zebflow 0.9.0

//! Pipeline node interfaces and built-in node implementations.
//!
//! This module is the **single source of truth** for how nodes are authored in Zebflow.
//! Every built-in node lives under [`basic`], in the folder of its DSL family:
//! `n.fs.save` is `basic/fs/save.rs`, `n.kv.get` is `basic/kv/get.rs`, and a
//! family with one node keeps it in that folder's `mod.rs`. Helpers several
//! families use live in [`shared`]. This doc is the living specification —
//! read it before creating or modifying any node.
//!
//! ---
//!
//! # Node Anatomy
//!
//! A complete node module exposes exactly four things:
//!
//! ```text
//! basic/<family>/<node>.rs
//! ├── pub fn definition() -> NodeDefinition   ← kind-level contract (schemas, flags, docs)
//! ├── pub struct Config { ... }               ← typed config with JsonSchema derive
//! ├── pub struct Node { ... }                 ← runtime instance
//! └── impl NodeHandler for Node               ← execution logic
//! ```
//!
//! Then add it to the family's `definitions()` in `basic/<family>/mod.rs`, which
//! `basic/mod.rs → builtin_node_definitions()` reads.  That's it.
//!
//! ---
//!
//! # 1. `definition()` — the node contract
//!
//! `definition()` is the **most important function in a node module**.  It populates
//! [`NodeDefinition`](crate::pipeline::NodeDefinition) — the kind-level contract that
//! drives everything: UI dialogs, DSL parsing, help text, LLM context, API docs.
//!
//! ```rust,ignore
//! pub fn definition() -> NodeDefinition {
//!     NodeDefinition {
//!         kind: NODE_KIND.to_string(),           // "n.category.action"
//!         title: "Human Title".to_string(),      // shown in UI catalogs
//!         description: "...".to_string(),        // shown in UI + fed to LLM
//!         config_schema: serde_json::json!({ ... }),  // or schemars::schema_for!(Config)
//!         input_schema:  serde_json::json!({ ... }),  // payload shape consumed
//!         output_schema: serde_json::json!({ ... }),  // payload shape produced
//!         input_pins:  vec!["in".to_string()],   // empty for trigger nodes
//!         output_pins: vec!["out".to_string()],  // ["out","error"], ["true","false"], etc.
//!         dsl_flags: vec![ ... ],                // see §4 below
//!         script_available: false,               // see §5 below
//!         script_bridge: None,
//!         ai_tool: Default::default(),           // see §6 below
//!     }
//! }
//! ```
//!
//! ## Module-level doc = node documentation
//!
//! Write the module `//!` doc block as if it IS the node's reference page.  Include:
//!
//! - One-line summary of what the node does.
//! - **Pipeline position** — is it a trigger? terminal? middle? never standalone?
//! - **User-facing config table** — only fields the user/agent sets. Never list
//!   platform-injected fields (markup, route, etc.).
//! - **Config schema** — the JSON Schema block so LLMs can read it directly.
//! - **Studio UI hint** — ASCII mockup of how the node dialog should look.
//! - **Input/output payload examples** — what flows in, what flows out.
//! - **DSL examples** — show the pipe chain with all flags.
//! - **Project-level settings** if the node is affected by `zebflow.yaml` settings.
//!
//! The module doc becomes the `description` field at `/docs/node` — **no separate
//! markdown file needed**.  The source IS the docs.
//!
//! ---
//!
//! # 2. Config struct
//!
//! ```rust,ignore
//! use serde::{Deserialize, Serialize};
//!
//! #[derive(Debug, Clone, Serialize, Deserialize, Default)]
//! pub struct Config {
//!     /// Required: path to the TSX page file relative to templates/.
//!     pub template_path: String,
//!     /// Optional: comma-separated external script URLs.
//!     #[serde(default)]
//!     pub load_scripts: String,
//! }
//! ```
//!
//! Rules:
//! - Field names are `snake_case` — they are the config JSON keys.
//! - Required fields have no `#[serde(default)]`.
//! - Optional fields have `#[serde(default)]` and use `Option<T>` or a defaultable type.
//! - Write a doc comment on every field — it becomes the `description` in `config_schema`.
//! - Platform-injected fields (markup, route, internal ids) live in the struct but must
//!   NOT appear in `dsl_flags` or `config_schema.required`.
//!
//! ---
//!
//! # 3. Config schema
//!
//! Populate `NodeDefinition::config_schema` by hand with a JSON Schema object:
//!
//! ```rust,ignore
//! config_schema: serde_json::json!({
//!     "type": "object",
//!     "required": ["template_path"],
//!     "properties": {
//!         "template_path": {
//!             "type": "string",
//!             "description": "TSX file relative to templates/. Example: pages/blog-home."
//!         },
//!         "load_scripts": {
//!             "type": "string",
//!             "description": "Comma-separated external script URLs. Each must match allow_list."
