pleme-doc-gen 0.1.41

//! Substrate self-introspection — typed capability surface per ecosystem.
//!
//! Per the operator direction 'standing on solid abstractions' +
//! the 9-layer compounding pattern's Layer 8 (directive): the
//! substrate now SELF-DESCRIBES which typed orchestration surfaces
//! it implements for each ecosystem.
//!
//! Operators run `pleme-doc-gen ecosystems` to see the matrix of
//! ecosystem × capability cells; agents read it as a JSON dispatch
//! table to know which CLI subcommands are safe to invoke against
//! which ecosystems.
//!
//! Five capability surfaces are cross-referenced:
//!   - forge_dispatch:    caixa::render handles the ecosystem keyword
//!   - reverse_extractor: reverse::enrich_from_manifest has a per-eco extractor
//!   - validator:         validator::validator_for has a per-eco Validator
//!   - green_ci_starter:  green_ci::starter_for has a per-eco GreenCiStarter
//!   - url_discover:      discover::detect_github_url recognises files for this eco

use std::collections::BTreeMap;

/// Typed view of one ecosystem's capabilities across the 5 substrate
/// orchestration surfaces. Field semantics:
///   - true   = the substrate has a real per-ecosystem impl
///   - false  = the substrate falls back to a default / no-op / unknown
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct EcosystemCapabilities {
    pub keyword: String,
    pub forge_dispatch: bool,
    pub reverse_extractor: bool,
    pub validator: bool,
    pub green_ci_starter: bool,
    pub url_discover: bool,
    /// Per ★★ CATALOG REFLECTION — the substrate self-describes which
    /// manifest fields it preserves on round-trip for this ecosystem.
    /// Sourced from fidelity::compared_fields_for().
    pub preserved_fields: Vec<String>,
}

impl EcosystemCapabilities {
    /// True iff EVERY substrate surface has a real impl for this eco.
    /// Operators can mass-publish against fully-capable ecosystems
    /// with maximum confidence.
    pub fn fully_capable(&self) -> bool {
        self.forge_dispatch
            && self.reverse_extractor
            && self.validator
            && self.green_ci_starter
            && self.url_discover
    }
}

/// Canonical list of every ecosystem the substrate knows about. The
/// list is the union of what the various dispatch tables recognise —
/// this module's job is to cross-reference + report per-eco capability.
pub const ALL_ECOSYSTEMS: &[&str] = &[
    // Rust
    "rust-single-crate", "rust-workspace",
    // Node / JS / TS
    "npm", "js-pnpm", "js-deno",
    // Python family
    "python", "python-pdm", "python-pipenv", "python-conda",
    // Helm / GH Action
    "helm", "github-action",
    // Compiled
    "go", "zig", "fortran-fpm",
    // JVM
    "java-maven", "java-gradle-kts", "scala-sbt", "clojure-deps",
    // .NET / Swift / Functional
    "dotnet-csproj", "swift-spm", "ocaml-dune",
    "haskell-cabal", "gleam", "racket-info",
    // BEAM / Ruby / Lua / Nim
    "elixir-mix", "ruby-gem", "lua-rockspec", "nim-nimble",
    // Polyglot / data
    "crystal", "dart", "composer", "julia", "r-description",
    "ada-alire",
    // C / C++
    "cpp-conan", "cpp-vcpkg", "cpp-meson", "cpp-cmake",
    // Nix-flake — consume-side typed ecosystem (detect + reverse +
    // minimal render). Forge / validator / green-ci / url-discover
    // not yet wired — substrate self-describes partial capability.
    "nix-flake",
    // Tatara-lisp library — typed pleme-io-native ecosystem for the
    // shared .tlisp helper repos like pleme-io/actions/_tlisp-stdlib.
    "tlisp-library",
];

/// Query the substrate's typed surface for one ecosystem. Returns
/// the EcosystemCapabilities cross-referenced across all 5 dispatchers.
pub fn query(keyword: &str) -> EcosystemCapabilities {
    EcosystemCapabilities {
        keyword: keyword.to_string(),
        forge_dispatch: has_forge_dispatch(keyword),
        reverse_extractor: has_reverse_extractor(keyword),
        validator: has_validator(keyword),
        green_ci_starter: has_green_ci_starter(keyword),
        url_discover: has_url_discover(keyword),
        preserved_fields: crate::fidelity::compared_fields_for(keyword)
            .iter().map(|s| (*s).to_string()).collect(),
    }
}

