Add GB2312 encoding, and build both encoding lists from one source

src/LogExpert.Core/Classes/Log/Streamreaders/PositionAwareStreamReaderBase.cs

@@ -1,4 +1,4 @@
-using System.Text;
+using System.Text;
 
 using LogExpert.Core.Entities;
 
@@ -221,8 +221,19 @@ public static (int length, Encoding? detectedEncoding) DetectPreambleLength (Str
         return (0, null);
     }
 
+    /// <summary>
+    /// Bytes to advance the position by per character read, or 0 for "measure the character".
+    /// </summary>
+    /// <remarks>
+    /// The fallback is keyed on <see cref="Encoding.IsSingleByte"/>, not on "everything else is one
+    /// byte": a variable-width code page such as GB2312 (one byte per ASCII character, two per Chinese
+    /// one) is neither a <see cref="UTF8Encoding"/> nor a <see cref="UnicodeEncoding"/>, and counting it
+    /// as one byte per character drifts the position on the first non-ASCII character.
+    /// </remarks>
     public static int GetPosIncPrecomputed (Encoding usedEncoding)
     {
+        ArgumentNullException.ThrowIfNull(usedEncoding);
+
         switch (usedEncoding)
         {
             case UTF8Encoding:
@@ -235,7 +246,7 @@ public static int GetPosIncPrecomputed (Encoding usedEncoding)
                 }
             default:
                 {
-                    return 1;
+                    return usedEncoding.IsSingleByte ? 1 : 0;
                 }
         }
     }

src/LogExpert.Core/Helpers/EncodingRegistry.cs

@@ -27,6 +27,58 @@ namespace LogExpert.Core.Helpers;
 /// </remarks>
 public static class EncodingRegistry
 {
+    /// <summary>
+    /// Code page of GB2312, the simplified-Chinese code page (issue #688).
+    /// </summary>
+    public const int CODE_PAGE_GB2312 = 936;
+
+    /// <summary>
+    /// The encodings LogExpert offers the user to choose from, in the order they are presented.
+    /// </summary>
+    /// <remarks>
+    /// The single list: the Preferences default-encoding combo box and the per-file View → Encoding
+    /// menu both draw their rows from it, so they cannot drift apart and adding an encoding is a
+    /// one-line change here. Three invariants, each pinned by a test:
+    /// <list type="bullet">
+    ///   <item>
+    ///     Every entry resolves from its own <see cref="Encoding.HeaderName"/>, because that name is
+    ///     what gets persisted (in the settings JSON and in a .lxp) and resolved again on the next
+    ///     start. An entry that did not would silently degrade to a fallback.
+    ///   </item>
+    ///   <item>
+    ///     No two entries share a code page. Both UIs label a row with its header name, so two entries
+    ///     for one code page are two rows the user cannot tell apart — which is exactly what
+    ///     <c>Encoding.Default</c> alongside <see cref="Encoding.UTF8"/> produced (both are code page
+    ///     65001 on .NET, differing only in the BOM they emit, which a read-side encoding never uses).
+    ///     Hence no <c>Encoding.Default</c> entry.
+    ///   </item>
+    ///   <item>
+    ///     New entries are appended rather than sorted in, so an existing user's row does not move
+    ///     under the cursor on upgrade.
+    ///   </item>
+    /// </list>
+    /// </remarks>
+    public static IReadOnlyList<Encoding> OfferedEncodings => _offeredEncodings.Value;
+
+    /// <remarks>
+    /// Built once: <see cref="Encoding"/> instances are immutable and the ones from
+    /// <see cref="GetEncoding(int)"/> are cached by .NET anyway, so the list is safe to share. Lazy
+    /// rather than a plain initialiser so the code pages resolve through
+    /// <see cref="EnsureRegistered"/> rather than depending on when this class is first touched.
+    /// </remarks>
+    private static readonly Lazy<IReadOnlyList<Encoding>> _offeredEncodings = new(
+        () =>
+        [
+            Encoding.ASCII,
+            Encoding.Latin1,
+            Encoding.UTF8,
+            Encoding.Unicode,
+            GetEncoding(1250),
+            GetEncoding(1252),
+            GetEncoding(CODE_PAGE_GB2312)
+        ],
+        LazyThreadSafetyMode.ExecutionAndPublication);
+
     /// <summary>
     /// Registers <see cref="CodePagesEncodingProvider"/> on first use.
     /// </summary>

