Raku: Fix detecting adverbs, plus a little cleanup

CIAvash · alecthomas · commit bb38ae204f9b · 2021-08-29T07:10:23.000+10:00
diff --git a/lexers/r/raku.go b/lexers/r/raku.go
@@ -49,8 +49,10 @@ func rakuRules() Rules {
 	const (
 		colonPairOpeningBrackets = `(?:<<|<|«|\(|\[|\{)`
 		colonPairClosingBrackets = `(?:>>|>|»|\)|\]|\})`
-		colonPairPattern         = `(?<colon>:)(?<key>\w[\w'-]*)(?<opening_delimiters>` + colonPairOpeningBrackets + `)`
-		namePattern              = `((?:(?!` + colonPairPattern + `)[\w':-])+)`
+		colonPairPattern         = `(?<!:)(?<colon>:)(?<key>\w[\w'-]*)(?<opening_delimiters>` + colonPairOpeningBrackets + `)`
+		colonPairLookahead       = `(?=(:['\w-]+` +
+			colonPairOpeningBrackets + `.+?` + colonPairClosingBrackets + `)?`
+		namePattern              = `((?:(?!` + colonPairPattern + `)(?:::|[\w':-]))+)`
 		variablePattern          = `[$@%&]+[.^:?=!~]?` + namePattern
 		globalVariablePattern    = `[$@%&]+\*` + namePattern
 	)
@@ -601,18 +603,10 @@ func rakuRules() Rules {
 			Include("colon-pair"),
 			// Token
 			{
-				// Token with adverbs
-				`(?<=(?:^|\s)(?:regex|token|rule)(\s+))(['\w:-]+)(?=:['\w-]+` +
-					colonPairOpeningBrackets + `.+?` + colonPairClosingBrackets + `\s*[({])`,
+				`(?<=(?:^|\s)(?:regex|token|rule)(\s+))` + namePattern + colonPairLookahead + `\s*[({])`,
 				NameFunction,
 				Push("token", "name-adverb"),
 			},
-			{
-				// Token without adverbs
-				`(?<=(?:^|\s)(?:regex|token|rule)(?:\s+))(['\w:-]+)`,
-				NameFunction,
-				Push("token"),
-			},
 			// Substitution
 			{`(?<=^|\b|\s)(?<!\.)(ss|S|s|TR|tr)\b(\s*)`, ByGroups(Keyword, Text), Push("substitution")},
 			{keywordsPattern, Keyword, nil},
@@ -626,18 +620,10 @@ func rakuRules() Rules {
 			},
 			//  Routine
 			{
-				// Routine with adverbs
-				`(?<=(?:^|\s)(?:sub|method|multi sub|multi)\s+)!?['\w:-]+(?=:['\w-]+` +
-					colonPairOpeningBrackets + `.+?` + colonPairClosingBrackets + `\s*[({])`,
+				`(?<=(?:^|\s)(?:sub|method|multi sub|multi)\s+)!?` + namePattern + colonPairLookahead + `\s*[({])`,
 				NameFunction,
 				Push("name-adverb"),
 			},
-			{
-				// Routine without adverbs
-				`(?<=(?:^|\s)(?:sub|submethod|method|multi)\s+)!?['\w:-]+`,
-				NameFunction,
-				nil,
-			},
 			// Constant
 			{`(?<=\bconstant\s+)` + namePattern, NameConstant, Push("name-adverb")},
 			// Namespace
@@ -655,22 +641,16 @@ func rakuRules() Rules {
 			},
 			// Function
 			{
-				`\b(?:\w['\w:-]*)(?=:['\w-]+` +
-					colonPairOpeningBrackets + `.+?` + colonPairClosingBrackets + `\()`,
+				`\b` + namePattern + colonPairLookahead + `\()`,
 				NameFunction,
 				Push("name-adverb"),
 			},
-			{`\b(?:\w['\w:-]*)(?=\()`, NameFunction, nil},
 			// Method
-			// Method with adverb
 			{
-				`(?<!\.\.[?^*+]?)(?<=(?:\.[?^*+&]?)|self!)['\w:-]+(?=:['\w-]+` +
-					colonPairOpeningBrackets + `.+?` + colonPairClosingBrackets + `$)`,
+				`(?<!\.\.[?^*+]?)(?<=(?:\.[?^*+&]?)|self!)` + namePattern + colonPairLookahead + `\b)`,
 				NameFunction,
 				Push("name-adverb"),
 			},
-			// Method without adverb
-			{`(?<!\.\.[?^*+]?)(?<=(?:\.[?^*+&]?)|self!)['\w:-]+`, NameFunction, nil},
 			// Indirect invocant
 			{namePattern + `(?=\s+\W?['\w:-]+:\W)`, NameFunction, Push("name-adverb")},
 			{`(?<=\W)(?:∅|i|e|𝑒|tau|τ|pi|π|Inf|∞)(?=\W)`, NameConstant, nil},
diff --git a/lexers/testdata/raku.actual b/lexers/testdata/raku.actual
@@ -32,6 +32,8 @@ sub f2($a,$b) {
 
 2.&f2(3);
 
+Module::function($var);
+
 #| Fibonacci with Multiple dispatch
 multi sub fib (0 --> 0) {}
 multi sub fib (1 --> 1) {}
diff --git a/lexers/testdata/raku.expected b/lexers/testdata/raku.expected
@@ -102,6 +102,11 @@
   {"type":"LiteralNumberInteger","value":"3"},
   {"type":"Punctuation","value":");"},
   {"type":"Text","value":"\n\n"},
+  {"type":"NameFunction","value":"Module::function"},
+  {"type":"Punctuation","value":"("},
+  {"type":"NameVariable","value":"$var"},
+  {"type":"Punctuation","value":");"},
+  {"type":"Text","value":"\n\n"},
   {"type":"Keyword","value":"#| "},
   {"type":"LiteralStringDoc","value":"Fibonacci with Multiple dispatch\n"},
   {"type":"Keyword","value":"multi"},
