comparison rust/hg-core/src/revlog/revlog.rs @ 48542:35c47015b9b7

rhg: Expose FilelogEntry that wraps RevlogEntry

This can be later extended to access metadata such as `uncompressed_len` without
necessarily resolving deltas.

Differential Revision: https://phab.mercurial-scm.org/D11961

48541:f2f57724d4eb

            NodeMapError::RevisionNotInIndex(_) => RevlogError::corrupted(),
        }
    }
}

impl RevlogError {
    fn corrupted() -> Self {
        RevlogError::Other(HgError::corrupted("corrupted revlog"))
    }
}

/// Read only implementation of revlog.
pub struct Revlog {

        rev: Revision,
    ) -> Result<Cow<[u8]>, RevlogError> {
        if rev == NULL_REVISION {
            return Ok(Cow::Borrowed(&[]));
        };
        self.get_entry(rev)?.data()
    }

    /// Check the hash of some given data against the recorded hash.
    pub fn check_hash(
        &self,

    /// and its deltas.
    #[timed]
    fn build_data_from_deltas(
        snapshot: RevlogEntry,
        deltas: &[RevlogEntry],
    ) -> Result<Vec<u8>, RevlogError> {
        let snapshot = snapshot.data_chunk()?;
        let deltas = deltas
            .iter()
            .rev()
            .map(RevlogEntry::data_chunk)
            .collect::<Result<Vec<Cow<'_, [u8]>>, RevlogError>>()?;
        let patches: Vec<_> =
            deltas.iter().map(|d| patch::PatchList::new(d)).collect();
        let patch = patch::fold_patch_lists(&patches);
        Ok(patch.apply(&snapshot))
    }

            ),
        }
    }

    /// Get an entry of the revlog.
    fn get_entry(&self, rev: Revision) -> Result<RevlogEntry, RevlogError> {
        let index_entry = self
            .index
            .get_entry(rev)
            .ok_or(RevlogError::InvalidRevision)?;
        let start = index_entry.offset();

    /// when resolving internal references within revlog, any errors
    /// should be reported as corruption, instead of e.g. "invalid revision"
    fn get_entry_internal(
        &self,
        rev: Revision,
    ) -> Result<RevlogEntry, RevlogError> {
        return self.get_entry(rev).map_err(|_| RevlogError::corrupted());
    }
}

/// The revlog entry's bytes and the necessary informations to extract
/// the entry's data.

    pub fn revision(&self) -> Revision {
        self.rev
    }

    /// The data for this entry, after resolving deltas if any.
    pub fn data(&self) -> Result<Cow<'a, [u8]>, RevlogError> {
        let mut entry = self.clone();
        let mut delta_chain = vec![];

        // The meaning of `base_rev_or_base_of_delta_chain` depends on
        // generaldelta. See the doc on `ENTRY_DELTA_BASE` in

        // TODO do not look twice in the index
        let index_entry = self
            .revlog
            .index
            .get_entry(self.rev)
            .ok_or(RevlogError::InvalidRevision)?;

        let data = if delta_chain.is_empty() {
            entry.data_chunk()?
        } else {
            Revlog::build_data_from_deltas(entry, &delta_chain)?.into()

            index_entry.hash().as_bytes(),
            &data,
        ) {
            Ok(data)
        } else {
            Err(RevlogError::corrupted())
        }
    }

    /// Extract the data contained in the entry.
    /// This may be a delta. (See `is_delta`.)
    fn data_chunk(&self) -> Result<Cow<'a, [u8]>, RevlogError> {
        if self.bytes.is_empty() {
            return Ok(Cow::Borrowed(&[]));
        }
        match self.bytes[0] {
            // Revision data is the entirety of the entry, including this

            // zlib (RFC 1950) data.
            b'x' => Ok(Cow::Owned(self.uncompressed_zlib_data()?)),
            // zstd data.
            b'\x28' => Ok(Cow::Owned(self.uncompressed_zstd_data()?)),
            // A proper new format should have had a repo/store requirement.
            _format_type => Err(RevlogError::corrupted()),
        }
    }

    fn uncompressed_zlib_data(&self) -> Result<Vec<u8>, RevlogError> {
        let mut decoder = ZlibDecoder::new(self.bytes);
        if self.is_delta() {
            let mut buf = Vec::with_capacity(self.compressed_len);
            decoder
                .read_to_end(&mut buf)
                .map_err(|_| RevlogError::corrupted())?;
            Ok(buf)
        } else {
            let mut buf = vec![0; self.uncompressed_len];
            decoder
                .read_exact(&mut buf)
                .map_err(|_| RevlogError::corrupted())?;
            Ok(buf)
        }
    }

    fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, RevlogError> {
        if self.is_delta() {
            let mut buf = Vec::with_capacity(self.compressed_len);
            zstd::stream::copy_decode(self.bytes, &mut buf)
                .map_err(|_| RevlogError::corrupted())?;
            Ok(buf)
        } else {
            let mut buf = vec![0; self.uncompressed_len];
            let len = zstd::block::decompress_to_buffer(self.bytes, &mut buf)
                .map_err(|_| RevlogError::corrupted())?;
            if len != self.uncompressed_len {
                Err(RevlogError::corrupted())
            } else {
                Ok(buf)
            }
        }
    }

