//! `ail` — CLI für AILang.
//!
//! Subcommands sind so geschnitten, dass jedes einzelne Tool dem LLM einen
//! kleinen, fokussierten Kontext liefert (manifest = Übersicht; describe =
//! Detail; emit-ir = exakte Maschinensicht; build = Pipeline-Validierung).

use anyhow::{Context, Result};
use clap::{Parser, Subcommand};
use std::path::{Path, PathBuf};

#[derive(Parser)]
#[command(name = "ail", version, about = "AILang toolchain")]
struct Cli {
    #[command(subcommand)]
    cmd: Cmd,
}

#[derive(Subcommand)]
enum Cmd {
    /// Lädt ein Modul und gibt eine kompakte Symboltabelle aus.
    Manifest {
        path: PathBuf,
        #[arg(long)]
        json: bool,
    },
    /// Gibt das Modul in Textform aus (Pretty-Printer).
    Render { path: PathBuf },
    /// Gibt eine einzelne Definition als JSON oder Pretty-Text aus.
    Describe {
        path: PathBuf,
        name: String,
        #[arg(long)]
        json: bool,
    },
    /// Listet, welche Symbole jede Definition aufruft (statisch).
    Deps {
        path: PathBuf,
        /// Nur für ein Symbol; ohne Argument: für alle.
        #[arg(long)]
        of: Option<String>,
        #[arg(long)]
        json: bool,
    },
    /// Typprüft ein Modul.
    Check {
        path: PathBuf,
        /// Strukturierte Diagnostics als JSON-Array auf stdout.
        /// Exit-Code 1, wenn mindestens ein Error gemeldet wird.
        #[arg(long)]
        json: bool,
    },
    /// Schreibt LLVM IR (.ll) für das Modul.
    EmitIr {
        path: PathBuf,
        #[arg(short, long)]
        out: Option<PathBuf>,
    },
    /// Komplette Pipeline: check + emit-ir + clang -> Binary.
    Build {
        path: PathBuf,
        #[arg(short, long)]
        out: Option<PathBuf>,
        /// Optimierung (z. B. `-O2`); default `-O0` für Debugbarkeit.
        #[arg(long, default_value = "-O0")]
        opt: String,
    },
    /// Listet eingebaute Operationen mit ihren Signaturen.
    Builtins {
        #[arg(long)]
        json: bool,
    },
    /// Semantischer Modul-Diff per Def-Hash.
    ///
    /// Vergleicht zwei Module rein strukturell auf Top-Level-Defs:
    /// pro Name werden die Hashes der canonical Bytes verglichen. Das
    /// Diff funktioniert auch, wenn ein Modul gerade nicht typecheckt —
    /// nur das Schema und die JSON-Form müssen ladbar sein.
    ///
    /// Exit-Code: 0 wenn keine Änderungen (außer `unchanged`), sonst 1.
    Diff {
        a: PathBuf,
        b: PathBuf,
        #[arg(long)]
        json: bool,
    },
    /// Lädt einen Workspace (Eintrittsmodul + transitive Imports) und
    /// listet alle erreichbaren Module mit Hash und Def-Anzahl.
    ///
    /// Iter 5a: nur das Listing. Cross-Module-Typcheck/Codegen folgt in
    /// 5b/5c; bestehende Subkommandos arbeiten weiter pro Einzelmodul.
    Workspace {
        entry: PathBuf,
        #[arg(long)]
        json: bool,
    },
}

