memstead_cli/commands/

schema.rs

//! `memstead schema validate <path>` — load a schema package from disk
//! and report whether it conforms to the engine's schema rules.
//!
//! Validation runs the same loader the engine uses at boot
//! (`memstead_schema::loader::load_schema_from_dir`), so a package that
//! validates here is one the engine will accept. Parse failures carry
//! the YAML layer's line/column in their message; structural failures
//! (undeclared relationship vocabulary, type/file mismatch, missing
//! `_default` weight, …) carry the engine's typed diagnostic.
//!
//! `memstead schema install <name|path>` copies a schema package into
//! the current workspace's local schema storage so a mem can pin it.
//! It resolves the source — a built-in name (`planning`, `planning@0.1.0`)
//! or a path to a package directory — validates it, and writes the
//! package (including any `mem-template.json`) under the folder
//! backend's fixed `<workspace>/.memstead/schemas/<name>@<version>/`
//! location. Installing a built-in forks it into local storage, which
//! shadows the built-in per the resolution order — the customization
//! entry point. Idempotent: re-running reproduces the same files.
//! Git-branch (mem-repo) workspaces are a destination too: install
//! routes through the engine's below-boot repair surface
//! (`memstead_git_branch::repair::install_schema_below_boot`), which
//! runs the same validation gate as the booted `Engine::install_schema`
//! and seals the package onto the `__MEMSTEAD:schemas/` ref.
//!
//! `memstead schema migrate <path>` rewrites the retired keys of an
//! authoring package into the current schema language — by exactly the
//! translations the loader applies to sealed content
//! (`memstead_schema::migrate`). Dry run by default: it prints one line
//! per rewrite and writes nothing; `--write` applies them in place,
//! comments and key order preserved. It never bumps `version` and never
//! touches a sealed copy.
//!
//! `validate` and `migrate` are flavour-agnostic and touch no workspace. `install`
//! never boots the workspace on either flavour — it is a named remedy
//! for boot-blocking states (repair-below-boot rule), so it operates on
//! configuration and schema storage only: the folder flavour writes
//! `.memstead/schemas/` directly, the mem-repo flavour writes the
//! engine-owned `__MEMSTEAD` ref through the engine's own repair
//! surface.

use std::path::{Path, PathBuf};

use clap::{Args as ClapArgs, Subcommand};
use serde_json::json;

use memstead_schema::SchemaRef;

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

#[derive(ClapArgs, Debug)]
#[command(args_conflicts_with_subcommands = true)]
pub struct Args {
    #[command(subcommand)]
    pub command: Option<SchemaCommand>,

    /// Render a built-in schema package's README as the package it
    /// ships in. `<REF>` is a bare built-in name (`planning`, resolved
    /// to its newest generation) or a pin (`planning@0.4.0`). Every
    /// `<name>@<x.y.z>` reference to the package's own name is
    /// rewritten to the resolved pin: sibling generations of a built-in
    /// carry the README of their first generation verbatim (sealed
    /// bytes, never edited), so the resolved manifest is the one source
    /// of the identity the reader sees. Markdown by default; `--json`
    /// carries `schema`, `name`, `version`, `origin` and the rendered
    /// `readme` (`null` when the package ships none). Unknown names
    /// refuse with `SCHEMA_NOT_FOUND`.
    #[arg(value_name = "REF")]
    pub reference: Option<String>,
}

#[derive(Subcommand, Debug)]
pub enum SchemaCommand {
    /// Scaffold a new schema package at `./<name>/` — a manifest plus
    /// one commented example type — that `memstead schema validate`
    /// passes unmodified. Prints the follow-up commands that take the
    /// package from folder to pinned mem.
    New(NewArgs),

    /// Validate a schema package directory (`schema.yaml` plus an
    /// optional `types/*.yaml`) against the engine's schema loader, as
    /// the package you are AUTHORING — always read in the current
    /// schema language, so retired keys (`optional:`,
    /// `propagating_relationships`) refuse here by design. A package
    /// already sealed under an older language can therefore be
    /// refused by this command and still load: sealed content is read
    /// under the generation it was sealed in, and that is the point of
    /// sealing. Validate what you write, not what you installed.
    /// Exits non-zero (`SCHEMA_VALIDATION_FAILED`) on any conformance
    /// error, with the YAML line/column in the message where the parse
    /// layer provides it.
    Validate(ValidateArgs),

    /// Install a schema package into the current folder workspace's
    /// `.memstead/schemas/<name>@<version>/` so a mem can pin it.
    /// `<source>` is a built-in name (`planning`, `planning@0.1.0`) or a
    /// path to a package directory. Validates before copying; idempotent.
    ///
    /// Repair note — engines 0.6.0 through 0.8.1 mis-stamped LEGACY
    /// packages as current-language at install, flipping their bare
    /// metadata fields from required to optional. If you installed a
    /// legacy package in that window, re-run the install with a fixed
    /// engine; the detection command and details are in the docs-site
    /// schema-authoring guide.
    Install(InstallArgs),

    /// Rewrite an authoring package's retired keys into the current
    /// schema language — the keys `schema validate` refuses
    /// (`propagating_relationships`, `optional:`, the retired `examples:`
    /// list, the exemplar-relation `to:`/`type:` spelling) — by exactly
    /// the translations the engine applies when it reads sealed
    /// content. Dry run by default: prints one line per rewrite and
    /// writes nothing. `--write` applies the rewrites in place, keeping
    /// comments, key order and spacing. A package that carried
    /// `optional:` was written when an absent key meant required, so
    /// its fields declaring neither key get `required: true` — the
    /// meaning they already have when sealed; delete the line where you
    /// did not mean it. Never bumps `version` (whether a spelling
    /// migration deserves one is yours to decide) and never touches a
    /// sealed copy inside a mem. Exits non-zero (`SCHEMA_MIGRATE_FAILED`)
    /// when a value cannot be rewritten mechanically, when the package
    /// fails to load for a reason no retired key explains, or when the
    /// rewrite would not reproduce the engine's own translation.
    Migrate(MigrateArgs),
}

#[derive(ClapArgs, Debug)]
pub struct MigrateArgs {
    /// Path to the schema package directory (the folder containing
    /// `schema.yaml`).
    pub path: PathBuf,

    /// Apply the rewrites in place. Without it the command is a dry run
    /// that reports what would change and writes nothing.
    #[arg(long)]
    pub write: bool,
}

#[derive(ClapArgs, Debug)]
pub struct NewArgs {
    /// Schema name. Grammar: starts with a lowercase letter, then
    /// lowercase letters, digits, and hyphens. The package is written
    /// to `./<name>/`.
    pub name: String,
}

#[derive(ClapArgs, Debug)]
pub struct ValidateArgs {
    /// Path to the schema package directory (the folder containing
    /// `schema.yaml`).
    pub path: PathBuf,
}

#[derive(ClapArgs, Debug)]
pub struct InstallArgs {
    /// Built-in schema name (`planning`, `planning@0.1.0`) or a path to
    /// a schema package directory.
    pub source: String,
}

pub fn run(ctx: &CliContext, args: Args) -> anyhow::Result<()> {
    // Publish the authoring meta-schemas into `.memstead/meta-schemas/` so an
    // editor validates authored schema YAML against them (resolved by each
    // package's `# yaml-language-server:` directive). This is the only place
    // that writes them: they used to be published at engine boot, which made
    // every read of a mem modify the directory it read. Only the three
    // schema-authoring subcommands publish; rendering a built-in
    // (`memstead schema <ref>`) is a read and stays one. Best-effort — a
    // read-only workspace, or none at all, still runs the command.
    if matches!(
        args.command,
        Some(
            SchemaCommand::New(_)
                | SchemaCommand::Validate(_)
                | SchemaCommand::Install(_)
                | SchemaCommand::Migrate(_)
        )
    ) && let Some((_, root)) = ctx.workspace_shape()
    {
        let _ = memstead_schema::meta_schema::publish_meta_schemas(&root);
    }
    match (args.command, args.reference) {
        (Some(SchemaCommand::New(a)), _) => scaffold_new(ctx, a),
        (Some(SchemaCommand::Validate(a)), _) => validate(ctx, a),
        (Some(SchemaCommand::Install(a)), _) => install(ctx, a),
        (Some(SchemaCommand::Migrate(a)), _) => migrate(ctx, a),
        (None, Some(reference)) => show_builtin(ctx, &reference),
        (None, None) => Err(CliError::new(
            ExitKind::Validation,
            "INVALID_INPUT",
            "memstead schema needs a built-in reference to render (`memstead schema \
             planning@0.4.0`) or a subcommand (`new`, `validate`, `install`); \
             `memstead schema --help` lists them"
                .to_string(),
        )
        .into()),
    }
}

