WIP: aquatic_ws: add half-crappy load test, fix infinite loop bug

aquatic_ws_protocol/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "aquatic_ws_protocol"
+version = "0.1.0"
+authors = ["Joakim Frostegård <joakim.frostegard@gmail.com>"]
+edition = "2018"
+license = "Apache-2.0"
+
+[lib]
+name = "aquatic_ws_protocol"
+
+[dependencies]
+hashbrown = { version = "0.8", features = ["serde"] }
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+tungstenite = "0.11"
+
+[dev-dependencies]

aquatic_ws_protocol/src/lib.rs

@@ -0,0 +1,300 @@
+use hashbrown::HashMap;
+use serde::{Serialize, Deserialize};
+
+mod serde_helpers;
+
+use serde_helpers::*;
+
+
+#[derive(Debug, Clone, Copy, Hash, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct PeerId(
+    #[serde(
+        deserialize_with = "deserialize_20_bytes",
+        serialize_with = "serialize_20_bytes"
+    )]
+    pub [u8; 20]
+);
+
+
+#[derive(Debug, Clone, Copy, Hash, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct InfoHash(
+    #[serde(
+        deserialize_with = "deserialize_20_bytes",
+        serialize_with = "serialize_20_bytes"
+    )]
+    pub [u8; 20]
+);
+
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct OfferId(
+    #[serde(
+        deserialize_with = "deserialize_20_bytes",
+        serialize_with = "serialize_20_bytes"
+    )]
+    pub [u8; 20]
+);
+
+
+/// Some kind of nested structure from https://www.npmjs.com/package/simple-peer
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct JsonValue(pub ::serde_json::Value);
+
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "lowercase")]
+pub enum AnnounceEvent {
+    Started,
+    Stopped,
+    Completed,
+    Update
+}
+
+
+impl Default for AnnounceEvent {
+    fn default() -> Self {
+        Self::Update
+    }
+}
+
+
+/// Apparently, these are sent to a number of peers when they are set
+/// in an AnnounceRequest
+/// action = "announce"
+#[derive(Debug, Clone, Serialize)]
+pub struct MiddlemanOfferToPeer {
+    /// Peer id of peer sending offer
+    /// Note: if equal to client peer_id, client ignores offer
+    pub peer_id: PeerId,
+    pub info_hash: InfoHash,
+    /// Gets copied from AnnounceRequestOffer
+    pub offer: JsonValue, 
+    /// Gets copied from AnnounceRequestOffer
+    pub offer_id: OfferId,
+}
+
+
+/// If announce request has answer = true, send this to peer with
+/// peer id == "to_peer_id" field
+/// Action field should be 'announce'
+#[derive(Debug, Clone, Serialize)]
+pub struct MiddlemanAnswerToPeer {
+    /// Note: if equal to client peer_id, client ignores answer
+    pub peer_id: PeerId,
+    pub info_hash: InfoHash,
+    pub answer: JsonValue,
+    pub offer_id: OfferId,
+}
+
+
+/// Element of AnnounceRequest.offers
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct AnnounceRequestOffer {
+    pub offer: JsonValue,
+    pub offer_id: OfferId,
+}
+
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct AnnounceRequest {
+    pub info_hash: InfoHash,
+    pub peer_id: PeerId,
+    /// Just called "left" in protocol. Is set to None in some cases, such as
+    /// when opening a magnet link
+    #[serde(rename = "left")]
+    pub bytes_left: Option<usize>,
+    /// Can be empty. Then, default is "update"
+    #[serde(default)]
+    pub event: AnnounceEvent,
+
+    /// Only when this is an array offers are sent to random peers
+    /// Length of this is number of peers wanted?
+    /// Max length of this is 10 in reference client code
+    /// Not sent when announce event is stopped or completed
+    pub offers: Option<Vec<AnnounceRequestOffer>>, 
+    /// Seems to only get sent by client when sending offers, and is also same
+    /// as length of offers vector (or at least never less)
+    /// Max length of this is 10 in reference client code
+    /// Could probably be ignored, `offers.len()` should provide needed info
+    pub numwant: Option<usize>,
+
+    /// If empty, send response before sending offers (or possibly "skip sending update back"?)