fn main() -> Result<()> {
    let cli = Cli::parse();
    match cli.cmd {
        Cmd::Manifest { path, json } => {
            let m = ailang_core::load_module(&path)?;
            if json {
                let entries: Vec<_> = m
                    .defs
                    .iter()
                    .map(|d| {
                        let h = ailang_core::def_hash(d);
                        let (kind, ty, effects) = match d {
                            ailang_core::Def::Fn(f) => {
                                let effects = match &f.ty {
                                    ailang_core::Type::Fn { effects, .. } => effects.clone(),
                                    _ => vec![],
                                };
                                (
                                    "fn",
                                    ailang_core::pretty::type_to_string(&f.ty),
                                    effects,
                                )
                            }
                            ailang_core::Def::Const(c) => (
                                "const",
                                ailang_core::pretty::type_to_string(&c.ty),
                                vec![],
                            ),
                            ailang_core::Def::Type(t) => {
                                let s = t
                                    .ctors
                                    .iter()
                                    .map(|c| {
                                        if c.fields.is_empty() {
                                            c.name.clone()
                                        } else {
                                            format!(
                                                "{}({})",
                                                c.name,
                                                c.fields
                                                    .iter()
                                                    .map(ailang_core::pretty::type_to_string)
                                                    .collect::<Vec<_>>()
                                                    .join(", ")
                                            )
                                        }
                                    })
                                    .collect::<Vec<_>>()
                                    .join(" | ");
                                ("type", s, vec![])
                            }
                        };
                        serde_json::json!({
                            "name": d.name(),
                            "kind": kind,
                            "type": ty,
                            "effects": effects,
                            "hash": h,
                        })
                    })
                    .collect();
                let out = serde_json::json!({
                    "module": m.name,
                    "schema": m.schema,
                    "symbols": entries,
                });
                println!("{}", serde_json::to_string_pretty(&out)?);
            } else {
                print!("{}", ailang_core::pretty::manifest(&m));
            }
        }
        Cmd::Render { path } => {
            let m = ailang_core::load_module(&path)?;
            print!("{}", ailang_core::pretty::module(&m));
        }
        Cmd::Describe { path, name, json } => {
            let m = ailang_core::load_module(&path)?;
            let def = m
                .defs
                .iter()
                .find(|d| d.name() == name)
                .with_context(|| format!("no def `{name}` in module `{}`", m.name))?;
            if json {
                let s = serde_json::to_string_pretty(def)?;
                println!("{s}");
            } else {
                // Pretty-form: render module mit nur dieser Def.
                let one = ailang_core::Module {
                    schema: m.schema.clone(),
                    name: m.name.clone(),
                    imports: vec![],
                    defs: vec![def.clone()],
                };
                let h = ailang_core::def_hash(def);
                println!("hash: {h}");
                print!("{}", ailang_core::pretty::module(&one));
            }
        }
        Cmd::Check { path, json } => {
            // Iter 5b: `ail check` lädt jetzt **immer** über
            // `load_workspace` und prüft cross-module. Für Module ohne
            // Imports verhält sich der Workspace-Loader äquivalent zu
            // `load_module` plus Hash-Konsistenz-Check des Eintrittsfiles
            // — damit ist der Pfad einheitlich.
            if json {
                // JSON-Modus: stdout enthält ausschließlich das Diagnostics-
                // Array. Workspace-Lade-Fehler werden als strukturierte
                // Diagnostics emittiert (Codes `module-not-found`,
                // `module-cycle`, `module-name-mismatch`, `schema-mismatch`).
                // Echte I/O-Fehler des Eintrittsfiles bleiben fatal.
                let diags = match ailang_core::load_workspace(&path) {
                    Ok(ws) => ailang_check::check_workspace(&ws),
                    Err(e) => match workspace_error_to_diagnostic(&e) {
                        Some(d) => vec![d],
                        None => return Err(anyhow::anyhow!(e)),
                    },
                };
                println!("{}", serde_json::to_string(&diags)?);
                if diags
                    .iter()
                    .any(|d| matches!(d.severity, ailang_check::Severity::Error))
                {
                    std::process::exit(1);
                }
            } else {
                let ws = ailang_core::load_workspace(&path)?;
                let diags = ailang_check::check_workspace(&ws);
                if !diags.is_empty() {
                    for d in &diags {
                        eprintln!(
                            "{}: [{}] {}{}",
                            match d.severity {
                                ailang_check::Severity::Error => "error",
                                ailang_check::Severity::Warning => "warning",
                            },
                            d.code,
                            d.def
                                .as_ref()
                                .map(|n| format!("{n}: "))
                                .unwrap_or_default(),
                            d.message,
                        );
                    }
                    std::process::exit(1);
                }
                let total: usize = ws.modules.values().map(|m| m.defs.len()).sum();
                println!(
                    "ok ({} symbols across {} modules)",
                    total,
                    ws.modules.len()
                );
            }
        }
        Cmd::EmitIr { path, out } => {
            // Iter 5c: Workspace-Lowering. Bei Single-Modul-Programmen ist
            // der Workspace effektiv ein Trivial-Workspace mit einem Modul.
            let ws = ailang_core::load_workspace(&path)?;
            let diags = ailang_check::check_workspace(&ws);
            if !diags.is_empty() {
                for d in &diags {
                    eprintln!(
                        "{}: [{}] {}{}",
                        match d.severity {
                            ailang_check::Severity::Error => "error",
                            ailang_check::Severity::Warning => "warning",
                        },
                        d.code,
                        d.def
                            .as_ref()
                            .map(|n| format!("{n}: "))
                            .unwrap_or_default(),
                        d.message,
                    );
                }
                std::process::exit(1);
            }
            let ir = ailang_codegen::lower_workspace(&ws)?;
            match out {
                Some(p) => {
                    std::fs::write(&p, ir)?;
                    eprintln!("wrote {}", p.display());
                }
                None => print!("{ir}"),
            }
        }
        Cmd::Build { path, out, opt } => {
            // Iter 5c: gleiche Pipeline wie `emit-ir`, aber clang ruft am Ende.
            let ws = ailang_core::load_workspace(&path)?;
            let diags = ailang_check::check_workspace(&ws);
            if !diags.is_empty() {
                for d in &diags {
                    eprintln!(
                        "{}: [{}] {}{}",
                        match d.severity {
                            ailang_check::Severity::Error => "error",
                            ailang_check::Severity::Warning => "warning",
                        },
                        d.code,
                        d.def
                            .as_ref()
                            .map(|n| format!("{n}: "))
                            .unwrap_or_default(),
                        d.message,
                    );
                }
                std::process::exit(1);
            }
            let ir = ailang_codegen::lower_workspace(&ws)?;
            let tmpdir = std::env::temp_dir().join(format!("ailang-{}", std::process::id()));
            std::fs::create_dir_all(&tmpdir)?;
            let ll_path = tmpdir.join(format!("{}.ll", ws.entry));
            std::fs::write(&ll_path, &ir)?;
            let out_bin = out.unwrap_or_else(|| {
                Path::new(".").join(&ws.entry).with_extension("")
            });
            let status = std::process::Command::new("clang")
                .arg(&opt)
                .arg("-o")
                .arg(&out_bin)
                .arg(&ll_path)
                .status()
                .context("running clang")?;
            if !status.success() {
                anyhow::bail!(
                    "clang failed (status {}); ll at {}",
                    status,
                    ll_path.display()
                );
            }
            eprintln!("built {}", out_bin.display());
        }
        Cmd::Builtins { json } => {
            let list = ailang_check::builtins::list();
            if json {
                let arr: Vec<_> = list
                    .iter()
                    .map(|(n, s)| serde_json::json!({ "name": n, "sig": s }))
                    .collect();
                println!("{}", serde_json::to_string_pretty(&arr)?);
            } else {
                for (n, sig) in list {
                    println!("{n:<16}  {sig}");
                }
            }
        }
        Cmd::Diff { a, b, json } => {
            let ma = ailang_core::load_module(&a)?;
            let mb = ailang_core::load_module(&b)?;
            let report = build_diff(&ma, &mb);
            if json {
                let v = diff_report_to_json(&report);
                println!("{}", serde_json::to_string_pretty(&v)?);
            } else {
                print!("{}", render_diff_text(&report));
            }
            if !report.is_identical() {
                std::process::exit(1);
            }
        }
        Cmd::Workspace { entry, json } => {
            let ws = ailang_core::load_workspace(&entry)?;
            // Alphabetisch über Modul-Namen iterieren (BTreeMap-Order ist
            // bereits sortiert; explizit absichern).
            let mut entries: Vec<(String, String, usize)> = ws
                .modules
                .iter()
                .map(|(name, m)| {
                    (
                        name.clone(),
                        ailang_core::module_hash(m),
                        m.defs.len(),
                    )
                })
                .collect();
            entries.sort_by(|a, b| a.0.cmp(&b.0));

