tdf#114939 package,comphelper: Try both real SHA1 and StarOffice SHA1

... when importing ODF documents.

In CreatePackageEncryptionData(), add a 3rd SHA1 password hash,
PackageSHA1CorrectEncryptionKey, to EncryptionData.

Use it in ZipPackageStream::getDataStream(), which has 3 fall-backs
for SHA1 bugs now.

Also add a CorrectSHA1DigestContext, to be used together with
PackageSHA1CorrectEncryptionKey, and rename the existing one to
StarOfficeSHA1DigestContext, to be used together with the existing
2 PackageSHA1{UTF8,MS1252}EncryptionKey.

The fallback won't be used very often anyway: for the password SHA1
to be wrong, you need a password between 52 and 55 bytes long,
and for the SHA1/1K checksum to be wrong, you need a file
smaller than 1K with compressed size mod 64 between 52 and 55;
all XML files have enough random "chaff" added to be too large.

Test that we can read both correct SHA1 and StarOffice SHA1.

Change-Id: I988fa489b5e40c7657f404f18538f637d54d28f1

12 files changed, 146 insertions, 26 deletions

diff --git a/comphelper/source/misc/storagehelper.cxx b/comphelper/source/misc/storagehelper.cxx
index de813848653b..dd52b7f6181a 100644
--- a/comphelper/source/misc/storagehelper.cxx
+++ b/comphelper/source/misc/storagehelper.cxx
@@ -50,6 +50,7 @@
 #include <ucbhelper/content.hxx>
 
 #include <comphelper/fileformat.h>
+#include <comphelper/hash.hxx>
 #include <comphelper/processfactory.hxx>
 #include <comphelper/documentconstants.hxx>
 #include <comphelper/storagehelper.hxx>
@@ -400,7 +401,8 @@ uno::Sequence< beans::NamedValue > OStorageHelper::CreatePackageEncryptionData(
         // MS_1252 encoding was used for SO60 document format password encoding,
         // this encoding supports only a minor subset of nonascii characters,
         // but for compatibility reasons it has to be used for old document formats
-        aEncryptionData.realloc( nSha1Ind + 2 );
+        aEncryptionData.realloc( nSha1Ind + 3 );
+        // these are StarOffice not-quite-SHA1
         aEncryptionData[nSha1Ind].Name = PACKAGE_ENCRYPTIONDATA_SHA1UTF8;
         aEncryptionData[nSha1Ind + 1].Name = PACKAGE_ENCRYPTIONDATA_SHA1MS1252;
 
@@ -424,6 +426,15 @@ uno::Sequence< beans::NamedValue > OStorageHelper::CreatePackageEncryptionData(
 
             aEncryptionData[nSha1Ind+nInd].Value <<= uno::Sequence< sal_Int8 >( reinterpret_cast<sal_Int8*>(pBuffer), RTL_DIGEST_LENGTH_SHA1 );
         }
+
+        // actual SHA1
+        aEncryptionData[nSha1Ind + 2].Name = PACKAGE_ENCRYPTIONDATA_SHA1CORRECT;
+        OString aByteStrPass = OUStringToOString(aPassword, RTL_TEXTENCODING_UTF8);
+        std::vector<unsigned char> const sha1(::comphelper::Hash::calculateHash(
+                reinterpret_cast<unsigned char const*>(aByteStrPass.getStr()), aByteStrPass.getLength(),
+                ::comphelper::HashType::SHA1));
+        aEncryptionData[nSha1Ind + 2].Value <<= uno::Sequence<sal_Int8>(
+                reinterpret_cast<sal_Int8 const*>(sha1.data()), sha1.size());
     }
 