/// `memstead schema <REF>`: the package's README rendered for the
/// package it ships in (see [`Args::reference`]). Built-ins resolve
/// first; a pinned reference that is not a built-in falls through to
/// the workspace's installed stores — the filesystem
/// `.memstead/schemas/<name>@<version>/` layout and the mem-repo's
/// `__MEMSTEAD:schemas/` ref — so a workspace-local schema's sealed
/// README (a contract carrier, not only docs) is readable through the
/// same sanctioned verb as a built-in's.
fn show_builtin(ctx: &CliContext, reference: &str) -> anyhow::Result<()> {
    let schema_ref = resolve_builtin_read_ref(reference)?;
    let version = schema_ref.version.to_string();
    let (origin, name, ver, readme_bytes) =
        match memstead_schema::builtins::builtin_package(&schema_ref.name, &version) {
            Some(pkg) => {
                let bytes = pkg
                    .files
                    .iter()
                    .find(|(path, _)| path == memstead_schema::builtins::PACKAGE_README_FILE)
                    .map(|(_, bytes)| bytes.to_vec());
                (
                    "builtin",
                    pkg.name.to_string(),
                    pkg.version.to_string(),
                    bytes,
                )
            }
            None => match workspace_installed_readme(ctx, &schema_ref) {
                Some(bytes) => (
                    "workspace",
                    schema_ref.name.clone(),
                    version.clone(),
                    Some(bytes),
                ),
                None => {
                    return Err(CliError::new(
                        ExitKind::Validation,
                        "SCHEMA_NOT_FOUND",
                        format!(
                            "no built-in or workspace-installed schema {}",
                            schema_ref.as_display()
                        ),
                    )
                    .into());
                }
            },
        };
    let readme = readme_bytes
        .as_deref()
        .and_then(|bytes| std::str::from_utf8(bytes).ok())
        .map(|text| memstead_schema::builtins::render_package_readme(&name, &ver, text));
    let pin = format!("{name}@{ver}");
    if ctx.json {
        print_json(&json!({
            "schema": pin,
            "name": name,
            "version": ver,
            "origin": origin,
            "readme": readme,
        }))?;
        return Ok(());
    }
    // Human label per origin — the wire `origin` stays the machine token.
    let label = if origin == "builtin" {
        "built-in"
    } else {
        "workspace-installed"
    };
    match readme {
        Some(text) => print_markdown(&format!(
            "<!-- {pin}: {label} package README, rendered for this generation -->\n{text}"
        )),
        None => print_markdown(&format!(
            "`{pin}` is a {label} package that ships no README.\n"
        )),
    }
    Ok(())
}

/// The sealed README of a workspace-installed schema package, from the
/// filesystem `.memstead/schemas/<name>@<version>/` layout first, then
/// the mem-repo's `__MEMSTEAD:schemas/` ref (where `memstead schema
/// install` writes on git-backed workspaces). `None` outside a
/// workspace or when the package (or its README) is absent.
fn workspace_installed_readme(ctx: &CliContext, schema_ref: &SchemaRef) -> Option<Vec<u8>> {
    let (_, root) = ctx.workspace_shape()?;
    let fs_readme = root
        .join(".memstead")
        .join("schemas")
        .join(format!("{}@{}", schema_ref.name, schema_ref.version))
        .join("README.md");
    if let Ok(bytes) = std::fs::read(&fs_readme) {
        return Some(bytes);
    }
    #[cfg(feature = "mem-repo")]
    {
        let gitdir = root.join("mem-repo").join(".git");
        if gitdir.is_dir()
            && let Ok(Some(bytes)) =
                memstead_git_branch::storage_memstead::read_schema_file_from_memstead_ref(
                    &gitdir,
                    &schema_ref.name,
                    &schema_ref.version.to_string(),
                    "README.md",
                )
        {
            return Some(bytes);
        }
    }
    None
}

/// Resolve a reference for a READ: a pin as given, a bare name to its
/// newest built-in generation (a read is safe to default; `install`
/// keeps refusing bare names so a pin is always explicit). A pin is
/// only parsed here — whether it names a built-in or a
/// workspace-installed package is the render path's lookup, so a
/// workspace-local pin is not refused before the workspace stores are
/// consulted. Bare names stay built-in-only: the workspace stores are
/// version-addressed, and defaulting a generation for a contract
/// carrier would hide which generation the reader got.
fn resolve_builtin_read_ref(reference: &str) -> anyhow::Result<SchemaRef> {
    if reference.contains('@') {
        return reference.parse::<SchemaRef>().map_err(|e: String| {
            CliError::new(
                ExitKind::Validation,
                "INVALID_INPUT",
                format!("invalid schema pin {reference:?}: {e}"),
            )
            .into()
        });
    }
    let reg = memstead_schema::SchemaRegistry::builtin();
    let mut versions = reg.available_versions(reference);
    versions.sort();
    match versions.pop() {
        Some(v) => Ok(SchemaRef::new(reference.to_string(), v)),
        None => Err(CliError::new(
            ExitKind::Validation,
            "SCHEMA_NOT_FOUND",
            format!(
                "no built-in schema named {reference:?}; built-ins: {}",
                builtin_names_joined(&reg)
            ),
        )
        .into()),
    }
}

fn builtin_names_joined(reg: &memstead_schema::SchemaRegistry) -> String {
    let mut names: Vec<String> = reg.identities().into_iter().map(|(n, _)| n).collect();
    names.sort();
    names.dedup();
    names.join(", ")
}

/// Version every scaffolded package starts at.
const SCAFFOLD_VERSION: &str = "0.1.0";

fn scaffold_new(ctx: &CliContext, args: NewArgs) -> anyhow::Result<()> {
    if let Err(reason) = memstead_schema::loader::validate_schema_name(&args.name) {
        let suggestion = suggest_schema_name(&args.name);
        return Err(CliError::new(
            ExitKind::Validation,
            "INVALID_INPUT",
            format!(
                "invalid schema name {name:?}: {reason} (lowercase letter first, \
                 then lowercase letters, digits, hyphens). \
                 Try: memstead schema new {suggestion}",
                name = args.name,
            ),
        )
        .with_details(json!({
            "name": args.name,
            "reason": reason,
            "suggestion": suggestion,
        }))
        .into());
    }

    let pkg_dir = PathBuf::from(&args.name);
    if pkg_dir.join("schema.yaml").is_file() {
        return Err(CliError::new(
            ExitKind::Validation,
            "SCHEMA_PACKAGE_EXISTS",
            format!(
                "{} already contains a schema package — `memstead schema new` \
                 never overwrites. Check it with: memstead schema validate {}",
                pkg_dir.display(),
                args.name,
            ),
        )
        .with_details(json!({ "path": pkg_dir }))
        .into());
    }
    if pkg_dir.is_dir()
        && let Some(entry) = std::fs::read_dir(&pkg_dir)
            .map_err(|e| {
                CliError::new(
                    ExitKind::Generic,
                    "IO_ERROR",
                    format!("read {}: {e}", pkg_dir.display()),
                )
            })?
            .next()
            .transpose()
            .map_err(|e| {
                CliError::new(
                    ExitKind::Generic,
                    "IO_ERROR",
                    format!("read {}: {e}", pkg_dir.display()),
                )
            })?
    {
        let found = entry.file_name().to_string_lossy().to_string();
        return Err(CliError::new(
            ExitKind::Validation,
            "TARGET_NOT_EMPTY",
            format!(
                "{} exists and is not empty (found `{found}`) — clear it or \
                     pick a different name: memstead schema new {}-schema",
                pkg_dir.display(),
                args.name,
            ),
        )
        .with_details(json!({ "path": pkg_dir, "found": [found] }))
        .into());
    }

    let manifest = scaffold_manifest(&args.name);
    let example_type = scaffold_example_type();
    std::fs::create_dir_all(pkg_dir.join("types")).map_err(|e| {
        CliError::new(
            ExitKind::Generic,
            "IO_ERROR",
            format!("create {}: {e}", pkg_dir.join("types").display()),
        )
    })?;
    for (rel, content) in [
        ("schema.yaml", &manifest),
        ("types/note.yaml", &example_type),
    ] {
        let dest = pkg_dir.join(rel);
        std::fs::write(&dest, content).map_err(|e| {
            CliError::new(
                ExitKind::Generic,
                "IO_ERROR",
                format!("write {}: {e}", dest.display()),
            )
        })?;
    }

