use crate::OutputFormat; use anyhow::{anyhow, Context as AnyhowContext, Result}; use bytesize::ByteSize; use git_features::progress::{self, Progress}; use git_object::{owned, Kind}; use git_odb::pack::{self, index}; use std::{io, path::Path, str::FromStr}; pub use index::verify::Mode; #[derive(Debug, Eq, PartialEq, Hash, Clone, Copy)] pub enum Algorithm { LessTime, LessMemory, } impl Algorithm { pub fn variants() -> &'static [&'static str] { &["less-time", "less-memory"] } } impl FromStr for Algorithm { type Err = String; fn from_str(s: &str) -> Result { let s_lc = s.to_ascii_lowercase(); Ok(match s_lc.as_str() { "less-memory" => Algorithm::LessMemory, "less-time" => Algorithm::LessTime, _ => return Err(format!("Invalid verification algorithm: '{}'", s)), }) } } impl From for index::traverse::Algorithm { fn from(v: Algorithm) -> Self { match v { Algorithm::LessMemory => index::traverse::Algorithm::Lookup, Algorithm::LessTime => index::traverse::Algorithm::DeltaTreeLookup, } } } /// A general purpose context for many operations provided here pub struct Context { /// If set, provide statistics to `out` in the given format pub output_statistics: Option, /// A stream to which to output operation results pub out: W1, /// A stream to which to errors pub err: W2, /// If set, don't use more than this amount of threads. /// Otherwise, usually use as many threads as there are logical cores. /// A value of 0 is interpreted as no-limit pub thread_limit: Option, pub mode: index::verify::Mode, pub algorithm: Algorithm, } impl Default for Context, Vec> { fn default() -> Self { Context { output_statistics: None, thread_limit: None, mode: index::verify::Mode::Sha1CRC32, algorithm: Algorithm::LessMemory, out: Vec::new(), err: Vec::new(), } } } #[allow(clippy::large_enum_variant)] enum EitherCache { Left(pack::cache::DecodeEntryNoop), Right(pack::cache::DecodeEntryLRU), } impl pack::cache::DecodeEntry for EitherCache { fn put(&mut self, offset: u64, data: &[u8], kind: Kind, compressed_size: usize) { match self { EitherCache::Left(v) => v.put(offset, data, kind, compressed_size), EitherCache::Right(v) => v.put(offset, data, kind, compressed_size), } } fn get(&mut self, offset: u64, out: &mut Vec) -> Option<(Kind, usize)> { match self { EitherCache::Left(v) => v.get(offset, out), EitherCache::Right(v) => v.get(offset, out), } } } pub fn pack_or_pack_index( path: impl AsRef, progress: Option

, Context { mut out, mut err, mode, output_statistics, thread_limit, algorithm, }: Context, ) -> Result<(owned::Id, Option)> where P: Progress + Send,

::SubProgress: Send, W1: io::Write, W2: io::Write, <

::SubProgress as Progress>::SubProgress: Send, { let path = path.as_ref(); let ext = path.extension().and_then(|ext| ext.to_str()).ok_or_else(|| { anyhow!( "Cannot determine data type on path without extension '{}', expecting default extensions 'idx' and 'pack'", path.display() ) })?; let res = match ext { "pack" => { let pack = git_odb::pack::data::File::at(path).with_context(|| "Could not open pack file")?; pack.verify_checksum(progress::DoOrDiscard::from(progress).add_child("Sha1 of pack")) .map(|id| (id, None))? } "idx" => { let idx = git_odb::pack::index::File::at(path).with_context(|| "Could not open pack index file")?; let packfile_path = path.with_extension("pack"); let pack = git_odb::pack::data::File::at(&packfile_path) .map_err(|e| { writeln!( err, "Could not find matching pack file at '{}' - only index file will be verified, error was: {}", packfile_path.display(), e ) .ok(); e }) .ok(); let cache = || -> EitherCache { if output_statistics.is_some() { // turn off acceleration as we need to see entire chains all the time EitherCache::Left(pack::cache::DecodeEntryNoop) } else { EitherCache::Right(pack::cache::DecodeEntryLRU::default()) } }; idx.verify_integrity( pack.as_ref().map(|p| (p, mode, algorithm.into())), thread_limit, progress, cache, ) .map(|(a, b, _)| (a, b)) .with_context(|| "Verification failure")? } ext => return Err(anyhow!("Unknown extension {:?}, expecting 'idx' or 'pack'", ext)), }; if let Some(stats) = res.1.as_ref() { match output_statistics { Some(OutputFormat::Human) => drop(print_statistics(&mut out, stats)), #[cfg(feature = "serde1")] Some(OutputFormat::Json) => serde_json::to_writer_pretty(out, stats)?, _ => {} }; } Ok(res) } fn print_statistics(out: &mut impl io::Write, stats: &index::traverse::Outcome) -> io::Result<()> { writeln!(out, "objects per delta chain length")?; let mut chain_length_to_object: Vec<_> = stats.objects_per_chain_length.iter().map(|(a, b)| (*a, *b)).collect(); chain_length_to_object.sort_by_key(|e| e.0); let mut total_object_count = 0; for (chain_length, object_count) in chain_length_to_object.into_iter() { total_object_count += object_count; writeln!(out, "\t{:>2}: {}", chain_length, object_count)?; } writeln!(out, "\t->: {}", total_object_count)?; let pack::data::decode::Outcome { kind: _, num_deltas, decompressed_size, compressed_size, object_size, } = stats.average; let width = 30; writeln!(out, "\naverages")?; #[rustfmt::skip] writeln!( out, "\t{: