/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */ /* * This file is part of the LibreOffice project. * * This Source Code Form is subject to the terms of the Mozilla Public * License, v. 2.0. If a copy of the MPL was not distributed with this * file, You can obtain one at http://mozilla.org/MPL/2.0/. * * This file incorporates work covered by the following license notice: * * Licensed to the Apache Software Foundation (ASF) under one or more * contributor license agreements. See the NOTICE file distributed * with this work for additional information regarding copyright * ownership. The ASF licenses this file to you under the Apache * License, Version 2.0 (the "License"); you may not use this file * except in compliance with the License. You may obtain a copy of * the License at http://www.apache.org/licenses/LICENSE-2.0 . */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "blowfishcontext.hxx" #include "sha1context.hxx" #include #include #include "XUnbufferedStream.hxx" #include "XBufferedThreadedStream.hxx" #include #include #include #include "MemoryByteGrabber.hxx" #include using namespace com::sun::star; using namespace com::sun::star::io; using namespace com::sun::star::uno; using namespace com::sun::star::lang; using namespace com::sun::star::packages; using namespace com::sun::star::packages::zip; using namespace com::sun::star::packages::zip::ZipConstants; using ZipUtils::Inflater; #if OSL_DEBUG_LEVEL > 0 #define THROW_WHERE SAL_WHERE #else #define THROW_WHERE "" #endif /** This class is used to read entries from a zip file */ ZipFile::ZipFile( rtl::Reference< comphelper::RefCountedMutex > aMutexHolder, uno::Reference < XInputStream > const &xInput, uno::Reference < XComponentContext > xContext, bool bInitialise, bool bForceRecovery, Checks const checks) : m_aMutexHolder(std::move( aMutexHolder )) , m_Checks(checks) , aGrabber( xInput ) , aInflater( true ) , xStream(xInput) , m_xContext (std::move( xContext )) , bRecoveryMode( bForceRecovery ) { if (bInitialise) { if ( bForceRecovery ) { recover(); } else if ( readCEN() == -1 ) { aEntries.clear(); m_EntriesInsensitive.clear(); throw ZipException(u"stream data looks to be broken"_ustr ); } } } ZipFile::~ZipFile() { aEntries.clear(); } void ZipFile::setInputStream ( const uno::Reference < XInputStream >& xNewStream ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); xStream = xNewStream; aGrabber.setInputStream ( xStream ); } uno::Reference< xml::crypto::XDigestContext > ZipFile::StaticGetDigestContextForChecksum( const uno::Reference< uno::XComponentContext >& xArgContext, const ::rtl::Reference< EncryptionData >& xEncryptionData ) { assert(xEncryptionData->m_oCheckAlg); // callers checked it already uno::Reference< xml::crypto::XDigestContext > xDigestContext; if (*xEncryptionData->m_oCheckAlg == xml::crypto::DigestID::SHA256_1K) { uno::Reference< uno::XComponentContext > xContext = xArgContext; if ( !xContext.is() ) xContext = comphelper::getProcessComponentContext(); uno::Reference< xml::crypto::XNSSInitializer > xDigestContextSupplier = xml::crypto::NSSInitializer::create( xContext ); xDigestContext.set(xDigestContextSupplier->getDigestContext( *xEncryptionData->m_oCheckAlg, uno::Sequence()), uno::UNO_SET_THROW); } else if (*xEncryptionData->m_oCheckAlg == xml::crypto::DigestID::SHA1_1K) { if (xEncryptionData->m_bTryWrongSHA1) { xDigestContext.set(StarOfficeSHA1DigestContext::Create(), uno::UNO_SET_THROW); } else { xDigestContext.set(CorrectSHA1DigestContext::Create(), uno::UNO_SET_THROW); } } return xDigestContext; } uno::Reference< xml::crypto::XCipherContext > ZipFile::StaticGetCipher( const uno::Reference< uno::XComponentContext >& xArgContext, const ::rtl::Reference< EncryptionData >& xEncryptionData, bool bEncrypt ) { uno::Reference< xml::crypto::XCipherContext > xResult; if (xEncryptionData->m_nDerivedKeySize < 0) { throw ZipIOException(u"Invalid derived key length!"_ustr ); } uno::Sequence< sal_Int8 > aDerivedKey( xEncryptionData->m_nDerivedKeySize ); if (!xEncryptionData->m_oPBKDFIterationCount && !xEncryptionData->m_oArgon2Args && xEncryptionData->m_nDerivedKeySize == xEncryptionData->m_aKey.getLength()) { // gpg4libre: no need to derive key, m_aKey is already // usable as symmetric session key aDerivedKey = xEncryptionData->m_aKey; } else if (xEncryptionData->m_oArgon2Args) { // apparently multiple lanes cannot be processed in parallel (the // implementation will clamp), but it doesn't make sense to have more // threads than CPUs uint32_t const threads(::comphelper::ThreadPool::getPreferredConcurrency()); // need to use context to set a fixed version argon2_context context = { .out = reinterpret_cast(aDerivedKey.getArray()), .outlen = ::sal::static_int_cast(aDerivedKey.getLength()), .pwd = reinterpret_cast(xEncryptionData->m_aKey.getArray()), .pwdlen = ::sal::static_int_cast(xEncryptionData->m_aKey.getLength()), .salt = reinterpret_cast(xEncryptionData->m_aSalt.getArray()), .saltlen = ::sal::static_int_cast(xEncryptionData->m_aSalt.getLength()), .secret = nullptr, .secretlen = 0, .ad = nullptr, .adlen = 0, .t_cost = ::sal::static_int_cast(::std::get<0>(*xEncryptionData->m_oArgon2Args)), .m_cost = ::sal::static_int_cast(::std::get<1>(*xEncryptionData->m_oArgon2Args)), .lanes = ::sal::static_int_cast(::std::get<2>(*xEncryptionData->m_oArgon2Args)), .threads = threads, .version = ARGON2_VERSION_13, .allocate_cbk = nullptr, .free_cbk = nullptr, .flags = ARGON2_DEFAULT_FLAGS }; // libargon2 validates all the arguments so don't need to do it here int const rc = argon2id_ctx(&context); if (rc != ARGON2_OK) { SAL_WARN("package", "argon2id_ctx failed to derive key: " << argon2_error_message(rc)); throw ZipIOException(u"argon2id_ctx failed to derive key"_ustr); } } else if ( rtl_Digest_E_None != rtl_digest_PBKDF2( reinterpret_cast< sal_uInt8* >( aDerivedKey.getArray() ), aDerivedKey.getLength(), reinterpret_cast< const sal_uInt8 * > (xEncryptionData->m_aKey.getConstArray() ), xEncryptionData->m_aKey.getLength(), reinterpret_cast< const sal_uInt8 * > ( xEncryptionData->m_aSalt.getConstArray() ), xEncryptionData->m_aSalt.getLength(), *xEncryptionData->m_oPBKDFIterationCount) ) { throw ZipIOException(u"Can not create derived key!"_ustr ); } if (xEncryptionData->m_nEncAlg == xml::crypto::CipherID::AES_CBC_W3C_PADDING || xEncryptionData->m_nEncAlg == xml::crypto::CipherID::AES_GCM_W3C) { uno::Reference< uno::XComponentContext > xContext = xArgContext; if ( !xContext.is() ) xContext = comphelper::getProcessComponentContext(); uno::Reference< xml::crypto::XNSSInitializer > xCipherContextSupplier = xml::crypto::NSSInitializer::create( xContext ); xResult = xCipherContextSupplier->getCipherContext( xEncryptionData->m_nEncAlg, aDerivedKey, xEncryptionData->m_aInitVector, bEncrypt, uno::Sequence< beans::NamedValue >() ); } else if ( xEncryptionData->m_nEncAlg == xml::crypto::CipherID::BLOWFISH_CFB_8 ) { xResult = BlowfishCFB8CipherContext::Create( aDerivedKey, xEncryptionData->m_aInitVector, bEncrypt ); } else { throw ZipIOException(u"Unknown cipher algorithm is requested!"