vendor/crc-fast/src/bin/checksum.rs at 91334b7fa574 · andymccall/oxygen

# 91334b7fa574 oxygen / vendor / crc-fast / src / bin / checksum.rs

23.1 KiB · Rust · last changed Strata: the cloud stores send through ureq, so every release target links 1 month ago source blame history raw

// Copyright 2025 Don MacAskill. Licensed under MIT or Apache-2.0.

//! This is a simple program to calculate a checksum from the command line

use crc_fast::{checksum, checksum_file, CrcAlgorithm};
use std::env;
use std::process::ExitCode;
use std::str::FromStr;

#[derive(Debug)]
struct Config {
    algorithm: String,
    file: Option<String>,
    string: Option<String>,
    format: OutputFormat,
    benchmark: Option<BenchmarkConfig>,
}

#[derive(Debug)]
struct BenchmarkConfig {
    size: Option<usize>,
    duration: f64,
}

#[derive(Debug)]
enum BenchmarkData {
    InMemory(Vec<u8>),
    File(String),
}

#[derive(Debug)]
struct BenchmarkRunner {
    algorithm: CrcAlgorithm,
    data: BenchmarkData,
    duration: f64,
}

#[derive(Debug)]
struct BenchmarkResult {
    iterations: u64,
    elapsed_seconds: f64,
    throughput_gibs: f64,
    time_per_iteration_nanos: f64,
    acceleration_target: String,
    data_size: u64,
}

#[derive(Debug, Clone)]
enum OutputFormat {
    Hex,
    Decimal,
}

impl BenchmarkConfig {
    fn validate(&self) -> Result<(), String> {
        if self.duration <= 0.0 {
            return Err("Duration must be greater than 0".to_string());
        }

        if let Some(size) = self.size {
            if size == 0 {
                return Err("Size must be greater than 0".to_string());
            }
        }

        Ok(())
    }
}

impl BenchmarkRunner {
    fn new(algorithm: CrcAlgorithm, data: BenchmarkData, duration: f64) -> Self {
        Self {
            algorithm,
            data,
            duration,
        }
    }

    fn run(&self) -> Result<BenchmarkResult, String> {
        use std::time::Instant;

        let start = Instant::now();
        let mut iterations = 0u64;

        while start.elapsed().as_secs_f64() < self.duration {
            match &self.data {
                BenchmarkData::InMemory(data) => {
                    std::hint::black_box(checksum(self.algorithm, data));
                }
                BenchmarkData::File(filename) => {
                    match checksum_file(self.algorithm, filename, None) {
                        Ok(result) => {
                            std::hint::black_box(result);
                        }
                        Err(e) => {
                            return Err(format!("Failed to read file during benchmark: {}", e));
                        }
                    }
                }
            }
            iterations += 1;
        }

        let elapsed = start.elapsed().as_secs_f64();
        let data_size = match &self.data {
            BenchmarkData::InMemory(data) => data.len() as u64,
            BenchmarkData::File(filename) => std::fs::metadata(filename)
                .map(|m| m.len())
                .map_err(|e| format!("Failed to get file size: {}", e))?,
        };

        let acceleration_target = crc_fast::get_calculator_target(self.algorithm);

        Ok(BenchmarkResult::new(
            iterations,
            elapsed,
            acceleration_target,
            data_size,
        ))
    }
}

impl BenchmarkResult {
    fn new(
        iterations: u64,
        elapsed_seconds: f64,
        acceleration_target: String,
        data_size: u64,
    ) -> Self {
        let throughput_gibs = if elapsed_seconds > 0.0 {
            (data_size as f64 * iterations as f64) / elapsed_seconds / (1024.0 * 1024.0 * 1024.0)
        } else {
            0.0
        };

        let time_per_iteration_nanos = if iterations > 0 {
            elapsed_seconds * 1_000_000_000.0 / iterations as f64
        } else {
            0.0
        };

        Self {
            iterations,
            elapsed_seconds,
            throughput_gibs,
            time_per_iteration_nanos,
            acceleration_target,
            data_size,
        }
    }
}

/// Fills a buffer with pseudo-random data using xorshift64* algorithm.
/// This is a deterministic PRNG that's tiny, fast, and suitable for benchmark data generation.
/// The seed is derived from the buffer size to provide some variability.
/// This solves the problem of having to use a full-featured PRNG like rand for benchmark data generation.
#[inline]
fn fill_pseudo_random(buf: &mut [u8], seed: u64) {
    let mut x = seed;
    for b in buf {
        x ^= x >> 12;
        x ^= x << 25;
        x ^= x >> 27;
        let r = x.wrapping_mul(0x2545_F491_4F6C_DD1D);
        *b = r as u8;
    }
}

fn generate_random_data(size: usize) -> Result<Vec<u8>, String> {
    // Check for reasonable size limits to prevent memory issues
    if size > 1_073_741_824 {
        // 1 GiB limit
        return Err("Data size too large (maximum 1 GiB)".to_string());
    }

    // Use vec! macro to avoid clippy warning about slow initialization
    let mut buf = vec![0u8; size];

    // Derive seed from size for deterministic but varied data
    let seed = 0x9E37_79B9_7F4A_7C15_u64.wrapping_add(size as u64);
    fill_pseudo_random(&mut buf, seed);

