//! The "version 2" disk representation of the dirstate

//!

//! # File format

//!

//! In dirstate-v2 format, the `.hg/dirstate` file is a "docket that starts

//! with a fixed-sized header whose layout is defined by the `DocketHeader`

//! struct, followed by the data file identifier.

//!

//! A separate `.hg/dirstate.{uuid}.d` file contains most of the data. That

//! file may be longer than the size given in the docket, but not shorter. Only

//! the start of the data file up to the given size is considered. The

//! fixed-size "root" of the dirstate tree whose layout is defined by the

//! `Root` struct is found at the end of that slice of data.

//!

//! Its `root_nodes` field contains the slice (offset and length) to

//! the nodes representing the files and directories at the root of the

//! repository. Each node is also fixed-size, defined by the `Node` struct.

//! Nodes in turn contain slices to variable-size paths, and to their own child

//! nodes (if any) for nested files and directories.

use crate::dirstate_tree::dirstate_map::{self, DirstateMap, NodeRef};

use crate::dirstate_tree::path_with_basename::WithBasename;

use crate::errors::HgError;

use crate::utils::hg_path::HgPath;

use crate::DirstateEntry;

use crate::DirstateError;

use crate::DirstateParents;

use crate::EntryState;

use bytes_cast::unaligned::{I32Be, I64Be, U16Be, U32Be};

use bytes_cast::BytesCast;

use format_bytes::format_bytes;

use std::borrow::Cow;

use std::convert::{TryFrom, TryInto};

use std::time::{Duration, SystemTime, UNIX_EPOCH};

/// Added at the start of `.hg/dirstate` when the "v2" format is used.

/// This a redundant sanity check more than an actual "magic number" since

/// `.hg/requires` already governs which format should be used.

pub const V2_FORMAT_MARKER: &[u8; 12] = b"dirstate-v2\n";

/// Keep space for 256-bit hashes

const STORED_NODE_ID_BYTES: usize = 32;

/// … even though only 160 bits are used for now, with SHA-1

const USED_NODE_ID_BYTES: usize = 20;

pub(super) const IGNORE_PATTERNS_HASH_LEN: usize = 20;

pub(super) type IgnorePatternsHash = [u8; IGNORE_PATTERNS_HASH_LEN];

// Must match `HEADER` in `mercurial/dirstateutils/docket.py`

#[derive(BytesCast)]

#[repr(C)]

struct DocketHeader {

    marker: [u8; V2_FORMAT_MARKER.len()],

    parent_1: [u8; STORED_NODE_ID_BYTES],

    parent_2: [u8; STORED_NODE_ID_BYTES],

    /// Counted in bytes

    data_size: Size,

    uuid_size: u8,

}

pub struct Docket<'on_disk> {

    header: &'on_disk DocketHeader,

    uuid: &'on_disk [u8],

}

#[derive(BytesCast)]

#[repr(C)]

    root_nodes: ChildNodes,

    nodes_with_entry_count: Size,

    nodes_with_copy_source_count: Size,

    /// How many bytes of this data file are not used anymore

    unreachable_bytes: Size,

    /// If non-zero, a hash of ignore files that were used for some previous

    /// run of the `status` algorithm.

    ///

    /// We define:

    ///

    /// * "Root" ignore files are `.hgignore` at the root of the repository if

    ///   it exists, and files from `ui.ignore.*` config. This set of files is

    ///   then sorted by the string representation of their path.

    /// * The "expanded contents" of an ignore files is the byte string made

    ///   by concatenating its contents with the "expanded contents" of other

    ///   files included with `include:` or `subinclude:` files, in inclusion

    ///   order. This definition is recursive, as included files can

    ///   themselves include more files.

    ///

    /// This hash is defined as the SHA-1 of the concatenation (in sorted

    /// order) of the "expanded contents" of each "root" ignore file.

