lex_types/

env.rs

//! Type environment: type-decl info and value-binding scopes.

use crate::types::*;
use indexmap::IndexMap;

#[derive(Debug, Clone)]
pub struct TypeDef {
    pub params: Vec<String>,
    pub kind: TypeDefKind,
}

#[derive(Debug, Clone)]
pub enum TypeDefKind {
    /// A union: variant name → optional payload.
    Union(IndexMap<String, Option<Ty>>),
    /// A record alias: `type Foo = { x :: Int }` etc.
    Alias(Ty),
    /// Built-in opaque (Map, Set, ...).
    Opaque,
}

#[derive(Debug, Clone, Default)]
pub struct TypeEnv {
    /// Type-name → definition.
    pub types: IndexMap<String, TypeDef>,
    /// Constructor name → owning type-name.
    pub ctor_to_type: IndexMap<String, String>,
}

impl TypeEnv {
    pub fn new_with_builtins() -> Self {
        let mut e = TypeEnv::default();
        // Result[T, E] = Ok(T) | Err(E)
        let mut r_variants = IndexMap::new();
        r_variants.insert("Ok".into(), Some(Ty::Var(0))); // T
        r_variants.insert("Err".into(), Some(Ty::Var(1))); // E
        e.types.insert("Result".into(), TypeDef {
            params: vec!["T".into(), "E".into()],
            kind: TypeDefKind::Union(r_variants),
        });
        e.ctor_to_type.insert("Ok".into(), "Result".into());
        e.ctor_to_type.insert("Err".into(), "Result".into());

        // Option[T] = Some(T) | None
        let mut o_variants = IndexMap::new();
        o_variants.insert("Some".into(), Some(Ty::Var(0))); // T
        o_variants.insert("None".into(), None);
        e.types.insert("Option".into(), TypeDef {
            params: vec!["T".into()],
            kind: TypeDefKind::Union(o_variants),
        });
        e.ctor_to_type.insert("Some".into(), "Option".into());
        e.ctor_to_type.insert("None".into(), "Option".into());

        // Nil = Unit (alias)
        e.types.insert("Nil".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Unit),
        });

        // Map, Set: opaque-ish. We just register the names so they parse as Cons.
        e.types.insert("Map".into(), TypeDef { params: vec!["K".into(), "V".into()], kind: TypeDefKind::Opaque });
        e.types.insert("Set".into(), TypeDef { params: vec!["T".into()], kind: TypeDefKind::Opaque });

        // SqlParam = PStr(Str) | PInt(Int) | PFloat(Float) | PBool(Bool) | PNull
        // Typed parameter binding for std.sql (#362). Replaces the v1 List[Str]
        // approach so callers don't have to stringify non-string values.
        let mut sp_variants = IndexMap::new();
        sp_variants.insert("PStr".into(),   Some(Ty::str()));
        sp_variants.insert("PInt".into(),   Some(Ty::int()));
        sp_variants.insert("PFloat".into(), Some(Ty::float()));
        sp_variants.insert("PBool".into(),  Some(Ty::bool()));
        sp_variants.insert("PNull".into(),  None);
        e.types.insert("SqlParam".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(sp_variants),
        });
        for ctor in &["PStr", "PInt", "PFloat", "PBool", "PNull"] {
            e.ctor_to_type.insert((*ctor).into(), "SqlParam".into());
        }

        // SqlTx: opaque transaction handle (#362). Backed by the same
        // Int registry key as Db; the type system enforces that commit/
        // rollback can only be called on a value from sql.begin, not on
        // a raw Db connection.
        e.types.insert("SqlTx".into(), TypeDef { params: vec![], kind: TypeDefKind::Opaque });

        // SqlError = { message :: Str, code :: Option[Str], detail :: Option[Str] }
        // Structured error shape returned by every `std.sql` op (#380).
        // `code` carries the SQLSTATE (Postgres) or the symbolic SQLite
        // error name (`SQLITE_BUSY`, `SQLITE_CONSTRAINT_UNIQUE`, …) so
        // dialect-aware retry / conflict-handling can avoid string
        // parsing. `message` is always populated; `detail` carries a
        // driver-side detail string when present.
        let mut se_fields = IndexMap::new();
        se_fields.insert("message".into(), Ty::str());
        se_fields.insert("code".into(), Ty::Con("Option".into(), vec![Ty::str()]));
        se_fields.insert("detail".into(), Ty::Con("Option".into(), vec![Ty::str()]));
        e.types.insert("SqlError".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(se_fields)),
        });