48542:35c47015b9b7

            NodeMapError::RevisionNotInIndex(_) => RevlogError::corrupted(),
        }
    }
}

fn corrupted() -> HgError {
    HgError::corrupted("corrupted revlog")
}

impl RevlogError {
    fn corrupted() -> Self {
        RevlogError::Other(corrupted())
    }
}

/// Read only implementation of revlog.
pub struct Revlog {

        rev: Revision,
    ) -> Result<Cow<[u8]>, RevlogError> {
        if rev == NULL_REVISION {
            return Ok(Cow::Borrowed(&[]));
        };
        Ok(self.get_entry(rev)?.data()?)
    }

    /// Check the hash of some given data against the recorded hash.
    pub fn check_hash(
        &self,

    /// and its deltas.
    #[timed]
    fn build_data_from_deltas(
        snapshot: RevlogEntry,
        deltas: &[RevlogEntry],
    ) -> Result<Vec<u8>, HgError> {
        let snapshot = snapshot.data_chunk()?;
        let deltas = deltas
            .iter()
            .rev()
            .map(RevlogEntry::data_chunk)
            .collect::<Result<Vec<_>, _>>()?;
        let patches: Vec<_> =
            deltas.iter().map(|d| patch::PatchList::new(d)).collect();
        let patch = patch::fold_patch_lists(&patches);
        Ok(patch.apply(&snapshot))
    }

            ),
        }
    }

    /// Get an entry of the revlog.
    pub fn get_entry(
        &self,
        rev: Revision,
    ) -> Result<RevlogEntry, RevlogError> {
        let index_entry = self
            .index
            .get_entry(rev)
            .ok_or(RevlogError::InvalidRevision)?;
        let start = index_entry.offset();

    /// when resolving internal references within revlog, any errors
    /// should be reported as corruption, instead of e.g. "invalid revision"
    fn get_entry_internal(
        &self,
        rev: Revision,
    ) -> Result<RevlogEntry, HgError> {
        return self.get_entry(rev).map_err(|_| corrupted());
    }
}

/// The revlog entry's bytes and the necessary informations to extract
/// the entry's data.

    pub fn revision(&self) -> Revision {
        self.rev
    }

    /// The data for this entry, after resolving deltas if any.
    pub fn data(&self) -> Result<Cow<'a, [u8]>, HgError> {
        let mut entry = self.clone();
        let mut delta_chain = vec![];

        // The meaning of `base_rev_or_base_of_delta_chain` depends on
        // generaldelta. See the doc on `ENTRY_DELTA_BASE` in

        // TODO do not look twice in the index
        let index_entry = self
            .revlog
            .index
            .get_entry(self.rev)
            .ok_or_else(corrupted)?;

        let data = if delta_chain.is_empty() {
            entry.data_chunk()?
        } else {
            Revlog::build_data_from_deltas(entry, &delta_chain)?.into()

            index_entry.hash().as_bytes(),
            &data,
        ) {
            Ok(data)
        } else {
            Err(corrupted())
        }
    }

    /// Extract the data contained in the entry.
    /// This may be a delta. (See `is_delta`.)
    fn data_chunk(&self) -> Result<Cow<'a, [u8]>, HgError> {
        if self.bytes.is_empty() {
            return Ok(Cow::Borrowed(&[]));
        }
        match self.bytes[0] {
            // Revision data is the entirety of the entry, including this

            // zlib (RFC 1950) data.
            b'x' => Ok(Cow::Owned(self.uncompressed_zlib_data()?)),
            // zstd data.
            b'\x28' => Ok(Cow::Owned(self.uncompressed_zstd_data()?)),
            // A proper new format should have had a repo/store requirement.
            _format_type => Err(corrupted()),
        }
    }

    fn uncompressed_zlib_data(&self) -> Result<Vec<u8>, HgError> {
        let mut decoder = ZlibDecoder::new(self.bytes);
        if self.is_delta() {
            let mut buf = Vec::with_capacity(self.compressed_len);
            decoder.read_to_end(&mut buf).map_err(|_| corrupted())?;
            Ok(buf)
        } else {
            let mut buf = vec![0; self.uncompressed_len];
            decoder.read_exact(&mut buf).map_err(|_| corrupted())?;
            Ok(buf)
        }
    }

    fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, HgError> {
        if self.is_delta() {
            let mut buf = Vec::with_capacity(self.compressed_len);
            zstd::stream::copy_decode(self.bytes, &mut buf)
                .map_err(|_| corrupted())?;
            Ok(buf)
        } else {
            let mut buf = vec![0; self.uncompressed_len];
            let len = zstd::block::decompress_to_buffer(self.bytes, &mut buf)
                .map_err(|_| corrupted())?;
            if len != self.uncompressed_len {
                Err(corrupted())
            } else {
                Ok(buf)
            }
        }
    }