//!         }
//!     }
//! }),
//! ```
//!
//! Consumers of `config_schema`:
//! - **Studio UI** — renders form fields from `properties` without any TypeScript hardcoding.
//! - **Register validator** — checks incoming node configs before saving to disk.
//! - **LLM agent** — reads schema to know what to put in node config when authoring pipelines.
//! - **`/docs/node`** — exposes schema in the API contract document.
//!
//! ---
//!
//! # 4. DSL flags
//!
//! Every user-facing config field gets a [`DslFlag`](crate::pipeline::model::DslFlag) entry.
//!
//! ```rust,ignore
//! use crate::pipeline::model::{DslFlag, DslFlagKind};
//!
//! dsl_flags: vec![
//!     DslFlag {
//!         flag: "--template-path".to_string(),
//!         config_key: "template_path".to_string(),
//!         description: "TSX page file relative to templates/, e.g. pages/blog-home.".to_string(),
//!         kind: DslFlagKind::Scalar,
//!         required: true,
//!     },
//!     DslFlag {
//!         flag: "--load-scripts".to_string(),
//!         config_key: "load_scripts".to_string(),
//!         description: "Comma-separated external script URLs. Each must match allow_list.".to_string(),
//!         kind: DslFlagKind::CommaSeparatedList,
//!         required: false,
//!     },
//! ],
//! ```
//!
//! ## Universal parsing rule
//!
//! The DSL parser applies: `--flag-name` → `flag_name` (replace `-` with `_`) for ALL flags,
//! declared or not.  `dsl_flags` exists only for documentation, help text, required-flag
//! validation, and LLM context.  **New nodes work in the DSL without touching `parser.rs`.**
//!
//! ## Flag kind reference
//!
//! | `DslFlagKind` | DSL syntax | Config result |
//! |---|---|---|
//! | `Scalar` | `--key value` | `"value"` |
//! | `CommaSeparatedList` | `--key a,b,c` or `--key a --key b` | `["a","b"]` |
//! | `RepeatedList` | `--key a --key b` or `--key a,b` | `["a","b"]` |
//! | `Bool` | `--silent` (no value) | `true` |
//!
//! Body content (after ` -- `) is always captured separately as the node body string.
//!
//! ## Naming convention
//!
//! Flag name = `--kebab-case` of the `snake_case` config key.  Always.  No abbreviations,
//! no aliases.  `--credential-id` not `--cred`.  `--template-path` not `--template`.
//! This makes flags derivable from the schema property names — LLMs can infer them.
//!
//! ---
//!
//! # 5. Script bridge — declared, and built by nothing
//!
//! Write `script_available: false` and `script_bridge: None`.  Every node does.
//!
//! A script cannot reach a node handler.  The `n` object handed to an `n.script` body
//! is assembled by `build_capabilities_expr` in `language/engines/deno_sandbox/pool.rs`
//! from pure `time` and `math` helpers; the sandbox exposes two host ops, neither of
//! which dispatches a node, and hides `Deno.core` from user code.  Declaring `true`
//! would grant nothing and would put a false "n.script access" claim in the node
//! catalog, which is why no definition does.
//!
//! The two fields survive because `NodeDefinition` is a frozen `zebflow.com/v1`
//! contract; see `docs/contracts/kinds/node-definition/README.md`.  Building the bridge
//! is a feature, and the egress rules in `docs/contracts/confinement.md` §3 have to
//! cover it before it exists.
//!
//! ---
//!
//! # 6. AI tool — invokable by LLM agents
//!
//! Set `ai_tool.registered: true` to register the node as a callable tool for Zebtune
//! and other LLM agents running inside the platform:
//!
//! ```rust,ignore
//! ai_tool: NodeAiToolDefinition {
//!     registered: true,
//!     tool_name: "query_database".to_string(),
//!     tool_description: "Run a SQL query against a project database connection and return rows.".to_string(),
//!     tool_input_schema: serde_json::json!({
//!         "type": "object",
//!         "required": ["credential_id", "query"],
//!         "properties": {
//!             "credential_id": { "type": "string", "description": "Database connection slug." },
//!             "query":         { "type": "string", "description": "SQL query to execute." }
//!         }
//!     }),
//! },
//! ```
//!
//! `tool_input_schema` is what the LLM sees — it may differ from `config_schema` if the
//! tool accepts a simpler or different interface than the full node config.
//!
//! ---
//!
//! # 7. `NodeHandler` implementation
//!
//! ```rust,ignore
//! use async_trait::async_trait;
//! use crate::pipeline::nodes::{NodeHandler, NodeExecutionInput, NodeExecutionOutput};
//! use crate::pipeline::PipelineError;
//!
//! pub struct Node { config: Config }
//!
//! impl Node {
//!     pub fn new(config: Config) -> Self { Self { config } }
//! }
//!
//! #[async_trait]
//! impl NodeHandler for Node {
//!     fn kind(&self) -> &'static str { NODE_KIND }
//!
//!     fn input_pins(&self)  -> &'static [&'static str] { &["in"] }
