storage-engine-playground/crates/plan-runner/src/main.rs

//! `plan-run` CLI: read a JSON plan from a file (or stdin if `-`), execute
//! it against the chosen backend, and print the resulting binding relation
//! as JSON on stdout.
//!
//! Backends:
//!
//! - `memory` (default): build tables straight from the plan's `facts`
//!   block, no `Storage` trait involved. Pure in-memory path.
//! - `memory-storage`: load the same facts through `storage::MemoryStorage`
//!   via the `Storage` trait, then materialize tables back out with
//!   `scan_as_table` before executing.
//! - `lmdb`, `redb`, `fjall`, `sqlite`: file-backed `Storage` adapters.
//!   Each invocation creates a fresh tempdir for the store and drops it on
//!   exit; the runner is one-shot, so persistent paths aren't needed.
//! - `geomerge`: CRDT-backed adapter. Constructed in-memory; alpha-status
//!   upstream.

use std::collections::HashMap;
use std::io::{self, Read};
use std::process::ExitCode;

use plan_runner::{JsonValue, Plan, build_tables, build_tables_via_storage, execute, parse_plan};
use storage::MemoryStorage;
use storage::adapters::fjall::FjallStorage;
use storage::adapters::geomerge::{ColumnKind, GeomergeStorage};
use storage::adapters::lmdb::LmdbStorage;
use storage::adapters::redb::RedbStorage;
use storage::adapters::sqlite::SqliteStorage;
use storage::table::Table;
use storage::value::Value;
use tempfile::TempDir;

#[derive(Debug, Clone, Copy)]
enum Backend {
    Memory,
    MemoryStorage,
    Lmdb,
    Redb,
    Fjall,
    Sqlite,
    Geomerge,
}

impl Backend {
    fn parse(s: &str) -> Option<Self> {
        match s {
            "memory" => Some(Self::Memory),
            "memory-storage" => Some(Self::MemoryStorage),
            "lmdb" => Some(Self::Lmdb),
            "redb" => Some(Self::Redb),
            "fjall" => Some(Self::Fjall),
            "sqlite" => Some(Self::Sqlite),
            "geomerge" => Some(Self::Geomerge),
            _ => None,
        }
    }
}

const BACKENDS_HELP: &str = "memory|memory-storage|lmdb|redb|fjall|sqlite|geomerge";

fn main() -> ExitCode {
    let mut backend = Backend::Memory;
    let mut input_path: Option<String> = None;
    let mut args = std::env::args().skip(1);
    while let Some(arg) = args.next() {
        match arg.as_str() {
            "--backend" => {
                let Some(value) = args.next() else {
                    eprintln!("--backend requires a value ({BACKENDS_HELP})");
                    return ExitCode::from(2);
                };
                let Some(parsed) = Backend::parse(&value) else {
                    eprintln!("unknown backend {value:?} (try {BACKENDS_HELP})");
                    return ExitCode::from(2);
                };
                backend = parsed;
            }
            other if input_path.is_none() => input_path = Some(other.to_string()),
            other => {
                eprintln!("unexpected argument: {other}");
                return ExitCode::from(2);
            }
        }
    }
    let Some(path) = input_path else {
        eprintln!("usage: plan-run [--backend {BACKENDS_HELP}] <plan.json | ->");
        return ExitCode::from(2);
    };

    let input = match read_input(&path) {
        Ok(s) => s,
        Err(err) => {
            eprintln!("failed to read {path}: {err}");
            return ExitCode::from(1);
        }
    };

    let plan = match parse_plan(&input) {
        Ok(p) => p,
        Err(err) => {
            eprintln!("parse error: {err}");
            return ExitCode::from(1);
        }
    };

    let tables = match build_tables_for(&plan, backend) {
        Ok(t) => t,
        Err(err) => {
            eprintln!("{err}");
            return ExitCode::from(1);
        }
    };

    let relation = match execute(&tables, &plan.query) {
        Ok(r) => r,
        Err(err) => {
            eprintln!("execute error: {err}");
            return ExitCode::from(1);
        }
    };

    let payload = serde_json::json!({
        "columns": relation.columns,
        "rows": relation
            .rows
            .iter()
            .map(|row| row.iter().map(value_to_json).collect::<Vec<_>>())
            .collect::<Vec<_>>(),
    });
    println!("{payload}");
    ExitCode::SUCCESS
}

