From c1f987beef517e5983835e81e4dc56c6285cf992 Mon Sep 17 00:00:00 2001
From: Tom Smeding <tom.smeding@gmail.com>
Date: Tue, 24 Mar 2020 21:24:50 +0100
Subject: Parse extended header (to an extent), and more v2.4 compat

---
 src/id3v2.rs | 111 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 108 insertions(+), 3 deletions(-)

(limited to 'src')

diff --git a/src/id3v2.rs b/src/id3v2.rs
index f65260d..d11d445 100644
--- a/src/id3v2.rs
+++ b/src/id3v2.rs
@@ -25,6 +25,78 @@ fn parse_id3v2_header(bytes: &[u8]) -> Option<(u16, u8, usize)> {
     }
 }
 
+/// Returns the number of bytes consumed of the body.
+fn parse_extended_header(body: &[u8], version_sub: u8) -> io::Result<usize> {
+    let extended_size = match version_sub {
+        3 => 4 + read_big_endian(&body[0..4], 8),
+        4 => read_big_endian(&body[0..4], 7),
+        _ => panic!("")
+    };
+
+    if body.len() < extended_size {
+        return Err("Header too small for extended header size field".ioerr());
+    }
+
+    match version_sub {
+        3 => if extended_size != 10 && extended_size != 14 {
+            // Error message uses the v2.3 size format, which does not include the size
+            // number itself.
+            return Err(
+                format!("Extended header has unrecognised length {} (not 6 or 10)",
+                        extended_size)
+                    .ioerr()
+            );
+        }
+
+        4 => if extended_size < 6 {
+            return Err(format!("Extended header size too small ({})", extended_size).ioerr());
+        }
+
+        _ => panic!("")
+    }
+
+    match version_sub {
+        3 => {
+            let flags = read_big_endian(&body[4..6], 8);
+            // First bit is "CRC present", which we do not care about (it's contained in
+            // the extended header, which we otherwise ignore anyhow).
+            if (flags & 0x7fff) != 0 {
+                return Err(
+                    format!("Unknown extended header flags set (flags bytes 0x{:04x})",
+                            flags)
+                        .ioerr()
+                );
+            }
+        }
+
+        4 => {
+            let num_flag_bytes = usize::from(body[4]);
+            if num_flag_bytes != 1 {
+                return Err(
+                    format!("Unknown number of extended flag bytes {}", num_flag_bytes)
+                        .ioerr()
+                );
+            }
+
+            let flags = body[5];
+            // 0x40: "update tag"; ignored
+            // 0x20: "CRC present"; ignored
+            // 0x10: tag restructions; ignored
+            if (flags & 0x8f) != 0 {
+                return Err(
+                    format!("Unknown extended flags (flags byte 0x{:02x})", flags)
+                        .ioerr()
+                );
+            }
+        }
+
+        _ => panic!("")
+    }
+
+    // Ignore the rest of the extended header
+    Ok(extended_size)
+}
+
 /// Encodes native string to ID3v2 string encoding (either Latin-1 or UCS-2 according to the
 /// characters used).
 fn encode_string(s: &str) -> io::Result<Vec<u8>> {
@@ -110,9 +182,17 @@ impl RawFrame {
         }
 
         let id = String::from_utf8(data[0..4].to_vec()).unwrap();
-        let size = read_big_endian(&data[4..8], 8);
+        let size = match tag_version_sub {
+            3 => read_big_endian(&data[4..8], 8),
+            4 => read_big_endian(&data[4..8], 7),
+            _ => panic!("")
+        };
         let flags = read_big_endian(&data[8..10], 8) as u16;
 
+        if flags != 0 {
+            return Err(format!("Frame flags not supported (flags bytes {:04x})", flags));
+        }
+
         let body = data[10..10+size].to_vec();
 
         Ok(Some((RawFrame { id, flags, tag_version_sub, body }, 10 + size)))
@@ -259,8 +339,29 @@ impl ID3v2 {
             }
         };
 
-        if flags != 0 {
-            return Err(format!("No ID3 header flags supported ({:x})", flags).ioerr());
+        if (flags & 0x80) != 0 {
+            return Err(format!("ID3 unsynchronisation not supported").ioerr());
+        }
+
+        let extended_header = (flags & 0x40) != 0;
+
+        if (flags & 0x20) != 0 {
+            return Err(
+                format!("Refusing to read ID3 tag in \"experimental\" stage, whatever that may mean")
+                    .ioerr()
+            );
+        }
+
+        // ID3v2.4 only
+        if (flags & 0x10) != 0 {
+            return Err(format!("3DI footer unsupported (ID3v2.4 section 3.4)").ioerr());
+        }
+
+        if (flags & 0x0f) != 0 {
+            return Err(
+                format!("Unknown ID3 header flags found (flags byte: 0x{:02x})", flags)
+                    .ioerr()
+            );
         }
 
         let body = {
@@ -273,6 +374,10 @@ impl ID3v2 {
         let mut frames = Vec::new();
         let mut cursor = 0;
 
+        if extended_header {
+            cursor += parse_extended_header(&body, version_sub)?;
+        }
+
         while cursor < body.len() {
             let tag = &body[cursor..cursor+4];
             if tag.len() < 4 { break; }  // not even enough bytes anymore
-- 
cgit v1.2.3-70-g09d2