    Ok(buf)
}

fn print_usage() {
    println!("Usage: checksum -a algorithm [-f file] [-s string] [--format hex|decimal]");
    println!(
        "       checksum -a algorithm -b [--size bytes] [--duration seconds] [-f file] [-s string]"
    );
    println!();
    println!("Example: checksum -a CRC-32/ISCSI -f myfile.txt");
    println!("Example: checksum -a CRC-64/NVME -s 'Hello, world!' --format decimal");
    println!("Example: checksum -a CRC-32/ISCSI -b --size 1048576 --duration 5.0");
    println!();
    println!("Options:");
    println!("  -a algorithm        Specify the checksum algorithm (required)");
    println!("  -f file             Calculate checksum for the specified file");
    println!("  -h, --help          Show this help message");
    println!("  -s string           Calculate checksum for the specified string");
    println!("  --format hex|decimal Output format (default: hex)");
    println!();
    println!("Benchmarking:");
    println!("  -b                  Enable benchmark mode");
    println!("  --duration seconds  Benchmark duration in seconds (default: 10.0)");
    println!("  --size bytes        Data size for random generation in benchmark mode (default: 1048576 [1MiB])");
    println!();
    println!();
    println!("Note: In normal mode, either -f or -s must be provided, but not both.");
    println!("      In benchmark mode (-b), -f or -s are optional for using specific data.");
}

fn parse_args() -> Result<Config, String> {
    let args: Vec<String> = env::args().collect();

    if args.len() == 1 {
        return Err("No arguments provided".to_string());
    }

    // Check for help flag
    if args.contains(&"-h".to_string()) || args.contains(&"--help".to_string()) {
        return Err("help".to_string());
    }

    let mut algorithm: Option<String> = None;
    let mut file: Option<String> = None;
    let mut string: Option<String> = None;
    let mut format = OutputFormat::Hex; // Default to hex
    let mut benchmark_mode = false;
    let mut benchmark_size: Option<usize> = None;
    let mut benchmark_duration = 10.0; // Default duration

    let mut i = 1; // Skip program name
    while i < args.len() {
        match args[i].as_str() {
            "-a" => {
                if i + 1 >= args.len() {
                    return Err("Missing algorithm after -a flag".to_string());
                }
                algorithm = Some(args[i + 1].clone());
                i += 2;
            }
            "-f" => {
                if i + 1 >= args.len() {
                    return Err("Missing filename after -f flag".to_string());
                }
                if string.is_some() {
                    return Err("Cannot specify both -f and -s flags".to_string());
                }
                file = Some(args[i + 1].clone());
                i += 2;
            }
            "-s" => {
                if i + 1 >= args.len() {
                    return Err("Missing string after -s flag".to_string());
                }
                if file.is_some() {
                    return Err("Cannot specify both -f and -s flags".to_string());
                }
                string = Some(args[i + 1].clone());
                i += 2;
            }
            "--format" => {
                if i + 1 >= args.len() {
                    return Err("Missing format after --format flag".to_string());
                }
                match args[i + 1].as_str() {
                    "hex" => format = OutputFormat::Hex,
                    "decimal" => format = OutputFormat::Decimal,
                    invalid => {
                        return Err(format!(
                            "Invalid format '{}'. Use 'hex' or 'decimal'",
                            invalid
                        ))
                    }
                }
                i += 2;
            }
            "-b" => {
                benchmark_mode = true;
                i += 1;
            }
            "--size" => {
                if i + 1 >= args.len() {
                    return Err("Missing size value after --size flag".to_string());
                }
                benchmark_size = Some(
                    args[i + 1]
                        .parse::<usize>()
                        .map_err(|_| format!("Invalid size value: {}", args[i + 1]))?,
                );
                i += 2;
            }
            "--duration" => {
                if i + 1 >= args.len() {
                    return Err("Missing duration value after --duration flag".to_string());
                }
                benchmark_duration = args[i + 1]
                    .parse::<f64>()
                    .map_err(|_| format!("Invalid duration value: {}", args[i + 1]))?;
                i += 2;
            }
            arg => {
                return Err(format!("Unknown argument: {}", arg));
            }
        }
    }

    // Validate required arguments
    let algorithm = algorithm.ok_or("Algorithm (-a) is required")?;

    // Validate mutual exclusivity between benchmark and normal modes
    if !benchmark_mode && (benchmark_size.is_some() || benchmark_duration != 10.0) {
        return Err("--size and --duration can only be used with -b flag".to_string());
    }

    // Create benchmark config if in benchmark mode
    let benchmark = if benchmark_mode {
        let config = BenchmarkConfig {
            size: benchmark_size,
            duration: benchmark_duration,
        };
        config.validate()?;
        Some(config)
    } else {
        None
    };

    // Validate input requirements based on mode
    if benchmark.is_none() {
        // Normal mode: require either file or string input
        if file.is_none() && string.is_none() {
            return Err(
                "Either -f (file) or -s (string) must be provided in normal mode".to_string(),
            );
        }
    }
    // Benchmark mode: file and string are optional (will use generated data if neither provided)

    Ok(Config {
        algorithm,
        file,
        string,
        format,
        benchmark,
    })
}

fn calculate_checksum(config: &Config) -> Result<(), String> {
    let algorithm = CrcAlgorithm::from_str(&config.algorithm)
        .map_err(|_| format!("Invalid algorithm: {}", config.algorithm))?;

    // Check if benchmark mode is enabled
    if let Some(benchmark_config) = &config.benchmark {
        return run_benchmark(config, benchmark_config, algorithm);
    }

    let checksum = if let Some(ref filename) = config.file {
        checksum_file(algorithm, filename, None).unwrap()
    } else if let Some(ref text) = config.string {
        checksum(algorithm, text.as_bytes())
    } else {
        return Err("No input provided for checksum calculation".to_string());
    };

    match config.format {
        OutputFormat::Hex => println!("{:#x?}", checksum),
        OutputFormat::Decimal => println!("{}", checksum),
    }

    Ok(())
}

fn run_benchmark(
    config: &Config,
    benchmark_config: &BenchmarkConfig,
    algorithm: CrcAlgorithm,
) -> Result<(), String> {
    // Determine data source and create BenchmarkData
    let data = if let Some(ref filename) = config.file {
        // Validate file exists before benchmarking
        if !std::path::Path::new(filename).exists() {
            return Err(format!("File not found: {}", filename));
        }
        BenchmarkData::File(filename.clone())
    } else if let Some(ref text) = config.string {
        BenchmarkData::InMemory(text.as_bytes().to_vec())
    } else {
        // Generate random data with specified size or default (1 MiB)
        let size = benchmark_config.size.unwrap_or(1_048_576);
        let random_data = generate_random_data(size)?;
        BenchmarkData::InMemory(random_data)
    };

