[R] 重构双押纠错的实现方式，重构部分函数名，增加一些测试

Author: Starrah

i18n/Locale.Designer.cs

@@ -83,7 +83,7 @@
     <value>bar time {0:W}({1:F2}s)</value>
   </data>
   <data name="MessageParsing" xml:space="preserve">
-    <value>parsing {0}</value>
+    <value>parsing "{0}"</value>
   </data>
   <data name="Error" xml:space="preserve">
     <value>ERROR:</value>

i18n/Locale.resx

@@ -83,7 +83,7 @@
     <value>第{0:W}小节({1:F2}秒)</value>
   </data>
   <data name="MessageParsing" xml:space="preserve">
-    <value>解析 {0} 时</value>
+    <value>解析 "{0}" 时</value>
   </data>
   <data name="Error" xml:space="preserve">
     <value>发生错误:</value>

i18n/Locale.zh-hans.resx

@@ -83,7 +83,7 @@
     <value>第{0:W}小節({1:F2}秒)</value>
   </data>
   <data name="MessageParsing" xml:space="preserve">
-    <value>解析 {0} 時</value>
+    <value>解析 "{0}" 時</value>
   </data>
   <data name="Error" xml:space="preserve">
     <value>發生錯誤：</value>

i18n/Locale.zh-hant.resx

@@ -46,8 +46,8 @@ chart: (notations COMMA)* CHART_END? EOF;
 notations: (bpmTag | absulouteStepTag | metTag)* noteGroup?;
 
 noteGroup: note eachNote*;
-FALSE_EACH: '`';
-eachNote: (sep+=('/' | FALSE_EACH))+ note;
+FALSE_EACH: '`'+;
+eachNote: sep=('/' | FALSE_EACH) note?;
 
 bpmTag: (lp+='(')+ number (rp+=')')+;
 absulouteStepTag: (lp+='{')+ '#' number (rp+='}')+;

parser/simai/Simai.g4

@@ -242,22 +242,28 @@ public sealed override object VisitNotations(P.NotationsContext context)
         return true;
     }
 
-    private void WarnMoreThanOneTokens(IList<IToken> ps)
+    private void AlertExtraToken(string extraStr)
     {
-        if (ps.Count <= 1) return;
-        var extraStr = "'" + string.Join("", ps.Skip(1).Select(x => x.Text)) + "'";
+        if (extraStr.First() != '\'') extraStr = "'" + extraStr + "'";
         if (StrictLevel == StrictLevelEnum.Strict)
         { // 严格模式，抛异常
             AddAlert(Error, string.Format(Locale.RecoverInlineExtraneousTokenStrict, extraStr));
             throw new ConversionException(alerts);
         }
         else AddAlert(Warning, string.Format(Locale.RecoverInlineExtraneousToken, extraStr));
     }
+
+    private void AlertIfMoreThanOneTokens(IList<IToken> ps)
+    {
+        if (ps.Count <= 1) return;
+        var extraStr = string.Join("", ps.Skip(1).Select(x => x.Text));
+        AlertExtraToken(extraStr);
+    }
 
-    private void WarnMoreParentheses(IList<IToken> lp, IList<IToken> rp)
+    private void AlertIfMoreParentheses(IList<IToken> lp, IList<IToken> rp)
     {
-        WarnMoreThanOneTokens(lp);
-        WarnMoreThanOneTokens(rp);
+        AlertIfMoreThanOneTokens(lp);
+        AlertIfMoreThanOneTokens(rp);
     }
 
