cp#1000039 DOC import: ignore symbol charset of the symbol font

Otherwise characters unhandled by our OpenSymbol font (like Arabic 0-9
numbers) won't be rendered using an other font, as no other font
provides the symbol charset. Do this in
SwWW8ImplReader::GetFontParams(), where we already have font name ->
font family mappings for a few well-known fonts.

The DOCX filter does the same for quite some time, and that's how Arabic
numbers in text using the Symbol font were rendered, instead of little
rectangles.

Change-Id: Ib794cac19ad7b073e39f3cbd7d4bad3994151c14

sw/qa/extras/ww8import/ww8import.cxx

@@ -243,6 +243,12 @@ DECLARE_WW8IMPORT_TEST(testBnc821208, "bnc821208.doc")
     CPPUNIT_ASSERT_EQUAL(beans::PropertyState_DEFAULT_VALUE, ePropertyState);
 }
 
+DECLARE_WW8IMPORT_TEST(testCp1000039, "cp1000039.doc")
+{
+    // This was RTL_TEXTENCODING_SYMBOL, causing "1" rendered as a placeholder rectangle.
+    CPPUNIT_ASSERT_EQUAL(sal_Int16(RTL_TEXTENCODING_DONTKNOW), getProperty<sal_Int16>(getRun(getParagraph(1), 1), "CharFontCharSet"));
+}
+
 #endif
 
 CPPUNIT_PLUGIN_IMPLEMENT();

sw/source/filter/ww8/ww8par6.cxx

@@ -3397,6 +3397,11 @@ bool SwWW8ImplReader::GetFontParams( sal_uInt16 nFCode, FontFamily& reFamily,
             reCharSet = rtl_getTextEncodingFromWindowsCharset( pF->chs );
     }
 
+    // Make sure charset is not set in case of Symbol, otherwise in case a
+    // character is missing in our OpenSymbol, the character won't be replaced.
+    if (reCharSet == RTL_TEXTENCODING_SYMBOL && rName == "Symbol")
+        reCharSet = RTL_TEXTENCODING_DONTKNOW;
+
     // make sure Font Family Code is set correctly
     // at least for the most important fonts
     // ( might be set wrong when Doc was not created by