        // AeadResult = { ciphertext :: Bytes, tag :: Bytes } — return
        // shape for every AEAD seal op in `std.crypto` (#382 AEAD slice).
        // The auth tag is split out from the ciphertext so callers don't
        // have to know each algorithm's tag length: AES-GCM and
        // ChaCha20-Poly1305 both happen to be 16 bytes today, but the
        // shape keeps that detail encapsulated.
        let mut ar_fields = IndexMap::new();
        ar_fields.insert("ciphertext".into(), Ty::bytes());
        ar_fields.insert("tag".into(), Ty::bytes());
        e.types.insert("AeadResult".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(ar_fields)),
        });

        // UdpDatagram = { data :: Bytes, host :: Str, port :: Int } —
        // what `net.udp_recv` hands back (#760).
        //
        // The sender's address is part of the value rather than something
        // the caller has to ask for separately, because with UDP it is not
        // optional detail: any host can send to an open socket, so a reply
        // that does not carry who sent it cannot be safely acted on. A
        // request/response caller must check it matches who they asked.
        let mut dg_fields = IndexMap::new();
        dg_fields.insert("data".into(), Ty::bytes());
        dg_fields.insert("host".into(), Ty::str());
        dg_fields.insert("port".into(), Ty::int());
        e.types.insert("UdpDatagram".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(dg_fields)),
        });

        // Iter[T]: lazy positional iterator (#364). Backed at runtime by a
        // (List[T], Int) tuple; the Int is the current cursor index. All
        // iter.* operations are compiler-inlined so no effect is needed.
        e.types.insert("Iter".into(), TypeDef { params: vec!["T".into()], kind: TypeDefKind::Opaque });

        // Stream[T]: opaque streaming iterator (#305 slice 3).
        // Built and consumed exclusively through the `stream.*` and
        // `agent.cloud_stream` effect builtins; the runtime
        // represents a Stream value as an opaque variant carrying a
        // handle id. Registered as Opaque so type-checking knows
        // `Stream[Str]` parses but doesn't unwrap it structurally.
        e.types.insert("Stream".into(), TypeDef { params: vec!["T".into()], kind: TypeDefKind::Opaque });

        // Tz = Utc | Local | Offset(Int) | Iana(Str).
        // Used by std.datetime; the variant-typed alternative to the
        // pre-v1 stringly Tz ("UTC" / "Local" / "+05:30" / IANA name).
        // Registered globally so users don't have to import a module
        // to mention `Utc` / `Iana("America/New_York")` etc.
        let mut tz_variants = IndexMap::new();
        tz_variants.insert("Utc".into(), None);
        tz_variants.insert("Local".into(), None);
        tz_variants.insert("Offset".into(), Some(Ty::int()));
        tz_variants.insert("Iana".into(), Some(Ty::str()));
        e.types.insert("Tz".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(tz_variants),
        });
        for ctor in &["Utc", "Local", "Offset", "Iana"] {
            e.ctor_to_type.insert((*ctor).into(), "Tz".into());
        }

        // HttpError = NetworkError(Str) | TimeoutError | TlsError(Str)
        //           | DecodeError(Str)
        // Used by std.http; structured failure shape so callers can
        // discriminate transport vs. timeout vs. TLS vs. body-decode
        // errors without parsing strings.
        let mut http_err_variants = IndexMap::new();
        http_err_variants.insert("NetworkError".into(), Some(Ty::str()));
        http_err_variants.insert("TimeoutError".into(), None);
        http_err_variants.insert("TlsError".into(), Some(Ty::str()));
        http_err_variants.insert("DecodeError".into(), Some(Ty::str()));
        e.types.insert("HttpError".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(http_err_variants),
        });
        for ctor in &["NetworkError", "TimeoutError", "TlsError", "DecodeError"] {
            e.ctor_to_type.insert((*ctor).into(), "HttpError".into());
        }

        // HttpRequest = { method, url, headers, body, timeout_ms }.
        // The std.http request shape. Anonymous record literals coerce
        // to this nominal alias at every position (per the §3.13
        // record-coercion rules), so users write
        // `{ method: "GET", url: u, headers: map.new(), body: None,
        // timeout_ms: None }` rather than a dedicated constructor —
        // builders (`http.with_header` etc.) are pure transforms over
        // the same shape.
        let mut req_fields = IndexMap::new();
        req_fields.insert("method".into(), Ty::str());
        req_fields.insert("url".into(), Ty::str());
        req_fields.insert("headers".into(), Ty::Con("Map".into(), vec![Ty::str(), Ty::str()]));
        req_fields.insert("body".into(), Ty::Con("Option".into(), vec![Ty::bytes()]));
        req_fields.insert("timeout_ms".into(), Ty::Con("Option".into(), vec![Ty::int()]));
        e.types.insert("HttpRequest".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(req_fields)),
        });

