From 9cd330e521fe467b3018bab8097aca3f655b9751 Mon Sep 17 00:00:00 2001
From: Aadhavan Srinivasan <aadhavan@twomorecents.org>
Date: Wed, 12 Feb 2025 23:04:10 -0500
Subject: [PATCH] More work on unicode character class support - fix bug where
 all characters aren't being matched

---
 regex/compile.go | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/regex/compile.go b/regex/compile.go
index dfee016..428df30 100644
--- a/regex/compile.go
+++ b/regex/compile.go
@@ -490,6 +490,45 @@ func shuntingYard(re string, flags ...ReFlag) ([]postfixNode, error) {
 				} else {
 					return nil, fmt.Errorf("not enough hex characters found in expression")
 				}
+			} else if re_postfix[i] == 'p' || re_postfix[i] == 'P' {
+				charClassInverted := (re_postfix[i] == 'P')
+				i++
+				if isUnicodeCharClassLetter(re_postfix[i]) {
+					chars, err := unicodeCharClassToRange(string(re_postfix[i]))
+					if err != nil {
+						return nil, err
+					}
+					var toAppend postfixNode
+					if re_postfix[i] == 'p' {
+						toAppend = newPostfixNode(chars...)
+					}
+					if re_postfix[i] == 'P' {
+						toAppend = newPostfixDotNode()
+						toAppend.except = append([]postfixNode{}, newPostfixNode(chars...))
+					}
+					outQueue = append(outQueue, toAppend)
+				} else if re_postfix[i] == '{' {
+					i++ // Skip opening bracket
+					unicodeCharClassStr := ""
+					for re_postfix[i] != '}' {
+						unicodeCharClassStr += string(re_postfix[i])
+						i++
+					}
+					chars, err := unicodeCharClassToRange(unicodeCharClassStr)
+					if err != nil {
+						return nil, err
+					}
+					var toAppend postfixNode
+					if !charClassInverted { // \p
+						toAppend = newPostfixNode(chars...)
+					} else { // \P
+						toAppend = newPostfixDotNode()
+						toAppend.except = append([]postfixNode{}, newPostfixNode(chars...))
+					}
+					outQueue = append(outQueue, toAppend)
+				} else {
+					return nil, fmt.Errorf("error parsing unicode character class in expression")
+				}
 			} else if re_postfix[i] == '0' { // Octal value
 				var octVal int64
 				var octValStr string