            if json {
                let arr: Vec<_> = entries
                    .iter()
                    .map(|(name, hash, defs)| {
                        serde_json::json!({
                            "name": name,
                            "hash": hash,
                            "defs": defs,
                        })
                    })
                    .collect();
                let out = serde_json::json!({
                    "entry": ws.entry,
                    "modules": arr,
                });
                println!("{}", serde_json::to_string_pretty(&out)?);
            } else {
                // Spaltenbreite an längstem Modulnamen ausrichten. Erste Zeile
                // markiert das Eintrittsmodul mit `*`.
                let name_width = entries
                    .iter()
                    .map(|(n, _, _)| n.len())
                    .max()
                    .unwrap_or(0)
                    .max(6);
                println!("entry: {}", ws.entry);
                for (name, hash, defs) in &entries {
                    let marker = if *name == ws.entry { "*" } else { " " };
                    println!(
                        "{marker} {:<width$}  {}  {:>3} defs",
                        name,
                        hash,
                        defs,
                        width = name_width,
                    );
                }
            }
        }
        Cmd::Deps { path, of, json } => {
            let m = ailang_core::load_module(&path)?;
            let mut entries = Vec::new();
            for d in &m.defs {
                if let Some(filter) = &of {
                    if d.name() != filter {
                        continue;
                    }
                }
                let mut refs: Vec<String> = collect_refs(d).into_iter().collect();
                refs.sort();
                entries.push((d.name().to_string(), refs));
            }
            if json {
                let arr: Vec<_> = entries
                    .iter()
                    .map(|(n, r)| serde_json::json!({ "name": n, "refs": r }))
                    .collect();
                println!("{}", serde_json::to_string_pretty(&arr)?);
            } else {
                for (n, refs) in entries {
                    if refs.is_empty() {
                        println!("{n:>20}  -");
                    } else {
                        println!("{n:>20}  -> {}", refs.join(", "));
                    }
                }
            }
        }
    }
    Ok(())
}

