From 3704c4a01219cca7273fa5fa0bc6751d98ad05ac Mon Sep 17 00:00:00 2001
From: cypherpunks <cypherpunks@torproject.org>
Date: Wed, 29 Aug 2018 13:32:52 +0000
Subject: [PATCH] string: add BOM helper

---
 src/lib/string/util_string.c | 13 +++++++++++++
 src/lib/string/util_string.h |  1 +
 2 files changed, 14 insertions(+)

diff --git a/src/lib/string/util_string.c b/src/lib/string/util_string.c
index b2b85d151d..e76e73046f 100644
--- a/src/lib/string/util_string.c
+++ b/src/lib/string/util_string.c
@@ -541,3 +541,16 @@ string_is_utf8(const char *str, size_t len)
   }
   return true;
 }
+
+/** As string_is_utf8(), but returns false if the string begins with a UTF-8
+ * byte order mark (BOM).
+ */
+int
+string_is_utf8_no_bom(const char *str, size_t len)
+{
+  if (len >= 3 && (!strcmpstart(str, "\uFEFF") ||
+                   !strcmpstart(str, "\uFFFE"))) {
+    return false;
+  }
+  return string_is_utf8(str, len);
+}
diff --git a/src/lib/string/util_string.h b/src/lib/string/util_string.h
index 746ece0d33..99467a27c3 100644
--- a/src/lib/string/util_string.h
+++ b/src/lib/string/util_string.h
@@ -53,5 +53,6 @@ const char *find_str_at_start_of_line(const char *haystack,
 int string_is_C_identifier(const char *string);
 
 int string_is_utf8(const char *str, size_t len);
+int string_is_utf8_no_bom(const char *str, size_t len);
 
 #endif /* !defined(TOR_UTIL_STRING_H) */