1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171

use std::{
    cmp::Ordering,
    path::{Path, PathBuf},
};

use crate::{
    data,
    index::PrefixLookupResult,
    multi_index::{EntryIndex, File, PackIndex, Version},
};

/// Represents an entry within a multi index file, effectively mapping object [`IDs`][git_hash::ObjectId] to pack data
/// files and the offset within.
#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone)]
#[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
pub struct Entry {
    /// The ID of the object.
    pub oid: git_hash::ObjectId,
    /// The offset to the object's header in the pack data file.
    pub pack_offset: data::Offset,
    /// The index of the pack matching our [`File::index_names()`] slice.
    pub pack_index: PackIndex,
}

/// Access methods
impl File {
    /// Returns the verion of the multi-index file.
    pub fn version(&self) -> Version {
        self.version
    }
    /// Returns the path from which the multi-index file was loaded.
    ///
    /// Note that it might have changed in the mean time, or might have been removed as well.
    pub fn path(&self) -> &Path {
        &self.path
    }
    /// Returns the amount of indices stored in this multi-index file. It's the same as [File::index_names().len()][File::index_names()],
    /// and returned as one past the highest known index.
    pub fn num_indices(&self) -> PackIndex {
        self.num_indices
    }
    /// Returns the total amount of objects available for lookup, and returned as one past the highest known entry index
    pub fn num_objects(&self) -> EntryIndex {
        self.num_objects
    }
    /// Returns the kind of hash function used for object ids available in this index.
    pub fn object_hash(&self) -> git_hash::Kind {
        self.object_hash
    }
    /// Returns the checksum over the entire content of the file (excluding the checksum itself).
    ///
    /// It can be used to validate it didn't change after creation.
    pub fn checksum(&self) -> git_hash::ObjectId {
        git_hash::ObjectId::from(&self.data[self.data.len() - self.hash_len..])
    }
    /// Return all names of index files (`*.idx`) whose objects we contain.
    ///
    /// The corresponding pack can be found by replacing the `.idx` extension with `.pack`.
    pub fn index_names(&self) -> &[PathBuf] {
        &self.index_names
    }
}

impl File {
    /// Return the object id at the given `index`, which ranges from 0 to [File::num_objects()].
    pub fn oid_at_index(&self, index: EntryIndex) -> &git_hash::oid {
        debug_assert!(index < self.num_objects, "index out of bounds");
        let index: usize = index as usize;
        let start = self.lookup_ofs + index * self.hash_len;
        git_hash::oid::from_bytes_unchecked(&self.data[start..][..self.hash_len])
    }

    /// Given a `prefix`, find an object that matches it uniquely within this index and return `Some(Ok(entry_index))`.
    /// If there is more than one object matching the object `Some(Err(())` is returned.
    ///
    /// Finally, if no object matches the index, the return value is `None`.
    // NOTE: pretty much the same things as in `index::File::lookup`, change things there
    //       as well.
    pub fn lookup_prefix(&self, prefix: git_hash::Prefix) -> Option<PrefixLookupResult> {
        let first_byte = prefix.as_oid().first_byte() as usize;
        let mut upper_bound = self.fan[first_byte];
        let mut lower_bound = if first_byte != 0 { self.fan[first_byte - 1] } else { 0 };

        // Bisect using indices
        while lower_bound < upper_bound {
            let mid = (lower_bound + upper_bound) / 2;
            let mid_sha = self.oid_at_index(mid);

            use std::cmp::Ordering::*;
            match prefix.cmp_oid(mid_sha) {
                Less => upper_bound = mid,
                Equal => {
                    let next = mid + 1;
                    if next < self.num_objects && prefix.cmp_oid(self.oid_at_index(next)) == Ordering::Equal {
                        return Some(Err(()));
                    }
                    if mid != 0 && prefix.cmp_oid(self.oid_at_index(mid - 1)) == Ordering::Equal {
                        return Some(Err(()));
                    }
                    return Some(Ok(mid));
                }
                Greater => lower_bound = mid + 1,
            }
        }
        None
    }

    /// Find the index ranging from 0 to [File::num_objects()] that belongs to data associated with `id`, or `None` if it wasn't found.
    ///
    /// Use this index for finding additional information via [`File::pack_id_and_pack_offset_at_index()`].
    pub fn lookup(&self, id: impl AsRef<git_hash::oid>) -> Option<EntryIndex> {
        let id = id.as_ref();
        let first_byte = id.first_byte() as usize;
        let mut upper_bound = self.fan[first_byte];
        let mut lower_bound = if first_byte != 0 { self.fan[first_byte - 1] } else { 0 };

        // Bisect using indices
        while lower_bound < upper_bound {
            let mid = (lower_bound + upper_bound) / 2;
            let mid_sha = self.oid_at_index(mid);

            use std::cmp::Ordering::*;
            match id.cmp(mid_sha) {
                Less => upper_bound = mid,
                Equal => return Some(mid),
                Greater => lower_bound = mid + 1,
            }
        }
        None
    }

    /// Given the `index` ranging from 0 to [File::num_objects()], return the pack index and its absolute offset into the pack.
    ///
    /// The pack-index refers to an entry in the [`index_names`][File::index_names()] list, from which the pack can be derived.
    pub fn pack_id_and_pack_offset_at_index(&self, index: EntryIndex) -> (PackIndex, data::Offset) {
        const OFFSET_ENTRY_SIZE: usize = 4 + 4;
        let index = index as usize;
        let start = self.offsets_ofs + index * OFFSET_ENTRY_SIZE;

        const HIGH_BIT: u32 = 1 << 31;

        let pack_index = crate::read_u32(&self.data[start..][..4]);
        let offset = &self.data[start + 4..][..4];
        let ofs32 = crate::read_u32(offset);
        let pack_offset = if (ofs32 & HIGH_BIT) == HIGH_BIT {
            // We determine if large offsets are actually larger than 4GB and if not, we don't use the high-bit to signal anything
            // but allow the presence of the large-offset chunk to signal what's happening.
            if let Some(offsets_64) = self.large_offsets_ofs {
                let from = offsets_64 + (ofs32 ^ HIGH_BIT) as usize * 8;
                crate::read_u64(&self.data[from..][..8])
            } else {
                ofs32 as u64
            }
        } else {
            ofs32 as u64
        };
        (pack_index, pack_offset)
    }

    /// Return an iterator over all entries within this file.
    pub fn iter(&self) -> impl Iterator<Item = Entry> + '_ {
        (0..self.num_objects).map(move |idx| {
            let (pack_index, pack_offset) = self.pack_id_and_pack_offset_at_index(idx);
            Entry {
                oid: self.oid_at_index(idx).to_owned(),
                pack_offset,
                pack_index,
            }
        })
    }
}