/// Wandelt einen `WorkspaceLoadError` in ein passendes Diagnostic für den
/// JSON-Modus von `ail check`. Reine I/O-Fehler haben kein Modul-Diagnostic-
/// Äquivalent (sie sind nicht der Pipeline-Sache eines Konsumenten); für
/// die liefern wir `None` und lassen den Aufrufer fatal scheitern.
fn workspace_error_to_diagnostic(
    e: &ailang_core::WorkspaceLoadError,
) -> Option<ailang_check::Diagnostic> {
    use ailang_core::WorkspaceLoadError as W;
    match e {
        W::Io { .. } => None,
        W::Schema { source, .. } => match source {
            ailang_core::Error::SchemaMismatch { expected, got } => Some(
                ailang_check::Diagnostic::error(
                    "schema-mismatch",
                    format!(
                        "schema mismatch: expected {expected:?}, got {got:?}"
                    ),
                )
                .with_ctx(serde_json::json!({
                    "expected": expected,
                    "actual": got,
                })),
            ),
            _ => None,
        },
        W::ModuleNotFound { name, expected_path } => Some(
            ailang_check::Diagnostic::error(
                "module-not-found",
                format!(
                    "module `{name}` not found (expected at {})",
                    expected_path.display()
                ),
            )
            .with_ctx(serde_json::json!({
                "module": name,
                "expected_path": expected_path.display().to_string(),
            })),
        ),
        W::ModuleNameMismatch {
            name_in_file,
            name_from_path,
        } => Some(
            ailang_check::Diagnostic::error(
                "module-name-mismatch",
                format!(
                    "module name mismatch: file says {name_in_file:?}, path implies {name_from_path:?}"
                ),
            )
            .with_ctx(serde_json::json!({
                "name_in_file": name_in_file,
                "name_from_path": name_from_path,
            })),
        ),
        W::Cycle { path } => Some(
            ailang_check::Diagnostic::error(
                "module-cycle",
                format!("import cycle: {}", path.join(" -> ")),
            )
            .with_ctx(serde_json::json!({
                "path": path,
            })),
        ),
        W::ModuleHashMismatch { name } => Some(
            ailang_check::Diagnostic::error(
                "module-hash-mismatch",
                format!("module `{name}` loaded twice with differing content"),
            )
            .with_ctx(serde_json::json!({
                "module": name,
            })),
        ),
    }
}