    /// (Note that computing this does not require actually concatenating byte

    /// strings into contiguous memory, instead SHA-1 hashing can be done

    /// incrementally.)

    ignore_patterns_hash: IgnorePatternsHash,

}

#[derive(BytesCast)]

#[repr(C)]

pub(super) struct Node {

    full_path: PathSlice,

    /// In bytes from `self.full_path.start`

    base_name_start: PathSize,

    copy_source: OptPathSlice,

    children: ChildNodes,

    pub(super) descendants_with_entry_count: Size,

    pub(super) tracked_descendants_count: Size,

    /// Depending on the value of `state`:

    ///

    /// * A null byte: `data` is not used.

    ///

    /// * A `n`, `a`, `r`, or `m` ASCII byte: `state` and `data` together

    ///   represent a dirstate entry like in the v1 format.

    ///

    /// * A `d` ASCII byte: the bytes of `data` should instead be interpreted

    ///   as the `Timestamp` for the mtime of a cached directory.

    ///

    ///   The presence of this state means that at some point, this path in

    ///   the working directory was observed:

    ///

    ///   - To be a directory

    ///   - With the modification time as given by `Timestamp`

    ///   - That timestamp was already strictly in the past when observed,

    ///     meaning that later changes cannot happen in the same clock tick

    ///     and must cause a different modification time (unless the system

    ///     clock jumps back and we get unlucky, which is not impossible but

    ///     but deemed unlikely enough).

    ///   - All direct children of this directory (as returned by

    ///     `std::fs::read_dir`) either have a corresponding dirstate node, or

    ///     are ignored by ignore patterns whose hash is in

    ///

    ///   This means that if `std::fs::symlink_metadata` later reports the

    ///   same modification time and ignored patterns haven’t changed, a run

    ///   of status that is not listing ignored   files can skip calling

    ///   `std::fs::read_dir` again for this directory,   iterate child

    ///   dirstate nodes instead.

    state: u8,

    data: Entry,

}

#[derive(BytesCast, Copy, Clone)]

#[repr(C)]

struct Entry {

    mode: I32Be,

    mtime: I32Be,

    size: I32Be,

}

/// Duration since the Unix epoch

#[derive(BytesCast, Copy, Clone, PartialEq)]

#[repr(C)]

pub(super) struct Timestamp {

    seconds: I64Be,

    /// In `0 .. 1_000_000_000`.

    ///

    /// This timestamp is later or earlier than `(seconds, 0)` by this many

    /// nanoseconds, if `seconds` is non-negative or negative, respectively.

    nanoseconds: U32Be,

}

/// Counted in bytes from the start of the file

///

/// NOTE: not supporting `.hg/dirstate` files larger than 4 GiB.

type Offset = U32Be;

/// Counted in number of items

///

/// NOTE: we choose not to support counting more than 4 billion nodes anywhere.

type Size = U32Be;

/// Counted in bytes

///

/// NOTE: we choose not to support file names/paths longer than 64 KiB.

type PathSize = U16Be;

/// A contiguous sequence of `len` times `Node`, representing the child nodes

/// of either some other node or of the repository root.

///

/// Always sorted by ascending `full_path`, to allow binary search.

/// Since nodes with the same parent nodes also have the same parent path,

/// only the `base_name`s need to be compared during binary search.

#[derive(BytesCast, Copy, Clone)]

#[repr(C)]

struct ChildNodes {

    start: Offset,

    len: Size,

}

/// A `HgPath` of `len` bytes

#[derive(BytesCast, Copy, Clone)]

#[repr(C)]

struct PathSlice {

    start: Offset,

    len: PathSize,

}

/// Either nothing if `start == 0`, or a `HgPath` of `len` bytes

type OptPathSlice = PathSlice;

/// Unexpected file format found in `.hg/dirstate` with the "v2" format.

///

/// This should only happen if Mercurial is buggy or a repository is corrupted.

