SoulStream/crates/persona-state/src/lib.rs

//! Persona profile + avatar-asset state — the shapes the bundled Freenet
//! contracts validate.
//!
//! BYTE-COMPATIBILITY IS THE CONTRACT. `ProfileState` and `MediaAssetV1`
//! must stay bincode-identical (field order, types, sig-body layouts) to
//! the wasm bundled in `src-tauri/assets/` (built from the F2/F2b spike
//! sources, keys pinned by tests there). Change anything here and every
//! published profile stops verifying — that is a migration event.
//!
//! Two signing conventions coexist under `soulstream.profile/v1` (an F2b
//! record): PROFILES sign their RAW sig-body; ASSETS sign blake3(sig-body).
//! `profile_sign_payload` / `asset_sign_payload` return exactly what the
//! identity delegate must be handed for each.

use ed25519_dalek::{Signature, Verifier, VerifyingKey};
use serde::{Deserialize, Serialize};

pub const SCHEMA_VERSION: u32 = 1;
pub const SIGN_DOMAIN: &str = "soulstream.profile/v1";

pub const PROFILE_PARAMS_FAMILY: &[u8] = b"soulstream.persona.v1";
pub const MEDIA_PARAMS_FAMILY: &[u8] = b"soulstream.media.v1";

pub const MAX_PROFILE_BYTES: usize = 4 * 1024;
pub const MAX_NAME_BYTES: usize = 64;
pub const MAX_BIO_BYTES: usize = 280;

/// v1 avatar rendition: one square PNG.
pub const V1_MEDIA_TYPE: &str = "image/png";
pub const V1_DIMENSION: u32 = 256;
pub const MAX_MEDIA_BYTES: usize = 96 * 1024;
pub const MAX_MEDIA_STATE_BYTES: usize = 128 * 1024;

// ------------------------------------------------------------------ profile

#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct MediaRefV1 {
    pub content_hash: [u8; 32],
    pub media_type: String,
    pub byte_length: u32,
    pub width: u16,
    pub height: u16,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct ProfileState {
    pub schema_version: u32,
    pub owner: [u8; 32],
    pub sequence: u64,
    pub display_name: String,
    pub bio: String,
    pub avatar: Option<MediaRefV1>,
    pub sig: Vec<u8>,
}

#[derive(Serialize)]
struct ProfileSigBody<'a> {
    schema_version: u32,
    owner: &'a [u8; 32],
    sequence: u64,
    display_name: &'a str,
    bio: &'a str,
    avatar: &'a Option<MediaRefV1>,
}

/// What the delegate signs for a profile: the RAW sig-body (F2 convention).
pub fn profile_sign_payload(p: &ProfileState) -> Vec<u8> {
    bincode::serialize(&ProfileSigBody {
        schema_version: p.schema_version,
        owner: &p.owner,
        sequence: p.sequence,
        display_name: &p.display_name,
        bio: &p.bio,
        avatar: &p.avatar,
    })
    .expect("serializes")
}

fn signable(domain: &str, payload: &[u8]) -> Vec<u8> {
    let mut msg = Vec::new();
    msg.extend_from_slice(b"soulstream.sig.v1");
    msg.push(0);
    msg.extend_from_slice(domain.as_bytes());
    msg.push(0);
    msg.extend_from_slice(blake3::hash(payload).as_bytes());
    msg
}

pub fn verify_sig(public: &[u8; 32], payload: &[u8], sig: &[u8]) -> bool {
    let Ok(vk) = VerifyingKey::from_bytes(public) else { return false };
    let Ok(sig) = Signature::from_slice(sig) else { return false };
    vk.verify(&signable(SIGN_DOMAIN, payload), &sig).is_ok()
}

pub fn make_profile_params(owner: &[u8; 32]) -> Vec<u8> {
    let mut p = Vec::with_capacity(PROFILE_PARAMS_FAMILY.len() + 32);
    p.extend_from_slice(PROFILE_PARAMS_FAMILY);
    p.extend_from_slice(owner);
    p
}

#[derive(Debug, PartialEq, Eq)]
pub enum ProfileInvalid {
    Oversized(usize),
    Malformed,
    WrongSchema(u32),
    WrongOwner,
    NameTooLong(usize),
    BioTooLong(usize),
    AvatarUnbounded,
    BadSignature,
}

impl core::fmt::Display for ProfileInvalid {
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        match self {
            ProfileInvalid::Oversized(n) => write!(f, "profile {n} bytes over cap {MAX_PROFILE_BYTES}"),
            ProfileInvalid::Malformed => write!(f, "bytes do not parse as a profile"),
            ProfileInvalid::WrongSchema(v) => write!(f, "unknown profile schema {v}"),
            ProfileInvalid::WrongOwner => write!(f, "profile owner differs from the address owner"),
            ProfileInvalid::NameTooLong(n) => write!(f, "display name {n} bytes over cap {MAX_NAME_BYTES}"),
            ProfileInvalid::BioTooLong(n) => write!(f, "bio {n} bytes over cap {MAX_BIO_BYTES}"),
            ProfileInvalid::AvatarUnbounded => write!(f, "avatar reference outside the v1 bounds"),
            ProfileInvalid::BadSignature => write!(f, "owner signature does not verify"),
        }
    }
}

fn avatar_bounded(r: &MediaRefV1) -> bool {
    r.media_type == V1_MEDIA_TYPE
        && r.byte_length as usize <= MAX_MEDIA_BYTES
        && r.width as u32 == V1_DIMENSION
        && r.height as u32 == V1_DIMENSION
}

