_list(Next, 16) ++ ";" | xml_safe_string(Rest, Quotes)]);
       (Quotes andalso ([Next] =:= "\"")) -> lists:flatten(["&quot;" | xml_safe_string(Rest, Quotes)]);
       ([Next] =:= "&") -> lists:flatten(["&amp;" | xml_safe_string(Rest, Quotes)]);
@@ -276,7 +276,12 @@ xml_safe_string_test() ->
    "foo \ngeorge" = xml_safe_string(<<"foo \ngeorge">>),
    "foo &lt;&#x5;&gt; = &amp; help" = xml_safe_string(lists:flatten(["foo <", 5, "> = & help"])),
    "&#xE;" = xml_safe_string(<<14>>),
-   "\"foo\"" = xml_safe_string("\"foo\"").
+   "\"foo\"" = xml_safe_string("\"foo\""),
+   "test&#xD;\n" = xml_safe_string("test\r\n").
+
+xml_safe_string_utf8_test() ->
+   String = unicode:characters_to_list(<<"バカの名前">>),
+   String = xml_safe_string(String).
 
 c14n_3_1_test() ->
    {Doc, _} = xmerl_scan:string("<?xml version=\"1.0\"?>\n\n<?xml-stylesheet   href=\"doc.xsl\"\n   type=\"text/xsl\"   ?>\n\n<doc>Hello, world!<!-- Comment 1 --></doc>\n\n<?pi-without-data     ?>\n\n<!-- Comment 2 -->\n\n<!-- Comment 3 -->", [{namespace_conformant, true}, {document, true}]),

src/xmerl_dsig.erl

@@ -58,7 +58,8 @@ sign(ElementIn, PrivateKey, CertBin) ->
 
    % first we need the digest, to generate our SignedInfo element
    CanonXml = xmerl_c14n:c14n(Element),
-   DigestValue = base64:encode_to_string(crypto:sha(CanonXml)),
+   DigestValue = base64:encode_to_string(
+      crypto:sha(unicode:characters_to_binary(CanonXml, unicode, utf8))),
 
    Ns = #xmlNamespace{nodes = [{"ds", 'http://www.w3.org/2000/09/xmldsig#'}]},
    SigInfo = esaml:build_nsinfo(Ns, #xmlElement{
@@ -86,7 +87,7 @@ sign(ElementIn, PrivateKey, CertBin) ->
 
    % now we sign the SignedInfo element...
    SigInfoCanon = xmerl_c14n:c14n(SigInfo),
-   Data = list_to_binary(SigInfoCanon),
+   Data = unicode:characters_to_binary(SigInfoCanon, unicode, utf8),
 
    Signature = public_key:sign(Data, sha, PrivateKey),
    Sig64 = base64:encode_to_string(Signature),
@@ -121,7 +122,8 @@ verify(Element, Fingerprints) ->
 
    DsNs = [{"ds", 'http://www.w3.org/2000/09/xmldsig#'}],
    [#xmlText{value = Sha64}] = xmerl_xpath:string("ds:Signature/ds:SignedInfo/ds:Reference/ds:DigestValue/text()", Element, [{namespace, DsNs}]),
-   CanonSha = crypto:sha(CanonXml),
+   CanonXmlUtf8 = unicode:characters_to_binary(CanonXml, unicode, utf8),
+   CanonSha = crypto:sha(CanonXmlUtf8),
    CanonSha2 = base64:decode(Sha64),
    if not (CanonSha =:= CanonSha2) ->
       {error, bad_digest};