    // Self-check with the engine loader — the scaffold's contract is
    // "validates clean as generated"; fail loudly here rather than at
    // the user's `schema validate` if a template edit ever breaks it.
    if let Err(e) = memstead_schema::loader::load_schema_from_dir(&pkg_dir)
        .and_then(|s| memstead_schema::check_reserved_metadata_keys(&s).map(|()| s))
        .and_then(|s| memstead_schema::check_section_formats(&s).map(|()| s))
    {
        return Err(CliError::new(
            ExitKind::Generic,
            crate::INTERNAL_CODE,
            format!(
                "scaffold bug: generated package at {} fails validation: {e} — \
                 please report this",
                pkg_dir.display(),
            ),
        )
        .into());
    }

    let next_steps = scaffold_next_steps(ctx, &args.name);
    if ctx.json {
        print_json(&json!({
            "ok": true,
            "schema": format!("{}@{SCAFFOLD_VERSION}", args.name),
            "path": pkg_dir,
            "files": ["schema.yaml", "types/note.yaml"],
            "next_steps": next_steps
                .iter()
                .map(|s| json!({ "command": s.command, "note": s.note }))
                .collect::<Vec<_>>(),
        }))?;
    } else {
        let steps: Vec<String> = next_steps
            .iter()
            .enumerate()
            .map(|(i, s)| match &s.note {
                Some(note) => format!("{}. `{}` — {note}", i + 1, s.command),
                None => format!("{}. `{}`", i + 1, s.command),
            })
            .collect();
        print_markdown(&format!(
            "# Schema package scaffolded\n\n`{name}@{SCAFFOLD_VERSION}` at `{dir}` \
             (schema.yaml + types/note.yaml, one commented example type).\n\n\
             Edit the package, then:\n\n{steps}\n",
            name = args.name,
            dir = pkg_dir.display(),
            steps = steps.join("\n"),
        ));
    }
    Ok(())
}

/// The follow-up command sequence printed by `schema new` — the rest of
/// the custom-schema flow is copy-paste from here. When the command
/// runs inside a single-writable-mem workspace, the pin step names the
/// actual mem (from the mount roster — the authoritative name source);
/// otherwise it keeps a `<mem>` placeholder. When the workspace still
/// carries the `memstead quickstart` seed entity, a delete step for it
/// precedes the pin: `mem set-schema` switches atomically only when
/// every entity conforms to the target, and the default-schema seed
/// never conforms to a fresh custom schema — without the delete the
/// verbatim flow ends in a dual-pin migration instead of a pinned mem.
fn scaffold_next_steps(ctx: &CliContext, name: &str) -> Vec<Step> {
    use memstead_base::workspace::MountCapability;
    use memstead_base::workspace_store::{FileWorkspaceStore, WorkspaceStoreAdapter};
    let workspace = ctx.workspace_shape().and_then(|(shape, root)| match shape {
        WorkspaceShape::Filesystem => FileWorkspaceStore::new().load(&root).ok().and_then(|ws| {
            let mut writable = ws
                .mounts
                .iter()
                .filter(|m| m.capability == MountCapability::Write);
            match (writable.next(), writable.next()) {
                (Some(only), None) => Some((only.mem.clone(), root.clone())),
                _ => None,
            }
        }),
        WorkspaceShape::MemRepo => None,
    });
    let mem = workspace
        .as_ref()
        .map(|(mem, _)| mem.clone())
        .unwrap_or_else(|| "<mem>".to_string());
    // Filesystem mems live at the workspace root; the quickstart seed
    // is the fixed-slug `welcome-to-memstead.md`.
    let quickstart_seed = workspace
        .as_ref()
        .filter(|(_, root)| root.join("welcome-to-memstead.md").is_file())
        .map(|(mem, _)| format!("{mem}--welcome-to-memstead"));
    // Full flavour: install into the current workspace, then re-pin the
    // mem in place.
    #[cfg(feature = "mem-repo")]
    {
        let mut steps = vec![
            Step::bare(format!("memstead schema validate {name}")),
            Step::bare(format!("memstead schema install {name}")),
        ];
        if let Some(seed_id) = quickstart_seed {
            steps.push(Step {
                command: format!("memstead delete {seed_id}"),
                note: Some(
                    "the quickstart seed — the pin below switches atomically only when \
                     every entity conforms to the new schema"
                        .to_string(),
                ),
            });
        }
        steps.push(Step::bare(format!(
            "memstead mem set-schema {mem} {name}@{SCAFFOLD_VERSION}"
        )));
        steps
    }
    // Lean flavour: no `mem set-schema`, so the custom schema gets a
    // fresh mem. Order matters — `init` pins without resolving, and the
    // engine only boots once the package is installed *inside the new
    // workspace*, so the install step comes right after init and points
    // back at the scaffolded package. When `schema new` ran inside an
    // existing workspace, the fresh mem must land OUTSIDE it (workspaces
    // don't nest, and `init`'s refusal there is a dead end on this
    // binary) — the printed paths anchor at the workspace root's parent,
    // absolute and quoted so the sequence stays verbatim-runnable.
    #[cfg(not(feature = "mem-repo"))]
    {
        let _ = (mem, quickstart_seed); // full-only context
        let (fresh_dir, install_source) = match ctx.workspace_shape() {
            Some((_, root)) => {
                let parent = root.parent().unwrap_or(&root).to_path_buf();
                let pkg = std::env::current_dir().unwrap_or_default().join(name);
                (
                    format!("\"{}\"", parent.join(format!("{name}-mem")).display()),
                    format!("\"{}\"", pkg.display()),
                )
            }
            None => (format!("{name}-mem"), format!("../{name}")),
        };
        vec![
            Step::bare(format!("memstead schema validate {name}")),
            Step {
                command: format!(
                    "mkdir {fresh_dir} && cd {fresh_dir} && memstead init --name {name}-mem \
                     --schema {name}@{SCAFFOLD_VERSION}"
                ),
                note: Some(
                    "this binary cannot re-pin an existing mem, so the schema gets a \
                     fresh one"
                        .to_string(),
                ),
            },
            Step {
                command: format!("memstead schema install {install_source}"),
                note: Some(
                    "run inside the new folder — the workspace boots once its pinned \
                     schema is installed"
                        .to_string(),
                ),
            },
        ]
    }
}

/// One printed follow-up step: a verbatim-runnable command plus an
/// optional explanation rendered outside the command so copy-paste
/// stays clean.
struct Step {
    command: String,
    note: Option<String>,
}

impl Step {
    fn bare(command: String) -> Self {
        Step {
            command,
            note: None,
        }
    }
}

/// Best-effort correction for an invalid schema name, offered in the
/// refusal message: lowercase, non-grammar characters to hyphens,
/// hyphen runs collapsed, leading non-letters and trailing hyphens
/// trimmed.
fn suggest_schema_name(raw: &str) -> String {
    let mut out = String::with_capacity(raw.len());
    for c in raw.to_lowercase().chars() {
        if c.is_ascii_lowercase() || c.is_ascii_digit() {
            out.push(c);
        } else if !out.ends_with('-') && !out.is_empty() {
            out.push('-');
        }
    }
    let trimmed: String = out
        .trim_matches('-')
        .chars()
        .skip_while(|c| !c.is_ascii_lowercase())
        .collect();
    let trimmed = trimmed.trim_matches('-');
    if trimmed.is_empty() {
        "my-schema".to_string()
    } else {
        trimmed.to_string()
    }
}

/// The generated `schema.yaml` — a minimal, valid manifest whose
/// comments teach each knob. Kept in one place with the example type
/// so the scaffold reads as a coherent package.
fn scaffold_manifest(name: &str) -> String {
    format!(
        r#"# Schema package scaffolded by `memstead schema new`.
# A schema package is one folder: this manifest plus one YAML file per
# entity type under types/. Re-check any time with:
#   memstead schema validate {name}

name: {name}
version: {SCAFFOLD_VERSION}

# Shown in schema catalogues (memstead_overview, the registry).
description: |
  Describe the subject this schema models and the types it declares.