fn collect_refs(def: &ailang_core::Def) -> std::collections::BTreeSet<String> {
    let mut out = std::collections::BTreeSet::new();
    match def {
        ailang_core::Def::Fn(f) => walk_term(&f.body, &mut out),
        ailang_core::Def::Const(c) => walk_term(&c.value, &mut out),
        ailang_core::Def::Type(td) => {
            // Eine Typedef referenziert die Typen ihrer Felder.
            for c in &td.ctors {
                for ft in &c.fields {
                    if let ailang_core::Type::Con { name } = ft {
                        out.insert(format!("type:{name}"));
                    }
                }
            }
        }
    }
    out
}

fn walk_term(t: &ailang_core::Term, out: &mut std::collections::BTreeSet<String>) {
    use ailang_core::Term;
    match t {
        Term::Lit { .. } => {}
        Term::Var { name } => {
            out.insert(name.clone());
        }
        Term::App { callee, args } => {
            walk_term(callee, out);
            for a in args {
                walk_term(a, out);
            }
        }
        Term::Let { value, body, .. } => {
            walk_term(value, out);
            walk_term(body, out);
        }
        Term::If { cond, then, else_ } => {
            walk_term(cond, out);
            walk_term(then, out);
            walk_term(else_, out);
        }
        Term::Do { op, args } => {
            // Effekt-Ops als `effect:io/print_int` markieren, damit man sie
            // von normalen Funktionsaufrufen trennen kann.
            out.insert(format!("effect:{op}"));
            for a in args {
                walk_term(a, out);
            }
        }
        Term::Ctor { type_name, ctor, args } => {
            out.insert(format!("ctor:{type_name}/{ctor}"));
            for a in args {
                walk_term(a, out);
            }
        }
        Term::Match { scrutinee, arms } => {
            walk_term(scrutinee, out);
            for arm in arms {
                if let ailang_core::ast::Pattern::Ctor { ctor, .. } = &arm.pat {
                    out.insert(format!("ctor:{ctor}"));
                }
                walk_term(&arm.body, out);
            }
        }
    }
}

// --- ail diff -------------------------------------------------------------

/// Rein struktureller Modul-Diff. Top-Level-Defs werden per `name`
/// identifiziert und per BLAKE3-16-Hex der canonical Bytes verglichen.
struct DiffReport {
    module_a: String,
    module_b: String,
    added: Vec<DiffEntry>,
    removed: Vec<DiffEntry>,
    changed: Vec<ChangedEntry>,
    unchanged: Vec<DiffEntry>,
}

struct DiffEntry {
    name: String,
    hash: String,
    kind: &'static str,
}

struct ChangedEntry {
    name: String,
    hash_a: String,
    hash_b: String,
    kind_a: &'static str,
    kind_b: &'static str,
}

impl DiffReport {
    fn is_identical(&self) -> bool {
        self.added.is_empty() && self.removed.is_empty() && self.changed.is_empty()
    }
}

fn build_diff(a: &ailang_core::Module, b: &ailang_core::Module) -> DiffReport {
    use std::collections::BTreeMap;

    let map_a: BTreeMap<&str, &ailang_core::Def> =
        a.defs.iter().map(|d| (ailang_core::def_name(d), d)).collect();
    let map_b: BTreeMap<&str, &ailang_core::Def> =
        b.defs.iter().map(|d| (ailang_core::def_name(d), d)).collect();

    let mut added = Vec::new();
    let mut removed = Vec::new();
    let mut changed = Vec::new();
    let mut unchanged = Vec::new();

    // Removed + (un)changed: alles aus A.
    for (name, def_a) in &map_a {
        let hash_a = ailang_core::def_hash(def_a);
        let kind_a = ailang_core::def_kind(def_a);
        match map_b.get(*name) {
            None => removed.push(DiffEntry {
                name: (*name).to_string(),
                hash: hash_a,
                kind: kind_a,
            }),
            Some(def_b) => {
                let hash_b = ailang_core::def_hash(def_b);
                let kind_b = ailang_core::def_kind(def_b);
                if hash_a == hash_b {
                    unchanged.push(DiffEntry {
                        name: (*name).to_string(),
                        hash: hash_a,
                        kind: kind_a,
                    });
                } else {
                    changed.push(ChangedEntry {
                        name: (*name).to_string(),
                        hash_a,
                        hash_b,
                        kind_a,
                        kind_b,
                    });
                }
            }
        }
    }