src/LogExpert.Resources/Resources.de.resx

@@ -1324,21 +1324,6 @@ Ein ausgewähltes Tool erscheint in der Iconbar. Alle anderen verfügbaren Tools
   <data name="LogTabWindow_UI_ToolStripMenuItem_encodingToolStripMenuItem" xml:space="preserve">
 		<value>Encoding</value>
 	</data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingASCIIToolStripMenuItem" xml:space="preserve">
-		<value>ASCII</value>
-	</data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingANSIToolStripMenuItem" xml:space="preserve">
-		<value>ANSI</value>
-	</data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingISO88591toolStripMenuItem" xml:space="preserve">
-		<value>ISO-8859-1</value>
-	</data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF8toolStripMenuItem" xml:space="preserve">
-		<value>UTF8</value>
-	</data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF16toolStripMenuItem" xml:space="preserve">
-		<value>Unicode</value>
-	</data>
   <data name="LogTabWindow_UI_ToolStripTextBox_timeshiftToolStripTextBox" xml:space="preserve">
 		<value>+00:00:00.000</value>
 	</data>

src/LogExpert.Resources/Resources.resx

@@ -1347,21 +1347,6 @@ Checked tools will appear in the icon bar. All other tools are available in the
   <data name="LogTabWindow_UI_ToolStripMenuItem_encodingToolStripMenuItem" xml:space="preserve">
         <value>Encoding</value>
     </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingASCIIToolStripMenuItem" xml:space="preserve">
-        <value>ASCII</value>
-    </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingANSIToolStripMenuItem" xml:space="preserve">
-        <value>ANSI</value>
-    </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingISO88591toolStripMenuItem" xml:space="preserve">
-        <value>ISO-8859-1</value>
-    </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF8toolStripMenuItem" xml:space="preserve">
-        <value>UTF8</value>
-    </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF16toolStripMenuItem" xml:space="preserve">
-        <value>Unicode</value>
-    </data>
   <data name="LogTabWindow_UI_ToolStripTextBox_timeshiftToolStripTextBox" xml:space="preserve">
         <value>+00:00:00.000</value>
     </data>

src/LogExpert.Resources/Resources.zh-CN.resx

@@ -1966,21 +1966,6 @@ YY[YY] = 年
   <data name="LogTabWindow_UI_MenuStrip_mainMenuStrip" xml:space="preserve">
     <value>菜单条1</value>
   </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingASCIIToolStripMenuItem" xml:space="preserve">
-    <value>ASCII码</value>
-  </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingANSIToolStripMenuItem" xml:space="preserve">
-    <value>美国国家标准协会</value>
-  </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingISO88591toolStripMenuItem" xml:space="preserve">
-    <value>ISO-8859-1</value>
-  </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF8toolStripMenuItem" xml:space="preserve">
-    <value>UTF8</value>
-  </data>
-  <data name="LogTabWindow_UI_ToolStripMenuItem_encodingUTF16toolStripMenuItem" xml:space="preserve">
-    <value>统一码</value>
-  </data>
   <data name="LogTabWindow_UI_ToolStripTextBox_timeshiftToolStripTextBox" xml:space="preserve">
     <value>+00:00:00.000</value>
   </data>

src/LogExpert.Tests/Dialogs/SettingsDialogEncodingListTests.cs

@@ -8,122 +8,64 @@
 namespace LogExpert.Tests.Dialogs;
 