//!     fn output_pins(&self) -> &'static [&'static str] { &["out"] }
//!
//!     async fn execute_async(
//!         &self,
//!         input: NodeExecutionInput,
//!     ) -> Result<NodeExecutionOutput, PipelineError> {
//!         // Read config: self.config.*
//!         // Read upstream payload: input.payload
//!         // Read owner/project: input.metadata.get("owner")
//!         // Read request id: input.metadata.get("request_id")
//!         Ok(NodeExecutionOutput {
//!             output_pins: vec!["out".to_string()],
//!             payload: serde_json::json!({ "result": "..." }),
//!             trace: vec![format!("n.my.node: done")],
//!         })
//!     }
//! }
//! ```
//!
//! Rules:
//! - Always include at least one `trace` entry — it shows in pipeline run logs.
//! - Emit `PipelineError::new("FW_MY_NODE_CODE", "message")` for recoverable errors.
//! - Use `output_pins` to control which downstream edges fire — only listed pins propagate.
//! - For conditional branching, emit `vec!["true"]` or `vec!["false"]` selectively.
//!
//! ---
//!
//! # 8. Registration
//!
//! Declare the file and add your `definition()` call to the family list in
//! `src/pipeline/nodes/basic/<family>/mod.rs` (a new family is a new folder,
//! declared in `basic/mod.rs` and appended to `family_definitions()` there):
//!
//! ```rust,ignore
//! pub mod my_node;
//!
//! pub fn definitions() -> Vec<NodeDefinition> {
//!     vec![
//!         // ... the family's other nodes ...
//!         my_node::definition(),
//!     ]
//! }
//! ```
//!
//! And add the dispatch arm in `src/pipeline/engines/basic.rs` inside the `match node.kind`
//! block so the engine knows how to build and execute your node instance.
//!
//! ---
//!
//! # Complete example
//!
//! ```rust,ignore
//! //! `n.example.echo` — passes the upstream payload through unchanged with a tag.
//! //!
//! //! # Pipeline position
//! //! Middleware node. Always between a trigger and a terminal.
//! //!
//! //! # User-facing config
//! //! | Field | Type | Required | Description |
//! //! |---|---|---|---|
//! //! | `tag` | string | ✓ | Label injected into the payload as `echo_tag` |
//! //!
//! //! # DSL
//! //! ```text
//! //! | n.trigger.webhook --path /ping
//! //! | n.example.echo --tag hello
//! //! ```
//!
//! use async_trait::async_trait;
//! use serde::{Deserialize, Serialize};
//! use serde_json::json;
//! use crate::pipeline::{PipelineError, NodeDefinition};
//! use crate::pipeline::model::{DslFlag, DslFlagKind};
//! use crate::pipeline::nodes::{NodeHandler, NodeExecutionInput, NodeExecutionOutput};
//!
//! pub const NODE_KIND: &str = "n.example.echo";
//!
//! pub fn definition() -> NodeDefinition {
//!     NodeDefinition {
//!         kind: NODE_KIND.to_string(),
//!         title: "Echo".to_string(),
//!         description: "Passes payload through, injecting echo_tag from config.".to_string(),
//!         config_schema: json!({
//!             "type": "object",
//!             "required": ["tag"],
//!             "properties": {
//!                 "tag": { "type": "string", "description": "Label to inject as echo_tag." }
//!             }
//!         }),
//!         input_schema:  json!({ "type": "object" }),
//!         output_schema: json!({ "type": "object", "properties": { "echo_tag": { "type": "string" } } }),
//!         input_pins:  vec!["in".to_string()],
//!         output_pins: vec!["out".to_string()],
//!         dsl_flags: vec![
//!             DslFlag {
//!                 flag: "--tag".to_string(),
//!                 config_key: "tag".to_string(),
//!                 description: "Label injected into the output payload as echo_tag.".to_string(),
//!                 kind: DslFlagKind::Scalar,
//!                 required: true,
//!             },
//!         ],
//!         script_available: false,
//!         script_bridge: None,
//!         ai_tool: Default::default(),
//!     }
//! }
//!
//! #[derive(Debug, Clone, Serialize, Deserialize, Default)]
//! pub struct Config { pub tag: String }
//!
//! pub struct Node { config: Config }
//! impl Node { pub fn new(config: Config) -> Self { Self { config } } }
//!
//! #[async_trait]
//! impl NodeHandler for Node {
//!     fn kind(&self) -> &'static str { NODE_KIND }
//!     fn input_pins(&self)  -> &'static [&'static str] { &["in"] }
//!     fn output_pins(&self) -> &'static [&'static str] { &["out"] }
//!     async fn execute_async(&self, input: NodeExecutionInput) -> Result<NodeExecutionOutput, PipelineError> {
//!         let mut payload = input.payload;
//!         payload["echo_tag"] = json!(self.config.tag.clone());
//!         Ok(NodeExecutionOutput {
//!             output_pins: vec!["out".to_string()],
//!             payload,
//!             trace: vec![format!("n.example.echo: tag={}", self.config.tag)],
//!         })
//!     }
//! }
//! ```