_ustr ); } return xResult; } void ZipFile::StaticFillHeader( const ::rtl::Reference< EncryptionData >& rData, sal_Int64 nSize, const OUString& aMediaType, sal_Int8 * & pHeader ) { // I think it's safe to restrict vector and salt length to 2 bytes ! sal_Int16 nIVLength = static_cast < sal_Int16 > ( rData->m_aInitVector.getLength() ); sal_Int16 nSaltLength = static_cast < sal_Int16 > ( rData->m_aSalt.getLength() ); sal_Int16 nDigestLength = static_cast < sal_Int16 > ( rData->m_aDigest.getLength() ); sal_Int16 nMediaTypeLength = static_cast < sal_Int16 > ( aMediaType.getLength() * sizeof( sal_Unicode ) ); // First the header *(pHeader++) = ( n_ConstHeader >> 0 ) & 0xFF; *(pHeader++) = ( n_ConstHeader >> 8 ) & 0xFF; *(pHeader++) = ( n_ConstHeader >> 16 ) & 0xFF; *(pHeader++) = ( n_ConstHeader >> 24 ) & 0xFF; // Then the version *(pHeader++) = ( n_ConstCurrentVersion >> 0 ) & 0xFF; *(pHeader++) = ( n_ConstCurrentVersion >> 8 ) & 0xFF; // Then the iteration Count sal_Int32 const nIterationCount = rData->m_oPBKDFIterationCount ? *rData->m_oPBKDFIterationCount : 0; *(pHeader++) = static_cast< sal_Int8 >(( nIterationCount >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nIterationCount >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nIterationCount >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nIterationCount >> 24 ) & 0xFF); sal_Int32 const nArgon2t = rData->m_oArgon2Args ? ::std::get<0>(*rData->m_oArgon2Args) : 0; *(pHeader++) = static_cast((nArgon2t >> 0) & 0xFF); *(pHeader++) = static_cast((nArgon2t >> 8) & 0xFF); *(pHeader++) = static_cast((nArgon2t >> 16) & 0xFF); *(pHeader++) = static_cast((nArgon2t >> 24) & 0xFF); sal_Int32 const nArgon2m = rData->m_oArgon2Args ? ::std::get<1>(*rData->m_oArgon2Args) : 0; *(pHeader++) = static_cast((nArgon2m >> 0) & 0xFF); *(pHeader++) = static_cast((nArgon2m >> 8) & 0xFF); *(pHeader++) = static_cast((nArgon2m >> 16) & 0xFF); *(pHeader++) = static_cast((nArgon2m >> 24) & 0xFF); sal_Int32 const nArgon2p = rData->m_oArgon2Args ? ::std::get<2>(*rData->m_oArgon2Args) : 0; *(pHeader++) = static_cast((nArgon2p >> 0) & 0xFF); *(pHeader++) = static_cast((nArgon2p >> 8) & 0xFF); *(pHeader++) = static_cast((nArgon2p >> 16) & 0xFF); *(pHeader++) = static_cast((nArgon2p >> 24) & 0xFF); // FIXME64: need to handle larger sizes // Then the size: *(pHeader++) = static_cast< sal_Int8 >(( nSize >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nSize >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nSize >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nSize >> 24 ) & 0xFF); // Then the encryption algorithm sal_Int32 nEncAlgID = rData->m_nEncAlg; *(pHeader++) = static_cast< sal_Int8 >(( nEncAlgID >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nEncAlgID >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nEncAlgID >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nEncAlgID >> 24 ) & 0xFF); // Then the checksum algorithm sal_Int32 nChecksumAlgID = rData->m_oCheckAlg ? *rData->m_oCheckAlg : 0; *(pHeader++) = static_cast< sal_Int8 >(( nChecksumAlgID >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nChecksumAlgID >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nChecksumAlgID >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nChecksumAlgID >> 24 ) & 0xFF); // Then the derived key size sal_Int32 nDerivedKeySize = rData->m_nDerivedKeySize; *(pHeader++) = static_cast< sal_Int8 >(( nDerivedKeySize >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nDerivedKeySize >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nDerivedKeySize >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nDerivedKeySize >> 24 ) & 0xFF); // Then the start key generation algorithm sal_Int32 nKeyAlgID = rData->m_nStartKeyGenID; *(pHeader++) = static_cast< sal_Int8 >(( nKeyAlgID >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nKeyAlgID >> 8 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nKeyAlgID >> 16 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nKeyAlgID >> 24 ) & 0xFF); // Then the salt length *(pHeader++) = static_cast< sal_Int8 >(( nSaltLength >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nSaltLength >> 8 ) & 0xFF); // Then the IV length *(pHeader++) = static_cast< sal_Int8 >(( nIVLength >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nIVLength >> 8 ) & 0xFF); // Then the digest length *(pHeader++) = static_cast< sal_Int8 >(( nDigestLength >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nDigestLength >> 8 ) & 0xFF); // Then the mediatype length *(pHeader++) = static_cast< sal_Int8 >(( nMediaTypeLength >> 0 ) & 0xFF); *(pHeader++) = static_cast< sal_Int8 >(( nMediaTypeLength >> 8 ) & 0xFF); // Then the salt content memcpy ( pHeader, rData->m_aSalt.getConstArray(), nSaltLength ); pHeader += nSaltLength; // Then the IV content memcpy ( pHeader, rData->m_aInitVector.getConstArray(), nIVLength ); pHeader += nIVLength; // Then the digest content memcpy ( pHeader, rData->m_aDigest.getConstArray(), nDigestLength ); pHeader += nDigestLength; // Then the mediatype itself memcpy ( pHeader, aMediaType.getStr(), nMediaTypeLength ); pHeader += nMediaTypeLength; } bool ZipFile::StaticFillData ( ::rtl::Reference< BaseEncryptionData > const & rData, sal_Int32 &rEncAlg, sal_Int32 &rChecksumAlg, sal_Int32 &rDerivedKeySize, sal_Int32 &rStartKeyGenID, sal_Int32 &rSize, OUString& aMediaType, const uno::Reference< XInputStream >& rStream ) { bool bOk = false; const sal_Int32 nHeaderSize = n_ConstHeaderSize - 4; Sequence < sal_Int8 > aBuffer ( nHeaderSize ); if ( nHeaderSize == rStream->readBytes ( aBuffer, nHeaderSize ) ) { sal_Int16 nPos = 0; sal_Int8 *pBuffer = aBuffer.getArray(); sal_Int16 nVersion = pBuffer[nPos++] & 0xFF; nVersion |= ( pBuffer[nPos++] & 0xFF ) << 8; if ( nVersion == n_ConstCurrentVersion ) { sal_Int32 nCount = pBuffer[nPos++] & 0xFF; nCount |= ( pBuffer[nPos++] & 0xFF ) << 8; nCount |= ( pBuffer[nPos++] & 0xFF ) << 16; nCount |= ( pBuffer[nPos++] & 0xFF ) << 24; if (nCount != 0) { rData->m_oPBKDFIterationCount.emplace(nCount); } else { rData->m_oPBKDFIterationCount.reset(); } sal_Int32 nArgon2t = pBuffer[nPos++] & 0xFF; nArgon2t |= ( pBuffer[nPos++] & 0xFF ) << 8; nArgon2t |= ( pBuffer[nPos++] & 0xFF ) << 16; nArgon2t |= ( pBuffer[nPos++] & 0xFF ) << 24; sal_Int32 nArgon2m = pBuffer[nPos++] & 0xFF; nArgon2m |= ( pBuffer[nPos++] & 0xFF ) << 8; nArgon2m |= ( pBuffer[nPos++] & 0xFF ) << 16; nArgon2m |= ( pBuffer[nPos++] & 0xFF ) << 24; sal_Int32 nArgon2p = pBuffer[nPos++] & 0xFF; nArgon2p |= ( pBuffer[nPos++] & 0xFF ) << 8; nArgon2p |= ( pBuffer[nPos++] & 0xFF ) << 16; nArgon2p |= ( pBuffer[nPos++] & 0xFF ) << 24; if (nArgon2t != 0 && nArgon2m != 0 && nArgon2p != 0) { rData->m_oArgon2Args.emplace(nArgon2t, nArgon2m, nArgon2p); } else { rData->m_oArgon2Args.reset(); } rSize = pBuffer[nPos++] & 0xFF; rSize |= ( pBuffer[nPos++] & 0xFF ) << 8; rSize |= ( pBuffer[nPos++] & 0xFF ) << 16; rSize |= ( pBuffer[nPos++] & 0xFF ) << 24; rEncAlg = pBuffer[nPos++] & 0xFF; rEncAlg |= ( pBuffer[nPos++] & 0xFF ) << 8; rEncAlg |= ( pBuffer[nPos++] & 0xFF ) << 16; rEncAlg |= ( pBuffer[nPos++] & 0xFF ) << 24; rChecksumAlg = pBuffer[nPos++] & 0xFF; rChecksumAlg |= ( pBuffer[nPos++] & 0xFF ) << 8; rChecksumAlg |= ( pBuffer[nPos++] & 0xFF ) << 16; rChecksumAlg |= ( pBuffer[nPos++] & 0xFF ) << 24; rDerivedKeySize = pBuffer[nPos++] & 0xFF; rDerivedKeySize |= ( pBuffer[nPos++] & 0xFF ) << 8; rDerivedKeySize |= ( pBuffer[nPos++] & 0xFF ) << 16; rDerivedKeySize |= ( pBuffer[nPos++] & 0xFF ) << 24; rStartKeyGenID = pBuffer[nPos++] & 0xFF; rStartKeyGenID |= ( pBuffer[nPos++] & 0xFF ) << 8; rStartKeyGenID |= ( pBuffer[nPos++] & 0xFF ) << 16; rStartKeyGenID |= ( pBuffer[nPos++] & 0xFF ) << 24; sal_Int16 nSaltLength = pBuffer[nPos++] & 0xFF; nSaltLength |= ( pBuffer[nPos++] & 0xFF ) << 8; sal_Int16 nIVLength = ( pBuffer[nPos++] & 0xFF ); nIVLength |= ( pBuffer[nPos++] & 0xFF ) << 8; sal_Int16 nDigestLength = pBuffer[nPos++] & 0xFF; nDigestLength |= ( pBuffer[nPos++] & 0xFF ) << 8; sal_Int16 nMediaTypeLength = pBuffer[nPos++] & 0xFF; nMediaTypeLength |= ( pBuffer[nPos++] & 0xFF ) << 8; if ( nSaltLength == rStream->readBytes ( aBuffer, nSaltLength ) ) { rData->m_aSalt.realloc ( nSaltLength ); memcpy ( rData->m_aSalt.getArray(), aBuffer.getConstArray(), nSaltLength ); if ( nIVLength == rStream->readBytes ( aBuffer, nIVLength ) ) { rData->m_aInitVector.realloc ( nIVLength ); memcpy ( rData->m_aInitVector.getArray(), aBuffer.getConstArray(), nIVLength ); if ( nDigestLength == rStream->readBytes ( aBuffer, nDigestLength ) ) { rData->m_aDigest.realloc ( nDigestLength ); memcpy ( rData->m_aDigest.getArray(), aBuffer.getConstArray(), nDigestLength ); if ( nMediaTypeLength == rStream->readBytes ( aBuffer, nMediaTypeLength ) ) { aMediaType = OUString( reinterpret_cast(aBuffer.getConstArray()), nMediaTypeLength / sizeof( sal_Unicode ) ); bOk = true; } } } } } } return bOk; } #if 0 // for debugging purposes void CheckSequence( const uno::Sequence< sal_Int8 >& aSequence ) { if ( aSequence.getLength() ) { sal_Int32* pPointer = *( (sal_Int32**)&aSequence ); sal_Int32 nSize = *( pPointer + 1 ); sal_Int32 nMemSize = *( pPointer - 2 ); sal_Int32 nUsedMemSize = ( nSize + 4 * sizeof( sal_Int32 ) ); OSL_ENSURE( nSize == aSequence.getLength() && nUsedMemSize + 7 - ( nUsedMemSize - 1 ) % 8 == nMemSize, "Broken Sequence!" ); } } #endif bool ZipFile::StaticHasValidPassword( const uno::Reference< uno::XComponentContext >& rxContext, const Sequence< sal_Int8 > &aReadBuffer, const ::rtl::Reference< EncryptionData > &rData ) { assert(rData->m_nEncAlg != xml::crypto::CipherID::AES_GCM_W3C); // should not be called for AEAD if ( !rData.is() || !rData->m_aKey.hasElements() ) return false; bool bRet = false; uno::Reference< xml::crypto::XCipherContext > xCipher( StaticGetCipher( rxContext, rData, false ), uno::UNO_SET_THROW ); uno::Sequence< sal_Int8 > aDecryptBuffer; uno::Sequence< sal_Int8 > aDecryptBuffer2; try { aDecryptBuffer = xCipher->convertWithCipherContext( aReadBuffer ); aDecryptBuffer2 = xCipher->finalizeCipherContextAndDispose(); } catch( uno::Exception& ) { // decryption with padding will throw the exception in finalizing if the buffer represent only part of the stream // it is no problem, actually this is why we read 32 additional bytes ( two of maximal possible encryption blocks ) } if ( aDecryptBuffer2.hasElements() ) { sal_Int32 nOldLen = aDecryptBuffer.getLength(); aDecryptBuffer.realloc( nOldLen + aDecryptBuffer2.getLength() ); memcpy( aDecryptBuffer.getArray() + nOldLen, aDecryptBuffer2.getConstArray(), aDecryptBuffer2.getLength() ); } if ( aDecryptBuffer.getLength() > n_ConstDigestLength ) aDecryptBuffer.realloc( n_ConstDigestLength ); uno::Sequence< sal_Int8 > aDigestSeq; uno::Reference< xml::crypto::XDigestContext > xDigestContext( StaticGetDigestContextForChecksum( rxContext, rData ), uno::UNO_SET_THROW ); xDigestContext->updateDigest( aDecryptBuffer ); aDigestSeq = xDigestContext->finalizeDigestAndDispose(); // If we don't have a digest, then we have to assume that the password is correct if ( rData->m_aDigest.hasElements() && ( aDigestSeq.getLength() != rData->m_aDigest.getLength() || 0 != memcmp ( aDigestSeq.getConstArray(), rData->m_aDigest.getConstArray(), aDigestSeq.getLength() ) ) ) { // We should probably tell the user that the password they entered was wrong } else bRet = true; return bRet; } uno::Reference ZipFile::checkValidPassword( ZipEntry const& rEntry, ::rtl::Reference const& rData, rtl::Reference const& rMutex) { if (rData.is() && rData->m_nEncAlg == xml::crypto::CipherID::AES_GCM_W3C) { try // the only way to find out: decrypt the whole stream, which will { // check the tag uno::Reference const xRet = createStreamForZipEntry(rMutex, rEntry, rData, UNBUFF_STREAM_DATA, true); // currently XBufferedStream reads the whole stream in its ctor (to // verify the tag) - in case this gets changed, explicitly seek here uno::Reference const xSeek(xRet, uno::UNO_QUERY_THROW); xSeek->seek(xSeek->getLength()); xSeek->seek(0); return xRet; } catch (uno::Exception const&) { return {}; } } else if (rData.is() && rData->m_aKey.hasElements()) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); css::uno::Reference < css::io::XSeekable > xSeek(xStream, UNO_QUERY_THROW); xSeek->seek( rEntry.nOffset ); sal_Int64 nSize = rEntry.nMethod == DEFLATED ? rEntry.nCompressedSize : rEntry.nSize; // Only want to read enough to verify the digest if ( nSize > n_ConstDigestDecrypt ) nSize = n_ConstDigestDecrypt; assert(nSize <= n_ConstDigestDecrypt && nSize >= 0 && "silence bogus coverity overflow_sink"); Sequence aReadBuffer(nSize); xStream->readBytes( aReadBuffer, nSize ); if (StaticHasValidPassword(m_xContext, aReadBuffer, rData)) { return createStreamForZipEntry( rMutex, rEntry, rData, UNBUFF_STREAM_DATA, true); } } return {}; } namespace { class XBufferedStream : public cppu::WeakImplHelper, public comphelper::ByteReader { std::vector maBytes; size_t mnPos; size_t remainingSize() const { return maBytes.size() - mnPos; } bool hasBytes() const { return mnPos < maBytes.size(); } public: XBufferedStream( const uno::Reference& xSrcStream ) : mnPos(0) { sal_Int32 nRemaining = xSrcStream->available(); maBytes.reserve(nRemaining); if (auto pByteReader = dynamic_cast< comphelper::ByteReader* >( xSrcStream.get() )) { maBytes.resize(nRemaining); sal_Int8* pData = maBytes.data(); while (nRemaining > 0) { sal_Int32 nRead = pByteReader->readSomeBytes(pData, nRemaining); nRemaining -= nRead; pData += nRead; } return; } const sal_Int32 nBufSize = 8192; uno::Sequence aBuf(nBufSize); while (nRemaining > 0) { const sal_Int32 nBytes = xSrcStream->readBytes(aBuf, std::min(nBufSize, nRemaining)); if (!nBytes) break; maBytes.insert(maBytes.end(), aBuf.begin(), aBuf.begin() + nBytes); nRemaining -= nBytes; } } virtual sal_Int32 SAL_CALL readBytes( uno::Sequence& rData, sal_Int32 nBytesToRead ) override { if (!hasBytes()) return 0; sal_Int32 nReadSize = std::min(nBytesToRead, remainingSize()); rData.realloc(nReadSize); auto pData = rData.getArray(); std::vector::const_iterator it = maBytes.cbegin(); std::advance(it, mnPos); for (sal_Int32 i = 0; i < nReadSize; ++i, ++it) pData[i] = *it; mnPos += nReadSize; return nReadSize; } virtual sal_Int32 readSomeBytes(sal_Int8* pData, sal_Int32 nBytesToRead) override { if (!hasBytes()) return 0; sal_Int32 nReadSize = std::min(nBytesToRead, remainingSize()); std::vector::const_iterator it = maBytes.cbegin(); std::advance(it, mnPos); for (sal_Int32 i = 0; i < nReadSize; ++i, ++it) pData[i] = *it; mnPos += nReadSize; return