Add substitution support for fish, nushell, powershell, and xonsh

Add ( and ) as wordbreaks to fish, nushell, powershell, and xonsh
classifiers. Each format's PostProcess reclassifies parens as
substitution delimiters. PowerShell and xonsh detect $ + ( (and
@ + ( for xonsh) adjacency to merge into a single opener token.

Assisted-by: Crush:glm-5.2

Author: rsteube-bot

format_fish.go

@@ -18,7 +18,8 @@ func (fishFormat) Classifier() tokenClassifier {
 	// Fish operators: |, ;, <, >, &, ?
 	// & is included for &&, ||, &, |&, &|, &>, &>>, &>?, >&, <>&
 	// ? is part of redirect operators (>? >>? <?) — deprecated glob char
-	t.addWordbreaks("|;<>&?")
+	// ( and ) are command substitution delimiters
+	t.addWordbreaks("|;<>&?()")
 	return t
 }
 
@@ -84,3 +85,20 @@ var fishDoubleQuoteEscapes = map[rune]bool{
 func (fishFormat) QuoteWord(s string) string { return fishQuoteWord(s) }
 func (fishFormat) TripleQuoteSupport() bool  { return false }
 func (fishFormat) RawPrefixSupport() bool    { return false }
+
+// PostProcess reclassifies ( and ) as substitution delimiters for fish
+// command substitution. Fish uses bare () for command substitution
+// (no $ prefix).
+func (fishFormat) PostProcess(tokens TokenSlice) TokenSlice {
+	result := make(TokenSlice, 0, len(tokens))
+	for _, t := range tokens {
+		if t.Type == WORDBREAK_TOKEN && t.Value == "(" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_OPEN
+		}
+		if t.Type == WORDBREAK_TOKEN && t.Value == ")" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_CLOSE
+		}
+		result = append(result, t)
+	}
+	return result
+}

format_nushell.go

@@ -33,7 +33,8 @@ func (nushellFormat) Classifier() tokenClassifier {
 
 	// Nushell operators: |, ;, >, <, >>
 	// No &&, ||, & — no POSIX list operators
-	t.addWordbreaks("|;<>")
+	// ( and ) are subexpression delimiters
+	t.addWordbreaks("|;<>()")
 	return t
 }
 
@@ -172,5 +173,16 @@ func (nushellFormat) PostProcess(tokens TokenSlice) TokenSlice {
 
 		result = append(result, t)
 	}
+
+	// Reclassify ( and ) as substitution delimiters
+	for i := range result {
+		t := &result[i]
+		if t.Type == WORDBREAK_TOKEN && t.Value == "(" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_OPEN
+		}
+		if t.Type == WORDBREAK_TOKEN && t.Value == ")" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_CLOSE
+		}
+	}
 	return result
 }

format_powershell.go

@@ -28,7 +28,8 @@ func (powershellFormat) Classifier() tokenClassifier {
 
 	// PowerShell operators: |, ;, >, >>, &&, ||, &
 	// Note: & is the call operator, not a background operator
-	t.addWordbreaks("|;&><")
+	// ( and ) are subexpression/substitution delimiters
+	t.addWordbreaks("|;&><()")
 	return t
 }
 
@@ -151,5 +152,37 @@ func (powershellFormat) PostProcess(tokens TokenSlice) TokenSlice {
 
 		result = append(result, t)
 	}
-	return result
+
+	// Second pass: merge $ + ( into substitution opener and reclassify ) as closer
+	final := make(TokenSlice, 0, len(result))
+	for i := 0; i < len(result); i++ {
+		t := result[i]
+		// Detect $ + ( adjacency → merge into WORDBREAK_SUBSTITUTION_OPEN
+		if t.Type == WORD_TOKEN && t.Value == "$" && i+1 < len(result) {
+			next := result[i+1]
+			if next.Type == WORDBREAK_TOKEN && next.Value == "(" && t.adjoins(next) {
+				merged := Token{
+					Type:          WORDBREAK_TOKEN,
+					Value:         "$(",
+					RawValue:      t.RawValue + next.RawValue,
+					Span:          Span{Start: t.Span.Start, End: next.Span.End},
+					State:         next.State,
+					WordbreakType: WORDBREAK_SUBSTITUTION_OPEN,
+				}
+				final = append(final, merged)
+				i++
+				continue
+			}
+		}
+		// Reclassify standalone ( as substitution opener
+		if t.Type == WORDBREAK_TOKEN && t.Value == "(" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_OPEN
+		}
+		// Reclassify ) as substitution closer
+		if t.Type == WORDBREAK_TOKEN && t.Value == ")" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_CLOSE
+		}
+		final = append(final, t)
+	}
+	return final
 }

format_xonsh.go

@@ -22,7 +22,8 @@ func XonshFormat() Format { return xonshFormat{} }
 func (xonshFormat) Classifier() tokenClassifier {
 	t := newBaseClassifier(escapeRunes)
 	// Xonsh operators: |, >, >>, <, ;, &&, ||, &
-	t.addWordbreaks("|<>&;")
+	// ( and ) are subprocess/eval delimiters
+	t.addWordbreaks("|<>&;()")
 	return t
 }
 
@@ -111,5 +112,37 @@ func (xonshFormat) PostProcess(tokens TokenSlice) TokenSlice {
 
 		result = append(result, t)
 	}
-	return result
+
+	// Second pass: merge $/@ + ( into substitution opener and reclassify ) as closer
+	final := make(TokenSlice, 0, len(result))
+	for i := 0; i < len(result); i++ {
+		t := result[i]
+		// Detect $ + ( or @ + ( adjacency → merge into WORDBREAK_SUBSTITUTION_OPEN
+		if t.Type == WORD_TOKEN && (t.Value == "$" || t.Value == "@") && i+1 < len(result) {
+			next := result[i+1]
+			if next.Type == WORDBREAK_TOKEN && next.Value == "(" && t.adjoins(next) {
+				merged := Token{
+					Type:          WORDBREAK_TOKEN,
+					Value:         t.Value + "(",
+					RawValue:      t.RawValue + next.RawValue,
+					Span:          Span{Start: t.Span.Start, End: next.Span.End},
+					State:         next.State,
+					WordbreakType: WORDBREAK_SUBSTITUTION_OPEN,
+				}
+				final = append(final, merged)
+				i++
+				continue
+			}
+		}
+		// Reclassify standalone ( as substitution opener
+		if t.Type == WORDBREAK_TOKEN && t.Value == "(" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_OPEN
+		}
+		// Reclassify ) as substitution closer
+		if t.Type == WORDBREAK_TOKEN && t.Value == ")" {
+			t.WordbreakType = WORDBREAK_SUBSTITUTION_CLOSE
+		}
+		final = append(final, t)
+	}
+	return final
 }