     public sealed override object VisitAbsulouteStepTag(P.AbsulouteStepTagContext context)
@@ -269,7 +275,7 @@ public sealed override object VisitAbsulouteStepTag(P.AbsulouteStepTagContext co
             absoluteTimeStepWarned = true;
         }
         currContext = context;
-        WarnMoreParentheses(context._lp, context._rp);
+        AlertIfMoreParentheses(context._lp, context._rp);
         absoluteTimeStep = (decimal)VisitNumber(context.number());
         var currentBpm = chart.BpmList.Last().Bpm;
         step = (Rational)absoluteTimeStep / (240 / (Rational)currentBpm);
@@ -280,7 +286,7 @@ public sealed override object VisitBpmTag(P.BpmTagContext context)
     {
         if (SubtreeHasException(context)) return false; // 如果本节点下有异常，则直接整个吞掉，（避免具体的规则遇到不完整子树、爆出更不可预测的错误）
         currContext = context;
-        WarnMoreParentheses(context._lp, context._rp);
+        AlertIfMoreParentheses(context._lp, context._rp);
         var bpm = (decimal)VisitNumber(context.number());
         chart.BpmList.Add(new BPM(now, bpm));
         if (absoluteTimeStep != null)
@@ -294,7 +300,7 @@ public sealed override object VisitMetTag(P.MetTagContext context)
     { // metTag指的是标记分音的tag，如{4}
         if (SubtreeHasException(context)) return false; // 如果本节点下有异常，则直接整个吞掉，（避免具体的规则遇到不完整子树、爆出更不可预测的错误）
         currContext = context;
-        WarnMoreParentheses(context._lp, context._rp);
+        AlertIfMoreParentheses(context._lp, context._rp);
         var quaver = int.Parse(context.@int().GetText());
         step = new Rational(1, quaver);
         absoluteTimeStep = null;
@@ -313,28 +319,31 @@ public sealed override object VisitNoteGroup(P.NoteGroupContext context)
             else if (child is P.EachNoteContext c2)
             {
                 noteC = c2.note();
-                
-                var separators = c2._sep;
-                if (separators.Count >= 2 && separators.All(x=>x.Type == L.FALSE_EACH))
+                if (noteC == null)
                 {
-                    // 出现连续多个反引号的情况，如"2``3"。
-                    // 这并不是标准的simai语法。但是，MajdataView中对此提供了支持，将每个`实现为128分音。
-                    // 因此，我们也支持这一特性，在遇到大于一个`时，不实现成FalseEachIndex，而是直接给予相同的实现、每个`错后128分音。
-                    var length = separators.Count * new Rational(1, 128);
-                    now = (now + length).CanonicalForm;
-                    extendedFalseEach += length;
-                    falseEachIdx = 0;
-                    if (!extendedFalseEachWarned)
-                    {
-                        AddAlert(Warning, Locale.ExtenedFalseEach, context);
-                        extendedFalseEachWarned = true;
-                    }
+                    AlertExtraToken(c2.sep.Text);
+                    continue;
                 }
-                else
+                if (c2.sep.Type == L.FALSE_EACH)
                 {
-                    WarnMoreThanOneTokens(separators);
-                    if (separators[0].Type == L.FALSE_EACH) falseEachIdx++;
+                    if (c2.sep.Text.Length >= 2)
+                    {
+                        // 出现连续多个反引号的情况，如"2``3"。
+                        // 这并不是标准的simai语法。但是，MajdataView中对此提供了支持，将每个`实现为128分音。
+                        // 因此，我们也支持这一特性，在遇到大于一个`时，不实现成FalseEachIndex，而是直接给予相同的实现、每个`错后128分音。
+                        var length = c2.sep.Text.Length * new Rational(1, 128);
+                        now = (now + length).CanonicalForm;
+                        extendedFalseEach += length;
+                        falseEachIdx = 0;
+                        if (!extendedFalseEachWarned)
+                        {
+                            AddAlert(Warning, Locale.ExtenedFalseEach, context);
+                            extendedFalseEachWarned = true;
+                        }
+                    }
+                    else falseEachIdx++; // 普通的伪双押
                 }
+                // else 是普通双押符号'/'。无需做任何特殊处理，正常解析noteContext就好。
             }
             else throw Utils.Fail();
 
@@ -459,7 +468,7 @@ public sealed override object VisitDuration(P.DurationContext? context)
             result.InvariantBar = 0;
             return result;
         }
-        WarnMoreParentheses(context._lp, context._rp);
+        AlertIfMoreParentheses(context._lp, context._rp);
         if (context.beats() != null) result.InvariantBar = (Rational)VisitBeats(context.beats());
         else result.Seconds = (Rational)(decimal)VisitNumber(context.number());        
         return result;
@@ -505,7 +514,7 @@ public sealed override object VisitSlideDuration(P.SlideDurationContext context)
         var result = new Duration(currNote!);
         Duration? waitTime = null;
         isRealExactWaitTime = false;
-        WarnMoreParentheses(context._lp, context._rp);
+        AlertIfMoreParentheses(context._lp, context._rp);
 
