remove UTF8 BOM if present

src/loader.rs

@@ -37,7 +37,8 @@ error_chain! {
     }
 }
 
-fn read_file_to_string<P: AsRef<Path>>(p: P) -> Result<String> {
+#[doc(hidden)]
+pub fn read_file_to_string<P: AsRef<Path>>(p: P) -> Result<String> {
     let p = p.as_ref();
     let mut f = File::open(p).chain_err(|| ErrorKind::IOError)?;
     let mut reader: Vec<u8> = Vec::new();
@@ -50,7 +51,20 @@ fn read_file_to_string<P: AsRef<Path>>(p: P) -> Result<String> {
     let coder = encoding::label::encoding_from_whatwg_label(whtwg_label);
     let file_content = match coder {
         Some(c) => match c.decode(&reader, encoding::DecoderTrap::Ignore) {
-            Ok(x) => x,
+            Ok(x) => {
+                // handle UTF8 BOM manually
+                if x.len() > 0 && vec!["utf-8", "utf8"].contains(&&*whtwg_label.to_lowercase()) {
+                    let mut chars = x.chars();
+                    let first = chars.next().unwrap();
+                    if first == '\u{feff}' {
+                        chars.as_str().to_string()
+                    } else {
+                        x
+                    }
+                } else {
+                    x
+                }
+            },
             Err(e) => bail!(ErrorKind::DecodingError(e.into_owned())),
         },
         None => bail!(ErrorKind::EncodingDetectionError),

tests/txt.rs

@@ -3,6 +3,7 @@ extern crate ultrastar_txt;
 use std::collections::HashMap;
 use ultrastar_txt::*;
 use url::Url;
+use std::path::Path;
 
 // usage:
 //    assert_error_kind!(some_err, ErrorKind::MyErrorType)
@@ -356,6 +357,12 @@ fn remote_url_audio() {
                Source::Remote(Url::parse("https://www.example.com/Testfile.mp3").unwrap()));
 }
 
+fn encoding_utf8_bom() {
+    let txt = read_file_to_string(Path::new("tests/txts/encoding_utf8_bom.txt")).unwrap();
+    let header = get_simple_txt_header();
+    assert_eq!(header, parse_txt_header_str(&txt).unwrap());
+}
+
 fn get_simple_txt_str() -> &'static str {
     include_str!("txts/simple_txt_with_all_features.txt")
 }

tests/txts/encoding_utf8_bom.txt

@@ -0,0 +1,26 @@
+#TITLE:Testsong
+#ARTIST:Testartist
+#MP3:Testfile.mp3
+#GAP:666
+#BPM:123
+#GENRE:Music
+#EDITION:Testmusic
+#COVER:Cover.jpg
+#BACKGROUND:BG.jpg
+#VIDEO:DLzxrzFCyOs.mp4
+#VIDEOGAP:777
+#RELATIVE:NO
+#LANGUAGE:en
+#YEAR:1337
+: 0 4 59 Test
+: 4 4 59 I
+: 8 4 59 'm
+* 12 4 59 test
+: 16 4 59 ing.
+- 20
+: 24 4 59 Test
+: 28 4 59 I
+: 32 4 59 'm
+F 36 4 59 test
+F 40 4 59 ing.
+E