+    /// Else, send MiddlemanAnswerToPeer to peer with "to_peer_id" as peer_id.
+    /// I think using Option is good, it seems like this isn't always set
+    /// (same as `offers`)
+    pub answer: Option<JsonValue>, 
+    /// Likely undefined if !(answer == true)
+    pub to_peer_id: Option<PeerId>, 
+    /// Sent if answer is set
+    pub offer_id: Option<OfferId>,
+}
+
+
+#[derive(Debug, Clone, Serialize)]
+pub struct AnnounceResponse {
+    pub info_hash: InfoHash,
+    /// Client checks if this is null, not clear why
+    pub complete: usize,
+    pub incomplete: usize,
+    #[serde(rename = "interval")]
+    pub announce_interval: usize, // Default 2 min probably
+}
+
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ScrapeRequest {
+    // If omitted, scrape for all torrents, apparently
+    // There is some kind of parsing here too which accepts a single info hash
+    // and puts it into a vector
+    #[serde(
+        rename = "info_hash",
+        deserialize_with = "deserialize_info_hashes",
+        default
+    )]
+    pub info_hashes: Vec<InfoHash>,
+}
+
+
+#[derive(Debug, Clone, Serialize)]
+pub struct ScrapeStatistics {
+    pub complete: usize,
+    pub incomplete: usize,
+    pub downloaded: usize,
+}
+
+
+#[derive(Debug, Clone, Serialize)]
+pub struct ScrapeResponse {
+    pub files: HashMap<InfoHash, ScrapeStatistics>,
+    // Looks like `flags` field is ignored in reference client
+    // pub flags: HashMap<String, usize>,
+}
+
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "lowercase")]
+pub enum Action {
+    Announce,
+    Scrape
+}
+
+
+/// Helper for serializing and deserializing messages
+#[derive(Debug, Clone, Serialize, Deserialize)]
+struct ActionWrapper<T> {
+    pub action: Action,
+    #[serde(flatten)]
+    pub inner: T,
+}
+
+
+impl <T>ActionWrapper<T> {
+    #[inline]
+    pub fn announce(t: T) -> Self {
+        Self {
+            action: Action::Announce,
+            inner: t
+        }
+    }
+    #[inline]
+    pub fn scrape(t: T) -> Self {
+        Self {
+            action: Action::Scrape,
+            inner: t
+        }
+    }
+}
+
+
+#[derive(Debug, Clone)]
+pub enum InMessage {
+    AnnounceRequest(AnnounceRequest),
+    ScrapeRequest(ScrapeRequest),
+}
+
+
+impl InMessage {
+    /// Try parsing as announce request first. If that fails, try parsing as
+    /// scrape request, or return None
+    #[inline]
+    pub fn from_ws_message(ws_message: tungstenite::Message) -> Option<Self> {
+        use tungstenite::Message::{Text, Binary};
+
+        let text = match ws_message {
+            Text(text) => Some(text),
+            Binary(bytes) => String::from_utf8(bytes).ok(),
+            _ => None
+        }?;
+
+        let res: Result<ActionWrapper<AnnounceRequest>, _> = serde_json::from_str(&text);
+
+        if let Ok(ActionWrapper { action: Action::Announce, inner }) = res {
+            return Some(InMessage::AnnounceRequest(inner));
+        }
+        
+        let res: Result<ActionWrapper<ScrapeRequest>, _> = serde_json::from_str(&text);
+
+        if let Ok(ActionWrapper { action: Action::Scrape, inner }) = res {
+            return Some(InMessage::ScrapeRequest(inner));
+        }
+
+        None
+    }
+
+    pub fn to_ws_message(&self) -> ::tungstenite::Message {
+        let text = match self {
+            InMessage::AnnounceRequest(r) => {
+                serde_json::to_string(&ActionWrapper::announce(r)).unwrap()
+            },
+            InMessage::ScrapeRequest(r) => {
+                serde_json::to_string(&ActionWrapper::scrape(r)).unwrap()
+            },
+        };
+
+        ::tungstenite::Message::from(text)
+    }
+}
+
+
+#[derive(Debug, Clone)]
+pub enum OutMessage {
+    AnnounceResponse(AnnounceResponse),
+    ScrapeResponse(ScrapeResponse),
+    Offer(MiddlemanOfferToPeer),