     return aEncryptionData;
diff --git a/include/comphelper/storagehelper.hxx b/include/comphelper/storagehelper.hxx
index 1f5e22cb14e3..778fb1c8ea25 100644
--- a/include/comphelper/storagehelper.hxx
+++ b/include/comphelper/storagehelper.hxx
@@ -32,6 +32,7 @@
 #define PACKAGE_ENCRYPTIONDATA_SHA256UTF8 "PackageSHA256UTF8EncryptionKey"
 #define PACKAGE_ENCRYPTIONDATA_SHA1UTF8   "PackageSHA1UTF8EncryptionKey"
 #define PACKAGE_ENCRYPTIONDATA_SHA1MS1252 "PackageSHA1MS1252EncryptionKey"
+#define PACKAGE_ENCRYPTIONDATA_SHA1CORRECT "PackageSHA1CorrectEncryptionKey"
 
 namespace com { namespace sun { namespace star {
     namespace beans { struct NamedValue; }
diff --git a/package/inc/EncryptionData.hxx b/package/inc/EncryptionData.hxx
index 0add43f143de..c7c6ffb3555e 100644
--- a/package/inc/EncryptionData.hxx
+++ b/package/inc/EncryptionData.hxx
@@ -50,14 +50,16 @@ public:
     sal_Int32 m_nCheckAlg;
     sal_Int32 m_nDerivedKeySize;
     sal_Int32 m_nStartKeyGenID;
+    bool m_bTryWrongSHA1;
 
-    EncryptionData( const BaseEncryptionData& aData, const css::uno::Sequence< sal_Int8 >& aKey, sal_Int32 nEncAlg, sal_Int32 nCheckAlg, sal_Int32 nDerivedKeySize, sal_Int32 nStartKeyGenID )
+    EncryptionData(const BaseEncryptionData& aData, const css::uno::Sequence< sal_Int8 >& aKey, sal_Int32 nEncAlg, sal_Int32 nCheckAlg, sal_Int32 nDerivedKeySize, sal_Int32 nStartKeyGenID, bool const bTryWrongSHA1)
     : BaseEncryptionData( aData )
     , m_aKey( aKey )
     , m_nEncAlg( nEncAlg )
     , m_nCheckAlg( nCheckAlg )
     , m_nDerivedKeySize( nDerivedKeySize )
     , m_nStartKeyGenID( nStartKeyGenID )
+    , m_bTryWrongSHA1(bTryWrongSHA1)
     {}
 
     EncryptionData( const EncryptionData& aData )
@@ -67,6 +69,7 @@ public:
     , m_nCheckAlg( aData.m_nCheckAlg )
     , m_nDerivedKeySize( aData.m_nDerivedKeySize )
     , m_nStartKeyGenID( aData.m_nStartKeyGenID )
+    , m_bTryWrongSHA1(aData.m_bTryWrongSHA1)
     {}
 };
 
diff --git a/package/inc/ZipPackageStream.hxx b/package/inc/ZipPackageStream.hxx
index 4620b5da64f2..67428c086948 100644
--- a/package/inc/ZipPackageStream.hxx
+++ b/package/inc/ZipPackageStream.hxx
@@ -83,9 +83,10 @@ public:
     bool IsFromManifest() const { return m_bFromManifest; }
     void SetFromManifest( bool bValue ) { m_bFromManifest = bValue; }
 
-    ::rtl::Reference< EncryptionData > GetEncryptionData( bool bWinEncoding = false );
+    enum class Bugs { None, WinEncodingWrongSHA1, WrongSHA1 };
+    ::rtl::Reference<EncryptionData> GetEncryptionData(Bugs bugs = Bugs::WrongSHA1);
 
-    css::uno::Sequence< sal_Int8 > GetEncryptionKey( bool bWinEncoding = false );
+    css::uno::Sequence<sal_Int8> GetEncryptionKey(Bugs bugs = Bugs::WrongSHA1);
 
     sal_Int32 GetStartKeyGenID();
 
