fetch_ml/native_rust/dataset_hash/tests/integration.rs

//! Integration tests for dataset_hash
//!
//! Tests file hashing with various file sizes and edge cases.

use std::fs;
use tempfile::TempDir;

// Import the crate
use dataset_hash::{collect_files, hash_directory_batch, hash_directory_combined, hash_file};

#[test]
fn test_hash_file_basic() {
    let temp = TempDir::new().unwrap();
    let file_path = temp.path().join("test.txt");
    fs::write(&file_path, "hello world").unwrap();
    
    let hash1 = hash_file(&file_path).unwrap();
    let hash2 = hash_file(&file_path).unwrap();
    
    // Hash should be deterministic
    assert_eq!(hash1, hash2);
    // BLAKE3 produces 64-char hex strings
    assert_eq!(hash1.len(), 64);
}

#[test]
fn test_hash_file_empty() {
    let temp = TempDir::new().unwrap();
    let file_path = temp.path().join("empty.txt");
    fs::write(&file_path, "").unwrap();
    
    let hash = hash_file(&file_path).unwrap();
    assert_eq!(hash.len(), 64);
}

#[test]
fn test_hash_file_large() {
    let temp = TempDir::new().unwrap();
    let file_path = temp.path().join("large.bin");
    
    // Create 10MB file
    let data = vec![0u8; 10 * 1024 * 1024];
    fs::write(&file_path, &data).unwrap();
    
    let hash = hash_file(&file_path).unwrap();
    assert_eq!(hash.len(), 64);
}

#[test]
fn test_hash_file_different_content() {
    let temp = TempDir::new().unwrap();
    let file1 = temp.path().join("file1.txt");
    let file2 = temp.path().join("file2.txt");
    
    fs::write(&file1, "content A").unwrap();
    fs::write(&file2, "content B").unwrap();
    
    let hash1 = hash_file(&file1).unwrap();
    let hash2 = hash_file(&file2).unwrap();
    
    assert_ne!(hash1, hash2);
}

#[test]
fn test_collect_files_excludes_hidden() {
    let temp = TempDir::new().unwrap();
    fs::write(temp.path().join("visible.txt"), "data").unwrap();
    fs::write(temp.path().join(".hidden"), "data").unwrap();
    
    // Create hidden directory and file inside it
    let hidden_dir = temp.path().join(".hidden_dir");
    fs::create_dir(&hidden_dir).unwrap();
    fs::write(hidden_dir.join("file.txt"), "data").unwrap();
    
    let files = collect_files(temp.path()).unwrap();
    assert_eq!(files.len(), 1);
    assert!(files[0].file_name().unwrap() == "visible.txt");
}

#[test]
fn test_collect_files_sorted() {
    let temp = TempDir::new().unwrap();
    fs::write(temp.path().join("z.txt"), "z").unwrap();
    fs::write(temp.path().join("a.txt"), "a").unwrap();
    fs::write(temp.path().join("m.txt"), "m").unwrap();
    
    let files = collect_files(temp.path()).unwrap();
    assert_eq!(files.len(), 3);
    assert!(files[0].file_name().unwrap() == "a.txt");
    assert!(files[1].file_name().unwrap() == "m.txt");
    assert!(files[2].file_name().unwrap() == "z.txt");
}

#[test]
fn test_hash_directory_batch() {
    let temp = TempDir::new().unwrap();
    fs::write(temp.path().join("a.txt"), "AAA").unwrap();
    fs::write(temp.path().join("b.txt"), "BBB").unwrap();
    
    let pairs = hash_directory_batch(temp.path()).unwrap();
    assert_eq!(pairs.len(), 2);
    
    // Verify each file has a hash
    for (path, hash) in &pairs {
        assert!(path.ends_with(".txt"));
        assert_eq!(hash.len(), 64);
    }
}

#[test]
fn test_hash_directory_combined() {
    let temp = TempDir::new().unwrap();
    fs::write(temp.path().join("a.txt"), "AAA").unwrap();
    fs::write(temp.path().join("b.txt"), "BBB").unwrap();
    
    let hash1 = hash_directory_combined(temp.path()).unwrap();
    let hash2 = hash_directory_combined(temp.path()).unwrap();
    
    // Combined hash should be deterministic
    assert_eq!(hash1, hash2);
    assert_eq!(hash1.len(), 64);
}

#[test]
fn test_hash_directory_combined_changes_with_content() {
    let temp = TempDir::new().unwrap();
    fs::write(temp.path().join("file.txt"), "content").unwrap();
    
    let hash1 = hash_directory_combined(temp.path()).unwrap();
    
    // Modify file
    fs::write(temp.path().join("file.txt"), "modified").unwrap();
    
    let hash2 = hash_directory_combined(temp.path()).unwrap();
    
    assert_ne!(hash1, hash2);
}

#[test]
fn test_collect_files_no_symlinks() {
    let temp = TempDir::new().unwrap();
    let real_file = temp.path().join("real.txt");
    let symlink = temp.path().join("link.txt");
    
    fs::write(&real_file, "data").unwrap();
    #[cfg(unix)]
    std::os::unix::fs::symlink(&real_file, &symlink).unwrap();
    
    let files = collect_files(temp.path()).unwrap();
    // Should only include the real file, not the symlink
    assert_eq!(files.len(), 1);
}

#[test]
fn test_hash_directory_empty() {
    let temp = TempDir::new().unwrap();
    
    let pairs = hash_directory_batch(temp.path()).unwrap();
    assert!(pairs.is_empty());
    
    let combined = hash_directory_combined(temp.path()).unwrap();
    // Empty directory should still produce a valid hash
    assert_eq!(combined.len(), 64);
}

#[test]
fn test_hash_directory_nested() {
    let temp = TempDir::new().unwrap();
    let subdir = temp.path().join("subdir");
    fs::create_dir(&subdir).unwrap();
    
    fs::write(temp.path().join("root.txt"), "root").unwrap();
    fs::write(subdir.join("nested.txt"), "nested").unwrap();
    
    let pairs = hash_directory_batch(temp.path()).unwrap();
    assert_eq!(pairs.len(), 2);
}