From cfe169ec31095376a58b5cb1384d3605962ef4dc Mon Sep 17 00:00:00 2001
From: Santiago Carmuega <santiago@carmuega.me>
Date: Wed, 9 Feb 2022 08:07:09 -0300
Subject: [PATCH] feat: Implement block cbor probing

---
 pallas-primitives/Cargo.toml     |  2 +-
 pallas-primitives/src/lib.rs     |  1 +
 pallas-primitives/src/probing.rs | 57 ++++++++++++++++++++++++++++++++
 3 files changed, 59 insertions(+), 1 deletion(-)
 create mode 100644 pallas-primitives/src/probing.rs

diff --git a/pallas-primitives/Cargo.toml b/pallas-primitives/Cargo.toml
index ec133f92..15ae3489 100644
--- a/pallas-primitives/Cargo.toml
+++ b/pallas-primitives/Cargo.toml
@@ -14,7 +14,7 @@ authors = [
 ]
 
 [dependencies]
-minicbor = { version = "0.13", features = ["std"] }
+minicbor = { version = "0.13", features = ["std", "half"] }
 minicbor-derive = "0.8.0"
 hex = "0.4.3"
 log = "0.4.14"
diff --git a/pallas-primitives/src/lib.rs b/pallas-primitives/src/lib.rs
index 0b85a7f8..fe32a294 100644
--- a/pallas-primitives/src/lib.rs
+++ b/pallas-primitives/src/lib.rs
@@ -4,6 +4,7 @@ mod framework;
 
 pub mod alonzo;
 pub mod byron;
+pub mod probing;
 pub mod utils;
 
 pub use framework::*;
diff --git a/pallas-primitives/src/probing.rs b/pallas-primitives/src/probing.rs
new file mode 100644
index 00000000..8065588c
--- /dev/null
+++ b/pallas-primitives/src/probing.rs
@@ -0,0 +1,57 @@
+//! Heuristics for detecting cbor content without decoding
+
+use minicbor::decode::{Token, Tokenizer};
+
+pub enum BlockInference {
+    Byron,
+    Shelley,
+    Inconclusive,
+}
+
+// Executes a very lightweight inspection of the initial tokens of the CBOR
+// payload and infers with a certain degree of confidence the type of Cardano
+// structure within.
+pub fn probe_block_cbor(cbor: &[u8]) -> BlockInference {
+    let mut tokenizer = Tokenizer::new(cbor);
+
+    if !matches!(tokenizer.next(), Some(Ok(Token::Array(2)))) {
+        return BlockInference::Inconclusive;
+    }
+
+    if !matches!(tokenizer.next(), Some(Ok(Token::U8(_)))) {
+        return BlockInference::Inconclusive;
+    }
+
+    //println!("{:?}", tokenizer.next());
+
+    match tokenizer.next() {
+        Some(Ok(Token::Array(3))) => BlockInference::Byron,
+        Some(Ok(Token::Array(5))) => BlockInference::Shelley,
+        _ => BlockInference::Inconclusive,
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn byron_block_detected() {
+        let block_str = include_str!("byron/test_data/test1.block");
+        let bytes = hex::decode(block_str).unwrap();
+
+        let inference = probe_block_cbor(bytes.as_slice());
+
+        assert!(matches!(inference, BlockInference::Byron));
+    }
+
+    #[test]
+    fn shelley_block_detected() {
+        let block_str = include_str!("alonzo/test_data/test1.block");
+        let bytes = hex::decode(block_str).unwrap();
+
+        let inference = probe_block_cbor(bytes.as_slice());
+
+        assert!(matches!(inference, BlockInference::Shelley));
+    }
+}