pub mod basic;
mod interface;
pub mod shared;

pub use interface::{NodeExecutionInput, NodeExecutionOutput, NodeHandler};

use std::collections::HashSet;

use crate::pipeline::model::{
    DslFlagKind, LayoutItem, NodeCapability, NodeDefinition, NodeFieldType,
};

/// Returns all built-in node definitions.
pub fn builtin_node_definitions() -> Vec<crate::pipeline::NodeDefinition> {
    basic::builtin_node_definitions()
}

/// What every native node kind can reach, keyed by kind.
///
/// This is the compiled-in half of the answer the package safety review needs,
/// and it is a constant of the build: nothing installed can add to it, remove
/// from it, or disagree with it. Bundle-provided kinds are resolved separately,
/// by deriving them from the nodes they compose.
///
/// A kind with an empty set is present and empty, never absent. "This node
/// reaches nothing" and "nobody asked this node" are different answers and the
/// review reports them differently.
pub fn native_node_capabilities()
-> &'static std::collections::BTreeMap<String, std::collections::BTreeSet<NodeCapability>> {
    static NATIVE: std::sync::LazyLock<
        std::collections::BTreeMap<String, std::collections::BTreeSet<NodeCapability>>,
    > = std::sync::LazyLock::new(|| {
        builtin_node_definitions()
            .into_iter()
            .map(|definition| {
                (
                    definition.kind,
                    definition.capabilities.into_iter().collect(),
                )
            })
            .collect()
    });
    &NATIVE
}

fn validate_layout_item(
    item: &LayoutItem,
    field_names: &HashSet<String>,
    errors: &mut Vec<String>,
) {
    match item {
        LayoutItem::Field(name) => {
            if !field_names.contains(name) {
                errors.push(format!("layout references unknown field '{}'", name));
            }
        }
        LayoutItem::Row { row } => {
            if row.is_empty() {
                errors.push("layout row must not be empty".to_string());
            }
            for child in row {
                validate_layout_item(child, field_names, errors);
            }
        }
        LayoutItem::Col { col } => {
            if col.is_empty() {
                errors.push("layout col must not be empty".to_string());
            }
            for child in col {
                validate_layout_item(child, field_names, errors);
            }
        }
    }
}