#[derive(Debug)]

pub struct DirstateV2ParseError;

impl From<DirstateV2ParseError> for HgError {

    fn from(_: DirstateV2ParseError) -> Self {

        HgError::corrupted("dirstate-v2 parse error")

    }

}

impl From<DirstateV2ParseError> for crate::DirstateError {

    fn from(error: DirstateV2ParseError) -> Self {

        HgError::from(error).into()

    }

}

impl<'on_disk> Docket<'on_disk> {

    pub fn parents(&self) -> DirstateParents {

        use crate::Node;

        let p1 = Node::try_from(&self.header.parent_1[..USED_NODE_ID_BYTES])

            .unwrap()

            .clone();

        let p2 = Node::try_from(&self.header.parent_2[..USED_NODE_ID_BYTES])

            .unwrap()

            .clone();

        DirstateParents { p1, p2 }

    }

    pub fn data_size(&self) -> usize {

        // This `unwrap` could only panic on a 16-bit CPU

        self.header.data_size.get().try_into().unwrap()

    }

    pub fn data_filename(&self) -> String {

        String::from_utf8(format_bytes!(b"dirstate.{}.d", self.uuid)).unwrap()

    }

}

pub fn read_docket(

    on_disk: &[u8],

) -> Result<Docket<'_>, DirstateV2ParseError> {

    let (header, uuid) =

        DocketHeader::from_bytes(on_disk).map_err(|_| DirstateV2ParseError)?;

    let uuid_size = header.uuid_size as usize;

    if header.marker == *V2_FORMAT_MARKER && uuid.len() == uuid_size {

        Ok(Docket { header, uuid })

    } else {

        Err(DirstateV2ParseError)

    }

}

pub(super) fn read<'on_disk>(

    on_disk: &'on_disk [u8],

) -> Result<DirstateMap<'on_disk>, DirstateV2ParseError> {

    if on_disk.is_empty() {

        return Ok(DirstateMap::empty(on_disk));

    }

    let dirstate_map = DirstateMap {

        on_disk,

        root: dirstate_map::ChildNodes::OnDisk(read_nodes(

            on_disk,

        )?),

    };

    Ok(dirstate_map)

}

impl Node {

    pub(super) fn full_path<'on_disk>(

        &self,

        on_disk: &'on_disk [u8],

    ) -> Result<&'on_disk HgPath, DirstateV2ParseError> {

        read_hg_path(on_disk, self.full_path)

    }

    pub(super) fn base_name_start<'on_disk>(

        &self,

    ) -> Result<usize, DirstateV2ParseError> {

        let start = self.base_name_start.get();

        if start < self.full_path.len.get() {

            let start = usize::try_from(start)

                // u32 -> usize, could only panic on a 16-bit CPU

                .expect("dirstate-v2 base_name_start out of bounds");

            Ok(start)

        } else {

            Err(DirstateV2ParseError)

        }

    }

    pub(super) fn base_name<'on_disk>(

        &self,

        on_disk: &'on_disk [u8],

    ) -> Result<&'on_disk HgPath, DirstateV2ParseError> {

        let full_path = self.full_path(on_disk)?;

        let base_name_start = self.base_name_start()?;

        Ok(HgPath::new(&full_path.as_bytes()[base_name_start..]))

    }

    pub(super) fn path<'on_disk>(

        &self,

        on_disk: &'on_disk [u8],

    ) -> Result<dirstate_map::NodeKey<'on_disk>, DirstateV2ParseError> {

        Ok(WithBasename::from_raw_parts(

            Cow::Borrowed(self.full_path(on_disk)?),

            self.base_name_start()?,

        ))

    }

    pub(super) fn has_copy_source<'on_disk>(&self) -> bool {

        self.copy_source.start.get() != 0

    }

    pub(super) fn copy_source<'on_disk>(

        &self,

        on_disk: &'on_disk [u8],

    ) -> Result<Option<&'on_disk HgPath>, DirstateV2ParseError> {

        Ok(if self.has_copy_source() {

            Some(read_hg_path(on_disk, self.copy_source)?)

        } else {

            None

        })

    }

    pub(super) fn node_data(

        &self,

    ) -> Result<dirstate_map::NodeData, DirstateV2ParseError> {

        let entry = |state| {

            dirstate_map::NodeData::Entry(self.entry_with_given_state(state))

        };

        match self.state {

            b'\0' => Ok(dirstate_map::NodeData::None),

            b'd' => Ok(dirstate_map::NodeData::CachedDirectory {

                mtime: *self.data.as_timestamp(),

            }),