Ensure that strings returned by wxMBConv_cf are in NFC form.

[wxWidgets.git] / src / common / convauto.cpp

diff --git a/src/common/convauto.cpp b/src/common/convauto.cpp
index c9ff7df9f68b3285948321bbeeda366d3a034993..756e81c233f2d71b77dfe0e7ae96f1ad6712f19c 100644 (file)
--- a/src/common/convauto.cpp
+++ b/src/common/convauto.cpp
@@ -23,8 +23,6 @@
     #pragma hdrstop
 #endif
 
     #pragma hdrstop
 #endif
 

-#if wxUSE_WCHAR_T
-

 #ifndef WX_PRECOMP
     #include "wx/wx.h"
 #endif //WX_PRECOMP
 #ifndef WX_PRECOMP
     #include "wx/wx.h"
 #endif //WX_PRECOMP


@@ -107,11 +105,13 @@ wxConvAuto::BOMType wxConvAuto::DetectBOM(const char *src, size_t srcLen)
 
             if ( src[0] == '\x00' && src[1] == '\x00' )
             {
 
             if ( src[0] == '\x00' && src[1] == '\x00' )
             {

-                // this could only be UTF-32BE
-                if ( srcLen == 3 && src[2] == '\xFE' )
-                    return BOM_Unknown;
-            }
+                // this could only be UTF-32BE, check that the data we have so
+                // far allows for it
+                if ( srcLen == 3 && src[2] != '\xFE' )
+                    return BOM_None;

 
 

+                return BOM_Unknown;
+            }

             break;
 
         default:
             break;
 
         default:


@@ -227,14 +227,13 @@ void wxConvAuto::SkipBOM(const char **src, size_t *len) const
         *len -= ofs;
 }
 
         *len -= ofs;
 }
 

-bool wxConvAuto::InitFromInput(const char **src, size_t *len)
+bool wxConvAuto::InitFromInput(const char *src, size_t len)

 {
 {

-    m_bomType = DetectBOM(*src, *len);
+    m_bomType = DetectBOM(src, len == wxNO_LEN ? strlen(src) : len);

     if ( m_bomType == BOM_Unknown )
         return false;
 
     InitFromBOM(m_bomType);
     if ( m_bomType == BOM_Unknown )
         return false;
 
     InitFromBOM(m_bomType);

-    SkipBOM(src, len);

 
     return true;
 }
 
     return true;
 }


@@ -253,7 +252,7 @@ wxConvAuto::ToWChar(wchar_t *dst, size_t dstLen,
 
     if ( !m_conv )
     {
 
     if ( !m_conv )
     {

-        if ( !self->InitFromInput(&src, &srcLen) )
+        if ( !self->InitFromInput(src, srcLen) )

         {
             // there is not enough data to determine whether we have a BOM or
             // not, so fail for now -- the caller is supposed to call us again
         {
             // there is not enough data to determine whether we have a BOM or
             // not, so fail for now -- the caller is supposed to call us again


@@ -261,9 +260,21 @@ wxConvAuto::ToWChar(wchar_t *dst, size_t dstLen,
             return wxCONV_FAILED;
         }
     }
             return wxCONV_FAILED;
         }
     }

-    else if ( !m_consumedBOM && dst )
+
+    if ( !m_consumedBOM )

     {
         SkipBOM(&src, &srcLen);
     {
         SkipBOM(&src, &srcLen);

+        if ( srcLen == 0 )
+        {
+            // there is nothing left except the BOM so we'd return 0 below but
+            // this is unexpected: decoding a non-empty string must either fail
+            // or return something non-empty, in particular this would break
+            // the code in wxTextInputStream::NextChar()
+            //
+            // so still return an error as we need some more data to be able to
+            // decode it
+            return wxCONV_FAILED;
+        }

     }
 
     // try to convert using the auto-detected encoding
     }
 
     // try to convert using the auto-detected encoding


@@ -286,8 +297,10 @@ wxConvAuto::ToWChar(wchar_t *dst, size_t dstLen,
         }
     }
 
         }
     }
 

-    if (rc != wxCONV_FAILED && dst && !m_consumedBOM)
+    // don't skip the BOM again the next time if we really consumed it
+    if ( rc != wxCONV_FAILED && dst && !m_consumedBOM )

         self->m_consumedBOM = true;
         self->m_consumedBOM = true;

+

     return rc;
 }
 
     return rc;
 }
 


@@ -303,5 +316,3 @@ wxConvAuto::FromWChar(char *dst, size_t dstLen,
 
     return m_conv->FromWChar(dst, dstLen, src, srcLen);
 }
 
     return m_conv->FromWChar(dst, dstLen, src, srcLen);
 }

-
-#endif // wxUSE_WCHAR_T