# Read by agents (and humans) choosing a schema for a new mem.
when_to_use: |
  Say when this schema fits — and when an author should reach for a
  different one.

# Optional: served to agents working in a mem pinned to this schema.
system_message: |
  You are working in a graph using the {name} schema. Prefer precise
  types, link generously, and keep sections in their declared shape.

# One entry per file under types/ — `note` matches types/note.yaml.
# Add a type by adding both the file and its entry here.
types:
  - note

relationships:
  # strict: only the definitions below are legal edge types.
  # open: any UPPER_SNAKE_CASE name is accepted; definitions add weights.
  mode: strict
  # Optional relationships-level declarations (engine 0.10.0+):
  #   acyclic_sets — acyclicity over the UNION of a rel-type set, for
  #                  cycles no single rel-type contains:
  #                    acyclic_sets:
  #                      - [GROUNDS, CONCLUDES]
  #   labelling    — name the attack rel-types and the engine serves
  #                  the grounded labelling (accepted/defeated/
  #                  undecided) with evidence; optional support walk
  #                  adds chain-shape statistics.
  definitions:
    - name: PART_OF
      description: Hierarchical containment — the source is structurally part of the target.
      default_weight: 3.0
      acyclic: true
    - name: RELATES_TO
      description: General association between two entities when no sharper type fits.
      default_weight: 1.0
      # Every key below is OPTIONAL, but its default is not always the
      # permissive one — uncomment what you need.
      #
      # Per-edge `--description` text. DEFAULT IS `forbidden`: leave this
      # out and every `memstead relate ... --description` on this type is
      # REFUSED with DESCRIPTION_NOT_PERMITTED.
      # per_edge_description: optional   # forbidden | optional | required
      #
      # Restrict which types this edge may join. Omit for "any type".
      # source_types: [note]
      # target_types: [note]
      #
      # cardinality_per_source: 1   # at most one such edge per source
      # manual_authoring: false     # true = engine-emitted only
    - name: REFERENCES
      description: Soft reference. Auto-emitted from body wiki-links — never author by hand.
      default_weight: 0.5
    # Required entry — the fallback weight for any relationship not
    # listed above.
    - name: _default
      description: Fallback weight for any relationship not otherwise specified.
      default_weight: 1.0

# Body wiki-links `[[target]]` auto-emit as REFERENCES relations.
# Remove this key to make unbacked wiki-links a validation error instead.
alias_target_rel_type: REFERENCES

# Community detection (graph clustering) tuning. REQUIRED — the block
# must be present; the values below are the defaults, keep them unless
# you know why you are changing them.
community:
  resolution: 1.0
  seed: 42

# The complete key reference for schema packages — every key the loader
# accepts, with its type and default — is the meta-schema shipped in
# your workspace at `.memstead/meta-schemas/schema-manifest.schema.json`.
# This scaffold teaches by example; that file is exhaustive.
"#
    )
}

/// The generated example type. One well-commented type teaches the
/// format; the built-in catalogue (`memstead schema install default`,
/// then `.memstead/schemas/default@*/types/`) shows ten more.
fn scaffold_example_type() -> String {
    r#"# One entity type = one file. `name` must match the filename stem
# and appear in the manifest's `types:` list.
#
# Keys marked REQUIRED must be present in every type file — deleting
# one fails `memstead schema validate`. Everything else is optional.

# REQUIRED.
name: note
# REQUIRED.
description: |
  A general-purpose note — replace this with your first real type.
# REQUIRED.
when_to_use: |
  Use while sketching the schema; rename or split into sharper types
  as the domain vocabulary firms up.

# REQUIRED. Sections are the entity's markdown body. `required: true`
# sections must be present on every create.
sections:
  - key: summary
    heading: Summary
    required: true
    search_weight: 40.0
    write_rules:
      - "One or two sentences. Must stand alone in a search result."
  - key: details
    heading: Details
    required: false
    search_weight: 10.0
    # catch_all: content under unmatched headings lands here.
    catch_all: true
    write_rules:
      - "Everything beyond the summary. Bullets over prose."

# REQUIRED (the key; it may be an empty list). Typed, filterable
# frontmatter fields — beyond the built-in
# type / created_date / last_modified / tags.
# One rule for fields and sections alike: absence of `required` means
# optional. `required: true` refuses a create that leaves the field
# unset — unless a default fills it (required + default = always
# present, never refused).
metadata_fields:
  - key: status
    # required + default_value: every entity carries a status, and the
    # default means a create never has to supply one.
    required: true
    description: Lifecycle state of the note.
    field_type: string
    default_value: active
    enum_values: [active, archived]
    filterable: equality
  - key: source
    # No `required` key: optional — an entity without a source is
    # admitted. Use health_required_fields or a constraint if missing
    # values should surface as findings instead.
    description: Where the note's content came from.
    field_type: string

# REQUIRED. Search ranking: how much a title match weighs.
title_weight: 100.0
# REQUIRED. Sections included in full-text search.
text_fields: [summary, details]
# REQUIRED. Which declared relationship expresses hierarchy for this type.
hierarchy_relationship: PART_OF
# One effect only: relate refuses a self-loop (from == to) on the rel-types
# listed here. Nothing propagates; for impact propagation declare a
# `status_propagation` constraint instead.
no_self_loop_relationships: [PART_OF]
# Fields `memstead update` may touch on this type.
updatable_fields: [title, summary, details, status, tags]
# Sections the health report treats as required.
health_required_fields: [summary]
# Days without modification before health flags the entity stale.
staleness_threshold_days: 180
# Further optional type-level declarations (engine 0.10.0+), shapes in
# the authoring guide and the generated type-definition.schema.json:
#   required_outgoing  — edge obligations (cardinality, warn/block
#                        severity, optional when_field/when_value pair
#                        arming a block on a metadata enum value)
#   must_reach         — reachability obligations over a relation set
#                        (direction out/in, terminal_types, max_depth);
#                        health-sweep only, always warn
#   constraints        — the five-form vocabulary (requires_when,
#                        unique, enum_from_neighbour, status_propagation
#                        with rel_type or rel_types)
#   signals            — edge_load counts with notice/warn thresholds,
#                        served with contributors on every read
# Prose guidance served to agents writing entities of this type.
write_rules:
  - "Notes are placeholders — split recurring shapes into dedicated types."
"#
    .to_string()
}

fn validate(ctx: &CliContext, args: ValidateArgs) -> anyhow::Result<()> {
    // A directory carrying `schema-format.json` is a SEALED package —
    // installer output, not authoring input. Reporting conformance
    // errors against it sends the author fixing a file the seal wrote;
    // name what it is instead. (The validate-vs-loader tolerance
    // asymmetry itself is by design and documented; this is only the
    // recognition hint.)
    if args.path.join("schema-format.json").is_file() {
        return Err(CliError::new(
            ExitKind::Validation,
            "SCHEMA_VALIDATION_FAILED",
            format!(
                "{} is a sealed schema package (it carries `schema-format.json`, the seal \
                 marker), not authoring input — `schema validate` checks the directories you \
                 author, before sealing. Validate the package's source directory instead, or \
                 install this package directly with `memstead schema install`.",
                args.path.display(),
            ),
        )
        .with_details(json!({
            "path": args.path,
            "reason": "sealed_package",
        }))
        .into());
    }
    match memstead_schema::loader::load_schema_from_dir(&args.path)
        .and_then(|s| memstead_schema::check_section_heading_roundtrip(&s).map(|()| s))
        .and_then(|s| memstead_schema::check_reserved_metadata_keys(&s).map(|()| s))
        .and_then(|s| memstead_schema::check_section_formats(&s).map(|()| s))
    {
        Ok(schema) => {
            // Exemplar gate — same validator the install/seal path
            // runs; the authoring pre-flight must not pass a package
            // installation would refuse.
            let schema = std::sync::Arc::new(schema);
            if let Err(defect) = memstead_base::Engine::validate_schema_exemplars(&schema) {
                return Err(CliError::new(
                    ExitKind::Validation,
                    "SCHEMA_VALIDATION_FAILED",
                    format!("schema at {} is invalid: {defect}", args.path.display()),
                )
                .with_details(json!({ "path": args.path, "error": defect }))
                .into());
            }
            let (name, version) = schema.id();
            let type_count = schema.types.len();
            if ctx.json {
                print_json(&json!({
                    "ok": true,
                    "schema": format!("{name}@{version}"),
                    "types": type_count,
                    "path": args.path,
                }))?;
            } else {
                print_markdown(&format!(
                    "# Schema valid\n\n`{name}@{version}` — {type_count} type(s) at `{}`\n",
                    args.path.display(),
                ));
            }
            Ok(())
        }
        Err(e) => Err(CliError::new(
            ExitKind::Validation,
            "SCHEMA_VALIDATION_FAILED",
            format!("schema at {} is invalid: {e}", args.path.display()),
        )
        .with_details(json!({
            "path": args.path,
            "error": e.to_string(),
        }))
        .into()),
    }
}

