https://bugs.webkit.org/show_bug.cgi?id=124466

Reviewed by Sam Weinig.

Source/WebCore: 

Using Vector<char> for crypto key data is somewhat non-idiomatic, and it gets simply
dangerous for bignums, because signed arithmetic is not appropriate for bignum digits.

* Modules/websockets/WebSocketHandshake.cpp:
(WebCore::generateSecWebSocketKey):
(WebCore::WebSocketHandshake::getExpectedWebSocketAccept):
No longer need to cast data to char* here.

* bindings/js/JSCryptoKeySerializationJWK.cpp:
* bindings/js/JSCryptoKeySerializationJWK.h:
* crypto/CryptoDigest.h:
* crypto/CryptoKey.h:
* crypto/keys/CryptoKeyAES.cpp:
* crypto/keys/CryptoKeyAES.h:
* crypto/keys/CryptoKeyDataOctetSequence.h:
* crypto/keys/CryptoKeyDataRSAComponents.cpp:
* crypto/keys/CryptoKeyDataRSAComponents.h:
* crypto/keys/CryptoKeyHMAC.cpp:
* crypto/keys/CryptoKeyHMAC.h:
* crypto/keys/CryptoKeyRSA.h:
* crypto/keys/CryptoKeySerializationRaw.cpp:
* crypto/keys/CryptoKeySerializationRaw.h:
* crypto/mac/CryptoAlgorithmAES_CBCMac.cpp:
* crypto/mac/CryptoAlgorithmHMACMac.cpp:
* crypto/mac/CryptoDigestMac.cpp:
* crypto/mac/CryptoKeyMac.cpp:
* crypto/parameters/CryptoAlgorithmRsaKeyGenParams.h:
Switched to Vector<uint8_t>.

* crypto/mac/CryptoKeyRSAMac.cpp:
(WebCore::getPublicKeyComponents): Extracted from buildAlgorithmDescription() and simplified.
(WebCore::CryptoKeyRSA::create): Switched to Vector<uint8_t>.
(WebCore::CryptoKeyRSA::buildAlgorithmDescription): No longer need to copy data just
to change type from Vector<char> to Vector<unsigned char>.
(WebCore::bigIntegerToUInt32): Ditto. No longer need to cast types when dealing with the bignum.
(WebCore::CryptoKeyRSA::generatePair): Improved an error message a little.

* fileapi/FileReaderLoader.cpp: (WebCore::FileReaderLoader::convertToDataURL):
* inspector/DOMPatchSupport.cpp: (WebCore::DOMPatchSupport::createDigest):
* inspector/InspectorPageAgent.cpp: (WebCore::InspectorPageAgent::archive):
* platform/graphics/cg/ImageBufferCG.cpp: (WebCore::CGImageToDataURL):
No longer need to cast data to char* here.

Source/WTF: 

Binary data can be UTF-8, in which case "char*" is idiomatic, or it can be arbitrary
binary data, in which case "uint8_t*" is more common.

Changed encode functions that took "const char *" to "const void*", and decode
functions that took "Vector<char>&" now take an adapter class.

The adapter relies on Vector<char> and Vector<uint8_t> classes having an identical layout.

* wtf/text/Base64.cpp:
(WTF::base64Encode):
(WTF::base64URLEncode):
(WTF::base64DecodeInternal):
(WTF::base64Decode):
(WTF::base64URLDecode):
* wtf/text/Base64.h:
(WTF::SignedOrUnsignedCharVectorAdapter):
(WTF::ConstSignedOrUnsignedCharVectorAdapter):
(WTF::base64Encode):
(WTF::base64URLEncode):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@159390 268f45cc-cd09-0410-ab3c-d52691b4dbfc