    // Create and run benchmark
    let runner = BenchmarkRunner::new(algorithm, data, benchmark_config.duration);
    let result = runner.run()?;

    // Display results with algorithm name
    display_benchmark_results(&result, &config.algorithm);

    Ok(())
}

// Format numbers with comma separators for better readability
fn format_number_with_commas(n: u64) -> String {
    let s = n.to_string();
    let mut result = String::new();
    let chars: Vec<char> = s.chars().collect();

    for (i, ch) in chars.iter().enumerate() {
        if i > 0 && (chars.len() - i).is_multiple_of(3) {
            result.push(',');
        }
        result.push(*ch);
    }

    result
}

fn display_benchmark_results(result: &BenchmarkResult, algorithm_name: &str) {
    println!("Algorithm: {}", algorithm_name);
    println!("Acceleration Target: {}", result.acceleration_target);

    // Format data size with appropriate units
    let (size_value, size_unit) = if result.data_size >= 1_048_576 {
        (result.data_size as f64 / 1_048_576.0, "MiB")
    } else if result.data_size >= 1024 {
        (result.data_size as f64 / 1024.0, "KiB")
    } else {
        (result.data_size as f64, "bytes")
    };

    println!(
        "Data Size: {} bytes ({:.1} {})",
        format_number_with_commas(result.data_size),
        size_value,
        size_unit
    );
    println!("Duration: {:.2} seconds", result.elapsed_seconds);
    println!(
        "Iterations: {}",
        format_number_with_commas(result.iterations)
    );
    println!("Throughput: {:.2} GiB/s", result.throughput_gibs);

    // Format time per iteration with appropriate units
    let (time_value, time_unit) = if result.time_per_iteration_nanos >= 1_000_000.0 {
        (result.time_per_iteration_nanos / 1_000_000.0, "ms")
    } else if result.time_per_iteration_nanos >= 1_000.0 {
        (result.time_per_iteration_nanos / 1_000.0, "μs")
    } else {
        (result.time_per_iteration_nanos, "ns")
    };

    println!("Time per iteration: {:.1} {}", time_value, time_unit);
}

fn main() -> ExitCode {
    match parse_args() {
        Ok(config) => {
            if let Err(e) = calculate_checksum(&config) {
                eprintln!("Error: {}", e);
                return ExitCode::from(1);
            }
        }
        Err(msg) => {
            if msg == "help" {
                print_usage();
                return ExitCode::SUCCESS;
            } else {
                eprintln!("Error: {}", msg);
                println!();
                print_usage();
                return ExitCode::from(1);
            }
        }
    }

    ExitCode::SUCCESS
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::str::FromStr;

    #[test]
    fn test_benchmark_config_validation_valid() {
        let config = BenchmarkConfig {
            size: Some(1024),
            duration: 5.0,
        };
        assert!(config.validate().is_ok());
    }

    #[test]
    fn test_benchmark_config_validation_zero_duration() {
        let config = BenchmarkConfig {
            size: Some(1024),
            duration: 0.0,
        };
        assert!(config.validate().is_err());
        assert_eq!(
            config.validate().unwrap_err(),
            "Duration must be greater than 0"
        );
    }

    #[test]
    fn test_benchmark_config_validation_negative_duration() {
        let config = BenchmarkConfig {
            size: Some(1024),
            duration: -1.0,
        };
        assert!(config.validate().is_err());
        assert_eq!(
            config.validate().unwrap_err(),
            "Duration must be greater than 0"
        );
    }

    #[test]
    fn test_benchmark_config_validation_zero_size() {
        let config = BenchmarkConfig {
            size: Some(0),
            duration: 5.0,
        };
        assert!(config.validate().is_err());
        assert_eq!(
            config.validate().unwrap_err(),
            "Size must be greater than 0"
        );
    }

    #[test]
    fn test_benchmark_config_validation_none_size() {
        let config = BenchmarkConfig {
            size: None,
            duration: 5.0,
        };
        assert!(config.validate().is_ok());
    }

    #[test]
    fn test_generate_random_data_valid_size() {
        let data = generate_random_data(1024).unwrap();
        assert_eq!(data.len(), 1024);
    }

    #[test]
    fn test_generate_random_data_large_size() {
        let result = generate_random_data(1_073_741_825); // 1 GiB + 1
        assert!(result.is_err());
        assert_eq!(result.unwrap_err(), "Data size too large (maximum 1 GiB)");
    }

    #[test]
    fn test_benchmark_result_calculation() {
        let result = BenchmarkResult::new(
            1000,               // iterations
            2.0,                // elapsed_seconds
            "test".to_string(), // acceleration_target
            1024,               // data_size
        );

        assert_eq!(result.iterations, 1000);
        assert_eq!(result.elapsed_seconds, 2.0);
        assert_eq!(result.data_size, 1024);

        // Throughput should be (1024 * 1000) / 2.0 / (1024^3) GiB/s
        let expected_throughput = (1024.0 * 1000.0) / 2.0 / (1024.0 * 1024.0 * 1024.0);
        assert!((result.throughput_gibs - expected_throughput).abs() < 1e-10);

        // Time per iteration should be 2.0 * 1e9 / 1000 nanoseconds
        let expected_time_per_iter = 2.0 * 1_000_000_000.0 / 1000.0;
        assert!((result.time_per_iteration_nanos - expected_time_per_iter).abs() < 1e-6);
    }

    #[test]
    fn test_benchmark_runner_creation() {
        let algorithm = CrcAlgorithm::from_str("CRC-32/ISCSI").unwrap();
        let data = BenchmarkData::InMemory(vec![1, 2, 3, 4]);
        let runner = BenchmarkRunner::new(algorithm, data, 1.0);

        assert_eq!(runner.duration, 1.0);
        assert_eq!(runner.algorithm, algorithm);
    }