fn migrate(ctx: &CliContext, args: MigrateArgs) -> anyhow::Result<()> {
    use memstead_schema::migrate::{MigrateError, migrate_package, next_steps, write_migration};

    let map_err = |e: MigrateError| -> anyhow::Error {
        let reason = match &e {
            MigrateError::NotAPackage { .. } => "not_a_package",
            MigrateError::SealedPackage { .. } => "sealed_package",
            MigrateError::Io { .. } | MigrateError::NotUtf8 { .. } => "io",
            MigrateError::UnmigratableValue { .. } => "unmigratable_value",
            MigrateError::PackageDoesNotLoad { .. } => "package_does_not_load",
            MigrateError::RewriteLeavesViolations { .. } => "rewrite_leaves_violations",
            MigrateError::Unfaithful { .. } => "unfaithful_rewrite",
        };
        CliError::new(
            ExitKind::Validation,
            "SCHEMA_MIGRATE_FAILED",
            format!("schema migrate at {}: {e}", args.path.display()),
        )
        .with_details(json!({ "path": args.path, "reason": reason, "error": e.to_string() }))
        .into()
    };

    let report = migrate_package(&args.path).map_err(map_err)?;
    let wrote = args.write && !report.is_noop();
    if wrote {
        write_migration(&report).map_err(map_err)?;
    }
    let steps = next_steps(&args.path, &report.schema);

    if ctx.json {
        let rewrites: Vec<serde_json::Value> = report
            .files
            .iter()
            .flat_map(|f| {
                f.rewrites.iter().map(move |r| {
                    json!({
                        "file": f.rel_path,
                        "line": r.line,
                        "key": r.key,
                        "path": r.path,
                        "action": r.action.to_string(),
                    })
                })
            })
            .collect();
        print_json(&json!({
            "ok": true,
            "schema": report.schema,
            "path": report.package,
            "dry_run": !args.write,
            "wrote": wrote,
            "rewrites": rewrites,
            "files_changed": report.changed_files().map(|f| &f.rel_path).collect::<Vec<_>>(),
            "legacy_polarity": report.legacy_polarity,
            "required_added": report.required_added.iter().map(|b| json!({
                "type": b.type_name, "field": b.field,
            })).collect::<Vec<_>>(),
            "next_steps": if report.is_noop() { Vec::new() } else { steps },
        }))?;
        return Ok(());
    }

    let mut out = String::new();
    if report.is_noop() {
        out.push_str(&format!(
            "# Schema migrate

Nothing to migrate: `{}` at `{}` carries no retired key. No file written.
",
            report.schema,
            report.package.display(),
        ));
        print_markdown(&out);
        return Ok(());
    }
    let mode = if wrote {
        "written"
    } else {
        "dry run, nothing written"
    };
    out.push_str(&format!(
        "# Schema migrate — {mode}

`{}` at `{}` — {} rewrite(s) in {} file(s)
",
        report.schema,
        report.package.display(),
        report.rewrite_count(),
        report.changed_files().count(),
    ));
    for file in report.changed_files() {
        out.push_str(&format!(
            "
## {}

",
            file.rel_path
        ));
        for r in &file.rewrites {
            out.push_str(&format!(
                "- {r}
"
            ));
        }
    }
    if report.legacy_polarity {
        out.push_str(
            "
## Polarity

This package carries the retired `optional:` key, so it was written when an absent key meant required — the meaning its sealed copies still have. The rewrite keeps that meaning: every metadata field declaring neither key gets `required: true`. Delete the line where you did not mean it.
",
        );
    }
    out.push_str(
        "
## Next steps

",
    );
    if !wrote {
        out.push_str(&format!(
            "1. Review the rewrites above, then apply them: `memstead schema migrate {} --write`
",
            args.path.display()
        ));
    }
    let offset = if wrote { 1 } else { 2 };
    for (i, step) in steps.iter().enumerate() {
        out.push_str(&format!(
            "{}. `{step}`
",
            i + offset
        ));
    }
    out.push_str(
        "
`version` was not bumped: whether a spelling migration deserves a new version is your call. Sealed copies inside mems are untouched; they keep loading as they are.
",
    );
    print_markdown(&out);
    Ok(())
}

fn install(ctx: &CliContext, args: InstallArgs) -> anyhow::Result<()> {
    let (shape, root) = ctx.workspace_shape().ok_or_else(|| {
        CliError::new(
            ExitKind::Generic,
            "NO_WORKSPACE",
            "not inside a Memstead workspace (no `.memstead/workspace.toml` in any \
             ancestor) — cd into your workspace first, or create one: memstead quickstart"
                .to_string(),
        )
    })?;
    let (schema_ref, files) = resolve_source(
        &args.source,
        ctx.workspace_shape().map(|(_, r)| r).as_deref(),
    )?;

    match shape {
        WorkspaceShape::Filesystem => {
            // Folder backend: write the package under `.memstead/schemas/`.
            // AS-GIVEN: the resolver decided the generation — an authored
            // directory source arrives stamped; a legacy builtin arrives
            // unmarked, and stamping it here would flip every bare
            // field's meaning in later seals (the plan-05 defect).
            let pkg_dir = root
                .join(".memstead")
                .join("schemas")
                .join(format!("{}@{}", schema_ref.name, schema_ref.version));
            write_package(&pkg_dir, &files)?;
            if ctx.json {
                print_json(&json!({
                    "ok": true,
                    "schema": format!("{}@{}", schema_ref.name, schema_ref.version),
                    "backend": "folder",
                    "path": pkg_dir,
                    "files": files.iter().map(|f| &f.archive_path).collect::<Vec<_>>(),
                }))?;
            } else {
                print_markdown(&format!(
                    "# Schema installed\n\n`{}@{}` → `{}` ({} file(s))\n",
                    schema_ref.name,
                    schema_ref.version,
                    pkg_dir.display(),
                    files.len(),
                ));
            }
            Ok(())
        }
        WorkspaceShape::MemRepo => install_to_git_branch(ctx, &schema_ref, &files),
    }
}

/// Install onto the git-branch backend — write the package onto the
/// workspace's `__MEMSTEAD:schemas/` ref through the engine's
/// below-boot repair surface (`memstead_git_branch::repair`), which
/// runs the same `validate_schema_package` gate and the same ref
/// writer as the booted `Engine::install_schema`. Deliberately never
/// boots the workspace: `schema install` is a named remedy for
/// boot-blocking states (an unresolvable pin whose package was never
/// installed), so it must work on exactly the workspace whose boot it
/// repairs — the plenum outage's failed escape route. Only present in
/// the `mem-repo`-featured build; the lean binary refuses (it has no
/// git-branch ref writer).
#[cfg(feature = "mem-repo")]
fn install_to_git_branch(
    ctx: &CliContext,
    schema_ref: &SchemaRef,
    files: &[memstead_schema::SchemaSourceFile],
) -> anyhow::Result<()> {
    let Some((_shape, root)) = ctx.workspace_shape() else {
        return Err(crate::setup::workspace_not_initialised_error(
            "No workspace found. Run from a directory containing `.memstead/workspace.toml`.",
        )
        .into());
    };
    let pairs: Vec<(String, Vec<u8>)> = files
        .iter()
        .map(|f| (f.archive_path.clone(), f.bytes.clone()))
        .collect();
    let commit = memstead_git_branch::repair::install_schema_below_boot(
        &root,
        &schema_ref.name,
        &schema_ref.version.to_string(),
        &pairs,
    )
    .map_err(|e| crate::setup::boot_error_to_cli(&root, e))?;
    if ctx.json {
        print_json(&json!({
            "ok": true,
            "schema": format!("{}@{}", schema_ref.name, schema_ref.version),
            "backend": "git-branch",
            "ref": format!("__MEMSTEAD:schemas/{}@{}", schema_ref.name, schema_ref.version),
            "commit": commit,
        }))?;
    } else {
        print_markdown(&format!(
            "# Schema installed\n\n`{}@{}` → `__MEMSTEAD:schemas/{}@{}` (commit `{}`)\n",
            schema_ref.name, schema_ref.version, schema_ref.name, schema_ref.version, commit,
        ));
    }
    Ok(())
}