/// Gen-2 validation (avatar-accepting) — the generation the bundled
/// contract enforces. Gen-1 states (avatar=None) validate identically.
pub fn validate_profile(owner: &[u8; 32], bytes: &[u8]) -> Result<ProfileState, ProfileInvalid> {
    if bytes.len() > MAX_PROFILE_BYTES {
        return Err(ProfileInvalid::Oversized(bytes.len()));
    }
    let p: ProfileState = bincode::deserialize(bytes).map_err(|_| ProfileInvalid::Malformed)?;
    if p.schema_version != SCHEMA_VERSION {
        return Err(ProfileInvalid::WrongSchema(p.schema_version));
    }
    if &p.owner != owner {
        return Err(ProfileInvalid::WrongOwner);
    }
    if p.display_name.len() > MAX_NAME_BYTES {
        return Err(ProfileInvalid::NameTooLong(p.display_name.len()));
    }
    if p.bio.len() > MAX_BIO_BYTES {
        return Err(ProfileInvalid::BioTooLong(p.bio.len()));
    }
    if let Some(r) = &p.avatar {
        if !avatar_bounded(r) {
            return Err(ProfileInvalid::AvatarUnbounded);
        }
    }
    if !verify_sig(&p.owner, &profile_sign_payload(&p), &p.sig) {
        return Err(ProfileInvalid::BadSignature);
    }
    Ok(p)
}

// ------------------------------------------------------------------ asset

#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct MediaAssetV1 {
    pub schema_version: u32,
    pub owner: [u8; 32],
    pub asset_id: [u8; 16],
    pub revision: u32,
    pub media_type: String,
    pub width: u32,
    pub height: u32,
    pub sanitized_static_bytes: Vec<u8>,
    pub sig: Vec<u8>,
}

#[derive(Serialize)]
struct AssetSigBody<'a> {
    schema_version: u32,
    owner: &'a [u8; 32],
    asset_id: &'a [u8; 16],
    revision: u32,
    media_type: &'a str,
    width: u32,
    height: u32,
    content_hash: &'a [u8; 32],
}

fn asset_sig_body(a: &MediaAssetV1) -> Vec<u8> {
    let content_hash = *blake3::hash(&a.sanitized_static_bytes).as_bytes();
    bincode::serialize(&AssetSigBody {
        schema_version: a.schema_version,
        owner: &a.owner,
        asset_id: &a.asset_id,
        revision: a.revision,
        media_type: &a.media_type,
        width: a.width,
        height: a.height,
        content_hash: &content_hash,
    })
    .expect("serializes")
}

/// What the delegate signs for an asset: blake3(sig-body) — 32 bytes.
pub fn asset_sign_payload(a: &MediaAssetV1) -> Vec<u8> {
    blake3::hash(&asset_sig_body(a)).as_bytes().to_vec()
}

pub fn make_media_params(owner: &[u8; 32], content_hash: &[u8; 32]) -> Vec<u8> {
    let mut p = Vec::with_capacity(MEDIA_PARAMS_FAMILY.len() + 64);
    p.extend_from_slice(MEDIA_PARAMS_FAMILY);
    p.extend_from_slice(owner);
    p.extend_from_slice(content_hash);
    p
}

#[derive(Debug, PartialEq, Eq)]
pub enum AssetInvalid {
    Oversized(usize),
    HashMismatch,
    Malformed,
    WrongSchema(u32),
    WrongOwner,
    WrongType(String),
    WrongDimensions(u32, u32),
    BytesOverBudget(usize),
    BadSignature,
}

impl core::fmt::Display for AssetInvalid {
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        match self {
            AssetInvalid::Oversized(n) => write!(f, "asset state {n} bytes over gate {MAX_MEDIA_STATE_BYTES}"),
            AssetInvalid::HashMismatch => write!(f, "bytes do not hash to the referenced content hash"),
            AssetInvalid::Malformed => write!(f, "bytes do not parse as a media asset"),
            AssetInvalid::WrongSchema(v) => write!(f, "unknown media schema {v}"),
            AssetInvalid::WrongOwner => write!(f, "asset owner differs from the reference owner"),
            AssetInvalid::WrongType(t) => write!(f, "media type {t:?} not accepted (v1: image/png)"),
            AssetInvalid::WrongDimensions(w, h) => write!(f, "declared {w}x{h} is not the 256x256 rendition"),
            AssetInvalid::BytesOverBudget(n) => write!(f, "{n} media bytes over the {MAX_MEDIA_BYTES} budget"),
            AssetInvalid::BadSignature => write!(f, "owner signature does not verify"),
        }
    }
}