    // Added: was nur in B vorkommt.
    for (name, def_b) in &map_b {
        if !map_a.contains_key(*name) {
            added.push(DiffEntry {
                name: (*name).to_string(),
                hash: ailang_core::def_hash(def_b),
                kind: ailang_core::def_kind(def_b),
            });
        }
    }

    // BTreeMap-Iteration ist bereits alphabetisch — keine extra-Sortierung
    // nötig, aber explizit absichern, falls die Reihenfolge der Quelle
    // jemals umgestellt wird.
    added.sort_by(|x, y| x.name.cmp(&y.name));
    removed.sort_by(|x, y| x.name.cmp(&y.name));
    changed.sort_by(|x, y| x.name.cmp(&y.name));
    unchanged.sort_by(|x, y| x.name.cmp(&y.name));

    DiffReport {
        module_a: a.name.clone(),
        module_b: b.name.clone(),
        added,
        removed,
        changed,
        unchanged,
    }
}

fn diff_report_to_json(r: &DiffReport) -> serde_json::Value {
    let entry = |e: &DiffEntry| {
        serde_json::json!({
            "name": e.name,
            "hash": e.hash,
            "kind": e.kind,
        })
    };
    let changed = |c: &ChangedEntry| {
        serde_json::json!({
            "name": c.name,
            "hash_a": c.hash_a,
            "hash_b": c.hash_b,
            "kind_a": c.kind_a,
            "kind_b": c.kind_b,
        })
    };
    serde_json::json!({
        "module_a": r.module_a,
        "module_b": r.module_b,
        "added":     r.added.iter().map(entry).collect::<Vec<_>>(),
        "removed":   r.removed.iter().map(entry).collect::<Vec<_>>(),
        "changed":   r.changed.iter().map(changed).collect::<Vec<_>>(),
        "unchanged": r.unchanged.iter().map(entry).collect::<Vec<_>>(),
    })
}

fn render_diff_text(r: &DiffReport) -> String {
    use std::fmt::Write;

    let mut out = String::new();
    let _ = writeln!(out, "diff: {} -> {}", r.module_a, r.module_b);

    if r.is_identical() && r.unchanged.is_empty() {
        let _ = writeln!(out, "no changes");
        return out;
    }

    // Einheitliche Spaltenbreite für Namens-/Kind-Spalte, damit Hashes
    // visuell aligned sind. Längster Name bestimmt die Breite.
    let name_width = r
        .added
        .iter()
        .map(|e| e.name.len())
        .chain(r.removed.iter().map(|e| e.name.len()))
        .chain(r.changed.iter().map(|e| e.name.len()))
        .chain(r.unchanged.iter().map(|e| e.name.len()))
        .max()
        .unwrap_or(0);

    for e in &r.added {
        let _ = writeln!(
            out,
            "+ {:<width$} ({})  {}",
            e.name,
            e.kind,
            e.hash,
            width = name_width
        );
    }
    for e in &r.removed {
        let _ = writeln!(
            out,
            "- {:<width$} ({})  {}",
            e.name,
            e.kind,
            e.hash,
            width = name_width
        );
    }
    for c in &r.changed {
        // Wenn sich der Kind geändert hat (z. B. const → fn), beide zeigen.
        let kind = if c.kind_a == c.kind_b {
            c.kind_a.to_string()
        } else {
            format!("{} -> {}", c.kind_a, c.kind_b)
        };
        let _ = writeln!(
            out,
            "~ {:<width$} ({})  {} -> {}",
            c.name,
            kind,
            c.hash_a,
            c.hash_b,
            width = name_width
        );
    }
    for e in &r.unchanged {
        let _ = writeln!(
            out,
            "  {:<width$} ({})  {} (unchanged)",
            e.name,
            e.kind,
            e.hash,
            width = name_width
        );
    }
    out
}