/// Validates that a node definition is complete enough to serve as API and docs source.
pub fn validate_node_definition_contract(def: &NodeDefinition) -> Result<(), Vec<String>> {
    let mut errors = Vec::new();
    let kind = def.kind.trim();
    if kind.is_empty() {
        errors.push("kind must not be empty".to_string());
    } else if !kind.starts_with("n.") {
        errors.push(format!("kind '{}' must start with 'n.'", kind));
    }
    if def.title.trim().is_empty() {
        errors.push("title must not be empty".to_string());
    }
    if def.description.trim().is_empty() {
        errors.push("description must not be empty".to_string());
    }

    for (pin_role, pins) in [
        ("input", def.input_pins.as_slice()),
        ("output", def.output_pins.as_slice()),
    ] {
        for pin in pins {
            let trimmed = pin.trim();
            if trimmed.is_empty() {
                errors.push(format!("{} pin must not be empty", pin_role));
            } else if trimmed != pin || trimmed.contains(char::is_whitespace) {
                errors.push(format!(
                    "{} pin '{}' must be a compact token",
                    pin_role, pin
                ));
            }
        }
    }

    let mut flag_keys = HashSet::new();
    for flag in &def.dsl_flags {
        if !flag.flag.starts_with("--") || flag.flag.trim() != flag.flag {
            errors.push(format!(
                "DSL flag '{}' must start with -- and have no surrounding space",
                flag.flag
            ));
        }
        if flag.config_key.trim().is_empty() {
            errors.push(format!(
                "DSL flag '{}' config_key must not be empty",
                flag.flag
            ));
        } else {
            flag_keys.insert(flag.config_key.clone());
        }
        if flag.description.trim().is_empty() {
            errors.push(format!(
                "DSL flag '{}' description must not be empty",
                flag.flag
            ));
        }
    }

    let mut field_names = HashSet::new();
    for field in &def.fields {
        if field.name.trim().is_empty() {
            errors.push("field name must not be empty".to_string());
            continue;
        }
        if !field_names.insert(field.name.clone()) {
            errors.push(format!("field '{}' is declared more than once", field.name));
        }
        if field.label.trim().is_empty() {
            errors.push(format!("field '{}' label must not be empty", field.name));
        }
        if !matches!(field.field_type, NodeFieldType::Section)
            && field.help.as_deref().unwrap_or("").trim().is_empty()
        {
            errors.push(format!("field '{}' help must not be empty", field.name));
        }
    }

    for item in &def.layout {
        validate_layout_item(item, &field_names, &mut errors);
    }

    if let Some(properties) = def
        .config_schema
        .get("properties")
        .and_then(|value| value.as_object())
    {
        for key in properties.keys() {
            if key == "title" || key == "timeout_secs" {
                continue;
            }
            if !field_names.contains(key) && !flag_keys.contains(key) {
                errors.push(format!(
                    "config property '{}' must be documented by a field or DSL flag",
                    key
                ));
            }
        }
    }