/// Structural verification of a fetched asset against a signed reference —
/// everything that must pass BEFORE any image decoding happens.
pub fn validate_asset(
    owner: &[u8; 32],
    content_hash: &[u8; 32],
    state_bytes: &[u8],
) -> Result<MediaAssetV1, AssetInvalid> {
    if state_bytes.len() > MAX_MEDIA_STATE_BYTES {
        return Err(AssetInvalid::Oversized(state_bytes.len()));
    }
    let a: MediaAssetV1 = bincode::deserialize(state_bytes).map_err(|_| AssetInvalid::Malformed)?;
    if a.schema_version != SCHEMA_VERSION {
        return Err(AssetInvalid::WrongSchema(a.schema_version));
    }
    if &a.owner != owner {
        return Err(AssetInvalid::WrongOwner);
    }
    if a.sanitized_static_bytes.len() > MAX_MEDIA_BYTES {
        return Err(AssetInvalid::BytesOverBudget(a.sanitized_static_bytes.len()));
    }
    if blake3::hash(&a.sanitized_static_bytes).as_bytes() != content_hash {
        return Err(AssetInvalid::HashMismatch);
    }
    if a.media_type != V1_MEDIA_TYPE {
        return Err(AssetInvalid::WrongType(a.media_type.clone()));
    }
    if a.width != V1_DIMENSION || a.height != V1_DIMENSION {
        return Err(AssetInvalid::WrongDimensions(a.width, a.height));
    }
    if !verify_sig(owner, &asset_sign_payload(&a), &a.sig) {
        return Err(AssetInvalid::BadSignature);
    }
    Ok(a)
}

// ------------------------------------------------------------------ tests

#[cfg(test)]
mod tests {
    use super::*;
    use ed25519_dalek::{Signer, SigningKey};

    fn kp(seed: u8) -> SigningKey {
        SigningKey::from_bytes(&[seed; 32])
    }

    fn sign_raw(key: &SigningKey, payload: &[u8]) -> Vec<u8> {
        key.sign(&signable(SIGN_DOMAIN, payload)).to_bytes().to_vec()
    }

    #[test]
    fn profile_roundtrip_both_generations() {
        let k = kp(1);
        let owner = k.verifying_key().to_bytes();
        let mut p = ProfileState {
            schema_version: SCHEMA_VERSION,
            owner,
            sequence: 1,
            display_name: "Kyle".into(),
            bio: String::new(),
            avatar: None,
            sig: vec![],
        };
        p.sig = sign_raw(&k, &profile_sign_payload(&p));
        let b = bincode::serialize(&p).unwrap();
        assert!(validate_profile(&owner, &b).is_ok());
        // With a bounded avatar ref.
        let mut p2 = p.clone();
        p2.sequence = 2;
        p2.avatar = Some(MediaRefV1 {
            content_hash: [9u8; 32],
            media_type: V1_MEDIA_TYPE.into(),
            byte_length: 50_000,
            width: 256,
            height: 256,
        });
        p2.sig = sign_raw(&k, &profile_sign_payload(&p2));
        assert!(validate_profile(&owner, &bincode::serialize(&p2).unwrap()).is_ok());
        // Unbounded avatar refused; tamper refused.
        let mut p3 = p2.clone();
        p3.avatar.as_mut().unwrap().width = 4096;
        p3.sig = sign_raw(&k, &profile_sign_payload(&p3));
        assert_eq!(
            validate_profile(&owner, &bincode::serialize(&p3).unwrap()),
            Err(ProfileInvalid::AvatarUnbounded)
        );
        let mut p4 = p2.clone();
        p4.display_name = "Mallory".into();
        assert_eq!(
            validate_profile(&owner, &bincode::serialize(&p4).unwrap()),
            Err(ProfileInvalid::BadSignature)
        );
    }

    #[test]
    fn asset_roundtrip_and_hash_binding() {
        let k = kp(1);
        let owner = k.verifying_key().to_bytes();
        let mut a = MediaAssetV1 {
            schema_version: SCHEMA_VERSION,
            owner,
            asset_id: [4u8; 16],
            revision: 1,
            media_type: V1_MEDIA_TYPE.into(),
            width: V1_DIMENSION,
            height: V1_DIMENSION,
            sanitized_static_bytes: vec![7u8; 1000],
            sig: vec![],
        };
        a.sig = sign_raw(&k, &asset_sign_payload(&a));
        let hash = *blake3::hash(&a.sanitized_static_bytes).as_bytes();
        let b = bincode::serialize(&a).unwrap();
        assert!(validate_asset(&owner, &hash, &b).is_ok());
        assert_eq!(validate_asset(&owner, &[0u8; 32], &b), Err(AssetInvalid::HashMismatch));
    }
}

/// blake3 of raw bytes — the content-addressing hash used across the
/// persona/media contract families.
pub fn content_hash(bytes: &[u8]) -> [u8; 32] {
    *blake3::hash(bytes).as_bytes()
}