1+ using Antlr4 . Runtime ;
2+ using Antlr4 . Runtime . Misc ;
3+ using MuConvert . Antlr ;
4+ using MuConvert . utils ;
5+ using static MuConvert . utils . Alert . LEVEL ;
6+
7+ namespace MuConvert . parser ;
8+
9+ public class ErrorListener ( SimaiParser simaiParser ) : BaseErrorListener , IAntlrErrorListener < int >
10+ {
11+ // 语法分析的错误报告函数
12+ public override void SyntaxError ( TextWriter output , IRecognizer recognizer , IToken offendingSymbol , int line , int charPositionInLine , string msg , RecognitionException ? e )
13+ {
14+ var parser = ( Parser ) recognizer ;
15+ if ( e == null )
16+ { // 被recoverInline了。此时给个警告就可以了。（PS:只有在宽松模式下才会触发，严格模式是禁止recoverInline、直接丢InputMismatch的）
17+ if ( msg . StartsWith ( "extraneous" ) )
18+ {
19+ simaiParser . alerts . Add ( new Alert ( Warning ,
20+ string . Format ( Locale . RecoverInlineExtraneousToken , GetTokenErrorDisplay ( offendingSymbol ) ) ,
21+ line : line , relevantNote : _contextText ( parser . Context ) ) ) ;
22+ return ;
23+ }
24+ else if ( msg . StartsWith ( "missing" ) )
25+ {
26+ simaiParser . alerts . Add ( new Alert ( Warning ,
27+ string . Format ( Locale . RecoverInlineMissingToken , GetTokenErrorDisplay ( offendingSymbol ) , parser . GetExpectedTokens ( ) . ToString ( parser . Vocabulary ) ) ,
28+ line : line , relevantNote : _contextText ( parser . Context ) ) ) ;
29+ return ;
30+ }
31+ }
32+
33+ Alert . LEVEL level = simaiParser . StrictLevel == SimaiParser . StrictLevelEnum . Lax ? Warning : Error ;
34+ string msgPostFix = simaiParser . StrictLevel == SimaiParser . StrictLevelEnum . Lax ? Locale . LaxTryfixReminder : "" ;
35+ string message ;
36+ switch ( e )
37+ {
38+ case InputMismatchException :
39+ message = string . Format ( Locale . InputMismatchException , GetTokenErrorDisplay ( offendingSymbol ) , e . GetExpectedTokens ( ) . ToString ( recognizer . Vocabulary ) ) ;
40+ message += msgPostFix ;
41+ break ;
42+ case NoViableAltException ne :
43+ var iSt = ( ITokenStream ) parser . InputStream ;
44+ message = string . Format ( Locale . NoViableAltException , EscapeWSAndQuote ( iSt == null ? "<unknown input>" : ( ne . StartToken . Type != - 1 ? iSt . GetText ( ne . StartToken , ne . OffendingToken ) : "<EOF>" ) ) ) ;
45+ message += msgPostFix ;
46+ break ;
47+ default :
48+ message = string . Format ( Locale . AntlrUnknownError , msg ) ;
49+ break ;
50+ }
51+ simaiParser . alerts . Add ( new Alert ( level , message , line : line , relevantNote : _contextText ( parser . Context ) ) ) ;
52+ }
53+
54+ // 词法分析的错误报告函数
55+ public void SyntaxError ( TextWriter output , IRecognizer recognizer , int offendingSymbol , int line , int charPositionInLine ,
56+ string msg , RecognitionException e )
57+ {
58+ var lexer = ( Lexer ) recognizer ;
59+ // 遵照Lexer.NotifyListeners的实现写出的，获得出错的token的详细内容的代码。
60+ var input = ( ICharStream ) lexer . InputStream ;
61+ string errContent = lexer . GetErrorDisplay ( input . GetText ( Interval . Of ( lexer . TokenStartCharIndex , input . Index ) ) ) ;
62+
63+ if ( simaiParser . StrictLevel == SimaiParser . StrictLevelEnum . Strict )
64+ { // 严格模式下，不准恢复，抛异常
65+ simaiParser . alerts . Add ( new Alert ( Error ,
66+ string . Format ( Locale . LexerNoViableAltExceptionStrict , errContent ) ,
67+ line : line , relevantNote : errContent ) ) ;
68+ throw new ParseCanceledException ( e ) ;
69+ }
70+ simaiParser . alerts . Add ( new Alert ( Warning ,
71+ string . Format ( Locale . LexerNoViableAltException , errContent ) ,
72+ line : line , relevantNote : errContent ) ) ;
73+ }
74+
75+ // 从context获得为适合放进relevantNote里的形式
76+ private string ? _contextText ( RuleContext ? context )
77+ {
78+ while ( true )
79+ {
80+ if ( context == null ) return null ;
81+ if ( context . GetText ( ) . Length >= 5 ) return context . GetText ( ) ;
82+ context = context . Parent ;
83+ }
84+ }
85+
86+ # region 用于暴露DefaultErrorStrategy内部的GetTokenErrorDisplay函数