+    Answer(MiddlemanAnswerToPeer),
+}
+
+
+impl OutMessage {
+    #[inline]
+    pub fn to_ws_message(self) -> tungstenite::Message {
+        let json = match self {
+            Self::AnnounceResponse(message) => {
+                serde_json::to_string(
+                    &ActionWrapper::announce(message)
+                ).unwrap()
+            },
+            Self::Offer(message) => {
+                serde_json::to_string(
+                    &ActionWrapper::announce(message)
+                ).unwrap()
+            },
+            Self::Answer(message) => {
+                serde_json::to_string(
+                    &ActionWrapper::announce(message)
+                ).unwrap()
+            },
+            Self::ScrapeResponse(message) => {
+                serde_json::to_string(
+                    &ActionWrapper::scrape(message)
+                ).unwrap()
+            },
+        };
+                
+        tungstenite::Message::from(json)
+    }
+}

aquatic_ws_protocol/src/serde_helpers.rs

@@ -0,0 +1,247 @@
+use serde::{Serializer, Deserializer, de::{Visitor, SeqAccess}};
+
+use super::InfoHash;
+
+
+pub fn serialize_20_bytes<S>(
+    data: &[u8; 20],
+    serializer: S
+) -> Result<S::Ok, S::Error> where S: Serializer {
+    let text: String = data.iter().map(|byte| *byte as char).collect();
+
+    serializer.serialize_str(&text)
+}
+
+
+struct TwentyByteVisitor;
+
+impl<'de> Visitor<'de> for TwentyByteVisitor {
+    type Value = [u8; 20];
+
+    fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+        formatter.write_str("string consisting of 20 bytes")
+    }
+
+    #[inline]
+    fn visit_str<E>(self, value: &str) -> Result<Self::Value, E>
+        where E: ::serde::de::Error,
+    {
+        // Value is encoded in nodejs reference client something as follows:
+        // ```
+        // var infoHash = 'abcd..'; // 40 hexadecimals
+        // Buffer.from(infoHash, 'hex').toString('binary');
+        // ```
+        // As I understand it:
+        // - the code above produces a UTF16 string of 20 chars, each having
+        //   only the "low byte" set (e.g., numeric value ranges from 0-255)
+        // - serde_json decodes this to string of 20 chars (tested), each in
+        //   the aforementioned range (tested), so the bytes can be extracted
+        //   by casting each char to u8.
+
+        let mut arr = [0u8; 20];
+        let mut char_iter = value.chars();
+
+        for a in arr.iter_mut(){
+            if let Some(c) = char_iter.next(){
+                if c as u32 > 255 {
+                    return Err(E::custom(format!(
+                        "character not in single byte range: {:#?}",
+                        c
+                    )));
+                }
+
+                *a = c as u8;
+            } else {
+                return Err(E::custom(format!("not 20 bytes: {:#?}", value)));
+            }
+        }
+
+        Ok(arr)
+    }
+}
+
+
+#[inline]
+pub fn deserialize_20_bytes<'de, D>(
+    deserializer: D
+) -> Result<[u8; 20], D::Error>
+    where D: Deserializer<'de>
+{
+    deserializer.deserialize_any(TwentyByteVisitor)
+}
+
+
+pub struct InfoHashVecVisitor;
+
+
+impl<'de> Visitor<'de> for InfoHashVecVisitor {
+    type Value = Vec<InfoHash>;
+
+    fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+        formatter.write_str("string or array of strings consisting of 20 bytes")
+    }
+
+    #[inline]
+    fn visit_str<E>(self, value: &str) -> Result<Self::Value, E>
+        where E: ::serde::de::Error,
+    {
+        match TwentyByteVisitor::visit_str::<E>(TwentyByteVisitor, value){
+            Ok(arr) => Ok(vec![InfoHash(arr)]),
+            Err(err) => Err(E::custom(format!("got string, but {}", err)))
+        }
+    }
+
+    #[inline]
+    fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
