From b27d4afa8d076bc4394b3160f716bf58f586ec11 Mon Sep 17 00:00:00 2001
From: Maxime Coste <mawww@kakoune.org>
Date: Tue, 20 Mar 2018 04:57:47 +1100
Subject: [PATCH] Regex: Only allow SyntaxCharacter and - to be escaped in a
 character class

Letting any character to be escaped is error prone as it looks like
\l could mean [:lower:] (as it used to with boost) when it only means
literal l.

Fix the haskell.kak file as well.

Fixes #1945
---
 rc/base/haskell.kak |  2 +-
 src/regex_impl.cc   | 17 +++++++----------
 2 files changed, 8 insertions(+), 11 deletions(-)

diff --git a/rc/base/haskell.kak b/rc/base/haskell.kak
index cf06fade..25c62861 100644
--- a/rc/base/haskell.kak
+++ b/rc/base/haskell.kak
@@ -39,7 +39,7 @@ add-highlighter shared/haskell/code regex (?<!')\b(case|do|else|if|in|let|mdo|of
 
 # matches uppercase identifiers:  Monad Control.Monad
 # not non-space separated dot:    Just.const
-add-highlighter shared/haskell/code regex \b([A-Z]['\w]*\.)*[A-Z]['\w]*(?!['\w])(?![.\l]) 0:variable
+add-highlighter shared/haskell/code regex \b([A-Z]['\w]*\.)*[A-Z]['\w]*(?!['\w])(?![.a-z]) 0:variable
 
 # matches infix identifier: `mod` `Apa._T'M`
 add-highlighter shared/haskell/code regex `\b([A-Z]['\w]*\.)*[\w]['\w]*` 0:operator
diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index 7870fb56..1c2dfd84 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -429,24 +429,21 @@ private:
             if (cp == '\\')
             {
                 auto it = find_if(character_class_escapes,
-                                  [cp = *m_pos](auto& t) { return t.cp == cp; });
+                                  [cp = *m_pos](auto&& t) { return t.cp == cp; });
                 if (it != std::end(character_class_escapes))
                 {
                     character_class.ctypes |= it->ctype;
                     ++m_pos;
                     continue;
                 }
-                else // its just an escaped character
+                else // its an escaped character
                 {
                     cp = *m_pos++;
-                    for (auto& control : control_escapes)
-                    {
-                        if (control.name == cp)
-                        {
-                            cp = control.value;
-                            break;
-                        }
-                    }
+                    auto it = find_if(control_escapes, [cp](auto&& t) { return t.name == cp; });
+                    if (it != std::end(control_escapes))
+                        cp = it->value;
+                    else if (not contains("^$\\.*+?()[]{}|-", cp)) // SyntaxCharacter and -
+                        parse_error(format("unknown character class escape '{}'", cp));
                 }
             }