nReadSize; } virtual sal_Int32 SAL_CALL readSomeBytes( ::css::uno::Sequence& rData, sal_Int32 nMaxBytesToRead ) override { return readBytes(rData, nMaxBytesToRead); } virtual void SAL_CALL skipBytes( sal_Int32 nBytesToSkip ) override { if (!hasBytes()) return; mnPos += nBytesToSkip; } virtual sal_Int32 SAL_CALL available() override { if (!hasBytes()) return 0; return remainingSize(); } virtual void SAL_CALL closeInput() override { } // XSeekable virtual void SAL_CALL seek( sal_Int64 location ) override { if ( location < 0 || o3tl::make_unsigned(location) > maBytes.size() ) throw IllegalArgumentException(THROW_WHERE, uno::Reference< uno::XInterface >(), 1 ); mnPos = location; } virtual sal_Int64 SAL_CALL getPosition() override { return mnPos; } virtual sal_Int64 SAL_CALL getLength() override { return maBytes.size(); } }; } uno::Reference< XInputStream > ZipFile::createStreamForZipEntry( const rtl::Reference< comphelper::RefCountedMutex >& aMutexHolder, ZipEntry const & rEntry, const ::rtl::Reference< EncryptionData > &rData, sal_Int8 nStreamMode, bool bIsEncrypted, const bool bUseBufferedStream, const OUString& aMediaType ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); rtl::Reference< XUnbufferedStream > xSrcStream = new XUnbufferedStream( m_xContext, aMutexHolder, rEntry, xStream, rData, nStreamMode, bIsEncrypted, aMediaType, bRecoveryMode); if (!bUseBufferedStream) return xSrcStream; uno::Reference xBufStream; #ifndef EMSCRIPTEN static const sal_Int32 nThreadingThreshold = 10000; // "encrypted-package" is the only data stream, no point in threading it if (nThreadingThreshold < xSrcStream->available() && rEntry.sPath != "encrypted-package" // tdf#160888 no threading for AEAD streams: // 1. the whole stream must be read immediately to verify tag // 2. XBufferedThreadedStream uses same m_aMutexHolder->GetMutex() // => caller cannot read without deadlock && (nStreamMode != UNBUFF_STREAM_DATA || !rData.is() || rData->m_nEncAlg != xml::crypto::CipherID::AES_GCM_W3C)) { xBufStream = new XBufferedThreadedStream(xSrcStream, xSrcStream->getSize()); } else #endif xBufStream = new XBufferedStream(xSrcStream); return xBufStream; } uno::Reference< XInputStream > ZipFile::StaticGetDataFromRawStream( const rtl::Reference& rMutexHolder, const uno::Reference& rxContext, const uno::Reference& xStream, const ::rtl::Reference &rData) { if (!rData.is()) throw ZipIOException(u"Encrypted stream without encryption data!"_ustr ); if (!rData->m_aKey.hasElements()) throw packages::WrongPasswordException(THROW_WHERE); uno::Reference xSeek(xStream, UNO_QUERY); if (!xSeek.is()) throw ZipIOException(u"The stream must be seekable!"_ustr); // if we have a digest, then this file is an encrypted one and we should // check if we can decrypt it or not SAL_WARN_IF(rData->m_nEncAlg != xml::crypto::CipherID::AES_GCM_W3C && !rData->m_aDigest.hasElements(), "package", "Can't detect password correctness without digest!"); if (rData->m_nEncAlg == xml::crypto::CipherID::AES_GCM_W3C) { // skip header xSeek->seek(n_ConstHeaderSize + rData->m_aInitVector.getLength() + rData->m_aSalt.getLength() + rData->m_aDigest.getLength()); try { // XUnbufferedStream does not support XSeekable so wrap it ::rtl::Reference const pRet( new XBufferedStream(new XUnbufferedStream(rMutexHolder, xStream, rData))); // currently XBufferedStream reads the whole stream in its ctor (to // verify the tag) - in case this gets changed, explicitly seek here pRet->seek(pRet->getLength()); pRet->seek(0); return pRet; } catch (uno::Exception const&) { throw packages::WrongPasswordException(THROW_WHERE); } } else if (rData->m_aDigest.hasElements()) { sal_Int32 nSize = sal::static_int_cast(xSeek->getLength()); if (nSize > n_ConstDigestLength + 32) nSize = n_ConstDigestLength + 32; // skip header xSeek->seek(n_ConstHeaderSize + rData->m_aInitVector.getLength() + rData->m_aSalt.getLength() + rData->m_aDigest.getLength()); // Only want to read enough to verify the digest Sequence aReadBuffer(nSize); xStream->readBytes(aReadBuffer, nSize); if (!StaticHasValidPassword(rxContext, aReadBuffer, rData)) throw packages::WrongPasswordException(THROW_WHERE); } return new XUnbufferedStream(rMutexHolder, xStream, rData); } ZipEnumeration ZipFile::entries() { return aEntries; } uno::Reference< XInputStream > ZipFile::getInputStream( ZipEntry& rEntry, const ::rtl::Reference< EncryptionData > &rData, bool bIsEncrypted, const rtl::Reference& aMutexHolder ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); if ( rEntry.nOffset <= 0 ) readLOC( rEntry ); // We want to return a rawStream if we either don't have a key or if the // key is wrong bool bNeedRawStream = rEntry.nMethod == STORED; if (bIsEncrypted && rData.is()) { uno::Reference const xRet(checkValidPassword(rEntry, rData, aMutexHolder)); if (xRet.is()) { return xRet; } bNeedRawStream = true; } return createStreamForZipEntry ( aMutexHolder, rEntry, rData, bNeedRawStream ? UNBUFF_STREAM_RAW : UNBUFF_STREAM_DATA, bIsEncrypted ); } uno::Reference< XInputStream > ZipFile::getDataStream( ZipEntry& rEntry, const ::rtl::Reference< EncryptionData > &rData, bool bIsEncrypted, const rtl::Reference& aMutexHolder ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); if ( rEntry.nOffset <= 0 ) readLOC( rEntry ); // An exception must be thrown in case stream is encrypted and // there is no key or the key is wrong bool bNeedRawStream = false; if ( bIsEncrypted ) { // in case no digest is provided there is no way // to detect password correctness if ( !rData.is() ) throw ZipException(u"Encrypted stream without encryption data!"_ustr ); // if we have a digest, then this file is an encrypted one and we should // check if we can decrypt it or not SAL_WARN_IF(rData->m_nEncAlg != xml::crypto::CipherID::AES_GCM_W3C && !rData->m_aDigest.hasElements(), "package", "Can't detect password correctness without digest!"); uno::Reference const xRet(checkValidPassword(rEntry, rData, aMutexHolder)); if (!xRet.is()) { throw packages::WrongPasswordException(THROW_WHERE); } return xRet; } else bNeedRawStream = ( rEntry.nMethod == STORED ); return createStreamForZipEntry ( aMutexHolder, rEntry, rData, bNeedRawStream ? UNBUFF_STREAM_RAW : UNBUFF_STREAM_DATA, bIsEncrypted ); } uno::Reference< XInputStream > ZipFile::getRawData( ZipEntry& rEntry, const ::rtl::Reference< EncryptionData >& rData, bool bIsEncrypted, const rtl::Reference& aMutexHolder, const bool bUseBufferedStream ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); if ( rEntry.nOffset <= 0 ) readLOC( rEntry ); return createStreamForZipEntry ( aMutexHolder, rEntry, rData, UNBUFF_STREAM_RAW, bIsEncrypted, bUseBufferedStream ); } uno::Reference< XInputStream > ZipFile::getWrappedRawStream( ZipEntry& rEntry, const ::rtl::Reference< EncryptionData >& rData, const OUString& aMediaType, const rtl::Reference& aMutexHolder ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); if ( !rData.is() ) throw packages::NoEncryptionException(THROW_WHERE ); if ( rEntry.nOffset <= 0 ) readLOC( rEntry ); return createStreamForZipEntry ( aMutexHolder, rEntry, rData, UNBUFF_STREAM_WRAPPEDRAW, true, true, aMediaType ); } sal_uInt64 ZipFile::readLOC(ZipEntry &rEntry) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); std::vector