        // HttpResponse = { status, headers, body }. Returned by every
        // `http.{send,get,post}` happy path; also the input to
        // `http.{json_body,text_body}`.
        let mut resp_fields = IndexMap::new();
        resp_fields.insert("status".into(), Ty::int());
        resp_fields.insert("headers".into(), Ty::Con("Map".into(), vec![Ty::str(), Ty::str()]));
        resp_fields.insert("body".into(), Ty::bytes());
        e.types.insert("HttpResponse".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(resp_fields)),
        });

        // Matrix = { rows :: Int, cols :: Int, data :: List[Float] }.
        // Used by std.math; runtime values are the F64Array fast lane,
        // not a real record. The alias makes math.* signatures readable
        // (`:: Matrix` instead of an inline record) and lets call sites
        // unify nominally. Field access via `m.rows` would type-check
        // but fail at runtime — use `math.rows / math.cols / math.get`.
        let mut mat_fields = IndexMap::new();
        mat_fields.insert("rows".into(), Ty::int());
        mat_fields.insert("cols".into(), Ty::int());
        mat_fields.insert("data".into(), Ty::List(Box::new(Ty::float())));
        e.types.insert("Matrix".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(mat_fields)),
        });

        // Request = { method :: Str, path :: Str, query :: Str, body :: Str,
        //             headers :: Map[Str, Str], path_params :: Map[Str, Str] }
        // Inbound request shape used by net.serve_fn handlers.
        // `path_params` is populated by `net.serve_routed` from `:name`
        // segments in the route pattern; empty under `net.serve_fn`.
        let mut net_req_fields = IndexMap::new();
        net_req_fields.insert("method".into(), Ty::str());
        net_req_fields.insert("path".into(), Ty::str());
        net_req_fields.insert("query".into(), Ty::str());
        net_req_fields.insert("body".into(), Ty::str());
        net_req_fields.insert("headers".into(), Ty::Con("Map".into(), vec![Ty::str(), Ty::str()]));
        net_req_fields.insert("path_params".into(), Ty::Con("Map".into(), vec![Ty::str(), Ty::str()]));
        e.types.insert("Request".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(net_req_fields)),
        });

        // Response = { status :: Int, body :: ResponseBody, headers :: Map[Str, Str] }
        // Outbound response shape returned by net.serve_fn handlers.
        // #375: `body` is now an ADT instead of a bare Str. Streaming
        // variants (BodyStream / BodyBytes) carry an `Iter[T]` that the
        // server drains chunk-by-chunk under chunked transfer-encoding.
        let mut rb_variants = IndexMap::new();
        rb_variants.insert("BodyStr".into(),    Some(Ty::str()));
        rb_variants.insert(
            "BodyStream".into(),
            Some(Ty::Con("Iter".into(), vec![Ty::str()])),
        );
        rb_variants.insert(
            "BodyBytes".into(),
            Some(Ty::Con("Iter".into(), vec![Ty::List(Box::new(Ty::int()))])),
        );
        e.types.insert("ResponseBody".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(rb_variants),
        });
        for ctor in &["BodyStr", "BodyStream", "BodyBytes"] {
            e.ctor_to_type.insert((*ctor).into(), "ResponseBody".into());
        }

        let mut net_resp_fields = IndexMap::new();
        net_resp_fields.insert("status".into(), Ty::int());
        net_resp_fields.insert("body".into(), Ty::Con("ResponseBody".into(), vec![]));
        net_resp_fields.insert("headers".into(), Ty::Con("Map".into(), vec![Ty::str(), Ty::str()]));
        e.types.insert("Response".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(net_resp_fields)),
        });

        // WsConn = { id :: Str, path :: Str, subprotocol :: Str }
        // Passed to every net.serve_ws_fn message handler.
        let mut ws_conn_fields = IndexMap::new();
        ws_conn_fields.insert("id".into(), Ty::str());
        ws_conn_fields.insert("path".into(), Ty::str());
        ws_conn_fields.insert("subprotocol".into(), Ty::str());
        e.types.insert("WsConn".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Alias(Ty::Record(ws_conn_fields)),
        });

