[frameworks/kcodecs] /: [KEncodingProber] Fix broken UTF16 filtering for MBCS

Git commit f3c71ebfc6694189954fe77b27c44691f46fabe1 by Stefan Brüns.
Committed on 31/07/2026 at 07:01.
Pushed by bruns into branch 'master'.

[KEncodingProber] Fix broken UTF16 filtering for MBCS

M  +1    -1    autotests/kencodingprobertest.cpp
M  +16   -5    src/probers/nsMBCSGroupProber.cpp

https://invent.kde.org/frameworks/kcodecs/-/commit/f3c71ebfc6694189954fe77b27c44691f46fabe1

diff --git a/autotests/kencodingprobertest.cpp b/autotests/kencodingprobertest.cpp
index 5bbc008..2919bf2 100644
--- a/autotests/kencodingprobertest.cpp
+++ b/autotests/kencodingprobertest.cpp
@@ -123,7 +123,6 @@ void KEncodingProberTest::testProbe()
     QEXPECT_FAIL("Konnichiwa UTF-16LE", "Too low UTF-16LE confidence, too high Win-1252", Abort);
     QEXPECT_FAIL("Konnichiwa UTF-16BE", "Too low UTF-16BE confidence, too high Win-1252", Abort);
     QEXPECT_FAIL("EnjoyPlasma Japanese UTF-16LE Universal", "Too low UTF-16LE confidence, too high Win-1252", Abort);
-    QEXPECT_FAIL("EnjoyPlasma Japanese UTF-16LE", "MBCS filter broken", Abort);
     QEXPECT_FAIL("EnjoyPlasma Japanese UTF-16BE Universal", "Too low UTF-16BE confidence, too high Win-1252", Abort);
     QCOMPARE(ep.encoding().toLower(), encoding);
 
@@ -133,6 +132,7 @@ void KEncodingProberTest::testProbe()
     QEXPECT_FAIL("UTF-16BE Unicode definite 2", "UTF-16 zero confidence", Abort);
     QEXPECT_FAIL("UTF-16LE Unicode definite 2", "UTF-16 zero confidence", Abort);
     QEXPECT_FAIL("utf-8 Japanese", "Too low UTF-8 confidence", Abort);
+    QEXPECT_FAIL("EnjoyPlasma Japanese UTF-16LE", "UTF-16 zero confidence", Abort);
     QEXPECT_FAIL("EnjoyPlasma Japanese UTF-16BE", "UTF-16 zero confidence", Abort);
     QCOMPARE_GE(ep.confidence(), 0.2);
 }
diff --git a/src/probers/nsMBCSGroupProber.cpp b/src/probers/nsMBCSGroupProber.cpp
index 8115d49..f2a82b6 100644
--- a/src/probers/nsMBCSGroupProber.cpp
+++ b/src/probers/nsMBCSGroupProber.cpp
@@ -121,7 +121,18 @@ nsProbingState nsMBCSGroupProber::HandleData(const char *aBuf, unsigned int aLen
         }
     }
 
-    for (unsigned int i = 0; i < NUM_OF_PROBERS; ++i) {
+    // The UTF16 probers need unmangled data
+    for (unsigned int i = NUM_OF_PROBERS - 2; i < NUM_OF_PROBERS; ++i) {
+        if (!mIsActive[i]) {
+            continue;
+        }
+        if (const auto st = mProbers[i]->HandleData(aBuf, aLen); st == eNotMe) {
+            mIsActive[i] = false;
+            mActiveNum--;
+        }
+    }
+
+    for (unsigned int i = 0; i < NUM_OF_PROBERS - 2; ++i) {
         if (!mIsActive[i]) {
             continue;
         }
@@ -133,13 +144,13 @@ nsProbingState nsMBCSGroupProber::HandleData(const char *aBuf, unsigned int aLen
         } else if (st == eNotMe) {
             mIsActive[i] = false;
             mActiveNum--;
-            if (mActiveNum == 0) {
-                mState = eNotMe;
-                break;
-            }
         }
     }
 
+    if (mActiveNum == 0) {
+        mState = eNotMe;
+    }
+
     free(highbyteBuf);
 
     return mState;