 /// <summary>
-/// The Preferences encoding dropdown is the only way to set <c>Preferences.DefaultEncoding</c>, so the
-/// list is asserted directly — building the dialog is not needed to know what it offers.
+/// The Preferences encoding combo box, which is the only way to set <c>Preferences.DefaultEncoding</c>.
+/// What it offers is <see cref="EncodingRegistry.OfferedEncodings"/> and is asserted in
+/// <c>OfferedEncodingsTests</c>; what is left here is the part only a real
+/// <see cref="ComboBox"/> can answer.
 /// </summary>
 [TestFixture]
 public class SettingsDialogEncodingListTests
 {
-    [Test]
-    [TestCase(1250, TestName = "GetAvailableEncodings_OffersWindows1250")]
-    [TestCase(1252, TestName = "GetAvailableEncodings_OffersWindows1252")]
-    public void GetAvailableEncodings_OffersLegacyCodePage (int codePage)
-    {
-        var encodings = SettingsDialog.GetAvailableEncodings();
-
-        Assert.That(encodings.Select(encoding => encoding.CodePage), Does.Contain(codePage));
-    }
-
     /// <summary>
-    /// Every offered encoding is saved as its name and resolved from that name on the next start, so a
-    /// name that cannot be resolved again would silently degrade to
-    /// <see cref="SettingsDialog.FallbackEncoding"/>.
+    /// The persist/restore round trip against a real <see cref="ComboBox"/>, configured the way the
+    /// dialog configures it: the reselect goes through <c>Items.IndexOf</c>, and WinForms — not NUnit —
+    /// has the final say on whether a row is reachable. Deliberately kept alongside the plain assertion
+    /// in <c>OfferedEncodingsTests</c> rather than replacing it: this one needs an STA apartment and a
+    /// WinForms control, and the invariant should still be pinned where that is unavailable.
     /// </summary>
     [Test]
-    public void GetAvailableEncodings_EveryEntryResolvesByItsPersistedName ()
-    {
-        var encodings = SettingsDialog.GetAvailableEncodings();
-
-        Assert.Multiple(() =>
-        {
-            foreach (var encoding in encodings)
-            {
-                Assert.That(
-                    EncodingRegistry.TryGetEncoding(encoding.HeaderName, out _),
-                    Is.True,
-                    $"'{encoding.HeaderName}' cannot be resolved back from a saved preference");
-            }
-        });
-    }
-
-    /// <summary>
-    /// The dropdown renders each entry as its <see cref="Encoding.HeaderName"/>, so two entries sharing a
-    /// code page are two rows the user cannot tell apart. <c>Encoding.Default</c> and
-    /// <see cref="Encoding.UTF8"/> are both code page 65001 on .NET and both read <c>utf-8</c>.
-    /// </summary>
-    [Test]
-    public void GetAvailableEncodings_NoTwoEntriesShareACodePage ()
+    [Apartment(ApartmentState.STA)]
+    public void EncodingComboBox_EveryOfferedRowIsReselectableAfterARestart ()
     {
-        var codePages = SettingsDialog.GetAvailableEncodings().Select(encoding => encoding.CodePage);
-
-        Assert.That(codePages, Is.Unique);
-    }
+        var encodings = EncodingRegistry.OfferedEncodings;
 
-    /// <summary>
-    /// The selection is persisted by name and restored with
-    /// <c>comboBoxEncoding.SelectedItem = EncodingRegistry.GetEncoding(name, …)</c>. WinForms locates that
-    /// item with <see cref="object.Equals(object)"/>, so an entry whose name resolves back to an instance
-    /// that is not equal to it can never be reselected — the row goes dead after the first restart.
-    /// </summary>
-    [Test]
-    public void GetAvailableEncodings_EveryEntryIsReselectableAfterARoundTrip ()
-    {
-        var encodings = SettingsDialog.GetAvailableEncodings();
+        using var comboBox = new ComboBox { DropDownStyle = ComboBoxStyle.DropDownList, FormattingEnabled = true };
+        comboBox.Items.AddRange([.. encodings]);
 
         Assert.Multiple(() =>
         {
-            foreach (var encoding in encodings)
+            for (var row = 0; row < encodings.Count; row++)
             {
-                var restored = EncodingRegistry.GetEncoding(encoding.HeaderName, SettingsDialog.FallbackEncoding);
+                comboBox.SelectedIndex = row;
+                var saved = ((Encoding)comboBox.SelectedItem).HeaderName;
+
+                comboBox.SelectedIndex = -1;
+                comboBox.SelectedItem = EncodingRegistry.GetEncoding(saved, SettingsDialog.FallbackEncoding);
 
-                Assert.That(
-                    restored,
-                    Is.EqualTo(encoding),
-                    $"picking '{encoding.HeaderName}' saves a name that reselects a different entry");
+                Assert.That(comboBox.SelectedIndex, Is.EqualTo(row), $"row {row} ('{saved}') is unreachable after a restart");
             }
         });
     }
 
