roteiro 0.0.6

//! Roteiro umbrella CLI. Wires the graph, spec, and render crates behind
//! subcommands; owns argument parsing, process I/O, and exit codes. See
//! ADR-0001 for the roadmap.
//!
//! @rto:0001

use clap::{Parser, Subcommand};

mod init;

#[derive(Parser)]
#[command(
    name = "roteiro",
    version,
    about = "Provenance-tagged codebase knowledge graph",
    long_about = "Roteiro — the pilot book for your codebase.\n\n\
        One SQLite store holding structure, intent, and context as a single \
        provenance-tagged knowledge graph, queryable by humans and AI agents \
        alike. Subcommands are scaffolds while the graph core lands; see \
        ADR-0001 and docs/BUILD_PLAN.md for the roadmap.",
    arg_required_else_help = true,
    propagate_version = true
)]
struct Cli {
    #[command(subcommand)]
    command: Command,
}

#[derive(Subcommand)]
enum Command {
    /// Scaffold Roteiro in the current repository (store, hooks, agent skill).
    Init,
    /// Incrementally update the graph for the current tree (content-addressed).
    ///
    /// By default this includes uncommitted edits to tracked files (a
    /// pre-commit preview); pass `--committed` to sync only the `HEAD` tree.
    Sync {
        /// Emit the sync report as JSON.
        #[arg(long)]
        json: bool,
        /// Sync only the committed `HEAD` tree, ignoring uncommitted edits.
        #[arg(long)]
        committed: bool,
    },
    /// Verify authored links against code and ADR states; non-zero on drift.
    Check {
        /// Emit the check report as JSON.
        #[arg(long)]
        json: bool,
    },
    /// Query the graph: explain a node, or list all nodes of a kind.
    Query {
        /// Node key to explain (e.g. `sym:rust:…#Store`, `adr:0001`, `file:…`).
        key: Option<String>,
        /// List all nodes of this kind instead of explaining a key.
        #[arg(long, conflicts_with = "key")]
        kind: Option<String>,
        /// Emit the result as JSON.
        #[arg(long)]
        json: bool,
    },
    /// One-shot import from lat.md, Graphify, or codegraph.
    Import {
        /// Source tool: lat | graphify | codegraph
        #[arg(long)]
        from: String,
    },
    /// Render the graph: docs site or Obsidian vault.
    Render {
        /// Target: docs | obsidian
        target: String,
    },
    /// Spec authoring (intent interview, house-style ADR scaffolding).
    Spec,
    /// Start the MCP server (built with `--features mcp`).
    #[cfg(feature = "mcp")]
    Serve,
}

fn main() -> anyhow::Result<()> {
    let cli = Cli::parse();
    let name = match cli.command {
        Command::Sync { json, committed } => return run_sync(json, committed),
        Command::Check { json } => return run_check(json),
        Command::Query { key, kind, json } => return run_query(key, kind, json),
        Command::Init => return run_init(),
        Command::Import { .. } => "import",
        Command::Render { .. } => "render",
        Command::Spec => "spec",
        #[cfg(feature = "mcp")]
        Command::Serve => "serve",
    };
    anyhow::bail!("`roteiro {name}` is not implemented yet (scaffold; see docs/BUILD_PLAN.md)")
}

/// Sync the graph for the current repository, optionally including uncommitted
/// edits to tracked files.
fn run_sync(json: bool, committed_only: bool) -> anyhow::Result<()> {
    use rto_graph::{ObjectCache, Registry, Repo, Store, sync, sync_worktree};

    let cwd = std::env::current_dir()?;
    let repo = Repo::discover(&cwd)?;

