Merge pull request #377 from hamishknight/named-captures-only

Author: hamishknight

Sources/_RegexParser/Regex/AST/MatchingOptions.swift

@@ -17,7 +17,7 @@ extension AST {
       case caseInsensitive          // i
       case allowDuplicateGroupNames // J
       case multiline                // m
-      case noAutoCapture            // n
+      case namedCapturesOnly        // n
       case singleLine               // s
       case reluctantByDefault       // U
       case extended                 // x

Sources/_RegexParser/Regex/Parse/LexicalAnalysis.swift

@@ -616,7 +616,7 @@ extension Source {
       case "i": return advanceAndReturn(.caseInsensitive)
       case "J": return advanceAndReturn(.allowDuplicateGroupNames)
       case "m": return advanceAndReturn(.multiline)
-      case "n": return advanceAndReturn(.noAutoCapture)
+      case "n": return advanceAndReturn(.namedCapturesOnly)
       case "s": return advanceAndReturn(.singleLine)
       case "U": return advanceAndReturn(.reluctantByDefault)
       case "x":
@@ -914,6 +914,10 @@ extension Source {
         }
         // TODO: (name:)
 
+        // If (?n) is set, a bare (...) group is non-capturing.
+        if context.syntax.contains(.namedCapturesOnly) {
+          return .nonCapture
+        }
         return .capture
       }
     }

Sources/_RegexParser/Regex/Parse/Parse.swift

@@ -287,23 +287,34 @@ extension Parser {
   private mutating func applySyntaxOptions(
     of opts: AST.MatchingOptionSequence
   ) {
-    // We skip this for multi-line, as extended syntax is always enabled there.
-    if context.syntax.contains(.multilineExtendedSyntax) { return }
+    func mapOption(_ option: SyntaxOptions,
+                   _ pred: (AST.MatchingOption) -> Bool) {
+      if opts.resetsCurrentOptions {
+        context.syntax.remove(option)
+      }
+      if opts.adding.contains(where: pred) {
+        context.syntax.insert(option)
+      }
+      if opts.removing.contains(where: pred) {
+        context.syntax.remove(option)
+      }
+    }
+    func mapOption(_ option: SyntaxOptions, _ kind: AST.MatchingOption.Kind) {
+      mapOption(option, { $0.kind == kind })
+    }
+
+    // (?n)
+    mapOption(.namedCapturesOnly, .namedCapturesOnly)
 
-    // Check if we're introducing or removing extended syntax.
+    // (?x), (?xx)
+    // We skip this for multi-line, as extended syntax is always enabled there.
     // TODO: PCRE differentiates between (?x) and (?xx) where only the latter
     // handles non-semantic whitespace in a custom character class. Other
     // engines such as Oniguruma, Java, and ICU do this under (?x). Therefore,
     // treat (?x) and (?xx) as the same option here. If we ever get a strict
     // PCRE mode, we will need to change this to handle that.
-    if opts.resetsCurrentOptions {
-      context.syntax.remove(.extendedSyntax)
-    }
-    if opts.adding.contains(where: \.isAnyExtended) {
-      context.syntax.insert(.extendedSyntax)
-    }
-    if opts.removing.contains(where: \.isAnyExtended) {
-      context.syntax.remove(.extendedSyntax)
+    if !context.syntax.contains(.multilineExtendedSyntax) {
+      mapOption(.extendedSyntax, \.isAnyExtended)
     }
   }
 

Sources/_RegexParser/Regex/Parse/SyntaxOptions.swift

@@ -63,6 +63,9 @@ public struct SyntaxOptions: OptionSet {
     return [Self(1 << 6), .extendedSyntax]
   }
 
+  /// `(?n)`
+  public static var namedCapturesOnly: Self { Self(1 << 7) }
+
   /*
 
     /// `<digit>*` == `[[:digit:]]*` == `\d*`

Sources/_StringProcessing/MatchingOptions.swift

@@ -135,7 +135,7 @@ extension MatchingOptions {
     case caseInsensitive
     case allowDuplicateGroupNames
     case multiline
-    case noAutoCapture
+    case namedCapturesOnly
     case singleLine
     case reluctantByDefault
 
@@ -174,8 +174,8 @@ extension MatchingOptions {
         self = .allowDuplicateGroupNames
       case .multiline:
         self = .multiline
-      case .noAutoCapture:
-        self = .noAutoCapture
+      case .namedCapturesOnly:
+        self = .namedCapturesOnly
       case .singleLine:
         self = .singleLine
       case .reluctantByDefault:

Tests/RegexTests/ParseTests.swift

@@ -906,7 +906,7 @@ extension RegexTests {
     ))
 
     let allOptions: [AST.MatchingOption.Kind] = [
-      .caseInsensitive, .allowDuplicateGroupNames, .multiline, .noAutoCapture,
+      .caseInsensitive, .allowDuplicateGroupNames, .multiline, .namedCapturesOnly,
       .singleLine, .reluctantByDefault, .extraExtended, .extended,
       .unicodeWordBoundaries, .asciiOnlyDigit, .asciiOnlyPOSIXProps,
       .asciiOnlySpace, .asciiOnlyWord, .textSegmentGraphemeMode,
@@ -973,6 +973,13 @@ extension RegexTests {
       "d"
     )), captures: [.cap])
 
+    parseTest("(?n)(?^:())(?<x>)()", concat(
+      changeMatchingOptions(matchingOptions(adding: .namedCapturesOnly)),
+      changeMatchingOptions(unsetMatchingOptions(), capture(empty())),
+      namedCapture("x", empty()),
+      nonCapture(empty())
+    ), captures: [.cap, .named("x")])
+
     // MARK: References
 
     // \1 ... \9 are always backreferences.