Add spelling inconsistencies check

3 files changed, 238 insertions, 23 deletions

diff --git a/grammar/grammar.go b/grammar/grammar.go
index 0cde454..82eff5d 100644
--- a/grammar/grammar.go
+++ b/grammar/grammar.go
@@ -137,6 +137,11 @@ func (b *GrammarBuilder) Build() (*Grammar, error) {
 		}
 	}
 
+	b.checkSpellingInconsistenciesOfUserDefinedIDs(b.AST)
+	if len(b.errs) > 0 {
+		return nil, b.errs
+	}
+
 	symTabAndLexSpec, err := b.genSymbolTableAndLexSpec(b.AST)
 	if err != nil {
 		return nil, err
@@ -316,6 +321,71 @@ func markUsedSymbols(mark map[string]bool, marked map[string]bool, prods map[str
 	}
 }
 
+func (b *GrammarBuilder) checkSpellingInconsistenciesOfUserDefinedIDs(root *spec.RootNode) {
+	var ids []string
+	{
+		for _, prod := range root.Productions {
+			ids = append(ids, prod.LHS)
+			for _, alt := range prod.RHS {
+				for _, elem := range alt.Elements {
+					if elem.Label != nil {
+						ids = append(ids, elem.Label.Name)
+					}
+				}
+			}
+		}
+		for _, prod := range root.LexProductions {
+			ids = append(ids, prod.LHS)
+		}
+		for _, dir := range root.Directives {
+			dirIDs := collectUserDefinedIDsFromDirective(dir)
+			if len(dirIDs) > 0 {
+				ids = append(ids, dirIDs...)
+			}
+		}
+	}
+
+	duplicated := mlspec.FindSpellingInconsistencies(ids)
+	if len(duplicated) == 0 {
+		return
+	}
+
+	for _, dup := range duplicated {
+		var s string
+		{
+			var b strings.Builder
+			fmt.Fprintf(&b, "%+v", dup[0])
+			for _, id := range dup[1:] {
+				fmt.Fprintf(&b, ", %+v", id)
+			}
+			s = b.String()
+		}
+
+		b.errs = append(b.errs, &verr.SpecError{
+			Cause:  semErrSpellingInconsistency,
+			Detail: s,
+		})
+	}
+}
+
+func collectUserDefinedIDsFromDirective(dir *spec.DirectiveNode) []string {
+	var ids []string
+	for _, param := range dir.Parameters {
+		if param.Group != nil {
+			for _, d := range param.Group {
+				dIDs := collectUserDefinedIDsFromDirective(d)
+				if len(dIDs) > 0 {
+					ids = append(ids, dIDs...)
+				}
+			}
+		}
+		if param.OrderedSymbol != "" {
+			ids = append(ids, param.OrderedSymbol)
+		}
+	}
+	return ids
+}
+
 type symbolTableAndLexSpec struct {
 	symTab      *symbolTable
 	anonPat2Sym map[string]symbol
diff --git a/grammar/grammar_test.go b/grammar/grammar_test.go
index e61e422..5a8bb4a 100644
--- a/grammar/grammar_test.go
+++ b/grammar/grammar_test.go
@@ -1078,6 +1078,149 @@ func TestGrammarBuilderSpecError(t *testing.T) {
 		errs    []*SemanticError
 	}
 
+	spellingInconsistenciesTests := []*specErrTest{
+		{
+			caption: "a spelling inconsistency appears among non-terminal symbols",
+			specSrc: `
+#name test;
+
+a1
+    : a_1
+    ;
+a_1
+    : foo
+    ;
+
+foo
+    : 'foo';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among terminal symbols",
+			specSrc: `
+#name test;
+
+s
+    : foo1 foo_1
+    ;
+
+foo1
+    : 'foo1';
+foo_1
+    : 'foo_1';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among non-terminal and terminal symbols",
+			specSrc: `
+#name test;
+
+a1
+    : a_1
+    ;
+
+a_1
+    : 'a_1';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among ordered symbols whose precedence is the same",
+			specSrc: `
+#name test;
+
+#prec (
+    #assign $p1 $p_1
+);
+
+s
+    : foo #prec $p1
+    | bar #prec $p_1
+    ;
+
+foo
+    : 'foo';
+bar
+    : 'bar';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among ordered symbols whose precedence is not the same",
+			specSrc: `
+#name test;
+
+#prec (
+    #assign $p1
+    #assign $p_1
+);
+
+s
+    : foo #prec $p1
+    | bar #prec $p_1
+    ;
+
+foo
+    : 'foo';
+bar
+    : 'bar';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among labels the same alternative contains",
+			specSrc: `
+#name test;
+
+s
+    : foo@l1 foo@l_1
+    ;
+
+foo
+    : 'foo';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among labels the same production contains",
+			specSrc: `
+#name test;
+
+s
+    : foo@l1
+    | bar@l_1
+    ;
+
+foo
+    : 'foo';
+bar
+    : 'bar';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+		{
+			caption: "a spelling inconsistency appears among labels different productions contain",
+			specSrc: `
+#name test;
+
+s
+    : foo@l1
+    ;
+a
+    : bar@l_1
+    ;
+
+foo
+    : 'foo';
+bar
+    : 'bar';
+`,
+			errs: []*SemanticError{semErrSpellingInconsistency},
+		},
+	}
+
 	prodTests := []*specErrTest{
 		{
 			caption: "a production `b` is unused",
@@ -3212,6 +3355,7 @@ bar
 	}
 