    #[test]
    fn test_benchmark_runner_execution_in_memory() {
        let algorithm = CrcAlgorithm::from_str("CRC-32/ISCSI").unwrap();
        let data = BenchmarkData::InMemory(vec![1, 2, 3, 4]);
        let runner = BenchmarkRunner::new(algorithm, data, 0.1); // Short duration for test

        let result = runner.run().unwrap();
        assert!(result.iterations > 0);
        assert!(result.elapsed_seconds > 0.0);
        assert_eq!(result.data_size, 4);
        assert!(result.throughput_gibs >= 0.0);
    }

    #[test]
    fn test_parse_args_benchmark_mode() {
        // We can't easily mock std::env::args in unit tests, so we'll test the parsing logic
        // by creating a Config directly and validating its structure
        let config = Config {
            algorithm: "CRC-32/ISCSI".to_string(),
            file: None,
            string: None,
            format: OutputFormat::Hex,
            benchmark: Some(BenchmarkConfig {
                size: Some(1024),
                duration: 5.0,
            }),
        };

        assert!(config.benchmark.is_some());
        let benchmark_config = config.benchmark.unwrap();
        assert_eq!(benchmark_config.size, Some(1024));
        assert_eq!(benchmark_config.duration, 5.0);
    }

    #[test]
    fn test_parse_args_normal_mode() {
        let config = Config {
            algorithm: "CRC-32/ISCSI".to_string(),
            file: Some("test.txt".to_string()),
            string: None,
            format: OutputFormat::Hex,
            benchmark: None,
        };

        assert!(config.benchmark.is_none());
        assert_eq!(config.file, Some("test.txt".to_string()));
    }

    #[test]
    fn test_data_source_selection_file() {
        // Test that file input creates File variant
        let config = Config {
            algorithm: "CRC-32/ISCSI".to_string(),
            file: Some("test.txt".to_string()),
            string: None,
            format: OutputFormat::Hex,
            benchmark: Some(BenchmarkConfig {
                size: None,
                duration: 1.0,
            }),
        };

        // This would be tested in the run_benchmark function
        // We can verify the config structure is correct for file input
        assert!(config.file.is_some());
        assert!(config.string.is_none());
    }

    #[test]
    fn test_data_source_selection_string() {
        let config = Config {
            algorithm: "CRC-32/ISCSI".to_string(),
            file: None,
            string: Some("test data".to_string()),
            format: OutputFormat::Hex,
            benchmark: Some(BenchmarkConfig {
                size: None,
                duration: 1.0,
            }),
        };

        assert!(config.file.is_none());
        assert!(config.string.is_some());
    }

    #[test]
    fn test_data_source_selection_generated() {
        let config = Config {
            algorithm: "CRC-32/ISCSI".to_string(),
            file: None,
            string: None,
            format: OutputFormat::Hex,
            benchmark: Some(BenchmarkConfig {
                size: Some(1024),
                duration: 1.0,
            }),
        };

        // When neither file nor string is provided, generated data should be used
        assert!(config.file.is_none());
        assert!(config.string.is_none());
        assert_eq!(config.benchmark.as_ref().unwrap().size, Some(1024));
    }

    #[test]
    fn test_throughput_calculation_accuracy() {
        // Test with known values to verify calculation accuracy
        let data_size = 1_048_576u64; // 1 MiB
        let iterations = 1000u64;
        let elapsed_seconds = 1.0;

        let result =
            BenchmarkResult::new(iterations, elapsed_seconds, "test".to_string(), data_size);

        // Expected throughput: (1 MiB * 1000) / 1 second = 1000 MiB/s = ~0.9537 GiB/s
        let expected_gibs =
            (data_size as f64 * iterations as f64) / elapsed_seconds / (1024.0 * 1024.0 * 1024.0);
        assert!((result.throughput_gibs - expected_gibs).abs() < 1e-10);
    }

    #[test]
    fn test_output_format_variants() {
        let hex_format = OutputFormat::Hex;
        let decimal_format = OutputFormat::Decimal;

        // Test that both variants can be created and are different
        assert!(matches!(hex_format, OutputFormat::Hex));
        assert!(matches!(decimal_format, OutputFormat::Decimal));
    }

    #[test]
    fn test_format_number_with_commas() {
        assert_eq!(format_number_with_commas(0), "0");
        assert_eq!(format_number_with_commas(123), "123");
        assert_eq!(format_number_with_commas(1234), "1,234");
        assert_eq!(format_number_with_commas(12345), "12,345");
        assert_eq!(format_number_with_commas(123456), "123,456");
        assert_eq!(format_number_with_commas(1234567), "1,234,567");
        assert_eq!(format_number_with_commas(12345678), "12,345,678");
        assert_eq!(format_number_with_commas(123456789), "123,456,789");
        assert_eq!(format_number_with_commas(1000000000), "1,000,000,000");
    }
}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

