From 9d34bf831ea33d73bd300a18aa3af5daa271ebc0 Mon Sep 17 00:00:00 2001
From: Mirek Fidler <mirek.fidler@gmail.com>
Date: Tue, 24 Aug 2021 18:05:56 +0200
Subject: [PATCH] Core: Utf8BOM (skips UTF8 file header)

---
 uppsrc/Core/Bom.cpp                   | 9 +++++++++
 uppsrc/Core/CharSet.h                 | 3 +++
 uppsrc/Core/src.tpp/CharSet_en-us.tpp | 6 ++++++
 3 files changed, 18 insertions(+)

diff --git a/uppsrc/Core/Bom.cpp b/uppsrc/Core/Bom.cpp
index 74e2e970c..ef5abb71b 100644
--- a/uppsrc/Core/Bom.cpp
+++ b/uppsrc/Core/Bom.cpp
@@ -2,6 +2,15 @@
 
 namespace Upp {
 
+bool Utf8BOM(Stream& in)
+{
+	int64 pos = in.GetPos();
+	if(in.Get() == 0xef && in.Get() == 0xbb && in.Get() == 0xbf)
+		return true;
+	in.Seek(pos);
+	return false;
+}
+
 static void sLoadBom(Stream& in, String *t, WString *wt, byte def_charset) 
 {
 	if(in.IsOpen()) {
diff --git a/uppsrc/Core/CharSet.h b/uppsrc/Core/CharSet.h
index a389fbb3a..e8d19a28b 100644
--- a/uppsrc/Core/CharSet.h
+++ b/uppsrc/Core/CharSet.h
@@ -251,6 +251,9 @@ bool    SaveFileBOM(const char *path, const WString& data);
 bool    SaveStreamBOMUtf8(Stream& out, const String& data);
 bool    SaveFileBOMUtf8(const char *path, const String& data);
 
+bool    Utf8BOM(Stream& in);
+
+
 // Deprecated
 
 word UnicodeCombine(word chr, word combine);
diff --git a/uppsrc/Core/src.tpp/CharSet_en-us.tpp b/uppsrc/Core/src.tpp/CharSet_en-us.tpp
index 831729c1f..5a9d59449 100644
--- a/uppsrc/Core/src.tpp/CharSet_en-us.tpp
+++ b/uppsrc/Core/src.tpp/CharSet_en-us.tpp
@@ -705,6 +705,12 @@ true on success.&]
 true on success.&]
 [s3; &]
 [s4; &]
+[s5;:Upp`:`:Utf8BOM`(Upp`:`:Stream`&`): [@(0.0.255) bool]_[* Utf8BOM]([_^Upp`:`:Stream^ Str
+eam][@(0.0.255) `&]_[*@3 in])&]
+[s2;%% Tests for and skips UTF`-8 BOM mark in the seekable Stream 
+[%-*@3 in].&]
+[s3;%% &]
+[s4; &]
 [s5;:FromUtf8`(const char`*`,int`): [_^WString^ WString]_[* FromUtf8]([@(0.0.255) const]_[@(0.0.255) c
 har]_`*[*@3 `_s], [@(0.0.255) int]_[*@3 len])&]
 [s2;%% Converts UTF`-8 to UNICODE string. Any wrong bytes and sequences