        // WsMessage = WsText(Str) | WsBinary(List[Int]) | WsPing | WsClose
        let mut ws_msg_variants = IndexMap::new();
        ws_msg_variants.insert("WsText".into(), Some(Ty::str()));
        ws_msg_variants.insert("WsBinary".into(), Some(Ty::List(Box::new(Ty::int()))));
        ws_msg_variants.insert("WsPing".into(), None);
        ws_msg_variants.insert("WsClose".into(), None);
        e.types.insert("WsMessage".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(ws_msg_variants),
        });
        for ctor in &["WsText", "WsBinary", "WsPing", "WsClose"] {
            e.ctor_to_type.insert((*ctor).into(), "WsMessage".into());
        }

        // WsAction = WsSend(Str) | WsSendBinary(List[Int]) | WsNoOp
        // Handlers return this to tell the runtime what to send back.
        // Connection close is handled automatically when the runtime receives
        // an incoming WsClose frame; handlers do not need to emit a close action.
        let mut ws_act_variants = IndexMap::new();
        ws_act_variants.insert("WsSend".into(), Some(Ty::str()));
        ws_act_variants.insert("WsSendBinary".into(), Some(Ty::List(Box::new(Ty::int()))));
        ws_act_variants.insert("WsNoOp".into(), None);
        e.types.insert("WsAction".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(ws_act_variants),
        });
        for ctor in &["WsSend", "WsSendBinary", "WsNoOp"] {
            e.ctor_to_type.insert((*ctor).into(), "WsAction".into());
        }

        // ConcError = AlreadyRegistered(Str) | NotRegistered(Str)
        // Returned by `conc.register` / `conc.unregister` (#444). A
        // third `TypeMismatch` variant is reserved for when the
        // SigId-tagged registry lands — see `conc_registry.rs` in
        // lex-bytecode for the deferred-design note.
        let mut ce_variants = IndexMap::new();
        ce_variants.insert("AlreadyRegistered".into(), Some(Ty::str()));
        ce_variants.insert("NotRegistered".into(), Some(Ty::str()));
        e.types.insert("ConcError".into(), TypeDef {
            params: vec![],
            kind: TypeDefKind::Union(ce_variants),
        });
        for ctor in &["AlreadyRegistered", "NotRegistered"] {
            e.ctor_to_type.insert((*ctor).into(), "ConcError".into());
        }

        // ConnRedis: opaque handle for std.redis connections (#533).
        // Backed at runtime by an Int into a process-wide registry,
        // same pattern as Db (std.sql) and Kv (std.kv).
        e.types.insert("ConnRedis".into(), TypeDef { params: vec![], kind: TypeDefKind::Opaque });

        e
    }

    pub fn add_user_type(&mut self, name: &str, decl: lex_ast::TypeDecl) -> Result<(), String> {
        match &decl.definition {
            lex_ast::TypeExpr::Union { variants } => {
                let mut vmap = IndexMap::new();
                for v in variants {
                    let payload = v.payload.as_ref().map(|p| ty_from_canon(p, &decl.params));
                    vmap.insert(v.name.clone(), payload);
                    self.ctor_to_type.insert(v.name.clone(), name.to_string());
                }
                self.types.insert(name.to_string(), TypeDef {
                    params: decl.params.clone(),
                    kind: TypeDefKind::Union(vmap),
                });
            }
            other => {
                let ty = ty_from_canon_env(other, &decl.params, self);
                self.types.insert(name.to_string(), TypeDef {
                    params: decl.params.clone(),
                    kind: TypeDefKind::Alias(ty),
                });
            }
        }
        Ok(())
    }
}