diff --git a/package/source/zipapi/ZipFile.cxx b/package/source/zipapi/ZipFile.cxx
index de4c1a364acd..3d4abb65a8f9 100644
--- a/package/source/zipapi/ZipFile.cxx
+++ b/package/source/zipapi/ZipFile.cxx
@@ -144,7 +144,16 @@ uno::Reference< xml::crypto::XDigestContext > ZipFile::StaticGetDigestContextFor
         xDigestContext.set( xDigestContextSupplier->getDigestContext( xEncryptionData->m_nCheckAlg, uno::Sequence< beans::NamedValue >() ), uno::UNO_SET_THROW );
     }
     else if ( xEncryptionData->m_nCheckAlg == xml::crypto::DigestID::SHA1_1K )
-        xDigestContext.set( SHA1DigestContext::Create(), uno::UNO_SET_THROW );
+    {
+        if (xEncryptionData->m_bTryWrongSHA1)
+        {
+            xDigestContext.set(StarOfficeSHA1DigestContext::Create(), uno::UNO_SET_THROW);
+        }
+        else
+        {
+            xDigestContext.set(CorrectSHA1DigestContext::Create(), uno::UNO_SET_THROW);
+        }
+    }
 
     return xDigestContext;
 }
diff --git a/package/source/zipapi/sha1context.cxx b/package/source/zipapi/sha1context.cxx
index f24064616edb..af3123e2dbd0 100644
--- a/package/source/zipapi/sha1context.cxx
+++ b/package/source/zipapi/sha1context.cxx
@@ -19,6 +19,7 @@
 
 #include <sal/config.h>
 
+#include <comphelper/hash.hxx>
 #include <com/sun/star/lang/DisposedException.hpp>
 #include <rtl/digest.h>
 #include <rtl/ref.hxx>
@@ -28,9 +29,9 @@
 using namespace ::com::sun::star;
 
 // static
-uno::Reference< xml::crypto::XDigestContext > SHA1DigestContext::Create()
+uno::Reference<xml::crypto::XDigestContext> StarOfficeSHA1DigestContext::Create()
 {
-    ::rtl::Reference< SHA1DigestContext > xResult = new SHA1DigestContext();
+    ::rtl::Reference<StarOfficeSHA1DigestContext> xResult = new StarOfficeSHA1DigestContext();
     xResult->m_pDigest = rtl_digest_createSHA1();
     if ( !xResult->m_pDigest )
         throw uno::RuntimeException("Can not create cipher!" );
@@ -38,7 +39,7 @@ uno::Reference< xml::crypto::XDigestContext > SHA1DigestContext::Create()
     return uno::Reference< xml::crypto::XDigestContext >( xResult.get() );
 }
 
-SHA1DigestContext::~SHA1DigestContext()
+StarOfficeSHA1DigestContext::~StarOfficeSHA1DigestContext()
 {
     if ( m_pDigest )
     {
@@ -47,7 +48,7 @@ SHA1DigestContext::~SHA1DigestContext()
     }
 }
 
-void SAL_CALL SHA1DigestContext::updateDigest( const uno::Sequence< ::sal_Int8 >& aData )
+void SAL_CALL StarOfficeSHA1DigestContext::updateDigest(const uno::Sequence<::sal_Int8>& aData)
 {
     ::osl::MutexGuard aGuard( m_aMutex );
     if ( !m_pDigest )
@@ -62,7 +63,7 @@ void SAL_CALL SHA1DigestContext::updateDigest( const uno::Sequence< ::sal_Int8 >
     }
 }
 
-uno::Sequence< ::sal_Int8 > SAL_CALL SHA1DigestContext::finalizeDigestAndDispose()
+uno::Sequence<::sal_Int8> SAL_CALL StarOfficeSHA1DigestContext::finalizeDigestAndDispose()
 {
     ::osl::MutexGuard aGuard( m_aMutex );
     if ( !m_pDigest )
@@ -83,4 +84,45 @@ uno::Sequence< ::sal_Int8 > SAL_CALL SHA1DigestContext::finalizeDigestAndDispose
     return aResult;
 }
 
