tdf#148851 Unmap + close file again in xdictionary dtor

The file opened and mapped to memory in
the `xdictionary` ctor was never unmapped
and closed again, resulting in a memory leak.

For Android Viewer, this resulted in the app
running out of memory after a while when
scrolling up and down in the Chinese sample
doc from tdf#148851.

On Android, closing the file is actually the relevant part,
because the content of files from `/assets` is automatically
loaded into memory when the files are opened (s. the call to
`openMemoryAsFile` in `openFilePath` in `sal/osl/unx/file.cxx`)
and that memory is only freed when the file is closed
again (s. `osl_closeFile`).

When using the sample doc, the file in question
was "file:///assets/share/dict_zh.data", which
has a size of 2.2 MB and is therefore actually
loaded into memory anew every time, since only
files below 50 KiB are handled by the file cache
introduced in

commit 26a46c1143e34e361d76d6459535c2056c59de77
Date:   Fri Dec 20 14:46:36 2019 +0000

android: file-cache to improve performance.

Fix the memory leak by unmapping and closing
the file again in the dtor.

Change-Id: I3388964877080d1f2b3cf2682a41549e0bfb850c
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/133581
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>

2 files changed, 28 insertions, 12 deletions

diff --git a/i18npool/inc/xdictionary.hxx b/i18npool/inc/xdictionary.hxx
index cc7364bd14fc..f0861408bbd6 100644
--- a/i18npool/inc/xdictionary.hxx
+++ b/i18npool/inc/xdictionary.hxx
@@ -18,6 +18,7 @@
  */
 #pragma once
 
+#include <osl/file.h>
 #include <sal/types.h>
 
 #include <com/sun/star/i18n/Boundary.hpp>
@@ -63,6 +64,11 @@ private:
 
     css::i18n::Boundary boundary;
     bool japaneseWordBreak;
+#ifdef DICT_JA_ZH_IN_DATAFILE
+    oslFileHandle m_aFileHandle;
+    sal_uInt64 m_nFileSize;
+    char* m_pMapping;
+#endif
 
 public:
     xdictionary(const char *lang);
diff --git a/i18npool/source/breakiterator/xdictionary.cxx b/i18npool/source/breakiterator/xdictionary.cxx
index dd1f83f8baa7..6c326f69e48f 100644
--- a/i18npool/source/breakiterator/xdictionary.cxx
+++ b/i18npool/source/breakiterator/xdictionary.cxx
@@ -62,6 +62,11 @@ sal_Unicode* getDataArea_zh();
 
 xdictionary::xdictionary(const char *lang) :
     japaneseWordBreak( false )
+#ifdef DICT_JA_ZH_IN_DATAFILE
+    , m_aFileHandle(nullptr),
+    m_nFileSize(-1),
+    m_pMapping(nullptr)
+#endif
 {
 
 #ifdef DICT_JA_ZH_IN_DATAFILE
@@ -76,21 +81,18 @@ xdictionary::xdictionary(const char *lang) :
         else if( strcmp( lang, "zh" ) == 0 )
             sUrl += "zh.data";
 
-        oslFileHandle aFileHandle;
-        sal_uInt64 nFileSize;
-        char *pMapping;
-        if( osl_openFile( sUrl.pData, &aFileHandle, osl_File_OpenFlag_Read ) == osl_File_E_None &&
-            osl_getFileSize( aFileHandle, &nFileSize) == osl_File_E_None &&
-            osl_mapFile( aFileHandle, (void **) &pMapping, nFileSize, 0, osl_File_MapFlag_RandomAccess ) == osl_File_E_None )
+        if( osl_openFile( sUrl.pData, &m_aFileHandle, osl_File_OpenFlag_Read ) == osl_File_E_None &&
+            osl_getFileSize( m_aFileHandle, &m_nFileSize) == osl_File_E_None &&
+            osl_mapFile( m_aFileHandle, (void **) &m_pMapping, m_nFileSize, 0, osl_File_MapFlag_RandomAccess ) == osl_File_E_None )
         {
             // We have the offsets to the parts of the file at its end, see gendict.cxx
-            sal_Int64 *pEOF = (sal_Int64*)(pMapping + nFileSize);
+            sal_Int64 *pEOF = (sal_Int64*)(m_pMapping + m_nFileSize);
 
-            data.existMark = (sal_uInt8*) (pMapping + pEOF[-1]);
-            data.index2 = (sal_Int32*) (pMapping + pEOF[-2]);
-            data.index1 = (sal_Int16*) (pMapping + pEOF[-3]);
-            data.lenArray = (sal_Int32*) (pMapping + pEOF[-4]);
-            data.dataArea = (sal_Unicode*) (pMapping + pEOF[-5]);
+            data.existMark = (sal_uInt8*) (m_pMapping + pEOF[-1]);
+            data.index2 = (sal_Int32*) (m_pMapping + pEOF[-2]);
+            data.index1 = (sal_Int16*) (m_pMapping + pEOF[-3]);
+            data.lenArray = (sal_Int32*) (m_pMapping + pEOF[-4]);
+            data.dataArea = (sal_Unicode*) (m_pMapping + pEOF[-5]);
         }
     }
 
@@ -131,6 +133,14 @@ xdictionary::~xdictionary()
             delete [] i.wordboundary;
         }
     }
+#ifdef DICT_JA_ZH_IN_DATAFILE
+    if (m_aFileHandle) {
+        if (m_pMapping) {
+            osl_unmapMappedFile(m_aFileHandle, m_pMapping, m_nFileSize);
+        }
+        osl_closeFile(m_aFileHandle);
+    }
+#endif
 }
 
 namespace {