aNameBuffer; std::vector aExtraBuffer; return readLOC_Impl(rEntry, aNameBuffer, aExtraBuffer); } // Pass in a shared name buffer to reduce the number of allocations // we do when reading the CEN. sal_uInt64 ZipFile::readLOC_Impl(ZipEntry &rEntry, std::vector& rNameBuffer, std::vector& rExtraBuffer) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); sal_Int64 nPos = -rEntry.nOffset; aGrabber.seek(nPos); std::array aHeader; if (aGrabber.readBytes(aHeader.data(), 30) != 30) throw new uno::RuntimeException(); MemoryByteGrabber headerMemGrabber(aHeader.data(), 30); sal_Int32 nTestSig = headerMemGrabber.ReadInt32(); if (nTestSig != LOCSIG) throw ZipIOException(u"Invalid LOC header (bad signature)"_ustr ); // Ignore all (duplicated) information from the local file header. // various programs produced "broken" zip files; even LO at some point. // Just verify the path and calculate the data offset and otherwise // rely on the central directory info. headerMemGrabber.ReadInt16(); // version - ignore any mismatch (Maven created JARs) sal_uInt16 const nLocFlag = headerMemGrabber.ReadUInt16(); // general purpose bit flag sal_uInt16 const nLocMethod = headerMemGrabber.ReadUInt16(); // compression method // Do *not* compare timestamps, since MSO 2010 can produce documents // with timestamp difference in the central directory entry and local // file header. headerMemGrabber.ReadInt32(); //time sal_uInt32 nLocCrc = headerMemGrabber.ReadUInt32(); //crc sal_uInt64 nLocCompressedSize = headerMemGrabber.ReadUInt32(); //compressed size sal_uInt64 nLocSize = headerMemGrabber.ReadUInt32(); //size sal_Int16 nPathLen = headerMemGrabber.ReadInt16(); sal_Int16 nExtraLen = headerMemGrabber.ReadInt16(); if (nPathLen < 0) { SAL_WARN("package", "bogus path len of: " << nPathLen); nPathLen = 0; } rEntry.nOffset = aGrabber.getPosition() + nPathLen + nExtraLen; sal_Int64 nEnd = {}; // avoid -Werror=maybe-uninitialized bool bBroken = false; try { // read always in UTF8, some tools seem not to set UTF8 bit // coverity[tainted_data] - we've checked negative lens, and up to max short is ok here rNameBuffer.reserve(nPathLen); sal_Int32 nRead = aGrabber.readBytes(rNameBuffer.data(), nPathLen); std::string_view aNameView(reinterpret_cast(rNameBuffer.data()), nRead); OUString sLOCPath( aNameView.data(), aNameView.size(), RTL_TEXTENCODING_UTF8 ); if ( rEntry.nPathLen == -1 ) // the file was created { rEntry.nPathLen = nPathLen; rEntry.sPath = sLOCPath; } if (rEntry.nPathLen != nPathLen || rEntry.sPath != sLOCPath) { SAL_INFO("package", "LOC inconsistent name: \"" << rEntry.sPath << "\""); bBroken = true; } bool isZip64{false}; ::std::optional oOffset64; if (nExtraLen != 0) { rExtraBuffer.reserve(nExtraLen); aGrabber.readBytes(rExtraBuffer.data(), nExtraLen); MemoryByteGrabber extraMemGrabber(rExtraBuffer.data(), nExtraLen); isZip64 = readExtraFields(extraMemGrabber, nExtraLen, nLocSize, nLocCompressedSize, oOffset64, &aNameView); } // Just plain ignore bits 1 & 2 of the flag field - they are either // purely informative, or even fully undefined (depending on method). // Also ignore bit 11 ("Language encoding flag"): tdf125300.docx is // example with mismatch - and the actual file names are compared in // any case and required to be UTF-8. if ((rEntry.nFlag & ~0x806U) != (nLocFlag & ~0x806U)) { SAL_INFO("package", "LOC inconsistent flag: \"" << rEntry.sPath << "\""); bBroken = true; } // TODO: "older versions with encrypted streams write mismatching DEFLATE/STORE" ??? if (rEntry.nMethod != nLocMethod) { SAL_INFO("package", "LOC inconsistent method: \"" << rEntry.sPath << "\""); bBroken = true; } if (o3tl::checked_add(rEntry.nOffset, rEntry.nCompressedSize, nEnd)) { throw ZipException(u"Integer-overflow"_ustr); } // read "data descriptor" - this can be 12, 16, 20, or 24 bytes in size if ((rEntry.nFlag & 0x08) != 0) { #if 0 // Unfortunately every encrypted ODF package entry hits this, // because ODF requires deflated entry with value STORED and OOo/LO // has always written compressed streams with data descriptor. // So it is checked later in ZipPackage::checkZipEntriesWithDD() if (nLocMethod == STORED) { SAL_INFO("package", "LOC STORED with data descriptor: \"" << rEntry.sPath << "\""); bBroken = true; } else #endif { decltype(nLocCrc) nDDCrc; decltype(nLocCompressedSize) nDDCompressedSize; decltype(nLocSize) nDDSize; aGrabber.seek(aGrabber.getPosition() + rEntry.nCompressedSize); sal_uInt32 nTemp = aGrabber.ReadUInt32(); if (nTemp == 0x08074b50) // APPNOTE says PK78 is optional??? { nDDCrc = aGrabber.ReadUInt32(); } else { nDDCrc = nTemp; } if (isZip64) { nDDCompressedSize = aGrabber.ReadUInt64(); nDDSize = aGrabber.ReadUInt64(); } else { nDDCompressedSize = aGrabber.ReadUInt32(); nDDSize = aGrabber.ReadUInt32(); } if (nEnd < aGrabber.getPosition()) { nEnd = aGrabber.getPosition(); } else { SAL_INFO("package", "LOC invalid size: \"" << rEntry.sPath << "\""); bBroken = true; } // tdf91429.docx has same values in LOC and in (superfluous) DD if ((nLocCrc == 0 || nLocCrc == nDDCrc) && (nLocCompressedSize == 0 || nLocCompressedSize == sal_uInt64(-1) || nLocCompressedSize == nDDCompressedSize) && (nLocSize == 0 || nLocSize == sal_uInt64(-1) || nLocSize == nDDSize)) { nLocCrc = nDDCrc; nLocCompressedSize = nDDCompressedSize; nLocSize = nDDSize; } else { SAL_INFO("package", "LOC non-0 with data descriptor: \"" << rEntry.sPath << "\""); bBroken = true; } } } // unit test file export64.zip has nLocCrc/nLocCS/nLocSize = 0 on mimetype if (nLocCrc != 0 && static_cast(rEntry.nCrc) != nLocCrc) { SAL_INFO("package", "LOC inconsistent CRC: \"" << rEntry.sPath << "\""); bBroken = true; } if (nLocCompressedSize != 0 && static_cast(rEntry.nCompressedSize) != nLocCompressedSize) { SAL_INFO("package", "LOC inconsistent compressed size: \"" << rEntry.sPath << "\""); bBroken = true; } if (nLocSize != 0 && static_cast(rEntry.nSize) != nLocSize) { SAL_INFO("package", "LOC inconsistent size: \"" << rEntry.sPath << "\""); bBroken = true; } if (oOffset64 && o3tl::make_unsigned(nPos) != *oOffset64) { SAL_INFO("package", "LOC inconsistent offset: \"" << rEntry.sPath << "\""); bBroken = true; } } catch(...) { bBroken = true; } if ( bBroken && !bRecoveryMode ) throw ZipIOException(u"The stream seems to be broken!"_ustr ); return nEnd; } std::tuple ZipFile::findCentralDirectory() { // this method is called in constructor only, no need for mutex std::vector < sal_Int8 > aBuffer; try { sal_Int64 const nLength = aGrabber.getLength(); if (nLength < ENDHDR) { throw ZipException(u"Zip too small!"_ustr); } sal_Int64 nPos = nLength - ENDHDR - ZIP_MAXNAMELEN; sal_Int64 nEnd = nPos >= 0 ? nPos : 0; aGrabber.seek( nEnd ); auto nSize = nLength - nEnd; aBuffer.reserve(nSize); if (nSize != aGrabber.readBytes(aBuffer.data(), nSize)) throw ZipException(u"Zip END signature not found!"_ustr ); const sal_Int8 *pBuffer = aBuffer.data(); sal_Int64 nEndPos = {}; nPos = nSize - ENDHDR; while ( nPos >= 0 ) { if (pBuffer[nPos] == 'P' && pBuffer[nPos+1] == 'K' && pBuffer[nPos+2] == 5 && pBuffer[nPos+3] == 6 ) { nEndPos = nPos + nEnd; break; } nPos--; if (nPos == 0) { throw ZipException(u"Zip END signature not found!"