/// Query every supported ecosystem. Returns capabilities in
/// ALL_ECOSYSTEMS declaration order.
pub fn query_all() -> Vec<EcosystemCapabilities> {
    ALL_ECOSYSTEMS.iter().map(|e| query(e)).collect()
}

// ─── Per-surface introspection helpers ─────────────────────────
// These mirror the dispatch tables in caixa.rs / reverse.rs /
// validator.rs / green_ci.rs / discover.rs. They MUST stay in sync
// with those tables — the ecosystems_match_canonical_keywords test
// in this module fails when a new ecosystem ships without a matching
// entry in ALL_ECOSYSTEMS.

fn has_forge_dispatch(keyword: &str) -> bool {
    matches!(keyword,
        "rust-single-crate" | "rust-workspace" | "npm" | "python" | "helm" |
        "github-action" | "go" | "crystal" | "dart" | "composer" | "julia" |
        "java-maven" | "dotnet-csproj" | "ocaml-dune" | "java-gradle-kts" |
        "swift-spm" | "elixir-mix" | "ruby-gem" | "zig" | "nim-nimble" |
        "scala-sbt" | "clojure-deps" | "r-description" | "lua-rockspec" |
        "cpp-conan" | "python-conda" | "python-pipenv" | "python-pdm" |
        "js-deno" | "js-pnpm" | "cpp-vcpkg" | "cpp-meson" | "cpp-cmake" |
        "fortran-fpm" | "gleam" | "ada-alire" | "haskell-cabal" | "racket-info"
    )
}

fn has_reverse_extractor(keyword: &str) -> bool {
    matches!(keyword,
        "rust-single-crate" | "rust-workspace" | "npm" | "js-pnpm" |
        "python" | "python-pdm" | "helm" | "go" | "ruby-gem" |
        "ocaml-dune" | "dotnet-csproj" | "swift-spm" | "elixir-mix" |
        "scala-sbt" | "clojure-deps" |
        "crystal" | "dart" | "composer" | "julia" |
        "zig" | "fortran-fpm" | "gleam" | "racket-info" |
        "java-maven" | "js-deno" | "cpp-vcpkg" | "python-conda" | "ada-alire" |
        "java-gradle-kts" | "cpp-cmake" | "cpp-meson" | "cpp-conan" |
        "haskell-cabal" | "nim-nimble" | "lua-rockspec" | "r-description" |
        "python-pipenv" | "github-action" | "nix-flake" | "tlisp-library"
    )
}

fn has_validator(keyword: &str) -> bool {
    // SubstrateOnlyValidator is a no-op floor; per-ecosystem typed
    // validators have actual manifest assertions.
    matches!(keyword,
        "rust-single-crate" | "rust-workspace" |
        "npm" | "js-pnpm" |
        "python" | "python-pdm" |
        "helm" |
        "go" |
        "ruby-gem" |
        "ocaml-dune" |
        "dotnet-csproj" |
        "swift-spm" |
        "elixir-mix" |
        "scala-sbt" |
        "clojure-deps" |
        "crystal" |
        "dart" |
        "composer" |
        "julia" |
        "zig" |
        "fortran-fpm" |
        "gleam" |
        "racket-info" |
        "java-maven" |
        "js-deno" |
        "cpp-vcpkg" |
        "python-conda" |
        "ada-alire" |
        "java-gradle-kts" |
        "cpp-cmake" |
        "cpp-meson" |
        "cpp-conan" |
        "haskell-cabal" |
        "nim-nimble" |
        "lua-rockspec" |
        "r-description" |
        "python-pipenv" |
        "github-action"
    )
}