    if errors.is_empty() {
        Ok(())
    } else {
        Err(errors)
    }
}

pub fn format_node_definition_markdown(def: &NodeDefinition) -> String {
    let mut s = String::new();
    s.push_str(&format!("### `{}` — {}\n\n", def.kind, def.title));
    s.push_str(def.description.trim());
    s.push_str("\n\n");
    let ip = if def.input_pins.is_empty() {
        "*(none — trigger / entry)*".to_string()
    } else {
        format!("`{}`", def.input_pins.join("`, `"))
    };
    let op = if def.output_pins.is_empty() {
        "*(dynamic / graph-defined)*".to_string()
    } else {
        format!("`{}`", def.output_pins.join("`, `"))
    };
    s.push_str(&format!(
        "- **Input pins:** {ip}\n- **Output pins:** {op}\n\n"
    ));
    if !def.dsl_flags.is_empty() {
        s.push_str("| DSL flag | Config key | Required | Kind | Description |\n");
        s.push_str("|----------|------------|----------|------|-------------|\n");
        for f in &def.dsl_flags {
            let req = if f.required { "yes" } else { "no" };
            let kind_s = match f.kind {
                DslFlagKind::Scalar => "scalar",
                DslFlagKind::CommaSeparatedList => "comma-list",
                DslFlagKind::RepeatedList => "repeated-list",
                DslFlagKind::Bool => "bool",
                DslFlagKind::KeyValuePairs => "key-value-pairs",
                DslFlagKind::SchemaField => "schema-field",
            };
            let desc = f.description.replace('|', "\\|");
            s.push_str(&format!(
                "| `{}` | `{}` | {} | {} | {} |\n",
                f.flag, f.config_key, req, kind_s, desc
            ));
        }
        s.push_str("\n");
    }
    if !def.input_schema.is_null() {
        s.push_str("**Input payload (schema):**\n```json\n");
        if let Ok(pretty) = serde_json::to_string_pretty(&def.input_schema) {
            s.push_str(&pretty);
        }
        s.push_str("\n```\n\n");
    }
    if !def.output_schema.is_null() {
        s.push_str("**Output payload (schema):**\n```json\n");
        if let Ok(pretty) = serde_json::to_string_pretty(&def.output_schema) {
            s.push_str(&pretty);
        }
        s.push_str("\n```\n\n");
    }
    for example in &def.examples {
        let title = if example.title.is_empty() { "Example" } else { example.title.as_str() };
        s.push_str(&format!("**{title}:**\n"));
        if !example.dsl.is_empty() {
            s.push_str(&format!("```\n| {}\n```\n", example.dsl));
        }
        if !example.description.is_empty() {
            s.push_str(example.description.trim());
            s.push_str("\n");
        }
        if !example.input.is_null() || !example.output.is_null() {
            s.push_str("```json\n");
            if !example.input.is_null() {
                s.push_str(&format!("// input\n{}\n", example.input));
            }
            if !example.output.is_null() {
                s.push_str(&format!("// output\n{}\n", example.output));
            }
            s.push_str("```\n");
        }
        s.push_str("\n");
    }
    s.push_str(&format!(
        "**MCP:** `help_nodes` with `kind=\"{}\"` for this section only.\n\n",
        def.kind
    ));
    s
}

pub fn kind_query_matches_def(def: &NodeDefinition, query: &str) -> bool {
    let q = query.trim();
    if q.is_empty() {
        return false;
    }
    let kn = def.kind.to_lowercase();
    let qn = q.to_lowercase();
    if kn == qn {
        return true;
    }
    if qn.starts_with("n.") {
        return kn == qn;
    }
    if kn == format!("n.{qn}") {
        return true;
    }
    kn.strip_prefix("n.").is_some_and(|tail| tail == qn)
}

/// One node section — same source as [`builtin_node_definitions`].
pub fn node_markdown_by_kind_query(query: &str) -> Option<String> {
    basic::builtin_node_definitions()
        .into_iter()
        .find(|d| kind_query_matches_def(d, query))
        .map(|d| format_node_definition_markdown(&d))
}

/// Full catalog for `help_pipeline` / `help_nodes` — generated from Rust `definition()`, not hand-written markdown.
pub fn builtin_nodes_markdown_reference() -> String {
    let mut s = String::from(
        "## Node kinds (live — from `builtin_node_definitions()`)\n\n\
         This block matches the pipeline editor / node API: titles, descriptions, pins, DSL flags, and input/output schemas.\n\n\
         - **Full catalog:** `help_nodes` with no `kind` (same as this section).\n\
         - **One kind:** `help_nodes` with `kind=\"n.script\"` (or `script`, `trigger.webhook`, etc.).\n\n\
         ---\n\n",
    );
    for def in basic::builtin_node_definitions() {
        s.push_str(&format_node_definition_markdown(&def));
        s.push_str("---\n\n");
    }
    s
}

