From 571c34cbf56799a506e1bcd265c1e3729af59f61 Mon Sep 17 00:00:00 2001
From: Hamish Knight <hamish_github@mediocremail.com>
Date: Thu, 26 May 2022 20:46:35 +0100
Subject: [PATCH] Parse Java character properties

These correspond to various `is`-prefixed
accessors on `java.lang.Character`. For now, parse
them, but mark them unsupported.
---
 Sources/_RegexParser/Regex/AST/Atom.swift     | 27 ++++++++++++++++++-
 .../CharacterPropertyClassification.swift     |  7 +++--
 Sources/_RegexParser/Regex/Parse/Sema.swift   |  2 ++
 .../_StringProcessing/ConsumerInterface.swift |  3 +++
 Tests/RegexTests/ParseTests.swift             |  4 +++
 5 files changed, 40 insertions(+), 3 deletions(-)

diff --git a/Sources/_RegexParser/Regex/AST/Atom.swift b/Sources/_RegexParser/Regex/AST/Atom.swift
index 680a29dee..992604852 100644
--- a/Sources/_RegexParser/Regex/AST/Atom.swift
+++ b/Sources/_RegexParser/Regex/AST/Atom.swift
@@ -450,6 +450,9 @@ extension AST.Atom.CharacterProperty {
     /// Some special properties implemented by PCRE and Oniguruma.
     case pcreSpecial(PCRESpecialCategory)
 
+    /// Some special properties implemented by Java.
+    case javaSpecial(JavaSpecial)
+
     public enum MapKind: Hashable {
       case lowercase
       case uppercase
@@ -457,7 +460,6 @@ extension AST.Atom.CharacterProperty {
     }
   }
 
-  // TODO: erm, separate out or fold into something? splat it in?
   public enum PCRESpecialCategory: String, Hashable {
     case alphanumeric     = "Xan"
     case posixSpace       = "Xps"
@@ -465,6 +467,29 @@ extension AST.Atom.CharacterProperty {
     case universallyNamed = "Xuc"
     case perlWord         = "Xwd"
   }
+
+  /// Special Java properties that correspond to methods on
+  /// `java.lang.Character`, with the `java` prefix replaced by `is`.
+  public enum JavaSpecial: String, Hashable, CaseIterable {
+    case alphabetic             = "javaAlphabetic"
+    case defined                = "javaDefined"
+    case digit                  = "javaDigit"
+    case identifierIgnorable    = "javaIdentifierIgnorable"
+    case ideographic            = "javaIdeographic"
+    case isoControl             = "javaISOControl"
+    case javaIdentifierPart     = "javaJavaIdentifierPart" // not a typo, that's actually the name
+    case javaIdentifierStart    = "javaJavaIdentifierStart" // not a typo, that's actually the name
+    case javaLetter             = "javaLetter"
+    case javaLetterOrDigit      = "javaLetterOrDigit"
+    case lowerCase              = "javaLowerCase"
+    case mirrored               = "javaMirrored"
+    case spaceChar              = "javaSpaceChar"
+    case titleCase              = "javaTitleCase"
+    case unicodeIdentifierPart  = "javaUnicodeIdentifierPart"
+    case unicodeIdentifierStart = "javaUnicodeIdentifierStart"
+    case upperCase              = "javaUpperCase"
+    case whitespace             = "javaWhitespace"
+  }
 }
 
 extension AST.Atom {
diff --git a/Sources/_RegexParser/Regex/Parse/CharacterPropertyClassification.swift b/Sources/_RegexParser/Regex/Parse/CharacterPropertyClassification.swift
index 2ff162390..fb122e027 100644
--- a/Sources/_RegexParser/Regex/Parse/CharacterPropertyClassification.swift
+++ b/Sources/_RegexParser/Regex/Parse/CharacterPropertyClassification.swift
@@ -764,12 +764,15 @@ extension Source {
       return .block(block)
     }
 
-    // PCRE special properties.
-    // TODO: Normalize?
+    // Special properties from other engines.
     typealias PCRESpecial = AST.Atom.CharacterProperty.PCRESpecialCategory
     if let pcreSpecial = PCRESpecial(rawValue: value) {
       return .pcreSpecial(pcreSpecial)
     }
+    typealias JavaSpecial = AST.Atom.CharacterProperty.JavaSpecial
+    if let javaSpecial = JavaSpecial(rawValue: value) {
+      return .javaSpecial(javaSpecial)
+    }
 
     // TODO: This should be versioned, and do we want a more lax behavior for
     // the runtime?
diff --git a/Sources/_RegexParser/Regex/Parse/Sema.swift b/Sources/_RegexParser/Regex/Parse/Sema.swift
index 99654991a..c803087be 100644
--- a/Sources/_RegexParser/Regex/Parse/Sema.swift
+++ b/Sources/_RegexParser/Regex/Parse/Sema.swift
@@ -175,6 +175,8 @@ extension RegexValidator {
       throw error(.unsupported("PCRE property"), at: loc)
     case .block:
       throw error(.unsupported("Unicode block property"), at: loc)
+    case .javaSpecial:
+      throw error(.unsupported("Java property"), at: loc)
     }
   }
 
diff --git a/Sources/_StringProcessing/ConsumerInterface.swift b/Sources/_StringProcessing/ConsumerInterface.swift
index 6828245b1..640fe3c93 100644
--- a/Sources/_StringProcessing/ConsumerInterface.swift
+++ b/Sources/_StringProcessing/ConsumerInterface.swift
@@ -520,6 +520,9 @@ extension AST.Atom.CharacterProperty {
 
       case .pcreSpecial(let s):
         throw Unsupported("TODO: map PCRE special: \(s)")
+
+      case .javaSpecial(let s):
+        throw Unsupported("TODO: map Java special: \(s)")
       }
     }()
 
diff --git a/Tests/RegexTests/ParseTests.swift b/Tests/RegexTests/ParseTests.swift
index 640186a2b..0daa4a457 100644
--- a/Tests/RegexTests/ParseTests.swift
+++ b/Tests/RegexTests/ParseTests.swift
@@ -1385,6 +1385,10 @@ extension RegexTests {
       parseTest(#"\p{is\#(b.rawValue)}"#, prop(.binary(b, value: true)), throwsError: .unchecked)
     }
 
+    for j in AST.Atom.CharacterProperty.JavaSpecial.allCases {
+      parseTest(#"\p{\#(j.rawValue)}"#, prop(.javaSpecial(j)), throwsError: .unsupported)
+    }
+
     // Try prefixing each block property with "in" to make sure we don't stomp
     // on any other property shorthands.
     for b in Unicode.Block.allCases {