+uno::Reference<xml::crypto::XDigestContext> CorrectSHA1DigestContext::Create()
+{
+    return new CorrectSHA1DigestContext();
+}
+
+struct CorrectSHA1DigestContext::Impl
+{
+    ::osl::Mutex m_Mutex;
+    ::comphelper::Hash m_Hash{::comphelper::HashType::SHA1};
+    bool m_bDisposed{false};
+};
+
+CorrectSHA1DigestContext::CorrectSHA1DigestContext()
+    : m_pImpl(new Impl)
+{
+}
+
+CorrectSHA1DigestContext::~CorrectSHA1DigestContext()
+{
+}
+
+void SAL_CALL CorrectSHA1DigestContext::updateDigest(const uno::Sequence<::sal_Int8>& rData)
+{
+    ::osl::MutexGuard aGuard(m_pImpl->m_Mutex);
+    if (m_pImpl->m_bDisposed)
+        throw lang::DisposedException();
+
+    m_pImpl->m_Hash.update(reinterpret_cast<unsigned char const*>(rData.getConstArray()), rData.getLength());
+}
+
+uno::Sequence<::sal_Int8> SAL_CALL CorrectSHA1DigestContext::finalizeDigestAndDispose()
+{
+    ::osl::MutexGuard aGuard(m_pImpl->m_Mutex);
+    if (m_pImpl->m_bDisposed)
+        throw lang::DisposedException();
+
+    m_pImpl->m_bDisposed = true;
+    std::vector<unsigned char> const sha1(m_pImpl->m_Hash.finalize());
+    return uno::Sequence<sal_Int8>(reinterpret_cast<sal_Int8 const*>(sha1.data()), sha1.size());
+}
+
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/package/source/zipapi/sha1context.hxx b/package/source/zipapi/sha1context.hxx
index ef9c433082d3..436dfcccbf7c 100644
--- a/package/source/zipapi/sha1context.hxx
+++ b/package/source/zipapi/sha1context.hxx
@@ -24,18 +24,19 @@
 #include <cppuhelper/implbase.hxx>
 #include <osl/mutex.hxx>
 
-class SHA1DigestContext : public cppu::WeakImplHelper< css::xml::crypto::XDigestContext >
+class StarOfficeSHA1DigestContext
+    : public cppu::WeakImplHelper<css::xml::crypto::XDigestContext>
 {
     ::osl::Mutex m_aMutex;
     void* m_pDigest;
 
-    SHA1DigestContext()
+    StarOfficeSHA1DigestContext()
     : m_pDigest( nullptr )
     {}
 
 public:
 
-    virtual ~SHA1DigestContext() override;
+    virtual ~StarOfficeSHA1DigestContext() override;
 
     static css::uno::Reference< css::xml::crypto::XDigestContext > Create();
 
@@ -44,6 +45,25 @@ public:
 
 };
 
+class CorrectSHA1DigestContext
+    : public cppu::WeakImplHelper<css::xml::crypto::XDigestContext>
+{
+    struct Impl;
+    std::unique_ptr<Impl> m_pImpl;
+
+    CorrectSHA1DigestContext();
+
+public:
+
+    virtual ~CorrectSHA1DigestContext() override;
+
+    static css::uno::Reference<css::xml::crypto::XDigestContext> Create();
+
+    virtual void SAL_CALL updateDigest(const css::uno::Sequence<::sal_Int8>& rData) override;
+    virtual css::uno::Sequence<::sal_Int8> SAL_CALL finalizeDigestAndDispose() override;
+
+};
+
 #endif
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/package/source/zippackage/ZipPackageStream.cxx b/package/source/zippackage/ZipPackageStream.cxx
index bd914b663406..c9e987aec1bd 100644
--- a/package/source/zippackage/ZipPackageStream.cxx
+++ b/package/source/zippackage/ZipPackageStream.cxx
@@ -198,26 +198,27 @@ sal_Int32 ZipPackageStream::GetBlockSize() const
     return GetEncryptionAlgorithm() == css::xml::crypto::CipherID::AES_CBC_W3C_PADDING ? 16 : 8;
 }
 