/// Convert canonical TypeExpr to internal Ty, treating type params as
/// fresh-numbered Vars (0..n in declaration order). When instantiating, we
/// substitute these out.
pub fn ty_from_canon(t: &lex_ast::TypeExpr, params: &[String]) -> Ty {
    match t {
        lex_ast::TypeExpr::Named { name, args } => {
            // type param?
            if let Some(idx) = params.iter().position(|p| p == name) {
                if !args.is_empty() {
                    // Type params don't take args.
                    return Ty::Con(name.clone(), args.iter().map(|a| ty_from_canon(a, params)).collect());
                }
                return Ty::Var(idx as u32);
            }
            // Primitives.
            match name.as_str() {
                "Int" => return Ty::int(),
                "Float" => return Ty::float(),
                "Bool" => return Ty::bool(),
                "Str" => return Ty::str(),
                "Bytes" => return Ty::bytes(),
                "Unit" | "Nil" => return Ty::Unit,
                "Never" => return Ty::Never,
                "List" if args.len() == 1 => return Ty::List(Box::new(ty_from_canon(&args[0], params))),
                // `Tuple[T0, T1, ...]` is the constructor surface for
                // tuples; canonicalize to the structural Ty::Tuple so
                // it unifies with `(T0, T1)` literal-tuple syntax and
                // with std.tuple's signatures.
                "Tuple" => return Ty::Tuple(args.iter().map(|a| ty_from_canon(a, params)).collect()),
                _ => {}
            }
            Ty::Con(name.clone(), args.iter().map(|a| ty_from_canon(a, params)).collect())
        }
        lex_ast::TypeExpr::Record { fields } => {
            let mut m = IndexMap::new();
            for f in fields { m.insert(f.name.clone(), ty_from_canon(&f.ty, params)); }
            Ty::Record(m)
        }
        lex_ast::TypeExpr::Tuple { items } => Ty::Tuple(items.iter().map(|t| ty_from_canon(t, params)).collect()),
        lex_ast::TypeExpr::Function { params: ps, effects, effect_row_var, ret } => {
            // Plumb effect args (#207).
            let effs = EffectSet {
                concrete: {
                    let mut s = std::collections::BTreeSet::new();
                    for e in effects {
                        let arg = e.arg.as_ref().map(|a| match a {
                            lex_ast::EffectArg::Str { value } => crate::types::EffectArg::Str(value.clone()),
                            lex_ast::EffectArg::Int { value } => crate::types::EffectArg::Int(*value),
                            lex_ast::EffectArg::Ident { value } => crate::types::EffectArg::Ident(value.clone()),
                        });
                        s.insert(crate::types::EffectKind { name: e.name.clone(), arg });
                    }
                    s
                },
                // Open-row tail: `[io | E]` where `E` is one of the enclosing
                // type/fn's `params`. Resolve it to that param's index — the
                // same id space as `Ty::Var(idx)`, but read back through the
                // separate effect-substitution map at instantiation, so a
                // type param and an effect-row param never collide.
                var: effect_row_var
                    .as_ref()
                    .and_then(|name| params.iter().position(|p| p == name))
                    .map(|i| i as u32),
            };
            Ty::Function {
                params: ps.iter().map(|t| ty_from_canon(t, params)).collect(),
                effects: effs,
                ret: Box::new(ty_from_canon(ret, params)),
            }
        }
        lex_ast::TypeExpr::Union { .. } => {
            // Unions on the RHS of type-decls; not in arbitrary positions.
            Ty::Unit
        }
        lex_ast::TypeExpr::Refined { base, .. } => {
            // #209 slice 1: refinement types unify structurally as
            // their base type. The predicate is parsed and stored in
            // the AST (so `lex-vcs` content-addressing picks up
            // refinement edits), but static discharge and runtime
            // residual checks land in slices 2 and 3 of #209. The
            // unification behavior here means a function declaring
            // `Int{x | x > 0}` interoperates with plain `Int` callers
            // — the predicate is informational until discharge is
            // wired up.
            ty_from_canon(base, params)
        }
        lex_ast::TypeExpr::RecordWithSpreads { .. } => {
            // Caller should use ty_from_canon_env for spread resolution.
            Ty::Unit
        }
    }
}

/// Like `ty_from_canon` but resolves `RecordWithSpreads` by looking up base
/// type names in `env`. Called from `add_user_type` and `function_scheme` so
/// that `{ ...Post, extra :: Int }` expands to a flat `Ty::Record`.
pub fn ty_from_canon_env(t: &lex_ast::TypeExpr, params: &[String], env: &TypeEnv) -> Ty {
    match t {
        lex_ast::TypeExpr::RecordWithSpreads { spreads, fields } => {
            let mut m = IndexMap::new();
            for spread_name in spreads {
                if let Some(td) = env.types.get(spread_name.as_str()) {
                    if let TypeDefKind::Alias(Ty::Record(spread_fields)) = &td.kind {
                        for (k, v) in spread_fields {
                            m.insert(k.clone(), v.clone());
                        }
                    }
                }
            }
            for f in fields {
                m.insert(f.name.clone(), ty_from_canon_env(&f.ty, params, env));
            }
            Ty::Record(m)
        }
        other => ty_from_canon(other, params),
    }
}