         if (context.waitTime() != null)
         {

parser/simai/SimaiParser.cs

@@ -10,11 +10,11 @@ namespace MuConvert.Tests;
 /// 另含带 <c>#</c>、<c>||</c> 行注释的谱面（见 <see cref="Comment_Cases"/> / <see cref="含有注释"/>），与 Preprocess 去注释行为对齐。
 /// 在 <see cref="SimaiParser.Preprocess"/> 尚未接入对应替换逻辑前，本文件中的测试失败或抛错属于预期行为（TDD）。
 /// </summary>
-public class Simai预处理纠错测试
+public class Simai纠错测试
 {
     private readonly ITestOutputHelper _output;
 
-    public Simai预处理纠错测试(ITestOutputHelper output) => _output = output;
+    public Simai纠错测试(ITestOutputHelper output) => _output = output;
 
     /// <summary>规范 simai → 完整 MA2 文本（含头与 BPM）。</summary>
     private string SimaiToMa2(string inote, bool DontTryFix = false)
@@ -109,6 +109,12 @@ public static IEnumerable<object[]> TryFix_Cases()
             "(120){4}1/2,2/3,4,E",
             "(120){4}1//2,2//3,4,E"
         ];
+        yield return
+        [
+            "1/ 后没有音符",
+            "(120){4}1,(90)2,3,",
+            "(120){4}1/,(90)2,3,"
+        ];
     }
 
     public static IEnumerable<object[]> Comment_Cases()
@@ -200,4 +206,71 @@ public void 含有注释(string description, string canonicalSimai, string malfo
         var actual = SimaiToMa2(malformedSimai, true);
         Assert.Equal(expected, actual);
     }
+
+    /// <summary>
+    /// 仅覆盖 <see cref="Lax_Scratch_Cases"/>（scratch NewFile1..6）；以 Lax 解析 <paramref name="malformedSimai"/>，
+    /// 结果应与规范串在 Strict 下得到的 MA2 一致。
+    /// </summary>
+    [Theory]
+    [MemberData(nameof(Lax_Scratch_Cases))]
+    public void Lax模式测试(string description, string canonicalSimai, string malformedSimai)
+    {
+        _output.WriteLine(description);
+
+        string expected;
+        try
+        {
+            expected = SimaiToMa2(canonicalSimai, true);
+        }
+        catch (Exception e)
+        {
+            throw new InvalidOperationException($"Canonical simai failed (fix test data): {description}", e);
+        }
+
+        var actual = SimaiToMa2Lax(malformedSimai);
+        Assert.Equal(expected, actual);
+    }
+
+    /// <summary>
+    /// 来自 <c>tests/testset/scratch/NewFile1..6.txt</c> 的片段；规范串由 Lax 解析后 <see cref="SimaiGenerator"/> 反写得到。
+    /// </summary>
+    public static IEnumerable<object[]> Lax_Scratch_Cases()
+    {
+        yield return
+        [
+            "无效 &",
+            "(120){4}1,2,3,",
+            "(120){4}1,2&,3,"
+        ];
+        yield return
+        [
+            "数字与 '(' BPM 缺逗号",
+            "(120){4}1,(120)2,3,",
+            "(120){4}1(120)2,3,"
+        ];
+        yield return
+        [
+            "无法解析的 '-2,'（Lax 丢弃）",
+            "(120){4},3,4,",
+            "(120){4}1-2,3,4,"
+        ];
+        yield return
+        [
+            "1/ 后没有音符",
+            "(120){4}1,(90)2,3,",
+            "(120){4}1/,(90)2,3,"
+        ];
+        yield return
+        [
+            "多余的 '['",
+            "(120){4}1,2,3,",
+            "(120){4}1[,2,3,"
+        ];
+        yield return
+        [
+            "Hold 时长损坏（Lax 丢弃该 Hold）",
+            "(120){4}1,,3,4-5[6:1],7,8,E",
+            "(120){4}1,2h[3:,3,4-5[6:1],7,8,E"
+        ];
+    }
 }