fn has_green_ci_starter(keyword: &str) -> bool {
    matches!(keyword,
        "rust-single-crate" | "rust-workspace" |
        "go" |
        "npm" | "js-pnpm" |
        "python" | "python-pdm" |
        "helm" |
        "ruby-gem" |
        "ocaml-dune" |
        "dotnet-csproj" |
        "swift-spm" |
        "elixir-mix" |
        "scala-sbt" |
        "clojure-deps" |
        "crystal" |
        "dart" |
        "composer" |
        "julia" |
        "zig" |
        "fortran-fpm" |
        "gleam" |
        "racket-info" |
        "java-maven" |
        "js-deno" |
        "cpp-vcpkg" |
        "python-conda" |
        "ada-alire" |
        "java-gradle-kts" |
        "cpp-cmake" |
        "cpp-meson" |
        "cpp-conan" |
        "haskell-cabal" |
        "nim-nimble" |
        "lua-rockspec" |
        "r-description" |
        "python-pipenv" |
        "github-action"
    )
}

fn has_url_discover(keyword: &str) -> bool {
    // discover::detect_github_url's RULES table — these are the
    // ecosystems URL-mode discovery infers from a GH repo's root
    // file listing without cloning.
    matches!(keyword,
        "rust-single-crate" | "rust-workspace" | "js-pnpm" | "js-deno" |
        "npm" | "python-pipenv" | "python-conda" | "python-pdm" | "python" |
        "helm" | "github-action" | "go" | "zig" | "fortran-fpm" |
        "java-gradle-kts" | "java-maven" | "scala-sbt" | "clojure-deps" |
        "dotnet-csproj" | "swift-spm" | "ocaml-dune" | "haskell-cabal" |
        "gleam" | "racket-info" | "elixir-mix" | "ruby-gem" | "lua-rockspec" |
        "nim-nimble" | "crystal" | "dart" | "composer" | "julia" |
        "r-description" | "ada-alire" | "cpp-conan" | "cpp-vcpkg" |
        "cpp-meson" | "cpp-cmake"
    )
}

/// Render capability matrix as a typed JSON document via json_ast
/// (no format!() of JSON syntax — dogfoods the prime directive at
/// the introspection emit layer).
pub fn to_json(caps: &[EcosystemCapabilities]) -> String {
    use crate::json_ast::Value;
    let items: Vec<Value> = caps.iter().map(|c| {
        let mut o = Value::obj();
        o.insert("keyword", Value::s(&c.keyword));
        o.insert("forge", Value::b(c.forge_dispatch));
        o.insert("reverse", Value::b(c.reverse_extractor));
        o.insert("validate", Value::b(c.validator));
        o.insert("green-ci", Value::b(c.green_ci_starter));
        o.insert("url-discover", Value::b(c.url_discover));
        o.insert("fully-capable", Value::b(c.fully_capable()));
        // Per ★★ CATALOG REFLECTION — expose the substrate's preserved-
        // field surface as typed data so operators can query coverage.
        o.insert("preserved-fields",
            Value::Array(c.preserved_fields.iter().map(Value::s).collect()));
        o.insert("preserved-field-count", Value::i(c.preserved_fields.len() as i64));
        o
    }).collect();
    crate::json_ast::render(&Value::Array(items))
}

/// Compact text-table for terminal viewing. Columns:
///   ecosystem  forge reverse validate green-ci url-disc  total
pub fn to_text_table(caps: &[EcosystemCapabilities]) -> String {
    let mut out = String::new();
    out.push_str("ecosystem               forge rev val gci url   fully    fields\n");
    out.push_str("─────────────────────── ───── ─── ─── ─── ─── ──────── ──────\n");
    for c in caps {
        let line = format!(
            "{:23} {:5} {:3} {:3} {:3} {:3}   {:8} {}\n",
            truncate(&c.keyword, 23),
            yn(c.forge_dispatch),
            yn(c.reverse_extractor),
            yn(c.validator),
            yn(c.green_ci_starter),
            yn(c.url_discover),
            if c.fully_capable() { "✓ full" } else { "" },
            c.preserved_fields.len(),
        );
        out.push_str(&line);
    }
    // Summary footer
    let totals = capability_totals(caps);
    out.push_str(&format!(
        "\n{} ecosystems total · {} fully-capable · forge={} reverse={} \
         validate={} green-ci={} url-disc={}\n",
        caps.len(),
        caps.iter().filter(|c| c.fully_capable()).count(),
        totals.get("forge").copied().unwrap_or(0),
        totals.get("reverse").copied().unwrap_or(0),
        totals.get("validate").copied().unwrap_or(0),
        totals.get("green-ci").copied().unwrap_or(0),
        totals.get("url-disc").copied().unwrap_or(0),
    ));
    out
}