-::rtl::Reference< EncryptionData > ZipPackageStream::GetEncryptionData( bool bUseWinEncoding )
+::rtl::Reference<EncryptionData> ZipPackageStream::GetEncryptionData(Bugs const bugs)
 {
     ::rtl::Reference< EncryptionData > xResult;
     if ( m_xBaseEncryptionData.is() )
         xResult = new EncryptionData(
             *m_xBaseEncryptionData,
-            GetEncryptionKey( bUseWinEncoding ),
+            GetEncryptionKey(bugs),
             GetEncryptionAlgorithm(),
             m_nImportedChecksumAlgorithm ? m_nImportedChecksumAlgorithm : m_rZipPackage.GetChecksumAlgID(),
             m_nImportedDerivedKeySize ? m_nImportedDerivedKeySize : m_rZipPackage.GetDefaultDerivedKeySize(),
-            GetStartKeyGenID() );
+            GetStartKeyGenID(),
+            bugs != Bugs::None);
 
     return xResult;
 }
 
-uno::Sequence< sal_Int8 > ZipPackageStream::GetEncryptionKey( bool bUseWinEncoding )
+uno::Sequence<sal_Int8> ZipPackageStream::GetEncryptionKey(Bugs const bugs)
 {
     uno::Sequence< sal_Int8 > aResult;
     sal_Int32 nKeyGenID = GetStartKeyGenID();
-    bUseWinEncoding = ( bUseWinEncoding || m_bUseWinEncoding );
+    bool const bUseWinEncoding = (bugs == Bugs::WinEncodingWrongSHA1 || m_bUseWinEncoding);
 