    // Graph DB is per-worktree (under the worktree git dir); the extraction
    // cache is shared across worktrees (under the common git dir).
    let store_dir = repo.git_dir().join("roteiro");
    std::fs::create_dir_all(&store_dir)?;
    let mut store = Store::open(&store_dir.join("graph.db"))?;
    let cache = ObjectCache::open(repo.common_dir().join("roteiro").join("objects"))?;

    let report = if committed_only {
        sync(&mut store, &repo, &cache, &Registry)?
    } else {
        sync_worktree(&mut store, &repo, &cache, &Registry)?
    };

    if json {
        println!("{}", serde_json::to_string_pretty(&report)?);
    } else {
        let tree = &report.tree[..report.tree.len().min(12)];
        let dirty = if report.blobs_dirty > 0 {
            format!(" +{} uncommitted", report.blobs_dirty)
        } else {
            String::new()
        };
        if report.no_op {
            println!(
                "up to date (tree {tree}{dirty}) — {} nodes, {} edges",
                report.nodes, report.edges
            );
        } else {
            println!(
                "synced tree {tree}{dirty} — {} blobs ({} extracted, {} cached) → {} nodes, {} edges",
                report.blobs_total,
                report.blobs_extracted,
                report.blobs_cached,
                report.nodes,
                report.edges
            );
        }
    }
    Ok(())
}

/// Open the repository and its per-worktree store and shared object cache.
fn open_graph() -> anyhow::Result<(rto_graph::Repo, rto_graph::Store, rto_graph::ObjectCache)> {
    use rto_graph::{ObjectCache, Repo, Store};
    let cwd = std::env::current_dir()?;
    let repo = Repo::discover(&cwd)?;
    let store_dir = repo.git_dir().join("roteiro");
    std::fs::create_dir_all(&store_dir)?;
    let store = Store::open(&store_dir.join("graph.db"))?;
    let cache = ObjectCache::open(repo.common_dir().join("roteiro").join("objects"))?;
    Ok((repo, store, cache))
}

/// Build the full graph into `store`: the derived code graph (via `sync`) plus
/// the authored ADR layer read from the `HEAD` tree. Returns the authored-layer
/// check report (used by `check`; ignored by `query`).
fn build_graph(
    repo: &rto_graph::Repo,
    store: &mut rto_graph::Store,
    cache: &rto_graph::ObjectCache,
) -> anyhow::Result<rto_spec::CheckReport> {
    use rto_graph::{Registry, sync};
    sync(store, repo, cache, &Registry)?;

    let mut docs = Vec::new();
    let mut annotations = Vec::new();
    let mut malformed = Vec::new();
    for blob in repo.walk_blobs()? {
        let bytes = repo.read_blob(&blob.oid)?;
        let text = String::from_utf8_lossy(&bytes);
        let file = std::path::Path::new(&blob.path);
        let is_md = file
            .extension()
            .and_then(|e| e.to_str())
            .is_some_and(|e| e.eq_ignore_ascii_case("md"));
        let name = file
            .file_name()
            .and_then(|n| n.to_str())
            .unwrap_or_default();
        let is_adr = blob.path.starts_with("docs/adr/") && is_md && name != "README.md";
        if is_adr {
            match rto_spec::parse_adr(&blob.path, &text) {
                Ok(doc) => docs.push(doc),
                // A malformed ADR is drift, not a skippable warning: it would let
                // the gate pass while silently dropping authored intent.
                Err(e) => malformed.push(rto_spec::Violation {
                    kind: rto_spec::ViolationKind::MalformedAdr,
                    message: format!("{}: cannot parse ADR: {e}", blob.path),
                }),
            }
        } else {
            annotations.extend(rto_spec::scan_annotations(&blob.path, &text));
        }
    }

    let mut report = rto_spec::run(store, &docs, &annotations)?;
    report.violations.extend(malformed);
    Ok(report)
}

