gix_commitgraph/file/
init.rs

1use gix_error::Result;
2use std::path::{Path, PathBuf};
3
4use gix_error::{ResultExt, bail, corruption, message};
5
6use crate::{
7    File,
8    file::{
9        BASE_GRAPHS_LIST_CHUNK_ID, COMMIT_DATA_CHUNK_ID, COMMIT_DATA_ENTRY_SIZE_SANS_HASH,
10        EXTENDED_EDGES_LIST_CHUNK_ID, FAN_LEN, HEADER_LEN, OID_FAN_CHUNK_ID, OID_LOOKUP_CHUNK_ID, SIGNATURE,
11    },
12};
13
14const MIN_FILE_SIZE: usize = HEADER_LEN
15    + gix_chunk::file::Index::size_for_entries(3 /*OIDF, OIDL, CDAT*/)
16    + FAN_LEN * 4 /* FANOUT TABLE CHUNK OIDF */
17    + gix_hash::Kind::shortest().len_in_bytes();
18
19impl File {
20    /// Try to parse the commit graph file at `path`.
21    pub fn at(path: impl AsRef<Path>) -> Result<File> {
22        Self::try_from(path.as_ref())
23    }
24
25    /// A lower-level constructor which constructs a new instance directly from the mapping in `data`,
26    /// assuming that it originated from `path`.
27    ///
28    /// Note that `path` is only used for verification of the hash its basename contains, but otherwise
29    /// is not of importance.
30    pub fn new(data: memmap2::Mmap, path: PathBuf) -> Result<File> {
31        let data_size = data.len();
32        if data_size < MIN_FILE_SIZE {
33            bail!(corruption("Commit-graph file too small even for an empty graph"));
34        }
35
36        let mut ofs = 0;
37        if &data[ofs..ofs + SIGNATURE.len()] != SIGNATURE {
38            bail!(corruption("Commit-graph file does not start with expected signature"));
39        }
40        ofs += SIGNATURE.len();
41
42        match data[ofs] {
43            1 => (),
44            x => {
45                bail!("Unsupported commit-graph file version: {x}".unsupported());
46            }
47        }
48        ofs += 1;
49
50        let object_hash = gix_hash::Kind::try_from(data[ofs])
51            .map_err(|v| message!("Commit-graph file uses unsupported hash version: {v}").unsupported_error())?;
52        ofs += 1;
53
54        let chunk_count = data[ofs];
55        // Can assert chunk_count >= MIN_CHUNKS here, but later OIDF+OIDL+CDAT presence checks make
56        // it redundant.
57        ofs += 1;
58
59        let base_graph_count = data[ofs];
60        ofs += 1;
61
62        let chunks = gix_chunk::file::Index::from_bytes(&data, ofs, u32::from(chunk_count)).or_raise(|| {
63            message!("Couldn't read commit-graph file with {chunk_count} chunks at offset {ofs}").corrupted()
64        })?;
65
66        let base_graphs_list_offset = chunks
67            .validated_usize_offset_by_id(BASE_GRAPHS_LIST_CHUNK_ID, |chunk_range| -> Result<_> {
68                let chunk_size = chunk_range.len();
69                if chunk_size % object_hash.len_in_bytes() != 0 {
70                    bail!("Commit-graph chunk {BASE_GRAPHS_LIST_CHUNK_ID:?} has invalid size: {msg}".corrupted(),
71                        msg = format!(
72                            "chunk size {} is not a multiple of {}",
73                            chunk_size,
74                            object_hash.len_in_bytes()
75                        ),
76                    );
77                }
78                let chunk_base_graph_count: u32 = (chunk_size / object_hash.len_in_bytes())
79                    .try_into()
80                    .expect("base graph count to fit in 32-bits");
81                if chunk_base_graph_count != u32::from(base_graph_count) {
82                    bail!("Commit-graph {BASE_GRAPHS_LIST_CHUNK_ID:?} chunk contains {chunk_base_graph_count} base graphs, but commit-graph file header claims {base_graph_count} base graphs".corrupted())
83                }
84                Ok(chunk_range.start)
85            })
86            .ok()
87            .transpose()?;
88
89        let (commit_data_offset, commit_data_count): (_, u32) = chunks
90            .validated_usize_offset_by_id(COMMIT_DATA_CHUNK_ID, |chunk_range| -> Result<_> {
91                let chunk_size = chunk_range.len();
92
93                let entry_size = object_hash.len_in_bytes() + COMMIT_DATA_ENTRY_SIZE_SANS_HASH;
94                if chunk_size % entry_size != 0 {
95                    bail!("Commit-graph chunk {COMMIT_DATA_CHUNK_ID:?} has invalid size: chunk size {chunk_size} is not a multiple of {entry_size}".corrupted())
96                }
97                Ok((
98                    chunk_range.start,
99                    (chunk_size / entry_size)
100                        .try_into()
101                        .expect("number of commits in CDAT chunk to fit in 32 bits"),
102                ))
103            })
104            .or_raise(|| corruption("Error getting offset for commit data chunk"))??;
105
106        let fan_offset = chunks
107            .validated_usize_offset_by_id(OID_FAN_CHUNK_ID, |chunk_range| -> Result<_> {
108                let chunk_size = chunk_range.len();
109
110                let expected_size = 4 * FAN_LEN;
111                if chunk_size != expected_size {
112                    bail!("Commit-graph chunk {OID_FAN_CHUNK_ID:?} has invalid size: expected chunk length {expected_size}, got {chunk_size}".corrupted())
113                }
114                Ok(chunk_range.start)
115            })
116            .or_raise(|| corruption("Error getting offset for OID fan chunk"))?