1	// Copyright 2025 Don MacAskill. Licensed under MIT or Apache-2.0.
2
3	//! This is a simple program to calculate a checksum from the command line
4
5	use crc_fast::{checksum, checksum_file, CrcAlgorithm};
6	use std::env;
7	use std::process::ExitCode;
8	use std::str::FromStr;
9
10	#[derive(Debug)]
11	struct Config {
12	algorithm: String,
13	file: Option<String>,
14	string: Option<String>,
15	format: OutputFormat,
16	benchmark: Option<BenchmarkConfig>,
17	}
18
19	#[derive(Debug)]
20	struct BenchmarkConfig {
21	size: Option<usize>,
22	duration: f64,
23	}
24
25	#[derive(Debug)]
26	enum BenchmarkData {
27	InMemory(Vec<u8>),
28	File(String),
29	}
30
31	#[derive(Debug)]
32	struct BenchmarkRunner {
33	algorithm: CrcAlgorithm,
34	data: BenchmarkData,
35	duration: f64,
36	}
37
38	#[derive(Debug)]
39	struct BenchmarkResult {
40	iterations: u64,
41	elapsed_seconds: f64,
42	throughput_gibs: f64,
43	time_per_iteration_nanos: f64,
44	acceleration_target: String,
45	data_size: u64,
46	}
47
48	#[derive(Debug, Clone)]
49	enum OutputFormat {
50	Hex,
51	Decimal,
52	}
53
54	impl BenchmarkConfig {
55	fn validate(&self) -> Result<(), String> {
56	if self.duration <= 0.0 {
57	return Err("Duration must be greater than 0".to_string());
58	}
59
60	if let Some(size) = self.size {
61	if size == 0 {
62	return Err("Size must be greater than 0".to_string());
63	}
64	}
65
66	Ok(())
67	}
68	}
69
70	impl BenchmarkRunner {
71	fn new(algorithm: CrcAlgorithm, data: BenchmarkData, duration: f64) -> Self {
72	Self {
73	algorithm,
74	data,
75	duration,
76	}
77	}
78
79	fn run(&self) -> Result<BenchmarkResult, String> {
80	use std::time::Instant;
81
82	let start = Instant::now();
83	let mut iterations = 0u64;
84
85	while start.elapsed().as_secs_f64() < self.duration {
86	match &self.data {
87	BenchmarkData::InMemory(data) => {
88	std::hint::black_box(checksum(self.algorithm, data));
89	}
90	BenchmarkData::File(filename) => {
91	match checksum_file(self.algorithm, filename, None) {
92	Ok(result) => {
93	std::hint::black_box(result);
94	}
95	Err(e) => {
96	return Err(format!("Failed to read file during benchmark: {}", e));
97	}
98	}
99	}
100	}
101	iterations += 1;
102	}
103
104	let elapsed = start.elapsed().as_secs_f64();
105	let data_size = match &self.data {
106	BenchmarkData::InMemory(data) => data.len() as u64,
107	BenchmarkData::File(filename) => std::fs::metadata(filename)
108	.map(\|m\| m.len())
109	.map_err(\|e\| format!("Failed to get file size: {}", e))?,
110	};
111
112	let acceleration_target = crc_fast::get_calculator_target(self.algorithm);
113
114	Ok(BenchmarkResult::new(
115	iterations,
116	elapsed,
117	acceleration_target,
118	data_size,
119	))
120	}
121	}
122
123	impl BenchmarkResult {
124	fn new(
125	iterations: u64,
126	elapsed_seconds: f64,
127	acceleration_target: String,
128	data_size: u64,
129	) -> Self {
130	let throughput_gibs = if elapsed_seconds > 0.0 {
131	(data_size as f64 * iterations as f64) / elapsed_seconds / (1024.0 * 1024.0 * 1024.0)
132	} else {
133	0.0
134	};
135
136	let time_per_iteration_nanos = if iterations > 0 {
137	elapsed_seconds * 1_000_000_000.0 / iterations as f64
138	} else {
139	0.0
140	};
141
142	Self {
143	iterations,
144	elapsed_seconds,
145	throughput_gibs,
146	time_per_iteration_nanos,
147	acceleration_target,
148	data_size,
149	}
150	}
151	}
152
153	/// Fills a buffer with pseudo-random data using xorshift64* algorithm.
154	/// This is a deterministic PRNG that's tiny, fast, and suitable for benchmark data generation.
155	/// The seed is derived from the buffer size to provide some variability.
156	/// This solves the problem of having to use a full-featured PRNG like rand for benchmark data generation.
157	#[inline]
158	fn fill_pseudo_random(buf: &mut [u8], seed: u64) {
159	let mut x = seed;
160	for b in buf {
161	x ^= x >> 12;
162	x ^= x << 25;
163	x ^= x >> 27;
164	let r = x.wrapping_mul(0x2545_F491_4F6C_DD1D);
165	*b = r as u8;
166	}
167	}
168
169	fn generate_random_data(size: usize) -> Result<Vec<u8>, String> {
170	// Check for reasonable size limits to prevent memory issues
171	if size > 1_073_741_824 {
172	// 1 GiB limit
173	return Err("Data size too large (maximum 1 GiB)".to_string());
174	}
175
176	// Use vec! macro to avoid clippy warning about slow initialization
177	let mut buf = vec![0u8; size];
178
179	// Derive seed from size for deterministic but varied data
180	let seed = 0x9E37_79B9_7F4A_7C15_u64.wrapping_add(size as u64);
181	fill_pseudo_random(&mut buf, seed);
182
183	Ok(buf)
184	}
185
186	fn print_usage() {
187	println!("Usage: checksum -a algorithm [-f file] [-s string] [--format hex\|decimal]");
188	println!(
189	" checksum -a algorithm -b [--size bytes] [--duration seconds] [-f file] [-s string]"
190	);
191	println!();
192	println!("Example: checksum -a CRC-32/ISCSI -f myfile.txt");
193	println!("Example: checksum -a CRC-64/NVME -s 'Hello, world!' --format decimal");
194	println!("Example: checksum -a CRC-32/ISCSI -b --size 1048576 --duration 5.0");
195	println!();
196	println!("Options:");
197	println!(" -a algorithm Specify the checksum algorithm (required)");
198	println!(" -f file Calculate checksum for the specified file");
199	println!(" -h, --help Show this help message");
200	println!(" -s string Calculate checksum for the specified string");