/// Validate the authored layer (ADR `[[…]]` links and `@rto:` annotations)
/// against the derived graph; exit non-zero on drift.
fn run_check(json: bool) -> anyhow::Result<()> {
    let (repo, mut store, cache) = open_graph()?;
    let report = build_graph(&repo, &mut store, &cache)?;

    if json {
        println!("{}", serde_json::to_string_pretty(&report)?);
    } else {
        for v in &report.violations {
            eprintln!("drift [{}]: {}", v.kind.label(), v.message);
        }
        println!(
            "checked {} ADR(s): {} link(s) ok, {} annotation(s) ok, {} violation(s)",
            report.adrs,
            report.links_ok,
            report.annotations_ok,
            report.violations.len(),
        );
    }

    if report.has_violations() {
        std::process::exit(1);
    }
    Ok(())
}

/// Scaffold Roteiro in the current repository: build the initial graph, install
/// the `post-checkout`/`post-merge` hooks, and add the `AGENTS.md` snippet.
fn run_init() -> anyhow::Result<()> {
    let (repo, mut store, cache) = open_graph()?;

    let report = build_graph(&repo, &mut store, &cache)?;
    let nodes = store.node_count()?;
    let edges = store.edge_count()?;

    // Hooks live under the common git dir so they are shared across worktrees.
    let hooks_dir = repo.common_dir().join("hooks");
    for name in init::MANAGED_HOOKS {
        match init::install_hook(&hooks_dir, name)? {
            init::HookOutcome::Installed => println!("installed hook: {name}"),
            init::HookOutcome::Updated => println!("refreshed hook: {name}"),
            init::HookOutcome::SkippedForeign => eprintln!(
                "warning: existing non-Roteiro `{name}` hook left untouched; \
                 add `roteiro sync --committed` to it to keep the graph fresh"
            ),
        }
    }

    if let Some(workdir) = repo.workdir() {
        let path = workdir.join("AGENTS.md");
        if init::ensure_agents(&path)? {
            println!("wrote Roteiro section to {}", path.display());
        }
    }

    println!("roteiro initialised — graph has {nodes} nodes, {edges} edges");
    if report.has_violations() {
        eprintln!(
            "note: {} authored-layer violation(s); run `roteiro check` for details",
            report.violations.len()
        );
    }
    Ok(())
}

/// Query the graph: explain a node's provenance-labelled neighbourhood, or list
/// all nodes of a kind. Builds the full (derived + authored) graph first so
/// results reflect the current source and ADRs.
fn run_query(key: Option<String>, kind: Option<String>, json: bool) -> anyhow::Result<()> {
    use rto_graph::{NodeKind, explain, list_kind};

    let (repo, mut store, cache) = open_graph()?;
    build_graph(&repo, &mut store, &cache)?;

    match (key, kind) {
        (Some(key), _) => {
            let Some(ex) = explain(&store, &key)? else {
                anyhow::bail!(
                    "no node with key `{key}` (try `roteiro query --kind <kind>` to list nodes)"
                );
            };
            if json {
                println!("{}", serde_json::to_string_pretty(&ex)?);
            } else {
                println!("{}  ({})  {}", ex.node.key, ex.node.kind, ex.node.name);
                if let Some(path) = &ex.node.path {
                    println!("  path: {path}");
                }
                if !ex.outgoing.is_empty() {
                    println!("  outgoing:");
                    for e in &ex.outgoing {
                        println!("    -[{}/{}]-> {}", e.kind, e.provenance, e.node);
                    }
                }
                if !ex.incoming.is_empty() {
                    println!("  incoming:");
                    for e in &ex.incoming {
                        println!("    <-[{}/{}]- {}", e.kind, e.provenance, e.node);
                    }
                }
            }
        }
        (None, Some(kind)) => {
            let listing = list_kind(&store, &NodeKind::from_token(&kind))?;
            if json {
                println!("{}", serde_json::to_string_pretty(&listing)?);
            } else {
                println!("{} ({}):", listing.kind, listing.nodes.len());
                for n in &listing.nodes {
                    println!("  {}  {}", n.key, n.name);
                }
            }
        }
        (None, None) => {
            anyhow::bail!("provide a node key to explain, or `--kind <kind>` to list nodes");
        }
    }
    Ok(())
}