bson 2.5.0

//! ObjectId

use std::{
    convert::TryInto,
    error,
    fmt,
    result,
    str::FromStr,
    sync::atomic::{AtomicUsize, Ordering},
    time::SystemTime,
};

use hex::{self, FromHexError};
use rand::{thread_rng, Rng};

use lazy_static::lazy_static;

const TIMESTAMP_SIZE: usize = 4;
const PROCESS_ID_SIZE: usize = 5;
const COUNTER_SIZE: usize = 3;

const TIMESTAMP_OFFSET: usize = 0;
const PROCESS_ID_OFFSET: usize = TIMESTAMP_OFFSET + TIMESTAMP_SIZE;
const COUNTER_OFFSET: usize = PROCESS_ID_OFFSET + PROCESS_ID_SIZE;

const MAX_U24: usize = 0xFF_FFFF;

lazy_static! {
    static ref OID_COUNTER: AtomicUsize = AtomicUsize::new(thread_rng().gen_range(0..=MAX_U24));
}

/// Errors that can occur during [`ObjectId`] construction and generation.
#[derive(Clone, Debug)]
#[non_exhaustive]
pub enum Error {
    /// An invalid character was found in the provided hex string. Valid characters are: `0...9`,
    /// `a...f`, or `A...F`.
    #[non_exhaustive]
    InvalidHexStringCharacter { c: char, index: usize, hex: String },

    /// An [`ObjectId`]'s hex string representation must be an exactly 12-byte (24-char)
    /// hexadecimal string.
    #[non_exhaustive]
    InvalidHexStringLength { length: usize, hex: String },
}

/// Alias for Result<T, oid::Error>.
pub type Result<T> = result::Result<T, Error>;

impl fmt::Display for Error {
    fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
        match self {
            Error::InvalidHexStringCharacter { c, index, hex } => {
                write!(
                    fmt,
                    "invalid character '{}' was found at index {} in the provided hex string: \
                     \"{}\"",
                    c, index, hex
                )
            }
            Error::InvalidHexStringLength { length, hex } => {
                write!(
                    fmt,
                    "provided hex string representation must be exactly 12 bytes, instead got: \
                     \"{}\", length {}",
                    hex, length
                )
            }
        }
    }
}

impl error::Error for Error {}

/// A wrapper around raw 12-byte ObjectId representations.
#[derive(Clone, Copy, PartialEq, PartialOrd, Eq, Ord, Hash)]
pub struct ObjectId {
    id: [u8; 12],
}

impl Default for ObjectId {
    fn default() -> Self {
        Self::new()
    }
}

impl FromStr for ObjectId {
    type Err = Error;

    fn from_str(s: &str) -> std::result::Result<Self, Self::Err> {
        Self::parse_str(s)
    }
}

impl From<[u8; 12]> for ObjectId {
    fn from(bytes: [u8; 12]) -> Self {
        Self { id: bytes }
    }
}

impl ObjectId {
    /// Generates a new [`ObjectId`], represented in bytes.
    /// See the [docs](http://www.mongodb.com/docs/manual/reference/object-id/)
    /// for more information.
    pub fn new() -> ObjectId {
        let timestamp = ObjectId::gen_timestamp();
        let process_id = ObjectId::gen_process_id();
        let counter = ObjectId::gen_count();

        let mut buf: [u8; 12] = [0; 12];
        buf[TIMESTAMP_OFFSET..(TIMESTAMP_SIZE + TIMESTAMP_OFFSET)]
            .clone_from_slice(&timestamp[..TIMESTAMP_SIZE]);
        buf[PROCESS_ID_OFFSET..(PROCESS_ID_SIZE + PROCESS_ID_OFFSET)]
            .clone_from_slice(&process_id[..PROCESS_ID_SIZE]);
        buf[COUNTER_OFFSET..(COUNTER_SIZE + COUNTER_OFFSET)]
            .clone_from_slice(&counter[..COUNTER_SIZE]);

        ObjectId::from_bytes(buf)
    }

    /// Constructs a new ObjectId wrapper around the raw byte representation.
    pub const fn from_bytes(bytes: [u8; 12]) -> ObjectId {
        ObjectId { id: bytes }
    }

    /// Creates an ObjectID using a 12-byte (24-char) hexadecimal string.
    pub fn parse_str(s: impl AsRef<str>) -> Result<ObjectId> {
        let s = s.as_ref();

        let bytes: Vec<u8> = hex::decode(s.as_bytes()).map_err(|e| match e {
            FromHexError::InvalidHexCharacter { c, index } => Error::InvalidHexStringCharacter {
                c,
                index,
                hex: s.to_string(),
            },
            FromHexError::InvalidStringLength | FromHexError::OddLength => {
                Error::InvalidHexStringLength {
                    length: s.len(),
                    hex: s.to_string(),
                }
            }
        })?;
        if bytes.len() != 12 {
            Err(Error::InvalidHexStringLength {
                length: s.len(),
                hex: s.to_string(),
            })
        } else {
            let mut byte_array: [u8; 12] = [0; 12];
            byte_array[..].copy_from_slice(&bytes[..]);
            Ok(ObjectId::from_bytes(byte_array))
        }
    }

    /// Retrieves the timestamp from an [`ObjectId`].
    pub fn timestamp(&self) -> crate::DateTime {
        let mut buf = [0; 4];
        buf.copy_from_slice(&self.id[0..4]);
        let seconds_since_epoch = u32::from_be_bytes(buf);

        // This doesn't overflow since u32::MAX * 1000 < i64::MAX
        crate::DateTime::from_millis(seconds_since_epoch as i64 * 1000)
    }