/// Build the input tables for `plan` using `backend`. Path-based adapters
/// allocate a fresh tempdir; it drops at the end of this function, which is
/// safe because `build_tables_via_storage` fully materializes the tables
/// into owned `Vec<Value>` before returning.
fn build_tables_for(plan: &Plan, backend: Backend) -> Result<HashMap<String, Table>, String> {
    match backend {
        Backend::Memory => build_tables(plan).map_err(|e| format!("build error: {e}")),
        Backend::MemoryStorage => {
            let mut storage = MemoryStorage::default();
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (memory-storage): {e}"))
        }
        Backend::Lmdb => {
            let dir = TempDir::new().map_err(|e| format!("tempdir: {e}"))?;
            let mut storage = LmdbStorage::open(dir.path())
                .map_err(|e| format!("failed to open lmdb backend: {e}"))?;
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (lmdb): {e}"))
        }
        Backend::Redb => {
            let dir = TempDir::new().map_err(|e| format!("tempdir: {e}"))?;
            let mut storage = RedbStorage::open(dir.path().join("data.redb"))
                .map_err(|e| format!("failed to open redb backend: {e}"))?;
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (redb): {e}"))
        }
        Backend::Fjall => {
            let dir = TempDir::new().map_err(|e| format!("tempdir: {e}"))?;
            let mut storage = FjallStorage::open(dir.path())
                .map_err(|e| format!("failed to open fjall backend: {e}"))?;
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (fjall): {e}"))
        }
        Backend::Sqlite => {
            let dir = TempDir::new().map_err(|e| format!("tempdir: {e}"))?;
            let mut storage = SqliteStorage::open(dir.path().join("data.sqlite"))
                .map_err(|e| format!("failed to open sqlite backend: {e}"))?;
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (sqlite): {e}"))
        }
        Backend::Geomerge => {
            let relations = plan
                .schema
                .iter()
                .map(|(name, &arity)| (name.clone(), infer_column_kinds(plan, name, arity)));
            let mut storage = GeomergeStorage::with_relations(relations)
                .map_err(|e| format!("failed to open geomerge backend: {e}"))?;
            build_tables_via_storage(plan, &mut storage)
                .map_err(|e| format!("build error (geomerge): {e}"))
        }
    }
}

/// Best-effort column type inference for `geomerge`'s synthesized theory.
/// The runner IR carries only arity, so we peek at the first fact row of
/// the relation. Columns without a sample default to `String`, which
/// matches every checked-in fixture (entity identities are encoded as
/// strings by the exporter).
fn infer_column_kinds(plan: &Plan, name: &str, arity: usize) -> Vec<ColumnKind> {
    let mut kinds = vec![ColumnKind::String; arity];
    let Some(rows) = plan.facts.get(name) else {
        return kinds;
    };
    let Some(first) = rows.first() else {
        return kinds;
    };
    for (i, cell) in first.iter().take(arity).enumerate() {
        kinds[i] = match cell {
            JsonValue::Int(_) => ColumnKind::Int,
            JsonValue::Str(_) => ColumnKind::String,
        };
    }
    kinds
}

fn read_input(path: &str) -> io::Result<String> {
    if path == "-" {
        let mut buf = String::new();
        io::stdin().read_to_string(&mut buf)?;
        Ok(buf)
    } else {
        std::fs::read_to_string(path)
    }
}

fn value_to_json(value: &Value) -> serde_json::Value {
    match value {
        Value::Int(n) => serde_json::Value::Number((*n).into()),
        Value::Str(s) => serde_json::Value::String(s.clone()),
        Value::Id(id) => serde_json::Value::String(id.to_string()),
    }
}