From 112e7da3f300ea06b3c445daf1318ec7ef7bc901 Mon Sep 17 00:00:00 2001
From: Tangent 128 <Tangent128@gmail.com>
Date: Thu, 12 Jan 2017 00:41:35 -0500
Subject: [PATCH] Implement EBML varint parser

---
 src/lib.rs | 87 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 87 insertions(+)

diff --git a/src/lib.rs b/src/lib.rs
index 9194270..4e305b4 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,10 +1,72 @@
 
 extern crate futures;
 
+#[derive(Debug, PartialEq)]
+pub enum Error {
+    CorruptVarint
+}
+
+#[derive(Debug, PartialEq)]
+pub enum EbmlVarint {
+    Value(u64),
+    Unknown
+}
+
+/// Try to parse an EBML varint starting at the start of the given slice.
+/// Returns an Err() if the format is corrupt.
+/// Returns Ok(None) if more bytes are needed to get a result.
+/// Returns Ok(Some((varint, next))) to return a varint value and
+/// the starting location fo the next 
+pub fn decode_varint(bytes: &[u8]) -> Result<Option<(EbmlVarint, usize)>, Error> {
+    let mut value: u64 = 0;
+    let mut value_length = 1;
+    let mut mask: u8 = 0x80;
+    let mut unknown_marker: u64 = !0;
+
+    if bytes.len() == 0 {
+        return Ok(None)
+    }
+
+    // get length marker bit from first byte & parse first byte
+    while mask > 0 {
+        if (mask & bytes[0]) != 0 {
+            value = (bytes[0] & !mask) as u64;
+            unknown_marker = (mask - 1) as u64;
+            break
+        }
+        value_length += 1;
+        mask = mask >> 1;
+    }
+
+    if mask == 0 {
+        return Err(Error::CorruptVarint)
+    }
+
+    // check we have enough data to parse
+    if value_length > bytes.len() {
+        return Ok(None)
+    }
+
+    // decode remaining bytes
+    for i in 1..value_length {
+        value = (value << 8) + (bytes[i] as u64);
+        unknown_marker = (unknown_marker << 8) + 0xFF;
+    }
+
+    // determine result
+    if value == unknown_marker {
+        Ok(Some((EbmlVarint::Unknown, value_length)))
+    } else {
+        Ok(Some((EbmlVarint::Value(value), value_length)))
+    }
+}
+
 #[cfg(test)]
 mod tests {
 
     use futures::future::{ok, Future};
+    use super::{decode_varint, Error};
+    use super::EbmlVarint::{Unknown, Value};
 
     #[test]
     fn hello_futures() {
@@ -15,4 +77,29 @@ mod tests {
 
         assert_eq!(string_result, "Hello, Futures!");
     }
+
+    #[test]
+    fn fail_corrupted_varints() {
+        assert_eq!(decode_varint(&[0]), Err(Error::CorruptVarint));
+        assert_eq!(decode_varint(&[0, 0, 0]), Err(Error::CorruptVarint));
+    }
+
+    #[test]
+    fn incomplete_varints() {
+        assert_eq!(decode_varint(&[]), Ok(None));
+        assert_eq!(decode_varint(&[0x40]), Ok(None));
+        assert_eq!(decode_varint(&[0x01, 0, 0]), Ok(None));
+    }
+
+    #[test]
+    fn parse_varints() {
+        assert_eq!(decode_varint(&[0xFF]), Ok(Some((Unknown, 1))));
+        assert_eq!(decode_varint(&[0x7F, 0xFF]), Ok(Some((Unknown, 2))));
+        assert_eq!(decode_varint(&[0x80]), Ok(Some((Value(0), 1))));
+        assert_eq!(decode_varint(&[0x81]), Ok(Some((Value(1), 1))));
+        assert_eq!(decode_varint(&[0x40, 52]), Ok(Some((Value(52), 2))));
+
+        // test extra data in buffer
+        assert_eq!(decode_varint(&[0x83, 0x11]), Ok(Some((Value(3), 1))));
+    }
 }