memstead-cli 0.11.0

use clap::Parser;

use memstead_base::Entity;
use memstead_base::EntityId;
use memstead_base::Store;
use memstead_base::chunking::apply_chunking;
use memstead_base::render;

use crate::CliError;
use crate::output::{ExitKind, print_markdown};
use crate::setup::{CliContext, CliEngine};

/// Read one entity as markdown.
#[derive(Parser, Debug)]
pub struct Args {
    /// Entity ID (e.g. `specs--my-entity`).
    pub id: String,

    /// Restrict output to specific section keys (repeatable).
    #[arg(long = "section", value_name = "KEY")]
    pub sections: Vec<String>,

    /// Append relations as a trailing JSON code block.
    #[arg(long)]
    pub include_relations: bool,

    /// Token budget for chunking. Omit for no chunking.
    #[arg(long)]
    pub token_budget: Option<usize>,

    /// 1-based chunk index to return (requires `--token-budget`).
    #[arg(long)]
    pub chunk: Option<usize>,

    /// Append the derived mutation-provenance block: created-by and
    /// last-modified-by with actor, client, declared role (or
    /// `unspecified`), and timestamp — read from the append-only
    /// mutation record, which no verb can edit after the fact.
    #[arg(long = "provenance")]
    pub provenance: bool,
}

pub fn run(ctx: &CliContext, args: Args) -> anyhow::Result<()> {
    let id = EntityId::canonical(&args.id);
    // The engine's `get_entity` returns `Option<&Entity>`, so the
    // typed `ENTITY_NOT_FOUND` code lives on the CLI side here —
    // pin it explicitly so the wire envelope matches what the engine
    // would emit from a write-path miss.
    let miss = |engine: &memstead_base::Engine| {
        // A quarantined mem's entities are deliberately absent — the
        // read names the quarantine (MEM_QUARANTINED with the boot
        // reason), not a phantom ENTITY_NOT_FOUND.
        if engine.quarantine_reason(id.mem()).is_some() {
            return CliError::from_engine_op(engine.unknown_mem_error(id.mem()));
        }
        CliError::new(
            ExitKind::NotFound,
            "ENTITY_NOT_FOUND",
            format!("Entity not found: {}", args.id),
        )
        .with_details(serde_json::json!({ "id": args.id }))
    };
    // Snapshot the store's outgoing edges alongside the entity so the
    // JSON envelope can resolve each relationship's `source` label after
    // the engine goes out of scope. The outgoing edges carry the
    // authoritative `EdgeSource` discriminator (`Explicit` /
    // `BodyLink` / `Hierarchy`); the entity's `relationships` vec
    // doesn't encode it.
    // Derived mutation-provenance block (agent-trust plan 13), only
    // when `--provenance` asked for it — default output stays
    // byte-unchanged. Unavailability (an archive seam with no
    // history) is stated, never fabricated.
    let provenance_block = |engine: &memstead_base::Engine| -> Option<serde_json::Value> {
        if !args.provenance {
            return None;
        }
        Some(match engine.entity_provenance(id.mem(), id.as_ref()) {
            Ok(prov) => serde_json::to_value(&prov).unwrap_or(serde_json::Value::Null),
            Err(e) => serde_json::json!({ "unavailable": e.to_string() }),
        })
    };
    // Load scope: everything this command renders lives in the target
    // mem's slice of the store — the entity, its sections, and its
    // OUTGOING edges (carried by its own source record) — plus mount
    // metadata. So a lazy workspace pays only the target mem's load,
    // the cold-path cut the plan names. Three forms are cross-mem and
    // take the full load, never a partial answer:
    // `--include-relations` (INCOMING edges can originate in any mem),
    // declared signals (an `in`-direction signal reads incoming edges,
    // and a neighbour pair reads the counterpart record either side of
    // the edge), and the labelling view (it counts the cross-mem edges
    // it excludes, and the support walk may cross mems). The scoped
    // cold path stays for every mem whose schema declares neither.
    let declares_cross_mem_serving = |schema: Option<std::sync::Arc<memstead_schema::Schema>>| {
        schema.is_some_and(|s| {
            s.manifest.relationships.labelling.is_some()
                || s.types.values().any(|td| !td.signals.is_empty())
        })
    };
    let engine_handle = if args.include_relations {
        ctx.cli_engine()?
    } else {
        let scoped = ctx.cli_engine_scoped(id.mem())?;
        let escalate = match &scoped {
            #[cfg(feature = "mem-repo")]
            CliEngine::MemRepo(engine) => declares_cross_mem_serving(engine.schema_for(id.mem())),
            CliEngine::Filesystem(engine) => {
                declares_cross_mem_serving(engine.schema_for(id.mem()))
            }
        };
        if escalate { ctx.cli_engine()? } else { scoped }
    };
    let (
        entity,
        output,
        outgoing_snapshot,
        incoming_snapshot,
        origin,
        provenance,
        signals,
        labelling,
    ) = match engine_handle {
        #[cfg(feature = "mem-repo")]
        CliEngine::MemRepo(engine) => {
            let entity = engine
                .get_entity(&id)
                .cloned()
                .ok_or_else(|| miss(&engine))?;
            let signals = engine.computed_signals(&entity);
            let labelling = engine.computed_labelling(&entity);
            let md = render_with_optional_relations(
                &entity,
                &id,
                engine.store(),
                &args,
                signals.as_deref(),
                labelling.as_ref(),
            );
            let outgoing = engine.store().outgoing(&id).to_vec();
            let incoming = args
                .include_relations
                .then(|| engine.store().incoming(&id).to_vec());
            let origin = engine.mem_origin_class(id.mem());
            let prov = provenance_block(&engine);
            (
                entity, md, outgoing, incoming, origin, prov, signals, labelling,
            )
        }
        CliEngine::Filesystem(engine) => {
            let entity = engine
                .get_entity(&id)
                .cloned()
                .ok_or_else(|| miss(&engine))?;
            let signals = engine.computed_signals(&entity);
            let labelling = engine.computed_labelling(&entity);
            let md = render_with_optional_relations(
                &entity,
                &id,
                engine.store(),
                &args,
                signals.as_deref(),
                labelling.as_ref(),
            );
            let outgoing = engine.store().outgoing(&id).to_vec();
            let incoming = args
                .include_relations
                .then(|| engine.store().incoming(&id).to_vec());
            let origin = engine.mem_origin_class(id.mem());
            let prov = provenance_block(&engine);
            (
                entity, md, outgoing, incoming, origin, prov, signals, labelling,
            )
        }
    };