201	println!(" --format hex\|decimal Output format (default: hex)");
202	println!();
203	println!("Benchmarking:");
204	println!(" -b Enable benchmark mode");
205	println!(" --duration seconds Benchmark duration in seconds (default: 10.0)");
206	println!(" --size bytes Data size for random generation in benchmark mode (default: 1048576 [1MiB])");
207	println!();
208	println!();
209	println!("Note: In normal mode, either -f or -s must be provided, but not both.");
210	println!(" In benchmark mode (-b), -f or -s are optional for using specific data.");
211	}
212
213	fn parse_args() -> Result<Config, String> {
214	let args: Vec<String> = env::args().collect();
215
216	if args.len() == 1 {
217	return Err("No arguments provided".to_string());
218	}
219
220	// Check for help flag
221	if args.contains(&"-h".to_string()) \|\| args.contains(&"--help".to_string()) {
222	return Err("help".to_string());
223	}
224
225	let mut algorithm: Option<String> = None;
226	let mut file: Option<String> = None;
227	let mut string: Option<String> = None;
228	let mut format = OutputFormat::Hex; // Default to hex
229	let mut benchmark_mode = false;
230	let mut benchmark_size: Option<usize> = None;
231	let mut benchmark_duration = 10.0; // Default duration
232
233	let mut i = 1; // Skip program name
234	while i < args.len() {
235	match args[i].as_str() {
236	"-a" => {
237	if i + 1 >= args.len() {
238	return Err("Missing algorithm after -a flag".to_string());
239	}
240	algorithm = Some(args[i + 1].clone());
241	i += 2;
242	}
243	"-f" => {
244	if i + 1 >= args.len() {
245	return Err("Missing filename after -f flag".to_string());
246	}
247	if string.is_some() {
248	return Err("Cannot specify both -f and -s flags".to_string());
249	}
250	file = Some(args[i + 1].clone());
251	i += 2;
252	}
253	"-s" => {
254	if i + 1 >= args.len() {
255	return Err("Missing string after -s flag".to_string());
256	}
257	if file.is_some() {
258	return Err("Cannot specify both -f and -s flags".to_string());
259	}
260	string = Some(args[i + 1].clone());
261	i += 2;
262	}
263	"--format" => {
264	if i + 1 >= args.len() {
265	return Err("Missing format after --format flag".to_string());
266	}
267	match args[i + 1].as_str() {
268	"hex" => format = OutputFormat::Hex,
269	"decimal" => format = OutputFormat::Decimal,
270	invalid => {
271	return Err(format!(
272	"Invalid format '{}'. Use 'hex' or 'decimal'",
273	invalid
274	))
275	}
276	}
277	i += 2;
278	}
279	"-b" => {
280	benchmark_mode = true;
281	i += 1;
282	}
283	"--size" => {
284	if i + 1 >= args.len() {
285	return Err("Missing size value after --size flag".to_string());
286	}
287	benchmark_size = Some(
288	args[i + 1]
289	.parse::<usize>()
290	.map_err(\|_\| format!("Invalid size value: {}", args[i + 1]))?,
291	);
292	i += 2;
293	}
294	"--duration" => {
295	if i + 1 >= args.len() {
296	return Err("Missing duration value after --duration flag".to_string());
297	}
298	benchmark_duration = args[i + 1]
299	.parse::<f64>()
300	.map_err(\|_\| format!("Invalid duration value: {}", args[i + 1]))?;
301	i += 2;
302	}
303	arg => {
304	return Err(format!("Unknown argument: {}", arg));
305	}
306	}
307	}
308
309	// Validate required arguments
310	let algorithm = algorithm.ok_or("Algorithm (-a) is required")?;
311
312	// Validate mutual exclusivity between benchmark and normal modes
313	if !benchmark_mode && (benchmark_size.is_some() \|\| benchmark_duration != 10.0) {
314	return Err("--size and --duration can only be used with -b flag".to_string());
315	}
316
317	// Create benchmark config if in benchmark mode
318	let benchmark = if benchmark_mode {
319	let config = BenchmarkConfig {
320	size: benchmark_size,
321	duration: benchmark_duration,
322	};
323	config.validate()?;
324	Some(config)
325	} else {
326	None
327	};
328
329	// Validate input requirements based on mode
330	if benchmark.is_none() {
331	// Normal mode: require either file or string input
332	if file.is_none() && string.is_none() {
333	return Err(
334	"Either -f (file) or -s (string) must be provided in normal mode".to_string(),
335	);
336	}
337	}
338	// Benchmark mode: file and string are optional (will use generated data if neither provided)
339
340	Ok(Config {
341	algorithm,
342	file,
343	string,
344	format,
345	benchmark,
346	})
347	}
348
349	fn calculate_checksum(config: &Config) -> Result<(), String> {
350	let algorithm = CrcAlgorithm::from_str(&config.algorithm)
351	.map_err(\|_\| format!("Invalid algorithm: {}", config.algorithm))?;
352
353	// Check if benchmark mode is enabled
354	if let Some(benchmark_config) = &config.benchmark {
355	return run_benchmark(config, benchmark_config, algorithm);
356	}
357
358	let checksum = if let Some(ref filename) = config.file {
359	checksum_file(algorithm, filename, None).unwrap()
360	} else if let Some(ref text) = config.string {
361	checksum(algorithm, text.as_bytes())
362	} else {
363	return Err("No input provided for checksum calculation".to_string());
364	};
365
366	match config.format {
367	OutputFormat::Hex => println!("{:#x?}", checksum),
368	OutputFormat::Decimal => println!("{}", checksum),
369	}
370
371	Ok(())
372	}
373
374	fn run_benchmark(
375	config: &Config,
376	benchmark_config: &BenchmarkConfig,
377	algorithm: CrcAlgorithm,
378	) -> Result<(), String> {
379	// Determine data source and create BenchmarkData