117            .or_raise(|| message("Error getting offset for OID fan chunk"))?;
118
119        let (oid_lookup_offset, oid_lookup_count): (_, u32) = chunks
120            .validated_usize_offset_by_id(OID_LOOKUP_CHUNK_ID, |chunk_range| -> Result<_> {
121                let chunk_size = chunk_range.len();
122
123                if chunk_size % object_hash.len_in_bytes() != 0 {
124                    bail!("Commit-graph chunk {OID_LOOKUP_CHUNK_ID:?} has invalid size: chunk size {chunk_size} is not a multiple of {hash_len}".corrupted(), hash_len = object_hash.len_in_bytes())
125                }
126                Ok((
127                    chunk_range.start,
128                    (chunk_size / object_hash.len_in_bytes())
129                        .try_into()
130                        .expect("number of commits in OIDL chunk to fit in 32 bits"),
131                ))
132            })
133            .or_raise(|| corruption("Error getting offset for OID lookup chunk"))?
134            .or_raise(|| message("Error getting offset for OID lookup chunk"))?;
135
136        let extra_edges_list_range = chunks.usize_offset_by_id(EXTENDED_EDGES_LIST_CHUNK_ID).ok();
137
138        let trailer = &data[chunks.highest_offset() as usize..];
139        if trailer.len() != object_hash.len_in_bytes() {
140            bail!(
141                "Expected commit-graph trailer to contain {} bytes, got {}".corrupted(),
142                object_hash.len_in_bytes(),
143                trailer.len()
144            );
145        }
146
147        if base_graph_count > 0 && base_graphs_list_offset.is_none() {
148            bail!("Chunk named {BASE_GRAPHS_LIST_CHUNK_ID:?} was not found in chunk file index".corrupted());
149        }
150
151        let (fan, _) = read_fan(&data[fan_offset..]);
152        if oid_lookup_count != fan[255] {
153            bail!(
154                "Commit-graph {OID_FAN_CHUNK_ID:?} chunk contains {chunk1_commits} commits, but {OID_LOOKUP_CHUNK_ID:?} chunk contains {chunk2_commits} commits".corrupted(),
155                chunk1_commits = fan[255],
156                chunk2_commits = oid_lookup_count,
157            );
158        }
159        if commit_data_count != fan[255] {
160            bail!(
161                "Commit-graph {OID_FAN_CHUNK_ID:?} chunk contains {chunk1_commits} commits, but {COMMIT_DATA_CHUNK_ID:?} chunk contains {chunk2_commits} commits".corrupted(),
162                chunk1_commits = fan[255],
163                chunk2_commits = commit_data_count,
164            );
165        }
166        Ok(File {
167            base_graph_count,
168            base_graphs_list_offset,
169            commit_data_offset,
170            data,
171            extra_edges_list_range,
172            fan,
173            oid_lookup_offset,
174            path,
175            hash_len: object_hash.len_in_bytes(),
176            object_hash,
177        })
178    }
179}
180
181impl TryFrom<&Path> for File {
182    type Error = gix_error::Error;
183
184    fn try_from(path: &Path) -> Result<Self> {
185        let data = std::fs::File::open(path)
186            .and_then(|file| {
187                // SAFETY: we have to take the risk of somebody changing the file underneath. Git never writes into the same file.
188                #[expect(unsafe_code)]
189                unsafe {
190                    memmap2::MmapOptions::new().map_copy_read_only(&file)
191                }
192            })
193            .or_raise(|| message!("Could not open commit-graph file at \"{path}\"", path = path.display()))?;
194        Self::new(data, path.to_owned())
195    }
196}
197
198// Copied from gix-odb/pack/index/init.rs
199fn read_fan(d: &[u8]) -> ([u32; FAN_LEN], usize) {
200    assert!(d.len() >= FAN_LEN * 4);
201
202    let mut fan = [0; FAN_LEN];
203    for (c, f) in d.as_chunks::<4>().0.iter().zip(fan.iter_mut()) {
204        *f = u32::from_be_bytes(*c);
205    }
206    (fan, FAN_LEN * 4)
207}
gix_commitgraph/file/init.rs

gix_commitgraph/file/
init.rs