fn truncate(s: &str, max: usize) -> String {
    if s.chars().count() <= max { return s.to_string(); }
    s.chars().take(max).collect()
}

fn yn(b: bool) -> &'static str { if b { "yes" } else { "no" } }

fn capability_totals(caps: &[EcosystemCapabilities]) -> BTreeMap<&'static str, usize> {
    let mut m = BTreeMap::new();
    m.insert("forge",    caps.iter().filter(|c| c.forge_dispatch).count());
    m.insert("reverse",  caps.iter().filter(|c| c.reverse_extractor).count());
    m.insert("validate", caps.iter().filter(|c| c.validator).count());
    m.insert("green-ci", caps.iter().filter(|c| c.green_ci_starter).count());
    m.insert("url-disc", caps.iter().filter(|c| c.url_discover).count());
    m
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn query_known_ecosystem_returns_typed_capabilities() {
        let c = query("rust-single-crate");
        assert!(c.forge_dispatch);
        assert!(c.reverse_extractor);
        assert!(c.validator);
        assert!(c.green_ci_starter);
        assert!(c.url_discover);
        assert!(c.fully_capable(), "rust-single-crate should be fully-capable");
    }

    #[test]
    fn query_unknown_ecosystem_returns_all_false() {
        let c = query("some-future-ecosystem-not-yet-supported");
        assert!(!c.forge_dispatch);
        assert!(!c.reverse_extractor);
        assert!(!c.validator);
        assert!(!c.green_ci_starter);
        assert!(!c.url_discover);
        assert!(!c.fully_capable());
    }

    #[test]
    fn query_all_returns_full_ecosystem_set() {
        let all = query_all();
        assert!(all.len() >= 36, "expected ≥ 36 ecosystems, got {}", all.len());
        // First entry should be the first one in ALL_ECOSYSTEMS.
        assert_eq!(all[0].keyword, ALL_ECOSYSTEMS[0]);
    }

    #[test]
    fn fully_capable_set_includes_the_4_green_ci_ecosystems() {
        let all = query_all();
        let fully: Vec<&str> = all.iter()
            .filter(|c| c.fully_capable())
            .map(|c| c.keyword.as_str())
            .collect();
        // These are the ecosystems with green-CI + validator + reverse
        // + forge + url-discover all present.
        for required in &["rust-single-crate", "rust-workspace", "npm",
                          "js-pnpm", "python", "python-pdm"] {
            assert!(fully.contains(required),
                "expected {required} to be fully-capable; got fully={fully:?}");
        }
    }

    #[test]
    fn to_text_table_renders_with_headers_and_totals() {
        let caps = vec![query("rust-single-crate"), query("go")];
        let t = to_text_table(&caps);
        assert!(t.contains("ecosystem"));
        assert!(t.contains("rust-single-crate"));
        assert!(t.contains("go"));
        assert!(t.contains("2 ecosystems total"));
        assert!(t.contains("fully-capable"));
    }

    #[test]
    fn to_json_emits_typed_capability_records() {
        let caps = vec![query("rust-single-crate")];
        let j = to_json(&caps);
        assert!(j.contains("\"keyword\": \"rust-single-crate\""));
        assert!(j.contains("\"forge\": true"));
        assert!(j.contains("\"fully-capable\": true"));
    }

    #[test]
    fn all_ecosystems_are_recognised_by_at_least_one_surface() {
        // The substrate-completeness invariant: every keyword in
        // ALL_ECOSYSTEMS MUST be recognised by AT LEAST ONE surface
        // (otherwise the keyword is dead — neither forgeable nor
        // discoverable nor validatable). This catches typos in
        // ALL_ECOSYSTEMS + drift between this module and the dispatch
        // tables.
        for eco in ALL_ECOSYSTEMS {
            let c = query(eco);
            assert!(
                c.forge_dispatch || c.reverse_extractor || c.validator ||
                c.green_ci_starter || c.url_discover,
                "{eco} is in ALL_ECOSYSTEMS but no substrate surface recognises it"
            );
        }
    }
}