#[cfg(test)]
mod tests {
    #[test]
    fn native_node_definitions_have_complete_contracts() {
        let mut failures = Vec::new();
        for def in super::builtin_node_definitions() {
            if let Err(error) = crate::contracts::kinds::validate_normalized_node_definition(&def) {
                failures.push(format!("{}: {}", def.kind, error));
            }
        }
        assert!(
            failures.is_empty(),
            "incomplete node definitions:\n{}",
            failures.join("\n")
        );
    }

    /// The canvas-preview flags are injected into every node, the two table
    /// nodes included: their row sample moved to `--preview-rows` /
    /// `preview_rows`, so `--preview` means one thing everywhere.
    #[test]
    fn every_node_takes_the_canvas_preview_flags() {
        let defs = super::builtin_node_definitions();
        let flags_of = |kind: &str| {
            defs.iter()
                .find(|d| d.kind == kind)
                .unwrap_or_else(|| panic!("{kind} missing"))
                .dsl_flags
                .iter()
                .map(|f| (f.flag.clone(), f.config_key.clone()))
                .collect::<Vec<_>>()
        };

        for kind in ["n.table.query", "n.table.convert", "n.script"] {
            let flags = flags_of(kind);
            assert!(
                flags.iter().any(|(flag, key)| flag == "--preview" && key == "preview.out"),
                "{kind} takes --preview: {flags:?}"
            );
            assert!(
                flags.iter().any(|(flag, key)| flag == "--preview-in" && key == "preview.in"),
                "{kind} takes --preview-in: {flags:?}"
            );
            assert!(
                !flags.iter().any(|(_, key)| key == "preview"),
                "{kind} no longer stores anything flat at `preview`: {flags:?}"
            );
        }
        for kind in ["n.table.query", "n.table.convert"] {
            let flags = flags_of(kind);
            assert!(
                flags.iter().any(|(flag, key)| flag == "--preview-rows" && key == "preview_rows"),
                "{kind} samples rows with --preview-rows: {flags:?}"
            );
        }
    }

    /// The source layout is the catalogue's family list: one folder per DSL
    /// family under `basic/`, one file per node, nothing loose beside
    /// `basic/mod.rs`. A family with one node keeps it in that folder's
    /// `mod.rs`; helpers several families share live in `nodes/shared/`.
    #[test]
    fn every_node_file_lives_in_its_family_folder() {
        let basic = std::path::Path::new(env!("CARGO_MANIFEST_DIR")).join("src/pipeline/nodes/basic");
        let families: std::collections::BTreeSet<String> = super::builtin_node_definitions()
            .iter()
            .filter_map(|def| def.kind.strip_prefix("n."))
            .map(|rest| rest.split('.').next().unwrap_or(rest).to_string())
            .collect();

        let mut failures = Vec::new();
        let mut folders = std::collections::BTreeSet::new();
        for entry in std::fs::read_dir(&basic).expect("basic/ is readable") {
            let entry = entry.expect("entry");
            let name = entry.file_name().to_string_lossy().to_string();
            if entry.path().is_dir() {
                if !families.contains(&name) {
                    failures.push(format!(
                        "basic/{name}/ is not a DSL family of the catalogue (families: {})",
                        families.iter().cloned().collect::<Vec<_>>().join(", ")
                    ));
                }
                folders.insert(name);
            } else if name != "mod.rs" {
                failures.push(format!(
                    "basic/{name} sits beside mod.rs; a node file is basic/<family>/<node>.rs"
                ));
            }
        }
        for family in &families {
            if !folders.contains(family) {
                failures.push(format!("family `{family}` has nodes in the catalogue but no basic/{family}/ folder"));
            }
        }
        assert!(failures.is_empty(), "node layout:\n{}", failures.join("\n"));
    }

    #[test]
    fn embedded_official_node_definitions_have_complete_contracts() {
        let mut failures = Vec::new();
        for def in crate::platform::services::node_registry::NodeRegistryService::embedded_official_definitions() {
            if let Err(error) = crate::contracts::kinds::validate_normalized_node_definition(&def) {
                failures.push(format!("{}: {}", def.kind, error));
            }
        }
        assert!(
            failures.is_empty(),
            "incomplete embedded official node definitions:\n{}",
            failures.join("\n")
        );
    }
}