MercuryToolbox/crates/tonl/tests/tonl_cli.rs

//! Integration tests for the `tonl` command.

use std::fs;
use std::path::PathBuf;

use assert_cmd::Command;
use predicates::prelude::*;
use serde_json::Value;
use tempfile::tempdir;

fn fixture(path: &str) -> PathBuf {
    let fixture = PathBuf::from(env!("CARGO_MANIFEST_DIR"))
        .join("..")
        .join("..")
        .join("fixtures")
        .join("json-family")
        .join("tonl")
        .join(path);
    assert!(
        fixture.exists(),
        "missing fixture `{path}` at {}",
        fixture.display()
    );
    fixture
}

fn cargo_tonl() -> Command {
    Command::cargo_bin("tonl").expect("binary")
}

fn fixture_text(path: &str) -> String {
    fs::read_to_string(fixture(path)).expect("fixture text")
}

#[test]
fn roundtrip_json_object_through_tonl() {
    let encoded = cargo_tonl()
        .args(["encode"])
        .arg(fixture("sample.json"))
        .assert()
        .success()
        .stdout(predicate::str::contains("name = \"Ada\""))
        .get_output()
        .stdout
        .clone();

    let decoded = cargo_tonl()
        .args(["decode"])
        .write_stdin(encoded)
        .assert()
        .success()
        .get_output()
        .stdout
        .clone();

    let expected: Value =
        serde_json::from_str(&fixture_text("sample.json")).expect("expected json");
    let actual: Value = serde_json::from_slice(&decoded).expect("decoded json");
    assert_eq!(actual, expected);
}

#[test]
fn invalid_input_reports_line_diagnostics() {
    cargo_tonl()
        .args(["decode"])
        .arg(fixture("invalid.tonl"))
        .assert()
        .failure()
        .code(2)
        .stderr(predicate::str::contains("line 3"))
        .stderr(predicate::str::contains("valid JSON value"));
}

#[test]
fn supports_path_and_stdin_input() {
    cargo_tonl()
        .args(["decode"])
        .arg(fixture("sample.tonl"))
        .assert()
        .success()
        .stdout(predicate::str::contains("\"name\":\"Ada\""));

    cargo_tonl()
        .args(["encode"])
        .write_stdin(fixture_text("sample.json"))
        .assert()
        .success()
        .stdout(predicate::str::contains("score = 42"));
}

#[test]
fn encode_json_wrapper_reports_tonl_text() {
    cargo_tonl()
        .args(["encode", "--json"])
        .arg(fixture("sample.json"))
        .assert()
        .success()
        .stdout(predicate::str::contains("\"format\":\"tonl\""))
        .stdout(predicate::str::contains("\"records\":1"))
        .stdout(predicate::str::contains("name = \\\"Ada\\\""));
}

#[test]
fn encode_toon_wrapper_reports_tonl_text() {
    cargo_tonl()
        .args(["encode", "--format", "toon"])
        .arg(fixture("sample.json"))
        .assert()
        .success()
        .stdout(predicate::str::contains("format: tonl"))
        .stdout(predicate::str::contains("records: 1"))
        .stdout(predicate::str::contains("name = \\\"Ada\\\""));
}

#[test]
fn query_filters_records_by_field_value() {
    let output = cargo_tonl()
        .args(["query", "--where", "active=true"])
        .arg(fixture("records.tonl"))
        .assert()
        .success()
        .get_output()
        .stdout
        .clone();

    let record: Value = serde_json::from_slice(&output).unwrap_or_else(|error| {
        panic!(
            "query output should be a JSON record: {error}\n{}",
            String::from_utf8_lossy(&output)
        )
    });
    assert_eq!(
        record["id"],
        "user-1",
        "stdout={}",
        String::from_utf8_lossy(&output)
    );
    assert_eq!(
        record["active"],
        true,
        "stdout={}",
        String::from_utf8_lossy(&output)
    );
}

#[test]
fn validate_reports_failures() {
    cargo_tonl()
        .args(["validate"])
        .arg(fixture("invalid.tonl"))
        .assert()
        .failure()
        .code(2)
        .stderr(predicate::str::contains("invalid TONL"));
}

#[test]
fn index_build_and_query_lookup_record() {
    let temp = tempdir().expect("tempdir");
    let index = temp.path().join("records.tonl-index.json");

    cargo_tonl()
        .args(["index", "build", "--key", "id", "--output"])
        .arg(&index)
        .arg(fixture("records.tonl"))
        .assert()
        .success()
        .stdout(predicate::str::contains("indexed 2 records"));

    cargo_tonl()
        .args(["index", "query"])
        .arg(&index)
        .arg("user-2")
        .assert()
        .success()
        .stdout(predicate::str::contains("\"name\":\"Bob\""))
        .stdout(predicate::str::contains("\"active\":false"));
}

#[test]
fn stream_query_smoke_filters_jsonl_records() {
    cargo_tonl()
        .args(["stream", "query", "--where", "active=false"])
        .arg(fixture("records.jsonl"))
        .assert()
        .success()
        .stdout(predicate::str::contains("\"id\":\"user-2\""))
        .stdout(predicate::str::contains("\"id\":\"user-1\"").not());
}

#[test]
fn etl_converts_jsonl_to_tonl_and_tonl_to_jsonl() {
    cargo_tonl()
        .args(["etl", "--from", "jsonl", "--to", "tonl"])
        .arg(fixture("records.jsonl"))
        .assert()
        .success()
        .stdout(predicate::str::contains("---"))
        .stdout(predicate::str::contains("id = \"user-1\""));

    cargo_tonl()
        .args(["etl", "--from", "tonl", "--to", "jsonl"])
        .arg(fixture("records.tonl"))
        .assert()
        .success()
        .stdout(predicate::str::contains("{\"id\":\"user-1\""))
        .stdout(predicate::str::contains("{\"id\":\"user-2\""));
}

#[test]
fn help_lists_full_tool_surface() {
    cargo_tonl()
        .arg("--help")
        .assert()
        .success()
        .stdout(predicate::str::contains("encode"))
        .stdout(predicate::str::contains("decode"))
        .stdout(predicate::str::contains("query"))
        .stdout(predicate::str::contains("validate"))
        .stdout(predicate::str::contains("index build"))
        .stdout(predicate::str::contains("index query"))
        .stdout(predicate::str::contains("stream query"))
        .stdout(predicate::str::contains("etl"));
}