sdext/source/pdfimport/filterdet.cxx | 48 +++++++++++++++++++++++++----------
1 file changed, 35 insertions(+), 13 deletions(-)
New commits:
commit f3fd83098f3020c9f69032911754a5c1d2fc4fd5
Author: Dr. David Alan Gilbert <d...@treblig.org>
AuthorDate: Thu Apr 10 23:57:29 2025 +0100
Commit: Tomaž Vajngerl <qui...@gmail.com>
CommitDate: Tue Apr 22 04:37:35 2025 +0200
tdf#55425,tdf#66580: sdext,pdfimport: Always stash password
Always stash the password in the detect phase, even if it's not
a hybrid. This lets the plain PDF import use that password.
Change-Id: Iebcd85428ebf81ea483543892687cf2ed56473ec
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/183992
Tested-by: Jenkins
Reviewed-by: Tomaž Vajngerl <qui...@gmail.com>
diff --git a/sdext/source/pdfimport/filterdet.cxx
b/sdext/source/pdfimport/filterdet.cxx
index 82474602a992..56d143d3a5a2 100644
--- a/sdext/source/pdfimport/filterdet.cxx
+++ b/sdext/source/pdfimport/filterdet.cxx
@@ -496,6 +496,19 @@ OUString SAL_CALL PDFDetector::detect( uno::Sequence<
beans::PropertyValue >& rF
aOutFilterName = "calc_pdf_addstream_import";
}
+ // Stash the password so that the importer can use it, even if we came to
the
+ // conclusion that it's not a hybrid, the PDF import side can use it.
+ if (!aPassword.isEmpty())
+ {
+ if (nPasswordPos == -1)
+ {
+ nPasswordPos = nAttribs;
+ rFilterData.realloc(++nAttribs);
+ rFilterData.getArray()[nPasswordPos].Name = "Password";
+ }
+ rFilterData.getArray()[nPasswordPos].Value <<= aPassword;
+ }
+
if (!aOutFilterName.isEmpty())
{
if( nFilterNamePos == -1 )
@@ -515,17 +528,6 @@ OUString SAL_CALL PDFDetector::detect( uno::Sequence<
beans::PropertyValue >& rF
pFilterData[nAttribs-1].Name = "EmbeddedSubstream";
pFilterData[nAttribs-1].Value <<= xEmbedStream;
}
- if (!aPassword.isEmpty())
- {
- if (nPasswordPos == -1)
- {
- nPasswordPos = nAttribs;
- rFilterData.realloc(++nAttribs);
- pFilterData = rFilterData.getArray();
- pFilterData[nPasswordPos].Name = "Password";
- }
- pFilterData[nPasswordPos].Value <<= aPassword;
- }
}
else
{
commit 0d4d2563091eccb125a3d0f11b2df858d059d6a7
Author: Dr. David Alan Gilbert <d...@treblig.org>
AuthorDate: Thu Apr 10 01:15:17 2025 +0100
Commit: Tomaž Vajngerl <qui...@gmail.com>
CommitDate: Tue Apr 22 04:37:20 2025 +0200
tdf#55425, tdf#66580: sdext,pdfimport: Extract the embedded file
At this point we're pretty sure the embedded file is a hybrid. Extract
it.
Change-Id: Id1aa42136692a1500a08455c6e7af5af3576b431
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/183952
Tested-by: Jenkins
Reviewed-by: Tomaž Vajngerl <qui...@gmail.com>
diff --git a/sdext/source/pdfimport/filterdet.cxx
b/sdext/source/pdfimport/filterdet.cxx
index 96fbfa5cc79b..82474602a992 100644
--- a/sdext/source/pdfimport/filterdet.cxx
+++ b/sdext/source/pdfimport/filterdet.cxx
@@ -302,9 +302,9 @@ constexpr FilenameMime aFilenameMimeMap[] = {
// be exactly one embedded file.
// This uses PDFium to do the legwork.
uno::Reference<io::XStream> getEmbeddedFile(const OUString& rInPDFFileURL,
- OUString& /*rOutMimetype*/,
+ OUString& rOutMimetype,
OUString& /*io_rPwd*/,
- const
uno::Reference<uno::XComponentContext>& /*xContext*/,
+ const
uno::Reference<uno::XComponentContext>& xContext,
const
uno::Sequence<beans::PropertyValue>& /*rFilterData*/,
bool /*bMayUseUI*/)
{
@@ -386,6 +386,26 @@ uno::Reference<io::XStream> getEmbeddedFile(const
OUString& rInPDFFileURL,
}
SAL_INFO("sdext.pdfimport", "getEmbeddedFile pdfium open");
+ std::vector<sal_uInt8> aExtractedFileBuf;
+ if (!pAttachment->getFile(aExtractedFileBuf))
+ {
+ break;
+ }
+ SAL_INFO("sdext.pdfimport", "getEmbeddedFile file buffer length: "
<< aExtractedFileBuf.size());
+ // Based on FileEmitContext above, we want to stash the data in a
TempFile
+ // but need an XStream
+ uno::Reference<io::XStream> xContextStream;
+ uno::Reference<io::XSeekable> xSeek;
+ xContextStream.set(io::TempFile::create(xContext),
uno::UNO_QUERY_THROW);
+ auto xOut = xContextStream->getOutputStream();
+ xSeek.set(xOut, uno::UNO_QUERY_THROW);
+ // writeBytes wants a Uno::Sequence rather than the std::vector
above, convert again
+ uno::Sequence<sal_Int8>
aExtractedFileSeq(reinterpret_cast<sal_Int8 *>(aExtractedFileBuf.data()),
aExtractedFileBuf.size());
+ xOut->writeBytes(aExtractedFileSeq);
+
+ xEmbed = xContextStream;
+ rOutMimetype = aMimetype;
+ SAL_INFO("sdext.pdfimport", "getEmbeddedFile returning stream");
} while(false);
osl_unmapMappedFile(fileHandle, pMemRawPdf, nFileSize);
