From 162f5a20095c6937030d23ee03fb8f72c51eefa1 Mon Sep 17 00:00:00 2001
From: tobias <tobias.schulz@hotmail.com>
Date: Sun, 6 Jun 2021 15:47:06 +0200
Subject: tdf#142669 Consider BOM on text encoding detection

Return a flag if the auto detected text has a BOM.
Save the flag in SwAsciiOptions so that BOM gets set correctly when
file is written.

Change-Id: I358c3ba243bc326a552c2dc24773c94f8319c700
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/116759
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
---
 sw/qa/extras/txtexport/data/UTF16LEBOMCRLF.txt | Bin 0 -> 18 bytes
 sw/qa/extras/txtexport/data/UTF16LECRLF.txt    | Bin 18 -> 16 bytes
 sw/qa/extras/txtexport/data/UTF8CRLF.txt       |   2 +
 sw/qa/extras/txtexport/txtexport.cxx           |  49 ++++++++++++++++---------
 4 files changed, 34 insertions(+), 17 deletions(-)
 create mode 100644 sw/qa/extras/txtexport/data/UTF16LEBOMCRLF.txt
 create mode 100644 sw/qa/extras/txtexport/data/UTF8CRLF.txt

(limited to 'sw/qa/extras')

diff --git a/sw/qa/extras/txtexport/data/UTF16LEBOMCRLF.txt b/sw/qa/extras/txtexport/data/UTF16LEBOMCRLF.txt
new file mode 100644
index 000000000000..be232521eafc
Binary files /dev/null and b/sw/qa/extras/txtexport/data/UTF16LEBOMCRLF.txt differ
diff --git a/sw/qa/extras/txtexport/data/UTF16LECRLF.txt b/sw/qa/extras/txtexport/data/UTF16LECRLF.txt
index be232521eafc..b74e964113de 100644
Binary files a/sw/qa/extras/txtexport/data/UTF16LECRLF.txt and b/sw/qa/extras/txtexport/data/UTF16LECRLF.txt differ
diff --git a/sw/qa/extras/txtexport/data/UTF8CRLF.txt b/sw/qa/extras/txtexport/data/UTF8CRLF.txt
new file mode 100644
index 000000000000..62d4d44677b6
--- /dev/null
+++ b/sw/qa/extras/txtexport/data/UTF8CRLF.txt
@@ -0,0 +1,2 @@
+フー
+バー
diff --git a/sw/qa/extras/txtexport/txtexport.cxx b/sw/qa/extras/txtexport/txtexport.cxx
index a5f989cb6689..0e52f51a4e34 100644
--- a/sw/qa/extras/txtexport/txtexport.cxx
+++ b/sw/qa/extras/txtexport/txtexport.cxx
@@ -20,19 +20,25 @@ public:
     }
 
 protected:
-    OString readExportedFile()
+    template <class T> std::vector<T> readMemoryStream()
     {
         SvMemoryStream aMemoryStream;
         SvFileStream aStream(maTempFile.GetURL(), StreamMode::READ);
         aStream.ReadStream(aMemoryStream);
-        const char* pData = static_cast<const char*>(aMemoryStream.GetData());
+        const T* pData = static_cast<const T*>(aMemoryStream.GetData());
+        return std::vector<T>(pData, pData + aMemoryStream.GetSize());
+    }
+
+    OString readExportedFile()
+    {
+        std::vector<char> aMemStream = readMemoryStream<char>();
 
         int offset = 0;
-        if (aMemoryStream.GetSize() > 2 && pData[0] == '\xEF' && pData[1] == '\xBB'
-            && pData[2] == '\xBF')
+        if (aMemStream.size() > 2 && aMemStream[0] == '\xEF' && aMemStream[1] == '\xBB'
+            && aMemStream[2] == '\xBF')
             offset = 3;
 
-        return OString(pData + offset, aMemoryStream.GetSize() - offset);
+        return OString(aMemStream.data() + offset, aMemStream.size() - offset);
     }
 };
 
@@ -64,25 +70,34 @@ DECLARE_TXTEXPORT_TEST(testBullets, "bullets.odt")
     CPPUNIT_ASSERT_EQUAL(aExpected, aData);
 }
 
-DECLARE_TXTEXPORT_TEST(testTdf120574_utf8, "UTF8BOMCRLF.txt")
+DECLARE_TXTEXPORT_TEST(testTdf120574_utf8bom, "UTF8BOMCRLF.txt")
 {
-    SvMemoryStream aMemoryStream;
-    SvFileStream aStream(maTempFile.GetURL(), StreamMode::READ);
-    aStream.ReadStream(aMemoryStream);
-    const char* pData = static_cast<const char*>(aMemoryStream.GetData());
-    OString aData(std::string_view(pData, aMemoryStream.GetSize()));
+    std::vector<char> aMemStream = readMemoryStream<char>();
+    OString aData(std::string_view(aMemStream.data(), aMemStream.size()));
     CPPUNIT_ASSERT_EQUAL(OString(u8"\uFEFFフー\r\nバー\r\n"), aData);
 }
 
-DECLARE_TXTEXPORT_TEST(testTdf120574_utf16le, "UTF16LECRLF.txt")
+DECLARE_TXTEXPORT_TEST(testTdf120574_utf16lebom, "UTF16LEBOMCRLF.txt")
 {
-    SvMemoryStream aMemoryStream;
-    SvFileStream aStream(maTempFile.GetURL(), StreamMode::READ);
-    aStream.ReadStream(aMemoryStream);
-    const sal_Unicode* pData = static_cast<const sal_Unicode*>(aMemoryStream.GetData());
-    OUString aData(pData, aMemoryStream.GetSize() / sizeof(sal_Unicode));
+    std::vector<sal_Unicode> aMemStream = readMemoryStream<sal_Unicode>();
+    OUString aData(aMemStream.data(), aMemStream.size() / sizeof(sal_Unicode));
     CPPUNIT_ASSERT_EQUAL(OUString(u"\uFEFFフー\r\nバー\r\n"), aData);
 }
+
+DECLARE_TXTEXPORT_TEST(testTdf142669_utf8, "UTF8CRLF.txt")
+{
+    std::vector<char> aMemStream = readMemoryStream<char>();
+    OString aData(std::string_view(aMemStream.data(), aMemStream.size()));
+    CPPUNIT_ASSERT_EQUAL(OString(u8"フー\r\nバー\r\n"), aData);
+}
+
+DECLARE_TXTEXPORT_TEST(testTdf142669_utf16le, "UTF16LECRLF.txt")
+{
+    std::vector<sal_Unicode> aMemStream = readMemoryStream<sal_Unicode>();
+    OUString aData(aMemStream.data(), aMemStream.size() / sizeof(sal_Unicode));
+    CPPUNIT_ASSERT_EQUAL(OUString(u"フー\r\nバー\r\n"), aData);
+}
+
 CPPUNIT_PLUGIN_IMPLEMENT();
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
-- 
cgit v1.2.3