gix_pack/data/output/
bytes.rs

1use std::io::Write;
2
3use crate::data::output;
4use crate::exact_vec;
5
6/// The error returned by `next()` in the [`FromEntriesIter`] iterator.
7#[allow(missing_docs)]
8#[derive(Debug, thiserror::Error)]
9pub enum Error<E>
10where
11    E: std::error::Error + 'static,
12{
13    #[error(transparent)]
14    Io(#[from] gix_hash::io::Error),
15    #[error(transparent)]
16    Input(E),
17}
18
19/// An implementation of [`Iterator`] to write [encoded entries][output::Entry] to an inner implementation each time
20/// `next()` is called.
21pub struct FromEntriesIter<I, W> {
22    /// An iterator for input [`output::Entry`] instances
23    pub input: I,
24    /// A way of writing encoded bytes.
25    output: gix_hash::io::Write<W>,
26    /// Our trailing hash when done writing all input entries
27    trailer: Option<gix_hash::ObjectId>,
28    /// The amount of objects in the iteration and the version of the packfile to be written.
29    /// Will be `None` to signal the header was written already.
30    header_info: Option<(crate::data::Version, u32)>,
31    /// The pack data version with which pack entries should be written.
32    entry_version: crate::data::Version,
33    /// The amount of written bytes thus far
34    written: u64,
35    /// Required to quickly find offsets by object IDs, as future objects may refer to those in the past to become a delta offset base.
36    /// It stores the pack offsets at which objects begin.
37    /// Additionally we store if an object was invalid, and if so we will not write it nor will we allow delta objects to it.
38    pack_offsets_and_validity: Vec<(u64, bool)>,
39    /// If we are done, no additional writes will occur
40    is_done: bool,
41}
42
43impl<I, W, E> FromEntriesIter<I, W>
44where
45    I: Iterator<Item = Result<Vec<output::Entry>, E>>,
46    W: std::io::Write,
47    E: std::error::Error + 'static,
48{
49    /// Create a new instance reading [entries][output::Entry] from an `input` iterator and write pack data bytes to
50    /// `output` writer, resembling a pack of `version` with exactly `num_entries` amount of objects contained in it.
51    /// `object_hash` is the kind of hash to use for the pack checksum and maybe other places, depending on the version.
52    ///
53    /// The input chunks are expected to be sorted already. You can use the [`InOrderIter`][gix_features::parallel::InOrderIter] to assure
54    /// this happens on the fly holding entire chunks in memory as long as needed for them to be dispensed in order.
55    ///
56    /// # Panics
57    ///
58    /// Not all combinations of `object_hash` and `version` are supported currently triggering assertion errors.
59    pub fn new(
60        input: I,
61        output: W,
62        num_entries: u32,
63        version: crate::data::Version,
64        object_hash: gix_hash::Kind,
65    ) -> Self {
66        assert!(
67            matches!(version, crate::data::Version::V2),
68            "currently only pack version 2 can be written",
69        );
70        FromEntriesIter {
71            input,
72            output: gix_hash::io::Write::new(output, object_hash),
73            trailer: None,
74            entry_version: version,
75            pack_offsets_and_validity: exact_vec(num_entries as usize),
76            written: 0,
77            header_info: Some((version, num_entries)),
78            is_done: false,
79        }
80    }
81
82    /// Consume this instance and return the `output` implementation.
83    ///
84    /// _Note_ that the `input` iterator can be moved out of this instance beforehand.
85    pub fn into_write(self) -> W {
86        self.output.inner
87    }
88
89    /// Returns the trailing hash over all written entries once done.
90    /// It's `None` if we are not yet done writing.
91    pub fn digest(&self) -> Option<gix_hash::ObjectId> {
92        self.trailer
93    }
94
95    fn next_inner(&mut self) -> Result<u64, Error<E>> {
96        let previous_written = self.written;
97        if let Some((version, num_entries)) = self.header_info.take() {
98            let header_bytes = crate::data::header::encode(version, num_entries);
99            self.output
100                .write_all(&header_bytes[..])
101                .map_err(gix_hash::io::Error::from)?;
102            self.written += header_bytes.len() as u64;
103        }
104        match self.input.next() {
105            Some(entries) => {
106                for entry in entries.map_err(Error::Input)? {
107                    if entry.is_invalid() {
108                        self.pack_offsets_and_validity.push((0, false));
109                        continue;
110                    }
111                    self.pack_offsets_and_validity.push((self.written, true));
112                    let header = entry.to_entry_header(self.entry_version, |index| {
113                        let (base_offset, is_valid_object) = self.pack_offsets_and_validity[index];
114                        if !is_valid_object {
115                            unreachable!("if you see this the object database is correct as a delta refers to a non-existing object")
116                        }
117                        self.written - base_offset
118                    });
119                    self.written += header
120                        .write_to(entry.decompressed_size as u64, &mut self.output)
121                        .map_err(gix_hash::io::Error::from)? as u64;
122                    self.written += std::io::copy(&mut &*entry.compressed_data, &mut self.output)
123                        .map_err(gix_hash::io::Error::from)?;
124                }
125            }
126            None => {
127                let digest = self
128                    .output
129                    .hash
130                    .clone()
131                    .try_finalize()
132                    .map_err(gix_hash::io::Error::from)?;
133                self.output
134                    .inner
135                    .write_all(digest.as_slice())
136                    .map_err(gix_hash::io::Error::from)?;
137                self.written += digest.as_slice().len() as u64;
138                self.output.inner.flush().map_err(gix_hash::io::Error::from)?;
139                self.is_done = true;
140                self.trailer = Some(digest);
141            }
142        }
143        Ok(self.written - previous_written)
144    }
145}
146
147impl<I, W, E> Iterator for FromEntriesIter<I, W>
148where
149    I: Iterator<Item = Result<Vec<output::Entry>, E>>,
150    W: std::io::Write,
151    E: std::error::Error + 'static,
152{
153    /// The amount of bytes written to `out` if `Ok` or the error `E` received from the input.
154    type Item = Result<u64, Error<E>>;
155
156    fn next(&mut self) -> Option<Self::Item> {
157        if self.is_done {
158            return None;
159        }
160        Some(match self.next_inner() {
161            Err(err) => {
162                self.is_done = true;
163                Err(err)
164            }
165            Ok(written) => Ok(written),
166        })
167    }
168}
gix_pack/data/output/bytes.rs

gix_pack/data/output/
bytes.rs