_ustr); } } aGrabber.seek(nEndPos + 4); sal_uInt16 const nEndDisk = aGrabber.ReadUInt16(); if (nEndDisk != 0) { // only single disk is supported! throw ZipException(u"invalid end (disk)"_ustr ); } sal_uInt16 const nEndDirDisk = aGrabber.ReadUInt16(); if (nEndDirDisk != 0) { throw ZipException(u"invalid end (directory disk)"_ustr ); } sal_uInt16 const nEndDiskEntries = aGrabber.ReadUInt16(); sal_uInt16 const nEndEntries = aGrabber.ReadUInt16(); if (nEndDiskEntries != nEndEntries) { throw ZipException(u"invalid end (entries)"_ustr ); } sal_Int32 const nEndDirSize = aGrabber.ReadInt32(); sal_Int32 const nEndDirOffset = aGrabber.ReadInt32(); // Zip64 end of central directory locator must immediately precede // end of central directory record if (20 <= nEndPos) { aGrabber.seek(nEndPos - 20); std::array aZip64EndLocator; if (20 != aGrabber.readBytes(aZip64EndLocator.data(), 20)) throw uno::RuntimeException(); MemoryByteGrabber loc64Grabber(aZip64EndLocator.data(), 20); if (loc64Grabber.ReadUInt8() == 'P' && loc64Grabber.ReadUInt8() == 'K' && loc64Grabber.ReadUInt8() == 6 && loc64Grabber.ReadUInt8() == 7) { sal_uInt32 const nLoc64Disk = loc64Grabber.ReadUInt32(); if (nLoc64Disk != 0) { throw ZipException(u"invalid Zip64 end locator (disk)"_ustr); } sal_Int64 const nLoc64End64Offset = loc64Grabber.ReadUInt64(); if (nEndPos < 20 + 56 || (nEndPos - 20 - 56) < nLoc64End64Offset || nLoc64End64Offset < 0) { throw ZipException(u"invalid Zip64 end locator (offset)"_ustr); } sal_uInt32 const nLoc64Disks = loc64Grabber.ReadUInt32(); if (nLoc64Disks != 1) { throw ZipException(u"invalid Zip64 end locator (number of disks)"_ustr); } aGrabber.seek(nLoc64End64Offset); std::vector aZip64EndDirectory(nEndPos - 20 - nLoc64End64Offset); aGrabber.readBytes(aZip64EndDirectory.data(), nEndPos - 20 - nLoc64End64Offset); MemoryByteGrabber end64Grabber(aZip64EndDirectory.data(), nEndPos - 20 - nLoc64End64Offset); if (end64Grabber.ReadUInt8() != 'P' || end64Grabber.ReadUInt8() != 'K' || end64Grabber.ReadUInt8() != 6 || end64Grabber.ReadUInt8() != 6) { throw ZipException(u"invalid Zip64 end (signature)"_ustr); } sal_Int64 const nEnd64Size = end64Grabber.ReadUInt64(); if (nEnd64Size != nEndPos - 20 - nLoc64End64Offset - 12) { throw ZipException(u"invalid Zip64 end (size)"_ustr); } end64Grabber.ReadUInt16(); // ignore version made by end64Grabber.ReadUInt16(); // ignore version needed to extract sal_uInt32 const nEnd64Disk = end64Grabber.ReadUInt32(); if (nEnd64Disk != 0) { throw ZipException(u"invalid Zip64 end (disk)"_ustr); } sal_uInt32 const nEnd64EndDisk = end64Grabber.ReadUInt32(); if (nEnd64EndDisk != 0) { throw ZipException(u"invalid Zip64 end (directory disk)"_ustr); } sal_uInt64 const nEnd64DiskEntries = end64Grabber.ReadUInt64(); sal_uInt64 const nEnd64Entries = end64Grabber.ReadUInt64(); if (nEnd64DiskEntries != nEnd64Entries) { throw ZipException(u"invalid Zip64 end (entries)"_ustr); } sal_Int64 const nEnd64DirSize = end64Grabber.ReadUInt64(); sal_Int64 const nEnd64DirOffset = end64Grabber.ReadUInt64(); if (nEndEntries != sal_uInt16(-1) && nEnd64Entries != nEndEntries) { throw ZipException(u"inconsistent Zip/Zip64 end (entries)"_ustr); } if (o3tl::make_unsigned(nEndDirSize) != sal_uInt32(-1) && nEnd64DirSize != nEndDirSize) { throw ZipException(u"inconsistent Zip/Zip64 end (size)"_ustr); } if (o3tl::make_unsigned(nEndDirOffset) != sal_uInt32(-1) && nEnd64DirOffset != nEndDirOffset) { throw ZipException(u"inconsistent Zip/Zip64 end (offset)"_ustr); } sal_Int64 end; if (o3tl::checked_add(nEnd64DirOffset, nEnd64DirSize, end) || nLoc64End64Offset < end || nEnd64DirOffset < 0 || nLoc64End64Offset - nEnd64DirSize != nEnd64DirOffset) { throw ZipException(u"Invalid Zip64 end (bad central directory size)"_ustr); } return { nEnd64Entries, nEnd64DirSize, nEnd64DirOffset }; } } sal_Int32 end; if (o3tl::checked_add(nEndDirOffset, nEndDirSize, end) || nEndPos < end || nEndDirOffset < 0 || nEndPos - nEndDirSize != nEndDirOffset) { throw ZipException(u"Invalid END header (bad central directory size)"_ustr); } return { nEndEntries, nEndDirSize, nEndDirOffset }; } catch ( IllegalArgumentException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } catch ( NotConnectedException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } catch ( BufferSizeExceededException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } } sal_Int32 ZipFile::readCEN() { // this method is called in constructor only, no need for mutex sal_Int32 nCenPos = -1; try { auto [nTotal, nCenLen, nCenOff] = findCentralDirectory(); nCenPos = nCenOff; // data before start of zip is not supported if ( nTotal > ZIP_MAXENTRIES ) throw ZipException(u"too many entries in ZIP File"_ustr ); if (nCenLen < nTotal * CENHDR) // prevent overflow with ZIP_MAXENTRIES throw ZipException(u"invalid END header (bad entry count)"_ustr ); if (nCenLen > SAL_MAX_INT32 || nCenLen < 0) throw ZipException(u"central directory too big"_ustr); aGrabber.seek(nCenPos); std::vector aCENBuffer; aCENBuffer.reserve(nCenLen); sal_Int64 nRead = aGrabber.readBytes ( aCENBuffer.data(), nCenLen ); if (nCenLen != nRead) throw ZipException (u"Error reading CEN into memory buffer!"_ustr ); MemoryByteGrabber aMemGrabber(aCENBuffer.data(), nCenLen); ZipEntry aEntry; sal_Int16 nCommentLen; ::std::vector> unallocated = { { 0, nCenPos } }; aEntries.reserve(nTotal); sal_Int64 nCount; std::vector aTempNameBuffer; std::vector aTempExtraBuffer; for (nCount = 0 ; nCount < nTotal; nCount++) { sal_Int32 nTestSig = aMemGrabber.ReadInt32(); if ( nTestSig != CENSIG ) throw ZipException(u"Invalid CEN header (bad signature)"_ustr ); sal_uInt16 versionMadeBy = aMemGrabber.ReadUInt16(); aEntry.nVersion = aMemGrabber.ReadInt16(); aEntry.nFlag = aMemGrabber.ReadInt16(); if ( ( aEntry.nFlag & 1 ) == 1 ) throw ZipException(u"Invalid CEN header (encrypted entry)"_ustr ); aEntry.nMethod = aMemGrabber.ReadInt16(); if ( aEntry.nMethod != STORED && aEntry.nMethod != DEFLATED) throw ZipException(u"Invalid CEN header (bad compression method)"_ustr ); aEntry.nTime = aMemGrabber.ReadInt32(); aEntry.nCrc = aMemGrabber.ReadInt32(); sal_uInt64 nCompressedSize = aMemGrabber.ReadUInt32(); sal_uInt64 nSize = aMemGrabber.ReadUInt32(); aEntry.nPathLen = aMemGrabber.ReadInt16(); aEntry.nExtraLen = aMemGrabber.ReadInt16(); nCommentLen = aMemGrabber.ReadInt16(); aMemGrabber.skipBytes ( 4 ); sal_uInt32 externalFileAttributes = aMemGrabber.ReadUInt32(); sal_uInt64 nOffset = aMemGrabber.ReadUInt32(); if ( aEntry.nPathLen < 0 ) throw ZipException(u"unexpected name length"_ustr ); if ( nCommentLen < 0 ) throw ZipException(u"unexpected comment length"_ustr ); if ( aEntry.nExtraLen < 0 ) throw ZipException(u"unexpected extra header info length"_ustr ); if (aEntry.nPathLen > aMemGrabber.remainingSize()) throw ZipException(u"name too long"_ustr); // read always in UTF8, some tools seem not to set UTF8 bit std::string_view aPathView(reinterpret_cast(aMemGrabber.getCurrentPos()), aEntry.nPathLen); aEntry.sPath = OUString( aPathView.data(), aPathView.size(), RTL_TEXTENCODING_UTF8 ); if ( !::comphelper::OStorageHelper::IsValidZipEntryFileName( aEntry.sPath, true ) ) throw ZipException(u"Zip entry has an invalid name."