380	let data = if let Some(ref filename) = config.file {
381	// Validate file exists before benchmarking
382	if !std::path::Path::new(filename).exists() {
383	return Err(format!("File not found: {}", filename));
384	}
385	BenchmarkData::File(filename.clone())
386	} else if let Some(ref text) = config.string {
387	BenchmarkData::InMemory(text.as_bytes().to_vec())
388	} else {
389	// Generate random data with specified size or default (1 MiB)
390	let size = benchmark_config.size.unwrap_or(1_048_576);
391	let random_data = generate_random_data(size)?;
392	BenchmarkData::InMemory(random_data)
393	};
394
395	// Create and run benchmark
396	let runner = BenchmarkRunner::new(algorithm, data, benchmark_config.duration);
397	let result = runner.run()?;
398
399	// Display results with algorithm name
400	display_benchmark_results(&result, &config.algorithm);
401
402	Ok(())
403	}
404
405	// Format numbers with comma separators for better readability
406	fn format_number_with_commas(n: u64) -> String {
407	let s = n.to_string();
408	let mut result = String::new();
409	let chars: Vec<char> = s.chars().collect();
410
411	for (i, ch) in chars.iter().enumerate() {
412	if i > 0 && (chars.len() - i).is_multiple_of(3) {
413	result.push(',');
414	}
415	result.push(*ch);
416	}
417
418	result
419	}
420
421	fn display_benchmark_results(result: &BenchmarkResult, algorithm_name: &str) {
422	println!("Algorithm: {}", algorithm_name);
423	println!("Acceleration Target: {}", result.acceleration_target);
424
425	// Format data size with appropriate units
426	let (size_value, size_unit) = if result.data_size >= 1_048_576 {
427	(result.data_size as f64 / 1_048_576.0, "MiB")
428	} else if result.data_size >= 1024 {
429	(result.data_size as f64 / 1024.0, "KiB")
430	} else {
431	(result.data_size as f64, "bytes")
432	};
433
434	println!(
435	"Data Size: {} bytes ({:.1} {})",
436	format_number_with_commas(result.data_size),
437	size_value,
438	size_unit
439	);
440	println!("Duration: {:.2} seconds", result.elapsed_seconds);
441	println!(
442	"Iterations: {}",
443	format_number_with_commas(result.iterations)
444	);
445	println!("Throughput: {:.2} GiB/s", result.throughput_gibs);
446
447	// Format time per iteration with appropriate units
448	let (time_value, time_unit) = if result.time_per_iteration_nanos >= 1_000_000.0 {
449	(result.time_per_iteration_nanos / 1_000_000.0, "ms")
450	} else if result.time_per_iteration_nanos >= 1_000.0 {
451	(result.time_per_iteration_nanos / 1_000.0, "μs")
452	} else {
453	(result.time_per_iteration_nanos, "ns")
454	};
455
456	println!("Time per iteration: {:.1} {}", time_value, time_unit);
457	}
458
459	fn main() -> ExitCode {
460	match parse_args() {
461	Ok(config) => {
462	if let Err(e) = calculate_checksum(&config) {
463	eprintln!("Error: {}", e);
464	return ExitCode::from(1);
465	}
466	}
467	Err(msg) => {
468	if msg == "help" {
469	print_usage();
470	return ExitCode::SUCCESS;
471	} else {
472	eprintln!("Error: {}", msg);
473	println!();
474	print_usage();
475	return ExitCode::from(1);
476	}
477	}
478	}
479
480	ExitCode::SUCCESS
481	}
482
483	#[cfg(test)]
484	mod tests {
485	use super::*;
486	use std::str::FromStr;
487
488	#[test]
489	fn test_benchmark_config_validation_valid() {
490	let config = BenchmarkConfig {
491	size: Some(1024),
492	duration: 5.0,
493	};
494	assert!(config.validate().is_ok());
495	}
496
497	#[test]
498	fn test_benchmark_config_validation_zero_duration() {
499	let config = BenchmarkConfig {
500	size: Some(1024),
501	duration: 0.0,
502	};
503	assert!(config.validate().is_err());
504	assert_eq!(
505	config.validate().unwrap_err(),
506	"Duration must be greater than 0"
507	);
508	}
509
510	#[test]
511	fn test_benchmark_config_validation_negative_duration() {
512	let config = BenchmarkConfig {
513	size: Some(1024),
514	duration: -1.0,
515	};
516	assert!(config.validate().is_err());
517	assert_eq!(
518	config.validate().unwrap_err(),
519	"Duration must be greater than 0"
520	);
521	}
522
523	#[test]
524	fn test_benchmark_config_validation_zero_size() {
525	let config = BenchmarkConfig {
526	size: Some(0),
527	duration: 5.0,
528	};
529	assert!(config.validate().is_err());
530	assert_eq!(
531	config.validate().unwrap_err(),
532	"Size must be greater than 0"
533	);
534	}
535
536	#[test]
537	fn test_benchmark_config_validation_none_size() {
538	let config = BenchmarkConfig {
539	size: None,
540	duration: 5.0,
541	};
542	assert!(config.validate().is_ok());
543	}
544
545	#[test]
546	fn test_generate_random_data_valid_size() {
547	let data = generate_random_data(1024).unwrap();
548	assert_eq!(data.len(), 1024);
549	}
550
551	#[test]
552	fn test_generate_random_data_large_size() {
553	let result = generate_random_data(1_073_741_825); // 1 GiB + 1
554	assert!(result.is_err());
555	assert_eq!(result.unwrap_err(), "Data size too large (maximum 1 GiB)");
556	}
557
558	#[test]
559	fn test_benchmark_result_calculation() {
560	let result = BenchmarkResult::new(
561	1000, // iterations
562	2.0, // elapsed_seconds
563	"test".to_string(), // acceleration_target
564	1024, // data_size
565	);
566
567	assert_eq!(result.iterations, 1000);
568	assert_eq!(result.elapsed_seconds, 2.0);
569	assert_eq!(result.data_size, 1024);
570
571	// Throughput should be (1024 * 1000) / 2.0 / (1024^3) GiB/s