#[cfg(not(feature = "mem-repo"))]
fn install_to_git_branch(
    _ctx: &CliContext,
    _schema_ref: &SchemaRef,
    _files: &[memstead_schema::SchemaSourceFile],
) -> anyhow::Result<()> {
    Err(CliError::new(
        ExitKind::Generic,
        "MEM_REPO_NOT_SUPPORTED",
        "this binary was built without git-branch support — use the `memstead` binary to \
         install a schema into a mem-repo workspace."
            .to_string(),
    )
    .into())
}

/// Resolve `<source>` (a path to a package dir, or a built-in name /
/// `name@version`) to its pin and the package files to write.
/// `workspace_root` (when the install runs inside one) makes the
/// authoring-provenance stamp portable: a path INSIDE the workspace is
/// stamped workspace-relative, so a clone on another machine (CI
/// included) resolves the same authoring dir instead of reporting a
/// machine-local absolute path as missing.
fn resolve_source(
    source: &str,
    workspace_root: Option<&Path>,
) -> anyhow::Result<(SchemaRef, Vec<memstead_schema::SchemaSourceFile>)> {
    let as_path = Path::new(source);
    if as_path.is_dir() {
        // Path source — validate with the engine loader before copying.
        // Includes the heading round-trip and reserved-metadata-key
        // gates: install is an authoring path, so a schema whose
        // headings cannot derive back to their keys, or that declares
        // a reserved `type`/`mem`/`id` metadata field, is refused
        // here, never sealed.
        let schema = memstead_schema::load_schema_from_dir(as_path)
            .and_then(|s| memstead_schema::check_section_heading_roundtrip(&s).map(|()| s))
            .and_then(|s| memstead_schema::check_reserved_metadata_keys(&s).map(|()| s))
            .and_then(|s| memstead_schema::check_section_formats(&s).map(|()| s))
            .map_err(|e| {
                CliError::new(
                    ExitKind::Validation,
                    "SCHEMA_VALIDATION_FAILED",
                    format!("package at {source} is invalid: {e}"),
                )
                .with_details(json!({ "path": source, "error": e.to_string() }))
            })?;
        // Exemplar gate — the same real-create-path validation the
        // mem-repo install runs via `validate_schema_package`; a
        // non-conformant exemplar never seals on either backend.
        let schema = std::sync::Arc::new(schema);
        if let Err(defect) = memstead_base::Engine::validate_schema_exemplars(&schema) {
            return Err(CliError::new(
                ExitKind::Validation,
                "SCHEMA_VALIDATION_FAILED",
                format!("package at {source} is invalid: {defect}"),
            )
            .with_details(json!({ "path": source, "error": defect }))
            .into());
        }
        let (name, version) = schema.id();
        let mut files = collect_dir_package(as_path)?;
        // Stamp the seal with its authoring provenance: the canonical
        // path this package was installed from. Detection basis for
        // the authoring-drift health axis; only path-sourced installs
        // get one (a built-in or archive source has no authoring dir).
        // The stamp stays workspace-local — package collectors and the
        // export paths exclude it by name.
        let authoring_path = as_path
            .canonicalize()
            .unwrap_or_else(|_| as_path.to_path_buf());
        // Workspace-relative when inside the workspace (portable across
        // clones — the stamped path resolves wherever the tree checks
        // out); absolute otherwise (honestly machine-pinned).
        let stamped = workspace_root
            .and_then(|root| root.canonicalize().ok())
            .and_then(|root| {
                authoring_path
                    .strip_prefix(&root)
                    .ok()
                    .map(|rel| rel.display().to_string())
            })
            .unwrap_or_else(|| authoring_path.display().to_string());
        files.push(memstead_schema::SchemaSourceFile {
            archive_path: memstead_schema::INSTALL_PROVENANCE_FILE.to_string(),
            bytes: serde_json::to_vec_pretty(&json!({
                "authoring_path": stamped,
            }))
            .expect("provenance stamp serialises"),
        });
        // A directory source is the AUTHORING tier and just validated
        // under the current language (the retired `optional:` key would
        // have refused above) — the one place a current-generation stamp
        // is verified, so it is minted here and only here. Builtin and
        // sealed sources below carry their marker (or its absence — the
        // legacy claim) as-found.
        let files = marked_package(files);
        Ok((SchemaRef::new(name, version), files))
    } else {
        // Name source — resolve against the built-in catalogue.
        let schema_ref = resolve_builtin_ref(source)?;
        let mut files =
            memstead_schema::collect_schema_source(None, None, &schema_ref).map_err(|e| {
                CliError::new(
                    ExitKind::Validation,
                    "SCHEMA_NOT_FOUND",
                    format!(
                        "could not collect source for {}: {e}",
                        schema_ref.as_display()
                    ),
                )
            })?;
        // Built-in packages may ship a `mem-template.json`; install it
        // alongside the schema so the scaffolding travels with the fork.
        if let Some(tpl) = memstead_schema::builtins::builtin_mem_template(&schema_ref.name) {
            files.push(memstead_schema::SchemaSourceFile {
                archive_path: "mem-template.json".to_string(),
                bytes: serde_json::to_vec_pretty(&tpl).unwrap_or_default(),
            });
        }
        Ok((schema_ref, files))
    }
}

/// Resolve a built-in source string (`planning` or `planning@0.1.0`) to
/// a concrete pin against the embedded catalogue.
fn resolve_builtin_ref(source: &str) -> anyhow::Result<SchemaRef> {
    let reg = memstead_schema::SchemaRegistry::builtin();
    if source.contains('@') {
        let r: SchemaRef = source.parse().map_err(|e: String| {
            CliError::new(
                ExitKind::Validation,
                "INVALID_INPUT",
                format!("invalid schema pin {source:?}: {e}"),
            )
        })?;
        if reg.get(&r.name, &r.version).is_none() {
            return Err(CliError::new(
                ExitKind::Validation,
                "SCHEMA_NOT_FOUND",
                format!(
                    "no built-in schema {source} — pass a path to install a non-built-in package"
                ),
            )
            .into());
        }
        Ok(r)
    } else {
        match reg.resolve_by_name(source) {
            Ok(Some(s)) => {
                let (n, v) = s.id();
                Ok(SchemaRef::new(n, v))
            }
            Ok(None) => Err(CliError::new(
                ExitKind::Validation,
                "SCHEMA_NOT_FOUND",
                format!(
                    "no built-in schema named {source:?} — pass a path to install a non-built-in \
                     package, or a `name@version` pin"
                ),
            )
            .into()),
            Err(e) => Err(CliError::new(
                ExitKind::Validation,
                "INVALID_INPUT",
                format!("built-in name {source:?} is ambiguous: {e}"),
            )
            .into()),
        }
    }
}

/// Collect the package files from an on-disk directory: `schema.yaml`,
/// `types/*.yaml`, and the optional `mem-template.json` / `README.md`.
fn collect_dir_package(dir: &Path) -> anyhow::Result<Vec<memstead_schema::SchemaSourceFile>> {
    use memstead_schema::SchemaSourceFile;
    let mut out = vec![SchemaSourceFile {
        archive_path: "schema.yaml".to_string(),
        bytes: std::fs::read(dir.join("schema.yaml"))?,
    }];
    let types = dir.join("types");
    if types.is_dir() {
        let mut paths: Vec<PathBuf> = std::fs::read_dir(&types)?
            .filter_map(|e| e.ok().map(|e| e.path()))
            .filter(|p| p.extension().and_then(|s| s.to_str()) == Some("yaml"))
            .collect();
        paths.sort();
        for p in paths {
            if let Some(name) = p.file_name().and_then(|s| s.to_str()) {
                out.push(SchemaSourceFile {
                    archive_path: format!("types/{name}"),
                    bytes: std::fs::read(&p)?,
                });
            }
        }
    }
    for opt in ["mem-template.json", "README.md"] {
        let p = dir.join(opt);
        if p.is_file() {
            out.push(SchemaSourceFile {
                archive_path: opt.to_string(),
                bytes: std::fs::read(&p)?,
            });
        }
    }
    Ok(out)
}