_ustr ); aMemGrabber.skipBytes(aEntry.nPathLen); if (aEntry.nExtraLen>0) { ::std::optional oOffset64; readExtraFields(aMemGrabber, aEntry.nExtraLen, nSize, nCompressedSize, oOffset64, &aPathView); if (oOffset64) { nOffset = *oOffset64; } } aEntry.nCompressedSize = nCompressedSize; aEntry.nSize = nSize; aEntry.nOffset = nOffset; if (o3tl::checked_multiply(aEntry.nOffset, -1, aEntry.nOffset)) throw ZipException(u"Integer-overflow"_ustr); if (aEntry.nMethod == STORED && aEntry.nCompressedSize != aEntry.nSize) { throw ZipException(u"entry STORED with inconsistent size"_ustr); } aMemGrabber.skipBytes(nCommentLen); // unfortunately readLOC is required now to check the consistency assert(aEntry.nOffset <= 0); sal_uInt64 const nStart{ o3tl::make_unsigned(-aEntry.nOffset) }; sal_uInt64 const nEnd = readLOC_Impl(aEntry, aTempNameBuffer, aTempExtraBuffer); assert(nStart < nEnd); for (auto it = unallocated.begin(); ; ++it) { if (it == unallocated.end()) { throw ZipException(u"overlapping entries"_ustr); } if (nStart < it->first) { throw ZipException(u"overlapping entries"_ustr); } else if (it->first == nStart) { if (it->second == nEnd) { unallocated.erase(it); break; } else if (nEnd < it->second) { it->first = nEnd; break; } else { throw ZipException(u"overlapping entries"_ustr); } } else if (nStart < it->second) { if (nEnd < it->second) { auto const temp{it->first}; it->first = nEnd; unallocated.insert(it, { temp, nStart }); break; } else if (nEnd == it->second) { it->second = nStart; break; } else { throw ZipException(u"overlapping entries"_ustr); } } } // Is this a FAT-compatible empty entry? if (aEntry.nSize == 0 && (versionMadeBy & 0xff00) == 0) { constexpr sal_uInt32 FILE_ATTRIBUTE_DIRECTORY = 16; if (externalFileAttributes & FILE_ATTRIBUTE_DIRECTORY) continue; // This is a directory entry, not a stream - skip it } if (aEntries.find(aEntry.sPath) != aEntries.end()) { SAL_INFO("package", "Duplicate CEN entry: \"" << aEntry.sPath << "\""); throw ZipException(u"Duplicate CEN entry"_ustr); } // this is required for OOXML, but not for ODF auto const lowerPath(aEntry.sPath.toAsciiLowerCase()); if (!m_EntriesInsensitive.insert(lowerPath).second && m_Checks == Checks::CheckInsensitive) { SAL_INFO("package", "Duplicate CEN entry (case insensitive): \"" << aEntry.sPath << "\""); throw ZipException(u"Duplicate CEN entry (case insensitive)"_ustr); } aEntries[aEntry.sPath] = aEntry; } if (nCount != nTotal) throw ZipException(u"Count != Total"_ustr ); if (!unallocated.empty()) { throw ZipException(u"Zip file has holes! It will leak!"_ustr); } } catch ( IllegalArgumentException & ) { // seek can throw this... nCenPos = -1; // make sure we return -1 to indicate an error } return nCenPos; } bool ZipFile::readExtraFields(MemoryByteGrabber& aMemGrabber, sal_Int16 nExtraLen, sal_uInt64& nSize, sal_uInt64& nCompressedSize, std::optional & roOffset, std::string_view const * pCENFilenameToCheck) { bool isZip64{false}; while (nExtraLen > 0) // Extensible data fields { sal_Int16 nheaderID = aMemGrabber.ReadInt16(); sal_uInt16 dataSize = aMemGrabber.ReadUInt16(); if (nheaderID == 1) // Load Zip64 Extended Information Extra Field { // Datasize should be 28byte but some files have less (maybe non standard?) nSize = aMemGrabber.ReadUInt64(); sal_uInt16 nReadSize = 8; if (dataSize >= 16) { nCompressedSize = aMemGrabber.ReadUInt64(); nReadSize = 16; if (dataSize >= 24 && roOffset) { roOffset.emplace(aMemGrabber.ReadUInt64()); nReadSize = 24; // 4 byte should be "Disk Start Number" but we not need it } } if (dataSize > nReadSize) aMemGrabber.skipBytes(dataSize - nReadSize); isZip64 = true; } // Info-ZIP Unicode Path Extra Field - pointless as we expect UTF-8 in CEN already else if (nheaderID == 0x7075 && pCENFilenameToCheck) // ignore in recovery mode { if (aMemGrabber.remainingSize() < dataSize) { SAL_INFO("package", "Invalid Info-ZIP Unicode Path Extra Field: invalid TSize"); throw ZipException(u"Invalid Info-ZIP Unicode Path Extra Field"_ustr); } auto const nVersion = aMemGrabber.ReadUInt8(); if (nVersion != 1) { SAL_INFO("package", "Invalid Info-ZIP Unicode Path Extra Field: unexpected Version"); throw ZipException(u"Invalid Info-ZIP Unicode Path Extra Field"_ustr); } // this CRC32 is actually of the pCENFilenameToCheck // so it's pointless to check it if we require the UnicodeName // to be equal to the CEN name anyway (and pCENFilenameToCheck // is already converted to UTF-16 here) (void) aMemGrabber.ReadUInt32(); // this is required to be UTF-8 std::string_view unicodePath(reinterpret_cast(aMemGrabber.getCurrentPos()), dataSize - 5); aMemGrabber.skipBytes(dataSize - 5); if (unicodePath != *pCENFilenameToCheck) { SAL_INFO("package", "Invalid Info-ZIP Unicode Path Extra Field: unexpected UnicodeName"); throw ZipException(u"Invalid Info-ZIP Unicode Path Extra Field"_ustr); } } else { aMemGrabber.skipBytes(dataSize); } nExtraLen -= dataSize + 4; } return isZip64; } // PK34: Local file header void ZipFile::HandlePK34(std::span data, sal_Int64 dataOffset, sal_Int64 totalSize) { ZipEntry aEntry; Sequence aTmpBuffer(data.data() + 4, 26); MemoryByteGrabber aMemGrabber(aTmpBuffer); aEntry.nVersion = aMemGrabber.ReadInt16(); aEntry.nFlag = aMemGrabber.ReadInt16(); if ((aEntry.nFlag & 1) == 1) return; aEntry.nMethod = aMemGrabber.ReadInt16(); if (aEntry.nMethod != STORED && aEntry.nMethod != DEFLATED) return; aEntry.nTime = aMemGrabber.ReadInt32(); aEntry.nCrc = aMemGrabber.ReadInt32(); sal_uInt64 nCompressedSize = aMemGrabber.ReadUInt32(); sal_uInt64 nSize = aMemGrabber.ReadUInt32(); aEntry.nPathLen = aMemGrabber.ReadInt16(); aEntry.nExtraLen = aMemGrabber.ReadInt16(); const sal_Int32 nDescrLength = (aEntry.nMethod == DEFLATED && (aEntry.nFlag & 8)) ? 16 : 0; const sal_Int64 nBlockHeaderLength = aEntry.nPathLen + aEntry.nExtraLen + 30 + nDescrLength; if (aEntry.nPathLen < 0 || aEntry.nExtraLen < 0 || dataOffset + nBlockHeaderLength > totalSize) return; // read always in UTF8, some tools seem not to set UTF8 bit if (o3tl::make_unsigned(30 + aEntry.nPathLen) <= data.size()) aEntry.sPath = OUString(reinterpret_cast(data.data() + 30), aEntry.nPathLen, RTL_TEXTENCODING_UTF8); else { std::vector aFileName(aEntry.nPathLen); aGrabber.seek(dataOffset + 30); aGrabber.readBytes(aFileName.data(), aEntry.nPathLen); aEntry.sPath = OUString(reinterpret_cast(aFileName.data()), aEntry.nPathLen, RTL_TEXTENCODING_UTF8); aEntry.nPathLen = aEntry.nPathLen; } aEntry.sPath = aEntry.sPath.replace('\\', '/'); // read 64bit header if (aEntry.nExtraLen > 0) { std::vector aExtraBuffer(aEntry.nExtraLen); if (o3tl::make_unsigned(30 + aEntry.nPathLen) + aEntry.nExtraLen <= data.size()) { auto it = data.begin() + 30 + aEntry.nPathLen; std::copy(it, it + aEntry.nExtraLen, aExtraBuffer.begin()); } else { aGrabber.seek(dataOffset + 30 + aEntry.nExtraLen); aGrabber.readBytes(aExtraBuffer.data(), aEntry.nExtraLen); } MemoryByteGrabber aMemGrabberExtra(aExtraBuffer.data(), aEntry.nExtraLen); if (aEntry.nExtraLen > 0) { ::std::optional oOffset64; readExtraFields(aMemGrabberExtra, aEntry.nExtraLen, nSize, nCompressedSize, oOffset64, nullptr); } } sal_Int64 nDataSize = (aEntry.nMethod == DEFLATED) ? nCompressedSize : nSize; sal_Int64 nBlockLength = nDataSize + nBlockHeaderLength; if (dataOffset + nBlockLength > totalSize) return; aEntry.nCompressedSize = nCompressedSize; aEntry.nSize = nSize; aEntry.nOffset = dataOffset + 30 + aEntry.nPathLen + aEntry.nExtraLen; if ((aEntry.nSize || aEntry.nCompressedSize) && !checkSizeAndCRC(aEntry)) { aEntry.nCrc = 0; aEntry.nCompressedSize = 0; aEntry.nSize = 0; } // Do not add this entry, if it is empty and is a directory of an already existing entry if (aEntry.nSize == 0 && aEntry.nCompressedSize == 0 && std::find_if(aEntries.begin(), aEntries.end(), [path = OUString(aEntry.sPath + "/")](const auto& r) { return r.first.startsWith(path); }) != aEntries.end()) return; auto const lowerPath(aEntry.sPath.toAsciiLowerCase()); if (m_EntriesInsensitive.find(lowerPath) != m_EntriesInsensitive.end()) { // this is required for OOXML, but not for ODF return; } m_EntriesInsensitive.insert(lowerPath); aEntries.emplace(aEntry.sPath, aEntry); // Drop any "directory" entry corresponding to this one's path; since we don't use // central directory, we don't see external file attributes, so sanitize here sal_Int32 i = 0; for (OUString subdir = aEntry.sPath.getToken(0, '/', i); i >= 0; subdir += OUString::Concat("/") + o3tl::getToken(aEntry.sPath, 0, '/', i)) { if (auto it = aEntries.find(subdir); it != aEntries.end()) { // if not empty, let it fail later in ZipPackage::getZipFileContents if (it->second.nSize == 0 && it->second.nCompressedSize == 0) aEntries.erase(it); } } } // PK78: Data descriptor void ZipFile::HandlePK78(std::span data, sal_Int64 dataOffset) { sal_Int64 nCompressedSize, nSize; Sequence aTmpBuffer(data.data() + 4, 12 + 8 + 4); MemoryByteGrabber aMemGrabber(aTmpBuffer); sal_Int32 nCRC32 = aMemGrabber.ReadInt32(); // FIXME64: find a better way to recognize if Zip64 mode is used // Now we check if the memory at +16 byte seems to be a signature // if not, then probably Zip64 mode is used here, except // if memory at +24 byte seems not to be a signature. // Normally Data Descriptor should followed by the next Local File header // that should start with PK34, except for the last file, then it may // followed by Central directory that start with PK12, or // followed by "archive decryption header" that don't have a signature. if ((data[16] == 'P' && data[17] == 'K' && data[19] == data[18] + 1 && (data[18] == 3 || data[18] == 1)) || !(data[24] == 'P' && data[25] == 'K' && data[27] == data[26] + 1 && (data[26] == 3 || data[26] == 1))) { nCompressedSize = aMemGrabber.ReadUInt32(); nSize = aMemGrabber.ReadUInt32(); } else { nCompressedSize = aMemGrabber.ReadUInt64(); nSize = aMemGrabber.ReadUInt64(); } for (auto& rEntry : aEntries) { // this is a broken package, accept this block not only for DEFLATED streams if ((rEntry.second.nFlag & 8) == 0) continue; sal_Int64 nStreamOffset = dataOffset - nCompressedSize; if (nStreamOffset == rEntry.second.nOffset && nCompressedSize > rEntry.second.nCompressedSize) { // only DEFLATED blocks need to be checked bool bAcceptBlock = (rEntry.second.nMethod == STORED && nCompressedSize == nSize); if (!bAcceptBlock) { sal_Int64 nRealSize = 0; sal_Int32 nRealCRC = 0; getSizeAndCRC(nStreamOffset, nCompressedSize, &nRealSize, &nRealCRC); bAcceptBlock = (nRealSize == nSize && nRealCRC == nCRC32); } if (bAcceptBlock) { rEntry.second.nCrc = nCRC32; rEntry.second.nCompressedSize = nCompressedSize; rEntry.second.nSize = nSize; } } #if 0 // for now ignore clearly broken streams else if( !rEntry.second.nCompressedSize ) { rEntry.second.nCrc = nCRC32; sal_Int32 nRealStreamSize = dataOffset - rEntry.second.nOffset; rEntry.second.nCompressedSize = nRealStreamSize; rEntry.second.nSize = nSize; } #endif } } void ZipFile::recover() { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); std::vector < sal_Int8 > aBuffer; const sal_Int64 nToRead = 32000; aBuffer.reserve(nToRead); try { const sal_Int64 nLength = aGrabber.getLength(); if (nLength < ENDHDR) return; aGrabber.seek( 0 ); sal_Int32 nRead; for( sal_Int64 nGenPos = 0; (nRead = aGrabber.readBytes( aBuffer.data(), nToRead )) && nRead > 16; ) { const sal_Int8 *pBuffer = aBuffer.data(); const sal_Int32 nBufSize = nRead; sal_Int64 nPos = 0; // the buffer should contain at least one header, // or if it is end of the file, at least the postheader with sizes and hash while( nPos < nBufSize - 30 || ( nBufSize < nToRead && nPos < nBufSize - 16 ) ) { if ( nPos < nBufSize - 30 && pBuffer[nPos] == 'P' && pBuffer[nPos+1] == 'K' && pBuffer[nPos+2] == 3 && pBuffer[nPos+3] == 4 ) { HandlePK34(std::span(pBuffer + nPos, nBufSize - nPos), nGenPos + nPos, nLength); nPos += 4; } else if (pBuffer[nPos] == 'P' && pBuffer[nPos+1] == 'K' && pBuffer[nPos+2] == 7 && pBuffer[nPos+3] == 8 ) { HandlePK78(std::span(pBuffer + nPos, nBufSize - nPos), nGenPos + nPos); nPos += 4; } else nPos++; } nGenPos += nPos; aGrabber.seek( nGenPos ); } } catch ( IllegalArgumentException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } catch ( NotConnectedException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } catch ( BufferSizeExceededException& ) { throw ZipException(u"Zip END signature not found!"_ustr ); } } bool ZipFile::checkSizeAndCRC( const ZipEntry& aEntry ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); sal_Int32 nCRC = 0; sal_Int64 nSize = 0; if( aEntry.nMethod == STORED ) return ( getCRC( aEntry.nOffset, aEntry.nSize ) == aEntry.nCrc ); if (aEntry.nCompressedSize < 0) { SAL_WARN("package", "bogus compressed size of: " << aEntry.nCompressedSize); return false; } getSizeAndCRC( aEntry.nOffset, aEntry.nCompressedSize, &nSize, &nCRC ); return ( aEntry.nSize == nSize && aEntry.nCrc == nCRC ); } sal_Int32 ZipFile::getCRC( sal_Int64 nOffset, sal_Int64 nSize ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); CRC32 aCRC; sal_Int64 nBlockSize = ::std::min(nSize, static_cast< sal_Int64 >(32000)); std::vector < sal_Int8 > aBuffer; aBuffer.reserve(nBlockSize); aGrabber.seek( nOffset ); sal_Int32 nRead; for (sal_Int64 ind = 0; (nRead = aGrabber.readBytes( aBuffer.data(), nBlockSize )) && ind * nBlockSize < nSize; ++ind) { aCRC.updateSegment(aBuffer.data(), nRead); } return aCRC.getValue(); } void ZipFile::getSizeAndCRC( sal_Int64 nOffset, sal_Int64 nCompressedSize, sal_Int64 *nSize, sal_Int32 *nCRC ) { ::osl::MutexGuard aGuard( m_aMutexHolder->GetMutex() ); CRC32 aCRC; sal_Int64 nRealSize = 0; ZipUtils::InflaterBytes aInflaterLocal( true ); sal_Int32 nBlockSize = static_cast< sal_Int32 > (::std::min( nCompressedSize, static_cast< sal_Int64 >( 32000 ) ) ); std::vector < sal_Int8 > aBuffer(nBlockSize); std::vector< sal_Int8 > aData( nBlockSize ); aGrabber.seek( nOffset ); sal_Int32 nRead; for ( sal_Int64 ind = 0; !aInflaterLocal.finished() && (nRead = aGrabber.readBytes( aBuffer.data(), nBlockSize )) && ind * nBlockSize < nCompressedSize; ind++ ) { sal_Int32 nLastInflated = 0; sal_Int64 nInBlock = 0; aInflaterLocal.setInput( aBuffer.data(), nRead ); do { nLastInflated = aInflaterLocal.doInflateSegment( aData.data(), nBlockSize, 0, nBlockSize ); aCRC.updateSegment( aData.data(), nLastInflated ); nInBlock += nLastInflated; } while( !aInflater.finished() && nLastInflated ); nRealSize += nInBlock; } *nSize = nRealSize; *nCRC = aCRC.getValue(); } /* vim:set shiftwidth=4 softtabstop=4 expandtab: */