     /// <summary>
-    /// <c>FillDialog</c> and <c>SavePreferences</c> fall back to this encoding when the persisted name is
-    /// unusable, so it has to be an offered entry — otherwise the dropdown shows no selection and OK
-    /// rewrites the preference to something that was never on the list.
-    /// </summary>
-    [Test]
-    public void GetAvailableEncodings_OffersTheFallbackEncoding ()
-    {
-        var encodings = SettingsDialog.GetAvailableEncodings();
-
-        Assert.That(encodings, Does.Contain(SettingsDialog.FallbackEncoding));
-    }
-
-    /// <summary>
-    /// The same round trip against a real <see cref="ComboBox"/>, because the reselect actually goes
-    /// through <c>Items.IndexOf</c> — WinForms, not NUnit, has the final say on whether a row is
-    /// reachable. Deliberately kept alongside the assertion above rather than replacing it: this one
-    /// needs an STA apartment and a WinForms control, and the invariant should still be pinned where
-    /// that is unavailable.
+    /// The combo box has no <c>DisplayMember</c> — the dialog only sets <c>ValueMember</c>, and WinForms
+    /// falls back to it for the display text. Pinned because without that fallback every row would render
+    /// as <c>Encoding.ToString()</c>, i.e. a .NET type name, and the rows the user picks between would be
+    /// indistinguishable for a different reason than the one issue #688 reported.
     /// </summary>
     [Test]
     [Apartment(ApartmentState.STA)]
-    public void GetAvailableEncodings_EveryOfferedRowIsReselectableInARealComboBox ()
+    public void EncodingComboBox_RendersEachRowAsItsHeaderName ()
     {
-        var encodings = SettingsDialog.GetAvailableEncodings();
-
-        using var comboBox = new ComboBox { FormattingEnabled = true };
-        comboBox.Items.AddRange([.. encodings]);
+        using var comboBox = new ComboBox { DropDownStyle = ComboBoxStyle.DropDownList, FormattingEnabled = true };
+        comboBox.ValueMember = "HeaderName";
+        comboBox.Items.AddRange([.. EncodingRegistry.OfferedEncodings]);
 
         Assert.Multiple(() =>
         {
-            for (var row = 0; row < encodings.Count; row++)
+            foreach (Encoding encoding in comboBox.Items)
             {
-                comboBox.SelectedIndex = row;
-                var saved = ((Encoding)comboBox.SelectedItem).HeaderName;
-
-                comboBox.SelectedIndex = -1;
-                comboBox.SelectedItem = EncodingRegistry.GetEncoding(saved, SettingsDialog.FallbackEncoding);
-
-                Assert.That(comboBox.SelectedIndex, Is.EqualTo(row), $"row {row} ('{saved}') is unreachable after a restart");
+                Assert.That(comboBox.GetItemText(encoding), Is.EqualTo(encoding.HeaderName));
             }
         });
     }

src/LogExpert.Tests/Encodings/EncodingJsonConverterTests.cs

@@ -20,6 +20,7 @@ public class EncodingJsonConverterTests
     [Test]
     [TestCase("windows-1250", 1250)]
     [TestCase("windows-1252", 1252)]
+    [TestCase("gb2312", 936)]
     public void ReadJson_LegacyCodePageName_ResolvesInsteadOfFallingBack (string encodingName, int expectedCodePage)
     {
         var encoding = Deserialize($"\"{encodingName}\"");

src/LogExpert.Tests/Encodings/EncodingRegistryTests.cs

@@ -18,6 +18,7 @@ public class EncodingRegistryTests
     [Test]
     [TestCase(1250)]
     [TestCase(1252)]
+    [TestCase(936)]
     public void GetEncoding_LegacyCodePage_Resolves (int codePage)
     {
         var encoding = EncodingRegistry.GetEncoding(codePage);
@@ -30,6 +31,7 @@ public void GetEncoding_LegacyCodePage_Resolves (int codePage)
     [TestCase("windows-1252")]
     [TestCase("utf-8")]
     [TestCase("iso-8859-1")]
+    [TestCase("gb2312")]
     public void TryGetEncoding_SupportedName_ReturnsTrueAndEncoding (string name)
     {
         var resolved = EncodingRegistry.TryGetEncoding(name, out var encoding);