/// Append the sealed format marker to a resolved package's file list
/// if absent. For the authoring (directory-source) resolver branch
/// ONLY — the one place current-language content is verified; sealed
/// and builtin sources travel as-found (see `with_format_marker`'s
/// contract in the schema crate).
fn marked_package(
    mut files: Vec<memstead_schema::SchemaSourceFile>,
) -> Vec<memstead_schema::SchemaSourceFile> {
    let marker = memstead_schema::loader::SCHEMA_FORMAT_MARKER_FILE;
    if !files.iter().any(|f| f.archive_path == marker) {
        files.push(memstead_schema::SchemaSourceFile {
            archive_path: marker.to_string(),
            bytes: memstead_schema::loader::SCHEMA_FORMAT_MARKER_CONTENT
                .as_bytes()
                .to_vec(),
        });
    }
    files
}

/// Write the resolved package files under `pkg_dir`, creating parent
/// directories. The `# yaml-language-server:` directive on each YAML is
/// rewritten to the installed-location form so an editor resolves it
/// against the workspace's published `.memstead/meta-schemas/` rather
/// than the package source's repo-relative path. Idempotent —
/// re-running reproduces identical files.
fn write_package(
    pkg_dir: &Path,
    files: &[memstead_schema::SchemaSourceFile],
) -> anyhow::Result<()> {
    for f in files {
        let dest = pkg_dir.join(&f.archive_path);
        if let Some(parent) = dest.parent() {
            std::fs::create_dir_all(parent).map_err(|e| {
                CliError::new(
                    ExitKind::Generic,
                    "IO_ERROR",
                    format!("could not create {}: {e}", parent.display()),
                )
            })?;
        }
        let bytes = retarget_yaml_directive(&f.archive_path, &f.bytes);
        std::fs::write(&dest, &bytes).map_err(|e| {
            CliError::new(
                ExitKind::Generic,
                "IO_ERROR",
                format!("could not write {}: {e}", dest.display()),
            )
        })?;
    }
    Ok(())
}

/// The installed-location `# yaml-language-server:` directive for a
/// package member, or `None` for non-YAML members (README,
/// mem-template.json). Paths are relative to the member's location
/// under `.memstead/schemas/<name>@<version>/` and resolve to the
/// workspace's `.memstead/meta-schemas/`, published by the schema
/// subcommands (see the note in `run`), never by engine boot.
fn directive_for(archive_path: &str) -> Option<&'static str> {
    if archive_path == "schema.yaml" {
        Some("# yaml-language-server: $schema=../../meta-schemas/schema-manifest.schema.json")
    } else if archive_path.starts_with("types/") && archive_path.ends_with(".yaml") {
        Some("# yaml-language-server: $schema=../../../meta-schemas/type-definition.schema.json")
    } else {
        None
    }
}

/// Replace a leading `# yaml-language-server:` directive (or prepend one)
/// so the installed YAML points at the workspace-published meta-schema.
/// Non-YAML members and non-UTF-8 bytes pass through verbatim.
fn retarget_yaml_directive(archive_path: &str, bytes: &[u8]) -> Vec<u8> {
    let Some(directive) = directive_for(archive_path) else {
        return bytes.to_vec();
    };
    let Ok(text) = std::str::from_utf8(bytes) else {
        return bytes.to_vec();
    };
    let body = if text.starts_with("# yaml-language-server:") {
        text.split_once('\n').map(|(_, rest)| rest).unwrap_or("")
    } else {
        text
    };
    format!("{directive}\n{body}").into_bytes()
}

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

    fn ctx() -> CliContext {
        CliContext {
            json: false,
            quiet: true,
            role: Default::default(),
            identity: None,
        }
    }

    /// An unsealed copy of the current built-in is treated as fresh
    /// authoring input, and the shipped default-1.3 still speaks the
    /// retired exemplar-relation spelling (`to:` / `type:`) — its
    /// bytes are sealed, so it converges only at the family's next
    /// minted-for-meaning bump. Until then, validating the copy
    /// REFUSES with the rename pointer (install-time strict); the
    /// sealed original keeps loading through the catalogue's translate
    /// path, pinned by the loader suite. A converged copy of the same
    /// content validates cleanly — proving the refusal is the spelling
    /// alone.
    #[test]
    fn validate_builtin_default_copy_refuses_retired_exemplar_spelling() {
        let src = Path::new(env!("CARGO_MANIFEST_DIR"))
            .join("../memstead-schema/builtins/schemas/default-1.3");
        assert!(src.join("schema.yaml").is_file(), "fixture moved: {src:?}");
        let dir = tempfile::tempdir().unwrap();
        let dst = dir.path().join("authoring");
        copy_dir_without_marker(&src, &dst);
        let err = validate(&ctx(), ValidateArgs { path: dst.clone() })
            .expect_err("legacy exemplar spelling refuses as authoring input");
        assert!(
            err.to_string().contains("rel_type"),
            "refusal carries the rename pointer: {err}"
        );

        // Mechanical rename to the mutation vocabulary — then the same
        // content validates cleanly.
        for entry in std::fs::read_dir(dst.join("types")).unwrap() {
            let path = entry.unwrap().path();
            let text = std::fs::read_to_string(&path).unwrap();
            let converged = text
                .replace("\n    - to: ", "\n    - target: ")
                .replace("\n      type: ", "\n      rel_type: ");
            std::fs::write(&path, converged).unwrap();
        }
        validate(&ctx(), ValidateArgs { path: dst })
            .expect("converged default builtin content must validate");
    }

    fn copy_dir_without_marker(src: &Path, dst: &Path) {
        std::fs::create_dir_all(dst).unwrap();
        for entry in std::fs::read_dir(src).unwrap() {
            let entry = entry.unwrap();
            let name = entry.file_name();
            if name == "schema-format.json" {
                continue;
            }
            let target = dst.join(&name);
            if entry.file_type().unwrap().is_dir() {
                copy_dir_without_marker(&entry.path(), &target);
            } else {
                std::fs::copy(entry.path(), &target).unwrap();
            }
        }
    }

    /// The migrate verb on the sealed default-1.3 copy (retired exemplar
    /// spelling): the dry run writes nothing and reports the rewrites;
    /// `--write` makes the copy validate as authoring input.
    #[test]
    fn migrate_dry_run_then_write_makes_legacy_copy_validate() {
        let src = Path::new(env!("CARGO_MANIFEST_DIR"))
            .join("../memstead-schema/builtins/schemas/default-1.3");
        let dir = tempfile::tempdir().unwrap();
        let dst = dir.path().join("authoring");
        copy_dir_without_marker(&src, &dst);
        let snapshot = |d: &Path| -> Vec<(String, Vec<u8>)> {
            let mut files = Vec::new();
            for entry in std::fs::read_dir(d.join("types")).unwrap() {
                let entry = entry.unwrap();
                files.push((
                    entry.file_name().to_string_lossy().into_owned(),
                    std::fs::read(entry.path()).unwrap(),
                ));
            }
            files.sort();
            files
        };
        let before = snapshot(&dst);
        validate(&ctx(), ValidateArgs { path: dst.clone() })
            .expect_err("legacy copy refuses before migration");

        migrate(
            &ctx(),
            MigrateArgs {
                path: dst.clone(),
                write: false,
            },
        )
        .expect("dry run computes");
        assert_eq!(snapshot(&dst), before, "dry run writes nothing");

        migrate(
            &ctx(),
            MigrateArgs {
                path: dst.clone(),
                write: true,
            },
        )
        .expect("write applies");
        assert_ne!(snapshot(&dst), before, "--write rewrites the files");
        validate(&ctx(), ValidateArgs { path: dst.clone() }).expect("migrated copy validates");