87+ private class _ES : DefaultErrorStrategy
88+ {
89+ public new string GetTokenErrorDisplay ( IToken t ) => base . GetTokenErrorDisplay ( t ) ;
90+ public new string EscapeWSAndQuote ( string s ) => base . EscapeWSAndQuote ( s ) ;
91+ }
92+ private _ES _es = new ( ) ; // 仅用作调用里面的方法
93+ private string GetTokenErrorDisplay ( IToken t ) => _es . GetTokenErrorDisplay ( t ) ;
94+ private string EscapeWSAndQuote ( string s ) => _es . EscapeWSAndQuote ( s ) ;
95+ # endregion
96+ }
97+
98+ /**
99+ * 最宽松的ErrorStrategy，尽全力恢复不让谱面整个垮掉
100+ */
101+ public class LaxErrorStrategy : DefaultErrorStrategy
102+ {
103+ protected override IToken SingleTokenDeletion ( Parser recognizer )
104+ {
105+ if ( recognizer . CurrentToken ? . Type == SimaiLexer . COMMA ) return null ! ; // 不准删逗号
106+ return base . SingleTokenDeletion ( recognizer ) ;
107+ }
108+
109+ private HashSet < int > insertionForbidden = [
110+ SimaiLexer . COMMA , SimaiLexer . KEY , SimaiLexer . SLIDE_TYPE , SimaiLexer . TOUCH_AREA , SimaiLexer . INT ,
111+ SimaiLexer . CHART_END , SimaiLexer . MODIFIER , SimaiLexer . FALSE_EACH
112+ ] ; // 逗号，和不确定的可能引起歧义的符号，一律不允许补充
113+
114+ protected override IToken GetMissingSymbol ( Parser recognizer )
115+ {
116+ IToken currentToken = recognizer . CurrentToken ;
117+
118+ // 不准插入insertionForbidden里提到的元素
119+ var insertionCandidates = GetExpectedTokens ( recognizer ) . ToList ( ) . Where ( x => ! insertionForbidden . Contains ( x ) ) . ToList ( ) ;
120+ if ( insertionCandidates . Count == 0 ) throw new InputMismatchException ( recognizer ) ; // 等价于SingleTokenInsertion返回false的情况，recoverInline失败、转交给上层recover处理
121+ int minElement = insertionCandidates [ 0 ] ;
122+
123+ string tokenText = minElement != - 1 ? $ "<missing { recognizer . Vocabulary . GetDisplayName ( minElement ) } >": "<missing EOF>" ;
124+ IToken current = currentToken ;
125+ IToken token = ( ( ITokenStream ) recognizer . InputStream ) . LT ( - 1 ) ;
126+ if ( current . Type == - 1 && token != null )
127+ current = token ;
128+ return this . ConstructToken ( ( ( ITokenStream ) recognizer . InputStream ) . TokenSource , minElement , tokenText , current ) ;
129+ }
130+
131+ private static Dictionary < string , int > _literals = Enumerable . Range ( 1 , SimaiLexer . ruleNames . Length )
132+ . ToDictionary ( i => SimaiLexer . DefaultVocabulary . GetLiteralName ( i ) , i => i ) ;
133+ private List < int > recoverySetAllowed = [
134+ SimaiLexer . COMMA , SimaiLexer . FALSE_EACH , _literals [ "'/'" ] , _literals [ "'('" ] , _literals [ "'{'" ]
135+ ] ; // recover时，为了确保整个吞掉不合法的音符，而不是出现残缺的东西导致parser报错，只准同步到上面这些字符当中
136+
137+ public override void Recover ( Parser recognizer , RecognitionException e )
138+ {
139+ if ( this . lastErrorIndex == recognizer . InputStream . Index && this . lastErrorStates != null && this . lastErrorStates . Contains ( recognizer . State ) )
140+ recognizer . Consume ( ) ;
141+ this . lastErrorIndex = recognizer . InputStream . Index ;
142+ if ( this . lastErrorStates == null )
143+ this . lastErrorStates = new IntervalSet ( Array . Empty < int > ( ) ) ;
144+ this . lastErrorStates . Add ( recognizer . State ) ;
145+ IntervalSet errorRecoverySet = this . GetErrorRecoverySet ( recognizer ) ;
146+
147+ // 和上面的recoverySetAllowed取交集
148+ errorRecoverySet = new IntervalSet ( errorRecoverySet . ToList ( ) . Where ( x => recoverySetAllowed . Contains ( x ) ) . ToArray ( ) ) ;
149+
150+ this . ConsumeUntil ( recognizer , errorRecoverySet ) ;
151+ }
152+ }
153+
154+ /**
155+ * 允许recoverInline，但是禁止大范围recover。
156+ */
157+ public class ModerateErrorStrategy : LaxErrorStrategy
158+ {
159+ private BailErrorStrategy _bail = new ( ) ;
160+
161+ public override void Recover ( Parser recognizer , RecognitionException e ) => _bail . Recover ( recognizer , e ) ; // 不准recover，只准recoverInline
162+ }
0 commit comments