From cee00eb13d6c931f44d0057c8aac7076138811fd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C3=89tienne=20Reuh=20Fildadut?= <fildadut@reuh.eu>
Date: Fri, 19 Jan 2024 14:37:06 +0100
Subject: [PATCH] [internal] normalize newlines and bom

---
 anselme/parser/init.lua | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/anselme/parser/init.lua b/anselme/parser/init.lua
index ce2694b..b9f7413 100644
--- a/anselme/parser/init.lua
+++ b/anselme/parser/init.lua
@@ -9,8 +9,16 @@ local function expect_end(exp, rem)
 	return exp
 end
 
+-- we require UTF-8 but life is full of disapointments
+-- remove BOM
+-- \r\n and \r -> \n
+local function normalize_encoding(str)
+	return str:gsub("^"..string.char(0xEF, 0xBB, 0xBF), "")
+	            :gsub("\r\n?", "\n")
+end
+
 -- parse code (string) with the associated source (Source)
 -- the returned AST tree is stateless and can be stored/evaluated/etc as you please
 return function(code, source)
-	return expect_end(block(Source:new(source, 1, 1), Options:new(), code))
+	return expect_end(block(Source:new(source, 1, 1), Options:new(), normalize_encoding(code)))
 end