From fbce951ee3991c63e4a7fcc2c65a2b84097a1427 Mon Sep 17 00:00:00 2001
From: Dan Balasescu <smoogipoo@smgi.me>
Date: Thu, 1 Aug 2024 20:17:19 +0900
Subject: [PATCH] Implement PHP string escaping semantics

---
 .../Php/PhpStringLiteralSyntaxNode.cs         | 115 ++++++++++++++++--
 .../Php/PhpTokeniser.cs                       |  23 ++++
 2 files changed, 129 insertions(+), 9 deletions(-)

diff --git a/LocalisationAnalyser.Tools/Php/PhpStringLiteralSyntaxNode.cs b/LocalisationAnalyser.Tools/Php/PhpStringLiteralSyntaxNode.cs
index 646a3d6..aaedea5 100644
--- a/LocalisationAnalyser.Tools/Php/PhpStringLiteralSyntaxNode.cs
+++ b/LocalisationAnalyser.Tools/Php/PhpStringLiteralSyntaxNode.cs
@@ -1,7 +1,10 @@
 // Copyright (c) ppy Pty Ltd <contact@ppy.sh>. Licensed under the MIT Licence.
 // See the LICENCE file in the repository root for full licence text.
 
+using System;
+using System.Globalization;
 using System.Text;
+using System.Text.RegularExpressions;
 
 namespace LocalisationAnalyser.Tools.Php
 {
@@ -10,6 +13,10 @@ namespace LocalisationAnalyser.Tools.Php
     /// </summary>
     public class PhpStringLiteralSyntaxNode : PhpLiteralSyntaxNode
     {
+        private static readonly Regex oct_pattern = new Regex("^([0-7]{1,3})", RegexOptions.Compiled);
+        private static readonly Regex hex_pattern = new Regex("^(x[0-9A-Fa-f]{1,2})", RegexOptions.Compiled);
+        private static readonly Regex uni_pattern = new Regex("^(u{[0-9A-Fa-f]+})", RegexOptions.Compiled);
+
         public readonly string Text;
 
         public PhpStringLiteralSyntaxNode(string text)
@@ -21,27 +28,23 @@ public static PhpStringLiteralSyntaxNode Parse(PhpTokeniser tokeniser)
         {
             tokeniser.SkipWhitespace();
 
-            char trivia = tokeniser.GetTrivia();
-
-            // Skip leading trivia.
+            char leader = tokeniser.GetTrivia();
             tokeniser.Advance();
 
             var stringBuilder = new StringBuilder();
-            bool isEscaping = false;
 
-            while (isEscaping || tokeniser.GetTrivia() != trivia)
+            while (tokeniser.GetTrivia() != leader)
             {
-                var token = tokeniser.GetTrivia();
+                char token = tokeniser.GetTrivia();
                 tokeniser.Advance();
 
-                if (token == '\\' && !isEscaping)
+                if (token == '\\')
                 {
-                    isEscaping = true;
+                    stringBuilder.Append(processEscapeSequence(leader, tokeniser));
                     continue;
                 }
 
                 stringBuilder.Append(token);
-                isEscaping = false;
             }
 
             // Skip trailing trivia.
@@ -50,5 +53,99 @@ public static PhpStringLiteralSyntaxNode Parse(PhpTokeniser tokeniser)
 
             return new PhpStringLiteralSyntaxNode(stringBuilder.ToString());
         }
+
+        private static string processEscapeSequence(char leader, PhpTokeniser tokeniser)
+        {
+            char trivia = tokeniser.GetTrivia();
+
+            // Base cases for \{leader} and \\, supported by both single- and double-quoted strings.
+            if (trivia == leader || trivia == '\\')
+            {
+                tokeniser.Advance();
+                return trivia.ToString();
+            }
+
+            // No other escape sequences are supported for single-quoted strings.
+            if (leader == '\'')
+                return @"\";
+
+            // Double-quoted strings have a few more cases...
+            switch (trivia)
+            {
+                case 'n':
+                    tokeniser.Advance();
+                    return "\n";
+
+                case 'r':
+                    tokeniser.Advance();
+                    return "\r";
+
+                case 't':
+                    tokeniser.Advance();
+                    return "\t";
+
+                case 'v':
+                    tokeniser.Advance();
+                    return "\v";
+
+                case 'e':
+                    tokeniser.Advance();
+                    return "\x1B";
+
+                case 'f':
+                    tokeniser.Advance();
+                    return "\f";
+
+                case '$':
+                    tokeniser.Advance();
+                    return "$";
+
+                case >= '0' and <= '7':
+                {
+                    Match match = oct_pattern.Match($"{trivia}{tokeniser.PeekNext(2)}");
+
+                    if (match.Success)
+                    {
+                        tokeniser.Advance(match.Length);
+
+                        unchecked
+                        {
+                            byte octValue = (byte)Convert.ToInt32(match.Value, 8);
+                            return ((char)octValue).ToString();
+                        }
+                    }
+
+                    break;
+                }
+
+                case 'x':
+                {
+                    Match match = hex_pattern.Match($"{trivia}{tokeniser.PeekNext(2)}");
+
+                    if (match.Success)
+                    {
+                        tokeniser.Advance(match.Length);
+                        return ((char)byte.Parse(match.Value[1..], NumberStyles.HexNumber, CultureInfo.InvariantCulture)).ToString();
+                    }
+
+                    break;
+                }
+
+                case 'u':
+                {
+                    Match match = uni_pattern.Match($"{trivia}{tokeniser.PeekNext(16)}");
+
+                    if (match.Success)
+                    {
+                        tokeniser.Advance(match.Length);
+                        return char.ConvertFromUtf32(Convert.ToInt32(match.Value[2..^1], 16));
+                    }
+
+                    break;
+                }
+            }
+
+            return @"\";
+        }
     }
 }
diff --git a/LocalisationAnalyser.Tools/Php/PhpTokeniser.cs b/LocalisationAnalyser.Tools/Php/PhpTokeniser.cs
index dc3cc71..98be66b 100644
--- a/LocalisationAnalyser.Tools/Php/PhpTokeniser.cs
+++ b/LocalisationAnalyser.Tools/Php/PhpTokeniser.cs
@@ -67,6 +67,16 @@ public void TryAdvance()
             Advance();
         }
 
+        /// <summary>
+        /// Advances by a number of trivia.
+        /// </summary>
+        /// <param name="length">The number of trivias to advance by.</param>
+        public void Advance(int length)
+        {
+            for (int i = 0; i < length; i++)
+                Advance();
+        }
+
         /// <summary>
         /// Advances to the next trivia.
         /// </summary>
@@ -104,6 +114,19 @@ public bool TryPeekNext(out char trivia)
             return true;
         }
 
+        /// <summary>
+        /// Peeks a number of future trivia.
+        /// </summary>
+        /// <param name="length">The length of trivia to peek.</param>
+        /// <returns>The trivia.</returns>
+        public string PeekNext(int length)
+        {
+            int startIndex = Math.Min(content.Length, currentIndex + 1);
+            int endIndex = Math.Min(content.Length, startIndex + length);
+
+            return content.AsSpan()[startIndex..endIndex].ToString();
+        }
+
         /// <summary>
         /// Skips all current whitespace and comments.
         /// </summary>