     if ( m_bHaveOwnKey && m_aStorageEncryptionKeys.getLength() )
     {
@@ -226,7 +227,11 @@ uno::Sequence< sal_Int8 > ZipPackageStream::GetEncryptionKey( bool bUseWinEncodi
             aNameToFind = PACKAGE_ENCRYPTIONDATA_SHA256UTF8;
         else if ( nKeyGenID == xml::crypto::DigestID::SHA1 )
         {
-            aNameToFind = bUseWinEncoding ? OUString(PACKAGE_ENCRYPTIONDATA_SHA1MS1252) : OUString(PACKAGE_ENCRYPTIONDATA_SHA1UTF8);
+            aNameToFind = bUseWinEncoding
+                ? OUString(PACKAGE_ENCRYPTIONDATA_SHA1MS1252)
+                : (bugs == Bugs::WrongSHA1)
+                    ? OUString(PACKAGE_ENCRYPTIONDATA_SHA1UTF8)
+                    : OUString(PACKAGE_ENCRYPTIONDATA_SHA1CORRECT);
         }
         else
             throw uno::RuntimeException(THROW_WHERE "No expected key is provided!" );
@@ -1007,12 +1012,23 @@ uno::Reference< io::XInputStream > SAL_CALL ZipPackageStream::getDataStream()
         uno::Reference< io::XInputStream > xResult;
         try
         {
-            xResult = m_rZipPackage.getZipFile().getDataStream( aEntry, GetEncryptionData(), m_bIsEncrypted, m_rZipPackage.GetSharedMutexRef() );
+            xResult = m_rZipPackage.getZipFile().getDataStream( aEntry, GetEncryptionData(Bugs::WrongSHA1), m_bIsEncrypted, m_rZipPackage.GetSharedMutexRef() );
         }
         catch( const packages::WrongPasswordException& )
         {
             if ( m_rZipPackage.GetStartKeyGenID() == xml::crypto::DigestID::SHA1 )
             {
+                SAL_WARN("package", "ZipPackageStream::getDataStream(): SHA1 mismatch, trying fallbacks...");
+                try
+                {   // tdf#114939 try without legacy StarOffice SHA1 bug
+                    xResult = m_rZipPackage.getZipFile().getDataStream( aEntry, GetEncryptionData(Bugs::None), m_bIsEncrypted, m_rZipPackage.GetSharedMutexRef() );
+                    return xResult;
+                }
+                catch (const packages::WrongPasswordException&)
+                {
+                    /* ignore and try next... */
+                }
+
                 try
                 {
                     // rhbz#1013844 / fdo#47482 workaround for the encrypted
@@ -1035,7 +1051,7 @@ uno::Reference< io::XInputStream > SAL_CALL ZipPackageStream::getDataStream()
                 // workaround for the encrypted documents generated with the old OOo1.x bug.
                 if ( !m_bUseWinEncoding )
                 {
-                    xResult = m_rZipPackage.getZipFile().getDataStream( aEntry, GetEncryptionData( true ), m_bIsEncrypted, m_rZipPackage.GetSharedMutexRef() );
+                    xResult = m_rZipPackage.getZipFile().getDataStream( aEntry, GetEncryptionData(Bugs::WinEncodingWrongSHA1), m_bIsEncrypted, m_rZipPackage.GetSharedMutexRef() );
                     m_bUseWinEncoding = true;
                 }
                 else
diff --git a/sw/qa/extras/inc/swmodeltestbase.hxx b/sw/qa/extras/inc/swmodeltestbase.hxx
index 7cf82ad81311..6a0d6a1ff316 100644
--- a/sw/qa/extras/inc/swmodeltestbase.hxx
+++ b/sw/qa/extras/inc/swmodeltestbase.hxx
@@ -688,11 +688,18 @@ protected:
             aMediaDescriptor["FilterOptions"] <<= maFilterOptions;
         if (pPassword)
         {
-            OUString sPassword = OUString::createFromAscii(pPassword);
-            css::uno::Sequence<css::beans::NamedValue> aEncryptionData {
-                { "OOXPassword", css::uno::makeAny(sPassword) }
-            };
-            aMediaDescriptor[utl::MediaDescriptor::PROP_ENCRYPTIONDATA()] <<= aEncryptionData;
+            if (strcmp(pFilter, "Office Open XML Text"))
+            {
+                aMediaDescriptor["Password"] <<= OUString::createFromAscii(pPassword);
+            }
+            else
+            {
+                OUString sPassword = OUString::createFromAscii(pPassword);
+                css::uno::Sequence<css::beans::NamedValue> aEncryptionData {
+                    { "OOXPassword", css::uno::makeAny(sPassword) }
+                };
+                aMediaDescriptor[utl::MediaDescriptor::PROP_ENCRYPTIONDATA()] <<= aEncryptionData;
+            }
         }
         xStorable->storeToURL(maTempFile.GetURL(), aMediaDescriptor.getAsConstPropertyValueList());
         uno::Reference<lang::XComponent> xComponent(xStorable, uno::UNO_QUERY);
diff --git a/sw/qa/extras/odfexport/data/sha1_correct.odt b/sw/qa/extras/odfexport/data/sha1_correct.odt
new file mode 100644
index 000000000000..01cbb0a073b2
--- /dev/null
+++ b/sw/qa/extras/odfexport/data/sha1_correct.odt
diff --git a/sw/qa/extras/odfexport/data/sha1_wrong.odt b/sw/qa/extras/odfexport/data/sha1_wrong.odt
new file mode 100644
index 000000000000..94032025b3d9
--- /dev/null
+++ b/sw/qa/extras/odfexport/data/sha1_wrong.odt
diff --git a/sw/qa/extras/odfexport/odfexport.cxx b/sw/qa/extras/odfexport/odfexport.cxx
index 4e57a415e786..d4f8991b56de 100644
--- a/sw/qa/extras/odfexport/odfexport.cxx
+++ b/sw/qa/extras/odfexport/odfexport.cxx
@@ -384,6 +384,16 @@ DECLARE_ODFEXPORT_TEST(testFramebackgrounds, "framebackgrounds.odt")
     }
 }
 
+DECLARE_SW_ROUNDTRIP_TEST(testSHA1Correct, "sha1_correct.odt", "1012345678901234567890123456789012345678901234567890", Test)
+{   // tdf#114939 this has both an affected password as well as content.xml
+    getParagraph(1, "012");
+}
+
+DECLARE_SW_ROUNDTRIP_TEST(testSHA1Wrong, "sha1_wrong.odt", "1012345678901234567890123456789012345678901234567890", Test)
+{   // tdf#114939 this has both an affected password as well as content.xml
+    getParagraph(1, "012");
+}
+
 DECLARE_ODFEXPORT_TEST(testOOoxmlEmbedded, "oooxml_embedded.sxw")
 {
     uno::Reference<text::XTextEmbeddedObjectsSupplier> xTEOSupplier(mxComponent, uno::UNO_QUERY);