        // A second run finds nothing and leaves the bytes alone.
        let after = snapshot(&dst);
        migrate(
            &ctx(),
            MigrateArgs {
                path: dst.clone(),
                write: true,
            },
        )
        .expect("noop run");
        assert_eq!(snapshot(&dst), after);
    }

    /// A sealed package is refused by the migrate verb the same way
    /// `validate` refuses it — the verb never edits a sealed copy.
    #[test]
    fn migrate_refuses_sealed_package() {
        let path = Path::new(env!("CARGO_MANIFEST_DIR"))
            .join("../memstead-schema/builtins/schemas/default-1.3");
        let err = migrate(&ctx(), MigrateArgs { path, write: true })
            .expect_err("sealed package must refuse");
        let cli = err.downcast_ref::<CliError>().expect("typed CLI error");
        assert_eq!(cli.code, "SCHEMA_MIGRATE_FAILED");
        assert_eq!(cli.details.as_ref().unwrap()["reason"], "sealed_package");
    }

    /// A directory carrying `schema-format.json` — a sealed package —
    /// is named as such instead of being conformance-checked as
    /// authoring input. The shipped builtin is exactly that shape.
    #[test]
    fn validate_names_sealed_package() {
        let path = Path::new(env!("CARGO_MANIFEST_DIR"))
            .join("../memstead-schema/builtins/schemas/default-1.3");
        let err = validate(&ctx(), ValidateArgs { path }).expect_err("sealed package must refuse");
        let cli = err
            .downcast_ref::<CliError>()
            .expect("error is a typed CliError");
        assert_eq!(cli.code, "SCHEMA_VALIDATION_FAILED");
        assert!(
            cli.message.contains("sealed schema package"),
            "message names the sealed package: {}",
            cli.message,
        );
        assert_eq!(
            cli.details.as_ref().unwrap()["reason"],
            json!("sealed_package"),
        );
    }

    /// A malformed `schema.yaml` refuses with the typed
    /// `SCHEMA_VALIDATION_FAILED` code carrying the path in `details`.
    #[test]
    fn validate_rejects_malformed_schema_with_typed_code() {
        let dir = tempfile::tempdir().unwrap();
        std::fs::write(dir.path().join("schema.yaml"), "name: [unterminated\n").unwrap();
        let err = validate(
            &ctx(),
            ValidateArgs {
                path: dir.path().to_path_buf(),
            },
        )
        .expect_err("malformed schema must refuse");
        let cli = err
            .downcast_ref::<CliError>()
            .expect("error is a typed CliError");
        assert_eq!(cli.code, "SCHEMA_VALIDATION_FAILED");
        assert_eq!(cli.kind, ExitKind::Validation);
        assert_eq!(
            cli.details.as_ref().unwrap()["path"],
            json!(dir.path()),
            "details echoes the offending path",
        );
    }

    /// A read resolves a bare name to the newest generation and a pin
    /// as given; an unknown name refuses typed with the roster.
    #[test]
    fn resolve_builtin_read_ref_defaults_bare_names_to_newest() {
        let newest = resolve_builtin_read_ref("planning").expect("bare planning reads");
        let mut all = memstead_schema::SchemaRegistry::builtin().available_versions("planning");
        all.sort();
        assert_eq!(Some(&newest.version), all.last());
        let pinned = resolve_builtin_read_ref("planning@0.1.0").expect("pin reads");
        assert_eq!(pinned.version.to_string(), "0.1.0");
        let err = resolve_builtin_read_ref("not-a-builtin").expect_err("unknown refuses");
        let cli = err.downcast_ref::<CliError>().unwrap();
        assert_eq!(cli.code, "SCHEMA_NOT_FOUND");
        assert!(
            cli.message.contains("planning"),
            "names the roster: {}",
            cli.message
        );
    }

    /// A bare built-in name resolves to its concrete pin; an explicit
    /// `name@version` is accepted; an unknown name refuses typed.
    #[test]
    fn resolve_builtin_ref_handles_name_pin_and_unknown() {
        // Every built-in ships multiple versions since the plan-06
        // vocabulary bump, so bare names are ambiguous by design and
        // pins resolve explicitly.
        let bare = resolve_builtin_ref("software@0.2.0").expect("software pin resolves");
        assert_eq!(bare.name, "software");
        let pinned = resolve_builtin_ref("planning@0.1.0").expect("explicit pin resolves");
        assert_eq!(pinned.name, "planning");
        assert_eq!(pinned.version.to_string(), "0.1.0");
        resolve_builtin_ref("planning@0.2.0").expect("bumped pin resolves");
        resolve_builtin_ref("planning").expect_err("bare planning is ambiguous");
        let err = resolve_builtin_ref("not-a-builtin").expect_err("unknown name refuses");
        assert_eq!(
            err.downcast_ref::<CliError>().unwrap().code,
            "SCHEMA_NOT_FOUND",
        );
    }

    /// Installing a built-in by name collects its schema files *and*
    /// its `mem-template.json`.
    #[test]
    fn resolve_source_for_builtin_includes_schema_and_template() {
        let (schema_ref, files) =
            resolve_source("planning@0.1.0", None).expect("planning source collects");
        assert_eq!(schema_ref.name, "planning");
        let paths: Vec<&str> = files.iter().map(|f| f.archive_path.as_str()).collect();
        assert!(paths.contains(&"schema.yaml"), "got {paths:?}");
        assert!(
            paths.contains(&"mem-template.json"),
            "built-in install must carry the mem-template.json, got {paths:?}",
        );
    }

    /// `collect_dir_package` + `write_package` round-trip a package
    /// (schema.yaml + types + template) onto disk verbatim.
    #[test]
    fn collect_and_write_package_round_trips() {
        let src = tempfile::tempdir().unwrap();
        std::fs::create_dir_all(src.path().join("types")).unwrap();
        std::fs::write(src.path().join("schema.yaml"), b"name: x\n").unwrap();
        std::fs::write(src.path().join("types/doc.yaml"), b"name: doc\n").unwrap();
        std::fs::write(src.path().join("mem-template.json"), b"{}\n").unwrap();

        let files = collect_dir_package(src.path()).unwrap();
        let dest = tempfile::tempdir().unwrap();
        let pkg = dest.path().join("x@0.1.0");
        write_package(&pkg, &files).unwrap();

        // YAML members gain the installed-location directive; bodies and
        // non-YAML members (mem-template.json) are preserved.
        let schema = std::fs::read_to_string(pkg.join("schema.yaml")).unwrap();
        assert_eq!(
            schema,
            "# yaml-language-server: $schema=../../meta-schemas/schema-manifest.schema.json\nname: x\n",
        );
        let doc = std::fs::read_to_string(pkg.join("types/doc.yaml")).unwrap();
        assert_eq!(
            doc,
            "# yaml-language-server: $schema=../../../meta-schemas/type-definition.schema.json\nname: doc\n",
        );
        assert_eq!(
            std::fs::read(pkg.join("mem-template.json")).unwrap(),
            b"{}\n"
        );
        // Idempotent: a second write reproduces identical files.
        write_package(&pkg, &files).unwrap();
        assert_eq!(
            std::fs::read_to_string(pkg.join("schema.yaml")).unwrap(),
            schema
        );
    }

    /// The directive retarget replaces an existing leading directive (it
    /// does not stack) and prepends one when absent; non-YAML and
    /// non-UTF-8 members pass through.
    #[test]
    fn retarget_yaml_directive_replaces_or_prepends() {
        // Existing (repo-relative) directive is replaced, body kept.
        let existing = b"# yaml-language-server: $schema=../../../generated/schema-manifest.schema.json\nname: y\n";
        let out = String::from_utf8(retarget_yaml_directive("schema.yaml", existing)).unwrap();
        assert_eq!(
            out,
            "# yaml-language-server: $schema=../../meta-schemas/schema-manifest.schema.json\nname: y\n",
        );
        // Absent directive is prepended.
        let bare = retarget_yaml_directive("types/t.yaml", b"name: t\n");
        assert_eq!(
            String::from_utf8(bare).unwrap(),
            "# yaml-language-server: $schema=../../../meta-schemas/type-definition.schema.json\nname: t\n",
        );
        // Non-YAML members untouched.
        assert_eq!(retarget_yaml_directive("README.md", b"# hi\n"), b"# hi\n");
    }
}