572	let expected_throughput = (1024.0 * 1000.0) / 2.0 / (1024.0 * 1024.0 * 1024.0);
573	assert!((result.throughput_gibs - expected_throughput).abs() < 1e-10);
574
575	// Time per iteration should be 2.0 * 1e9 / 1000 nanoseconds
576	let expected_time_per_iter = 2.0 * 1_000_000_000.0 / 1000.0;
577	assert!((result.time_per_iteration_nanos - expected_time_per_iter).abs() < 1e-6);
578	}
579
580	#[test]
581	fn test_benchmark_runner_creation() {
582	let algorithm = CrcAlgorithm::from_str("CRC-32/ISCSI").unwrap();
583	let data = BenchmarkData::InMemory(vec![1, 2, 3, 4]);
584	let runner = BenchmarkRunner::new(algorithm, data, 1.0);
585
586	assert_eq!(runner.duration, 1.0);
587	assert_eq!(runner.algorithm, algorithm);
588	}
589
590	#[test]
591	fn test_benchmark_runner_execution_in_memory() {
592	let algorithm = CrcAlgorithm::from_str("CRC-32/ISCSI").unwrap();
593	let data = BenchmarkData::InMemory(vec![1, 2, 3, 4]);
594	let runner = BenchmarkRunner::new(algorithm, data, 0.1); // Short duration for test
595
596	let result = runner.run().unwrap();
597	assert!(result.iterations > 0);
598	assert!(result.elapsed_seconds > 0.0);
599	assert_eq!(result.data_size, 4);
600	assert!(result.throughput_gibs >= 0.0);
601	}
602
603	#[test]
604	fn test_parse_args_benchmark_mode() {
605	// We can't easily mock std::env::args in unit tests, so we'll test the parsing logic
606	// by creating a Config directly and validating its structure
607	let config = Config {
608	algorithm: "CRC-32/ISCSI".to_string(),
609	file: None,
610	string: None,
611	format: OutputFormat::Hex,
612	benchmark: Some(BenchmarkConfig {
613	size: Some(1024),
614	duration: 5.0,
615	}),
616	};
617
618	assert!(config.benchmark.is_some());
619	let benchmark_config = config.benchmark.unwrap();
620	assert_eq!(benchmark_config.size, Some(1024));
621	assert_eq!(benchmark_config.duration, 5.0);
622	}
623
624	#[test]
625	fn test_parse_args_normal_mode() {
626	let config = Config {
627	algorithm: "CRC-32/ISCSI".to_string(),
628	file: Some("test.txt".to_string()),
629	string: None,
630	format: OutputFormat::Hex,
631	benchmark: None,
632	};
633
634	assert!(config.benchmark.is_none());
635	assert_eq!(config.file, Some("test.txt".to_string()));
636	}
637
638	#[test]
639	fn test_data_source_selection_file() {
640	// Test that file input creates File variant
641	let config = Config {
642	algorithm: "CRC-32/ISCSI".to_string(),
643	file: Some("test.txt".to_string()),
644	string: None,
645	format: OutputFormat::Hex,
646	benchmark: Some(BenchmarkConfig {
647	size: None,
648	duration: 1.0,
649	}),
650	};
651
652	// This would be tested in the run_benchmark function
653	// We can verify the config structure is correct for file input
654	assert!(config.file.is_some());
655	assert!(config.string.is_none());
656	}
657
658	#[test]
659	fn test_data_source_selection_string() {
660	let config = Config {
661	algorithm: "CRC-32/ISCSI".to_string(),
662	file: None,
663	string: Some("test data".to_string()),
664	format: OutputFormat::Hex,
665	benchmark: Some(BenchmarkConfig {
666	size: None,
667	duration: 1.0,
668	}),
669	};
670
671	assert!(config.file.is_none());
672	assert!(config.string.is_some());
673	}
674
675	#[test]
676	fn test_data_source_selection_generated() {
677	let config = Config {
678	algorithm: "CRC-32/ISCSI".to_string(),
679	file: None,
680	string: None,
681	format: OutputFormat::Hex,
682	benchmark: Some(BenchmarkConfig {
683	size: Some(1024),
684	duration: 1.0,
685	}),
686	};
687
688	// When neither file nor string is provided, generated data should be used
689	assert!(config.file.is_none());
690	assert!(config.string.is_none());
691	assert_eq!(config.benchmark.as_ref().unwrap().size, Some(1024));
692	}
693
694	#[test]
695	fn test_throughput_calculation_accuracy() {
696	// Test with known values to verify calculation accuracy
697	let data_size = 1_048_576u64; // 1 MiB
698	let iterations = 1000u64;
699	let elapsed_seconds = 1.0;
700
701	let result =
702	BenchmarkResult::new(iterations, elapsed_seconds, "test".to_string(), data_size);
703
704	// Expected throughput: (1 MiB * 1000) / 1 second = 1000 MiB/s = ~0.9537 GiB/s
705	let expected_gibs =
706	(data_size as f64 * iterations as f64) / elapsed_seconds / (1024.0 * 1024.0 * 1024.0);
707	assert!((result.throughput_gibs - expected_gibs).abs() < 1e-10);
708	}
709
710	#[test]
711	fn test_output_format_variants() {
712	let hex_format = OutputFormat::Hex;
713	let decimal_format = OutputFormat::Decimal;
714
715	// Test that both variants can be created and are different
716	assert!(matches!(hex_format, OutputFormat::Hex));
717	assert!(matches!(decimal_format, OutputFormat::Decimal));
718	}
719
720	#[test]
721	fn test_format_number_with_commas() {
722	assert_eq!(format_number_with_commas(0), "0");
723	assert_eq!(format_number_with_commas(123), "123");
724	assert_eq!(format_number_with_commas(1234), "1,234");
725	assert_eq!(format_number_with_commas(12345), "12,345");
726	assert_eq!(format_number_with_commas(123456), "123,456");
727	assert_eq!(format_number_with_commas(1234567), "1,234,567");
728	assert_eq!(format_number_with_commas(12345678), "12,345,678");
729	assert_eq!(format_number_with_commas(123456789), "123,456,789");
730	assert_eq!(format_number_with_commas(1000000000), "1,000,000,000");
731	}
732	}

Keyboard