sc/source/filter/inc/orcus_utils.hxx | 16 +++++++++++
sc/source/filter/orcus/filterdetect.cxx | 18 ++-----------
sc/source/filter/orcus/orcusfiltersimpl.cxx | 32 -----------------------
sc/source/filter/orcus/utils.cxx | 38 ++++++++++++++++++++++++++++
4 files changed, 58 insertions(+), 46 deletions(-)
New commits:
commit 0923c08b2890a3d85694421babd6a512cafe6231
Author: Kohei Yoshida <kohei.yosh...@collabora.com>
AuthorDate: Tue Aug 26 21:43:44 2025 -0400
Commit: Kohei Yoshida <kohei.yosh...@collabora.com>
CommitDate: Wed Aug 27 13:38:17 2025 +0200
Create temp file to store input stream during type detection
This is also to avoid potential memory usage spikes on large files
when detecting their types.
Change-Id: I194620cfab0677f72beae1774e16d77395e6ec03
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/190261
Reviewed-by: Kohei Yoshida <ko...@libreoffice.org>
Tested-by: Jenkins
diff --git a/sc/source/filter/inc/orcus_utils.hxx
b/sc/source/filter/inc/orcus_utils.hxx
index ca7e73db59f2..c50f3e771f6d 100644
--- a/sc/source/filter/inc/orcus_utils.hxx
+++ b/sc/source/filter/inc/orcus_utils.hxx
@@ -8,9 +8,25 @@
#pragma once
#include <rtl/ustring.hxx>
+#include <unotools/tempfile.hxx>
+#include <com/sun/star/io/XInputStream.hpp>
#include <orcus/stream.hpp>
orcus::file_content toFileContent(const OUString& rPath);
+/**
+ * Stream copied to a temporary file with a filepath.
+ */
+class CopiedTempStream
+{
+ utl::TempFileNamed maTemp;
+
+public:
+ CopiedTempStream(SvStream& rSrc);
+ CopiedTempStream(const css::uno::Reference<css::io::XInputStream>& xSrc);
+
+ OUString getFileName() const { return maTemp.GetFileName(); }
+};
+
/* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sc/source/filter/orcus/filterdetect.cxx
b/sc/source/filter/orcus/filterdetect.cxx
index d8bfc2fe23cc..285d5a01d8cd 100644
--- a/sc/source/filter/orcus/filterdetect.cxx
+++ b/sc/source/filter/orcus/filterdetect.cxx
@@ -14,8 +14,8 @@
#include <cppuhelper/supportsservice.hxx>
#include <unotools/mediadescriptor.hxx>
-
#include <tools/stream.hxx>
+#include <orcus_utils.hxx>
#include <orcus/format_detection.hpp>
@@ -71,20 +71,10 @@ OUString
OrcusFormatDetect::detect(css::uno::Sequence<css::beans::PropertyValue>
css::uno::Reference<css::io::XInputStream> xInputStream(
aMediaDescriptor[utl::MediaDescriptor::PROP_INPUTSTREAM],
css::uno::UNO_QUERY);
- SvMemoryStream aContent(xInputStream->available());
-
- static const sal_Int32 nBytes = 4096;
- css::uno::Sequence<sal_Int8> aSeq(nBytes);
- bool bEnd = false;
- while (!bEnd)
- {
- sal_Int32 nReadBytes = xInputStream->readBytes(aSeq, nBytes);
- bEnd = (nReadBytes != nBytes);
- aContent.WriteBytes(aSeq.getConstArray(), nReadBytes);
- }
- std::string_view aStream(static_cast<const char*>(aContent.GetData()),
aContent.GetSize());
- orcus::format_t eFormat = orcus::detect(aStream);
+ CopiedTempStream aTemp(xInputStream);
+ auto aContent = toFileContent(aTemp.getFileName());
+ orcus::format_t eFormat = orcus::detect(aContent.str());
switch (eFormat)
{
diff --git a/sc/source/filter/orcus/orcusfiltersimpl.cxx
b/sc/source/filter/orcus/orcusfiltersimpl.cxx
index c0e2b2facd61..f0a09099d45b 100644
--- a/sc/source/filter/orcus/orcusfiltersimpl.cxx
+++ b/sc/source/filter/orcus/orcusfiltersimpl.cxx
@@ -20,7 +20,6 @@
#include <svl/itemset.hxx>
#include <rtl/ustring.hxx>
#include <sal/log.hxx>
-#include <unotools/tempfile.hxx>
#include <orcus/format_detection.hpp>
#include <orcus/orcus_import_ods.hpp>
@@ -31,37 +30,6 @@ using namespace com::sun::star;
namespace
{
-/**
- * Stream copied to a temporary file with a filepath.
- */
-class CopiedTempStream
-{
- utl::TempFileNamed maTemp;
-
-public:
- CopiedTempStream(SvStream& rSrc)
- {
- maTemp.EnableKillingFile();
- SvStream* pDest = maTemp.GetStream(StreamMode::WRITE);
-
- rSrc.Seek(0);
-
- const std::size_t nReadBuffer = 1024 * 32;
- std::size_t nRead = 0;
-
- do
- {
- char pData[nReadBuffer];
- nRead = rSrc.ReadBytes(pData, nReadBuffer);
- pDest->WriteBytes(pData, nRead);
- } while (nRead == nReadBuffer);
-
- maTemp.CloseStream();
- }
-
- OUString getFileName() const { return maTemp.GetFileName(); }
-};
-
uno::Reference<task::XStatusIndicator> getStatusIndicator(const SfxMedium&
rMedium)
{
uno::Reference<task::XStatusIndicator> xStatusIndicator;
diff --git a/sc/source/filter/orcus/utils.cxx b/sc/source/filter/orcus/utils.cxx
index d3df1513cc5c..89846684bc75 100644
--- a/sc/source/filter/orcus/utils.cxx
+++ b/sc/source/filter/orcus/utils.cxx
@@ -17,4 +17,42 @@ orcus::file_content toFileContent(const OUString& rPath)
#endif
}
+CopiedTempStream::CopiedTempStream(SvStream& rSrc)
+{
+ maTemp.EnableKillingFile();
+ SvStream* pDest = maTemp.GetStream(StreamMode::WRITE);
+
+ rSrc.Seek(0);
+
+ const std::size_t nReadBuffer = 1024 * 32;
+ std::size_t nRead = 0;
+
+ do
+ {
+ char pData[nReadBuffer];
+ nRead = rSrc.ReadBytes(pData, nReadBuffer);
+ pDest->WriteBytes(pData, nRead);
+ } while (nRead == nReadBuffer);
+
+ maTemp.CloseStream();
+}
+
+CopiedTempStream::CopiedTempStream(const
css::uno::Reference<css::io::XInputStream>& xSrc)
+{
+ maTemp.EnableKillingFile();
+ SvStream* pDest = maTemp.GetStream(StreamMode::WRITE);
+
+ const sal_Int32 nBytes = 1024 * 32;
+ css::uno::Sequence<sal_Int8> aSeq(nBytes);
+
+ for (bool bEnd = false; !bEnd;)
+ {
+ sal_Int32 nReadBytes = xSrc->readBytes(aSeq, nBytes);
+ bEnd = (nReadBytes != nBytes);
+ pDest->WriteBytes(aSeq.getConstArray(), nReadBytes);
+ }
+
+ maTemp.CloseStream();
+}
+
/* vim:set shiftwidth=4 softtabstop=4 expandtab: */
