git_pack/data/file/decode/
header.rs

1use crate::{
2    data,
3    data::{delta, file::decode::Error, File},
4};
5
6/// A return value of a resolve function, which given an [`ObjectId`][git_hash::ObjectId] determines where an object can be found.
7#[derive(Debug, PartialEq, Eq, Hash, Ord, PartialOrd, Clone)]
8#[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
9pub enum ResolvedBase {
10    /// Indicate an object is within this pack, at the given entry, and thus can be looked up locally.
11    InPack(data::Entry),
12    /// Indicates the object of `kind` was found outside of the pack.
13    OutOfPack {
14        /// The kind of object we found when reading the header of the out-of-pack base.
15        kind: git_object::Kind,
16        /// The amount of deltas encountered if the object was packed as well.
17        num_deltas: Option<u32>,
18    },
19}
20
21/// Additional information and statistics about a successfully decoded object produced by [`File::decode_header()`].
22///
23/// Useful to understand the effectiveness of the pack compression or the cost of decompression.
24#[derive(Debug, PartialEq, Eq, Hash, Ord, PartialOrd, Clone, Copy)]
25#[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
26pub struct Outcome {
27    /// The kind of resolved object.
28    pub kind: git_object::Kind,
29    /// The decompressed size of the object.
30    pub object_size: u64,
31    /// The amount of deltas in the chain of objects that had to be resolved beforehand.
32    pub num_deltas: u32,
33}
34
35/// Obtain object information quickly.
36impl File {
37    /// Resolve the object header information starting at `entry`, following the chain of entries as needed.
38    ///
39    /// The `entry` determines which object to decode, and is commonly obtained with the help of a pack index file or through pack iteration.
40    ///
41    /// `resolve` is a function to lookup objects with the given [`ObjectId`][git_hash::ObjectId], in case the full object id
42    /// is used to refer to a base object, instead of an in-pack offset.
43    pub fn decode_header(
44        &self,
45        mut entry: data::Entry,
46        resolve: impl Fn(&git_hash::oid) -> Option<ResolvedBase>,
47    ) -> Result<Outcome, Error> {
48        use crate::data::entry::Header::*;
49        let mut num_deltas = 0;
50        let mut first_delta_decompressed_size = None::<u64>;
51        loop {
52            match entry.header {
53                Tree | Blob | Commit | Tag => {
54                    return Ok(Outcome {
55                        kind: entry.header.as_kind().expect("always valid for non-refs"),
56                        object_size: first_delta_decompressed_size.unwrap_or(entry.decompressed_size),
57                        num_deltas,
58                    });
59                }
60                OfsDelta { base_distance } => {
61                    num_deltas += 1;
62                    if first_delta_decompressed_size.is_none() {
63                        first_delta_decompressed_size = Some(self.decode_delta_object_size(&entry)?);
64                    }
65                    entry = self.entry(entry.base_pack_offset(base_distance))
66                }
67                RefDelta { base_id } => {
68                    num_deltas += 1;
69                    if first_delta_decompressed_size.is_none() {
70                        first_delta_decompressed_size = Some(self.decode_delta_object_size(&entry)?);
71                    }
72                    match resolve(base_id.as_ref()) {
73                        Some(ResolvedBase::InPack(base_entry)) => entry = base_entry,
74                        Some(ResolvedBase::OutOfPack {
75                            kind,
76                            num_deltas: origin_num_deltas,
77                        }) => {
78                            return Ok(Outcome {
79                                kind,
80                                object_size: first_delta_decompressed_size.unwrap_or(entry.decompressed_size),
81                                num_deltas: origin_num_deltas.unwrap_or_default() + num_deltas,
82                            })
83                        }
84                        None => return Err(Error::DeltaBaseUnresolved(base_id)),
85                    }
86                }
87            };
88        }
89    }
90
91    #[inline]
92    fn decode_delta_object_size(&self, entry: &data::Entry) -> Result<u64, Error> {
93        let mut buf = [0_u8; 32];
94        let used = self.decompress_entry_from_data_offset_2(entry.data_offset, &mut buf)?.1;
95        let buf = &buf[..used];
96        let (_base_size, offset) = delta::decode_header_size(buf);
97        let (result_size, _offset) = delta::decode_header_size(&buf[offset..]);
98        Ok(result_size)
99    }
100}
101
102#[cfg(test)]
103mod tests {
104    use super::*;
105
106    #[test]
107    fn size_of_decode_entry_outcome() {
108        assert_eq!(
109            std::mem::size_of::<Outcome>(),
110            16,
111            "this shouldn't change without use noticing as it's returned a lot"
112        );
113    }
114}
git_pack/data/file/decode/header.rs

git_pack/data/file/decode/
header.rs