 	var tests []*specErrTest
+	tests = append(tests, spellingInconsistenciesTests...)
 	tests = append(tests, prodTests...)
 	tests = append(tests, nameDirTests...)
 	tests = append(tests, precDirTests...)
diff --git a/grammar/semantic_error.go b/grammar/semantic_error.go
index 794d8da..589e324 100644
--- a/grammar/semantic_error.go
+++ b/grammar/semantic_error.go
@@ -15,27 +15,28 @@ func (e *SemanticError) Error() string {
 }
 
 var (
-	semErrNoGrammarName       = newSemanticError("name is missing")
-	semErrDuplicateAssoc      = newSemanticError("associativity and precedence cannot be specified multiple times for a symbol")
-	semErrUndefinedPrec       = newSemanticError("symbol must has precedence")
-	semErrUndefinedOrdSym     = newSemanticError("undefined ordered symbol")
-	semErrUnusedProduction    = newSemanticError("unused production")
-	semErrUnusedTerminal      = newSemanticError("unused terminal")
-	semErrTermCannotBeSkipped = newSemanticError("a terminal used in productions cannot be skipped")
-	semErrNoProduction        = newSemanticError("a grammar needs at least one production")
-	semErrUndefinedSym        = newSemanticError("undefined symbol")
-	semErrDuplicateProduction = newSemanticError("duplicate production")
-	semErrDuplicateTerminal   = newSemanticError("duplicate terminal")
-	semErrDuplicateFragment   = newSemanticError("duplicate fragment")
-	semErrDuplicateName       = newSemanticError("duplicate names are not allowed between terminals and non-terminals")
-	semErrErrSymIsReserved    = newSemanticError("symbol 'error' is reserved as a terminal symbol")
-	semErrDuplicateLabel      = newSemanticError("a label must be unique in an alternative")
-	semErrInvalidLabel        = newSemanticError("a label must differ from terminal symbols or non-terminal symbols")
-	semErrDirInvalidName      = newSemanticError("invalid directive name")
-	semErrDirInvalidParam     = newSemanticError("invalid parameter")
-	semErrDuplicateDir        = newSemanticError("a directive must not be duplicated")
-	semErrDuplicateElem       = newSemanticError("duplicate element")
-	semErrAmbiguousElem       = newSemanticError("ambiguous element")
-	semErrInvalidProdDir      = newSemanticError("invalid production directive")
-	semErrInvalidAltDir       = newSemanticError("invalid alternative directive")
+	semErrNoGrammarName         = newSemanticError("name is missing")
+	semErrSpellingInconsistency = newSemanticError("the identifiers are treated as the same. please use the same spelling")
+	semErrDuplicateAssoc        = newSemanticError("associativity and precedence cannot be specified multiple times for a symbol")
+	semErrUndefinedPrec         = newSemanticError("symbol must has precedence")
+	semErrUndefinedOrdSym       = newSemanticError("undefined ordered symbol")
+	semErrUnusedProduction      = newSemanticError("unused production")
+	semErrUnusedTerminal        = newSemanticError("unused terminal")
+	semErrTermCannotBeSkipped   = newSemanticError("a terminal used in productions cannot be skipped")
+	semErrNoProduction          = newSemanticError("a grammar needs at least one production")
+	semErrUndefinedSym          = newSemanticError("undefined symbol")
+	semErrDuplicateProduction   = newSemanticError("duplicate production")
+	semErrDuplicateTerminal     = newSemanticError("duplicate terminal")
+	semErrDuplicateFragment     = newSemanticError("duplicate fragment")
+	semErrDuplicateName         = newSemanticError("duplicate names are not allowed between terminals and non-terminals")
+	semErrErrSymIsReserved      = newSemanticError("symbol 'error' is reserved as a terminal symbol")
+	semErrDuplicateLabel        = newSemanticError("a label must be unique in an alternative")
+	semErrInvalidLabel          = newSemanticError("a label must differ from terminal symbols or non-terminal symbols")
+	semErrDirInvalidName        = newSemanticError("invalid directive name")
+	semErrDirInvalidParam       = newSemanticError("invalid parameter")
+	semErrDuplicateDir          = newSemanticError("a directive must not be duplicated")
+	semErrDuplicateElem         = newSemanticError("duplicate element")
+	semErrAmbiguousElem         = newSemanticError("ambiguous element")
+	semErrInvalidProdDir        = newSemanticError("invalid production directive")
+	semErrInvalidAltDir         = newSemanticError("invalid alternative directive")
 )