    let chunked = match args.token_budget {
        Some(budget) => apply_chunking(
            &output,
            budget,
            args.chunk,
            &[("_hash", &entity.content_hash)],
        )
        .map_err(|e| CliError::new(ExitKind::Generic, "CHUNK_OUT_OF_RANGE", e))?,
        None => output.to_string(),
    };

    if ctx.json {
        // The CLI `--json`
        // shape mirrors the MCP `structured_content` envelope —
        // typed fields (`_hash`, `id`, `mem`, `type`, sections,
        // relationships) rather than a
        // `{ markdown: "..." }` flat shape that would force agents to
        // string-scrape frontmatter for `_hash`. Greenfield
        // justifies the wire shape break.
        let sections_filter = if args.sections.is_empty() {
            None
        } else {
            Some(args.sections.as_slice())
        };
        let rendered_body_tokens = memstead_base::chunking::estimate_tokens(&output);
        let full_tokens = if sections_filter.is_some() {
            let full_body = render::render_entity_markdown(&entity, None);
            Some(memstead_base::chunking::estimate_tokens(&full_body))
        } else {
            None
        };
        let mut envelope = render::build_entity_envelope(
            &entity,
            rendered_body_tokens,
            full_tokens,
            sections_filter,
            None,
            origin,
            &outgoing_snapshot,
            incoming_snapshot.as_deref(),
            signals.as_deref(),
            labelling.as_ref(),
        );
        if let (Some(prov), Some(obj)) = (&provenance, envelope.as_object_mut()) {
            obj.insert("mutation_provenance".into(), prov.clone());
        }
        crate::output::print_json(&envelope)?;
    } else {
        let mut text = chunked.clone();
        if let Some(prov) = &provenance {
            let render_rec = |label: &str, key: &str| -> Option<String> {
                let r = prov.get(key)?;
                Some(format!(
                    "- {label}: {} ({}), role {}, at {}",
                    r["client"].as_str().unwrap_or("unknown client"),
                    r["actor"].as_str().unwrap_or("unknown actor"),
                    r["role"].as_str().unwrap_or("unspecified"),
                    r["timestamp"],
                ))
            };
            text.push_str(
                "

## Mutation provenance
",
            );
            match prov.get("unavailable") {
                Some(reason) => {
                    text.push_str(&format!(
                        "- unavailable: {}
",
                        reason.as_str().unwrap_or("")
                    ));
                }
                None => {
                    if let Some(l) = render_rec("created by", "created_by") {
                        text.push_str(&l);
                        text.push('\n');
                    } else {
                        text.push_str(
                            "- created by: not recorded (story truncated)
",
                        );
                    }
                    if let Some(l) = render_rec("last modified by", "last_modified_by") {
                        text.push_str(&l);
                        text.push('\n');
                    }
                    if let Some(state) = prov.get("check_state").and_then(|v| v.as_str()) {
                        text.push_str(&format!("- check state: {state}\n"));
                    }
                }
            }
        }
        print_markdown(&text);
    }
    Ok(())
}

/// Render an entity's markdown body and, when `--include-relations` is
/// set, append the outgoing/incoming JSON block. Engine-agnostic: both
/// flavours expose a `&Store`.
fn render_with_optional_relations(
    entity: &Entity,
    id: &EntityId,
    store: &Store,
    args: &Args,
    signals: Option<&[memstead_base::ops::signals::ComputedSignal]>,
    labelling: Option<&memstead_base::ops::labelling::LabellingView>,
) -> String {
    let sections_filter = if args.sections.is_empty() {
        None
    } else {
        Some(args.sections.as_slice())
    };
    let mut md =
        render::render_entity_markdown_with_signals(entity, sections_filter, signals, labelling);
    if args.include_relations {
        let outgoing = store.outgoing(id).to_vec();
        let incoming = store.incoming(id).to_vec();
        let rel_json = render::render_relations_json(id.as_ref(), &outgoing, &incoming);
        md.push_str("\n## Relations (JSON)\n\n```json\n");
        md.push_str(&serde_json::to_string_pretty(&rel_json).unwrap_or_else(|_| "{}".to_string()));
        md.push_str("\n```\n");
    }
    md
}