MercuryToolbox/crates/chunkcat/tests/chunkcat_cli.rs

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

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

use assert_cmd::Command;
use predicates::prelude::*;
use serde_json::Value;
use tempfile::{TempDir, tempdir};

const SAMPLE_RS: &str = "reading/sample.rs";

fn cargo_command() -> Command {
    Command::cargo_bin("chunkcat").expect("binary")
}

fn fixture(path: &str) -> PathBuf {
    PathBuf::from(env!("CARGO_MANIFEST_DIR"))
        .join("..")
        .join("..")
        .join("fixtures")
        .join(path)
}

fn temp_file(name: &str, contents: impl AsRef<[u8]>) -> (TempDir, PathBuf) {
    let dir = tempdir().expect("tempdir");
    let path = dir.path().join(name);
    fs::write(&path, contents).expect("fixture");
    (dir, path)
}

fn pwsh_command(script: impl AsRef<str>) -> Command {
    let mut command = Command::new("pwsh");
    command
        .arg("-NoProfile")
        .arg("-Command")
        .arg(script.as_ref());
    command
}

fn ps_quote(value: impl std::fmt::Display) -> String {
    format!("'{}'", value.to_string().replace('\'', "''"))
}

fn display_path(path: &Path) -> String {
    path.display().to_string()
}

#[test]
fn lists_chunk_inventory_in_text_mode() {
    let mut command = cargo_command();
    command
        .arg(fixture(SAMPLE_RS))
        .arg("--max-lines")
        .arg("8")
        .arg("--overlap")
        .arg("2")
        .arg("--inventory")
        .assert()
        .success()
        .stdout(predicate::str::contains("chunks=6"))
        .stdout(predicate::str::contains("0 lines=1:8"))
        .stdout(predicate::str::contains("5 lines=31:35"));
}

#[test]
fn renders_selected_chunk_with_line_numbers() {
    let mut command = cargo_command();
    command
        .arg(fixture(SAMPLE_RS))
        .arg("--max-lines")
        .arg("8")
        .arg("--overlap")
        .arg("2")
        .arg("--chunk")
        .arg("2")
        .assert()
        .success()
        .stdout(predicate::str::contains("chunk=2 lines=13:20"))
        .stdout(predicate::str::contains("13: pub enum Mode {"))
        .stdout(predicate::str::contains("20:         Mode::Safe"));
}

#[test]
fn emits_json_inventory_and_selected_chunk_payloads() {
    let mut inventory = cargo_command();
    let inventory_output = inventory
        .arg("--json")
        .arg(fixture(SAMPLE_RS))
        .arg("--max-lines")
        .arg("8")
        .arg("--overlap")
        .arg("2")
        .assert()
        .success()
        .get_output()
        .stdout
        .clone();
    let inventory_json = serde_json::from_slice::<Value>(&inventory_output).expect("inventory");
    assert_eq!(inventory_json["path"], display_path(&fixture(SAMPLE_RS)));
    assert_eq!(inventory_json["total_lines"], 35);
    assert_eq!(inventory_json["chunks"][0]["start_line"], 1);
    assert_eq!(inventory_json["chunks"][5]["end_line"], 35);
    assert!(inventory_json["selected_chunk"].is_null());

    let mut selected = cargo_command();
    let selected_output = selected
        .arg("--json")
        .arg(fixture(SAMPLE_RS))
        .arg("--max-lines")
        .arg("8")
        .arg("--overlap")
        .arg("2")
        .arg("--chunk")
        .arg("4")
        .assert()
        .success()
        .get_output()
        .stdout
        .clone();
    let selected_json = serde_json::from_slice::<Value>(&selected_output).expect("selected");
    assert_eq!(selected_json["selected_chunk"]["index"], 4);
    assert_eq!(selected_json["selected_chunk"]["start_line"], 25);
    assert_eq!(selected_json["selected_chunk"]["lines"][0]["number"], 25);
    assert_eq!(
        selected_json["selected_chunk"]["lines"][0]["text"],
        "    match mode {"
    );
}

#[test]
fn supports_powershell_pipeline_for_path_input() {
    let binary = assert_cmd::cargo::cargo_bin("chunkcat");
    let input = fixture(SAMPLE_RS);
    let script = format!(
        "{} | & {} --json --max-lines 10 --chunk 1",
        ps_quote(input.display()),
        ps_quote(binary.display())
    );

    let mut command = pwsh_command(script);
    command
        .assert()
        .success()
        .stdout(predicate::str::contains("\"selected_chunk\""))
        .stdout(predicate::str::contains("\"index\":1"));
}

#[test]
fn utf8_bom_file_does_not_pollute_first_chunk_line() {
    let (_dir, path) = temp_file("bom.txt", "\u{feff}alpha\nbeta\n");

    let mut command = cargo_command();
    command
        .arg(&path)
        .arg("--max-lines")
        .arg("2")
        .arg("--chunk")
        .arg("0")
        .assert()
        .success()
        .stdout(predicate::str::contains("1: alpha"))
        .stdout(predicate::str::contains("\u{feff}alpha").not());
}

#[test]
fn invalid_utf8_file_reports_text_read_error() {
    let (_dir, path) = temp_file("invalid-utf8.txt", [0x61, 0x80, 0x0A]);

    let mut command = cargo_command();
    command
        .arg(&path)
        .arg("--max-lines")
        .arg("2")
        .arg("--chunk")
        .arg("0")
        .assert()
        .failure()
        .stderr(predicate::str::contains("looks like a binary file"))
        .stderr(predicate::str::contains("invalid-utf8.txt"));
}

#[test]
fn help_includes_examples_and_pipeline_usage() {
    let mut command = cargo_command();
    command
        .arg("--help")
        .assert()
        .success()
        .stdout(predicate::str::contains(
            "chunkcat .\\fixtures\\reading\\sample.rs --max-lines 8",
        ))
        .stdout(predicate::str::contains("--inventory"))
        .stdout(predicate::str::contains("ConvertFrom-Json"))
        .stdout(predicate::str::contains("--chunk"));
}