    /// Returns the raw byte representation of an ObjectId.
    pub const fn bytes(&self) -> [u8; 12] {
        self.id
    }

    /// Convert this [`ObjectId`] to its hex string representation.
    pub fn to_hex(self) -> String {
        hex::encode(self.id)
    }

    /// Generates a new timestamp representing the current seconds since epoch.
    /// Represented in Big Endian.
    fn gen_timestamp() -> [u8; 4] {
        let timestamp: u32 = SystemTime::now()
            .duration_since(SystemTime::UNIX_EPOCH)
            .expect("system clock is before 1970")
            .as_secs()
            .try_into()
            .unwrap(); // will succeed until 2106 since timestamp is unsigned
        timestamp.to_be_bytes()
    }

    /// Generate a random 5-byte array.
    fn gen_process_id() -> [u8; 5] {
        lazy_static! {
            static ref BUF: [u8; 5] = thread_rng().gen();
        }

        *BUF
    }

    /// Gets an incremental 3-byte count.
    /// Represented in Big Endian.
    fn gen_count() -> [u8; 3] {
        let u_counter = OID_COUNTER.fetch_add(1, Ordering::SeqCst);

        // Mod result instead of OID_COUNTER to prevent threading issues.
        let u = u_counter % (MAX_U24 + 1);

        // Convert usize to writable u64, then extract the first three bytes.
        let u_int = u as u64;

        let buf = u_int.to_be_bytes();
        let buf_u24: [u8; 3] = [buf[5], buf[6], buf[7]];
        buf_u24
    }
}

impl fmt::Display for ObjectId {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str(&self.to_hex())
    }
}

impl fmt::Debug for ObjectId {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.debug_tuple("ObjectId").field(&self.to_hex()).finish()
    }
}

#[cfg(test)]
use crate::tests::LOCK;

#[test]
fn count_generated_is_big_endian() {
    let _guard = LOCK.run_exclusively();
    let start = 1_122_866;
    OID_COUNTER.store(start, Ordering::SeqCst);

    // Test count generates correct value 1122866
    let count_bytes = ObjectId::gen_count();

    let mut buf: [u8; 4] = [0; 4];
    buf[1..=COUNTER_SIZE].clone_from_slice(&count_bytes[..COUNTER_SIZE]);

    let count = u32::from_be_bytes(buf);
    assert_eq!(start as u32, count);

    // Test OID formats count correctly as big endian
    let oid = ObjectId::new();

    assert_eq!(0x11u8, oid.bytes()[COUNTER_OFFSET]);
    assert_eq!(0x22u8, oid.bytes()[COUNTER_OFFSET + 1]);
    assert_eq!(0x33u8, oid.bytes()[COUNTER_OFFSET + 2]);
}

#[test]
fn test_counter_overflow_u24_max() {
    let _guard = LOCK.run_exclusively();
    let start = MAX_U24;
    OID_COUNTER.store(start, Ordering::SeqCst);
    let oid = ObjectId::new();
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET]);
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET + 1]);
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET + 2]);
    // Test counter overflows to 0 when set to MAX_24 + 1
    let oid_new = ObjectId::new();
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET]);
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET + 1]);
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET + 2]);
}

#[test]
fn test_counter_overflow_usize_max() {
    let _guard = LOCK.run_exclusively();
    let start = usize::max_value();
    OID_COUNTER.store(start, Ordering::SeqCst);
    // Test counter overflows to u24_max when set to usize_max
    let oid = ObjectId::new();
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET]);
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET + 1]);
    assert_eq!(0xFFu8, oid.bytes()[COUNTER_OFFSET + 2]);
    // Test counter overflows to 0 when set to usize_max + 1
    let oid_new = ObjectId::new();
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET]);
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET + 1]);
    assert_eq!(0x00u8, oid_new.bytes()[COUNTER_OFFSET + 2]);
}

#[cfg(test)]
mod test {
    use time::macros::datetime;

    #[test]
    fn test_display() {
        let id = super::ObjectId::parse_str("53e37d08776f724e42000000").unwrap();

        assert_eq!(format!("{}", id), "53e37d08776f724e42000000")
    }

    #[test]
    fn test_debug() {
        let id = super::ObjectId::parse_str("53e37d08776f724e42000000").unwrap();

        assert_eq!(
            format!("{:?}", id),
            "ObjectId(\"53e37d08776f724e42000000\")"
        );
        assert_eq!(
            format!("{:#?}", id),
            "ObjectId(\n    \"53e37d08776f724e42000000\",\n)"
        );
    }

    #[test]
    fn test_timestamp() {
        let id = super::ObjectId::parse_str("000000000000000000000000").unwrap();
        // "Jan 1st, 1970 00:00:00 UTC"
        assert_eq!(datetime!(1970-01-01 0:00 UTC), id.timestamp().to_time_0_3());

        let id = super::ObjectId::parse_str("7FFFFFFF0000000000000000").unwrap();
        // "Jan 19th, 2038 03:14:07 UTC"
        assert_eq!(
            datetime!(2038-01-19 3:14:07 UTC),
            id.timestamp().to_time_0_3()
        );

        let id = super::ObjectId::parse_str("800000000000000000000000").unwrap();
        // "Jan 19th, 2038 03:14:08 UTC"
        assert_eq!(
            datetime!(2038-01-19 3:14:08 UTC),
            id.timestamp().to_time_0_3()
        );

        let id = super::ObjectId::parse_str("FFFFFFFF0000000000000000").unwrap();
        // "Feb 7th, 2106 06:28:15 UTC"
        assert_eq!(
            datetime!(2106-02-07 6:28:15 UTC),
            id.timestamp().to_time_0_3()
        );
    }
}