Parse data from either UTF-8 or UTF-16

musicxml/src/parser/mod.rs

@@ -1,6 +1,6 @@
 use crate::elements::{ScorePartwise, ScoreTimewise};
 use alloc::{collections::BTreeMap, string::String, vec::Vec};
-use musicxml_internal::{ElementDeserializer, ElementSerializer, XmlElement};
+use musicxml_internal::{bytes_to_string, ElementDeserializer, ElementSerializer, XmlElement};
 
 #[cfg(feature = "std")]
 extern crate std;
@@ -45,7 +45,7 @@ fn get_musicxml_contents(data: Vec<u8>) -> Result<String, String> {
       Err(String::from("Cannot find MusicXML file in compressed archive"))?;
     }
   } else {
-    contents = String::from_utf8(data).or_else(|_| Err(String::from("Invalid UTF-8 data in MusicXML content")))?;
+    contents = bytes_to_string(&data)?;
   }
   Ok(contents)
 }

musicxml/src/parser/zip_parser.rs

@@ -4,6 +4,7 @@ use alloc::{collections::BTreeMap, vec::Vec};
 use crc32fast;
 use miniz_oxide::deflate::compress_to_vec;
 use miniz_oxide::inflate::decompress_to_vec;
+use musicxml_internal::bytes_to_string;
 
 #[cfg(feature = "std")]
 use std::time::{SystemTime, UNIX_EPOCH};
@@ -247,7 +248,7 @@ impl<'a> ZipArchive<'a> {
     let decoded_data =
       decompress_to_vec(&self.zip_data.content[file.relative_offset..(file.relative_offset + file.compressed_size)])
         .map_err(|e| e.to_string())?;
-    String::from_utf8(decoded_data).map_err(|e| e.to_string())
+    bytes_to_string(&decoded_data).map_err(|e| e.to_string())
   }
 
   pub fn iter(&self) -> impl Iterator<Item = &String> {

musicxml_internal/src/lib.rs

@@ -2,9 +2,25 @@
 
 extern crate alloc;
 
+use alloc::borrow::ToOwned;
 use alloc::string::{String, ToString};
 use alloc::vec::Vec;
 
+pub fn bytes_to_string(bytes: &[u8]) -> Result<String, String> {
+  String::from_utf8(bytes.to_owned()).or_else(|_| {
+    let convert = if bytes[0] == 0xFF && bytes[1] == 0xFE {
+      u16::from_le_bytes
+    } else {
+      u16::from_be_bytes
+    };
+    let u16_bytes: Vec<u16> = bytes
+      .chunks_exact(2)
+      .map(|bytes| convert([bytes[0], bytes[1]]))
+      .collect();
+    String::from_utf16(&u16_bytes).map_err(|_| String::from("Invalid UTF-8 or UTF-16 data in MusicXML content"))
+  })
+}
+
 #[derive(Debug, Default, PartialEq, Eq)]
 pub struct XmlElement {
   pub name: String,