+        where A: SeqAccess<'de>
+    {
+        let mut info_hashes: Self::Value = Vec::new();
+
+        while let Ok(Some(value)) = seq.next_element::<&str>(){
+            let arr = TwentyByteVisitor::visit_str(
+                TwentyByteVisitor, value
+            )?;
+
+            info_hashes.push(InfoHash(arr));
+        }
+
+        Ok(info_hashes)
+    }
+
+    #[inline]
+    fn visit_none<E>(self) -> Result<Self::Value, E>
+        where E: ::serde::de::Error
+    {
+        Ok(vec![])
+    }
+}
+
+
+/// Empty vector is returned if value is null or any invalid info hash
+/// is present
+#[inline]
+pub fn deserialize_info_hashes<'de, D>(
+    deserializer: D
+) -> Result<Vec<InfoHash>, D::Error>
+    where D: Deserializer<'de>,
+{
+    Ok(deserializer.deserialize_any(InfoHashVecVisitor).unwrap_or_default())
+}
+
+
+#[cfg(test)]
+mod tests {
+    use serde::Deserialize;
+
+    use super::*;
+
+    fn info_hash_from_bytes(bytes: &[u8]) -> InfoHash {
+        let mut arr = [0u8; 20];
+
+        assert!(bytes.len() == 20);
+
+        arr.copy_from_slice(&bytes[..]);
+
+        InfoHash(arr)
+    }
+
+    #[test]
+    fn test_deserialize_20_bytes(){
+        let input = r#""aaaabbbbccccddddeeee""#;
+
+        let expected = info_hash_from_bytes(b"aaaabbbbccccddddeeee");
+        let observed: InfoHash = serde_json::from_str(input).unwrap();
+
+        assert_eq!(observed, expected);
+
+        let input = r#""aaaabbbbccccddddeee""#;
+        let res_info_hash: Result<InfoHash, _> = serde_json::from_str(input);
+
+        assert!(res_info_hash.is_err());
+
+        let input = r#""aaaabbbbccccddddeee𝕊""#;
+        let res_info_hash: Result<InfoHash, _> = serde_json::from_str(input);
+
+        assert!(res_info_hash.is_err());
+    }
+
+    #[test]
+    fn test_serde_20_bytes(){
+        let info_hash = info_hash_from_bytes(b"aaaabbbbccccddddeeee");
+
+        let out = serde_json::to_string(&info_hash).unwrap();
+        let info_hash_2 = serde_json::from_str(&out).unwrap();
+
+        assert_eq!(info_hash, info_hash_2);
+    }
+
+    #[derive(Debug, PartialEq, Eq, Deserialize)]
+    struct Test {
+        #[serde(deserialize_with = "deserialize_info_hashes", default)]
+        info_hashes: Vec<InfoHash>,
+    }
+
+
+    #[test]
+    fn test_deserialize_info_hashes_vec(){
+        let input = r#"{
+            "info_hashes": ["aaaabbbbccccddddeeee", "aaaabbbbccccddddeeee"]
+        }"#;
+
+        let expected = Test {
+            info_hashes: vec![
+                info_hash_from_bytes(b"aaaabbbbccccddddeeee"),
+                info_hash_from_bytes(b"aaaabbbbccccddddeeee"),
+            ]
+        };
+
+        let observed: Test = serde_json::from_str(input).unwrap();
+
+        assert_eq!(observed, expected);
+    }
+
+    #[test]
+    fn test_deserialize_info_hashes_str(){
+        let input = r#"{
+            "info_hashes": "aaaabbbbccccddddeeee"
+        }"#;
+
+        let expected = Test {
+            info_hashes: vec![
+                info_hash_from_bytes(b"aaaabbbbccccddddeeee"),
+            ]
+        };
+
+        let observed: Test = serde_json::from_str(input).unwrap();
+
+        assert_eq!(observed, expected);
+    }
+
+    #[test]
+    fn test_deserialize_info_hashes_null(){
+        let input = r#"{
+            "info_hashes": null
+        }"#;
+
+        let expected = Test {
+            info_hashes: vec![]
+        };
+
+        let observed: Test = serde_json::from_str(input).unwrap();
+
+        assert_eq!(observed, expected);
+    }
+
+    #[test]
+    fn test_deserialize_info_hashes_missing(){
+        let input = r#"{}"#;
+
+        let expected = Test {
+            info_hashes: vec![]
+        };
+
+        let observed: Test = serde_json::from_str(input).unwrap();
+
+        assert_eq!(observed, expected);
+    }
+}