Files
krata/crates/oci/src/fetch.rs

375 lines
12 KiB
Rust
Raw Normal View History

use crate::{
progress::{OciBoundProgress, OciProgressPhase},
schema::OciSchema,
};
use super::{
name::ImageName,
registry::{OciPlatform, OciRegistryClient},
};
2024-03-13 14:20:22 +00:00
use std::{
fmt::Debug,
io::SeekFrom,
os::unix::fs::MetadataExt,
2024-03-13 14:20:22 +00:00
path::{Path, PathBuf},
pin::Pin,
};
2024-03-08 14:44:45 +00:00
use anyhow::{anyhow, Result};
2024-03-08 14:44:45 +00:00
use async_compression::tokio::bufread::{GzipDecoder, ZstdDecoder};
use log::debug;
2024-03-13 14:20:22 +00:00
use oci_spec::image::{
Descriptor, DescriptorBuilder, ImageConfiguration, ImageIndex, ImageManifest, MediaType,
ToDockerV2S2,
2024-03-13 14:20:22 +00:00
};
use serde::de::DeserializeOwned;
2024-03-08 14:44:45 +00:00
use tokio::{
fs::{self, File},
io::{AsyncRead, AsyncReadExt, AsyncSeekExt, BufReader, BufWriter},
2024-03-08 14:44:45 +00:00
};
2024-03-13 14:20:22 +00:00
use tokio_stream::StreamExt;
2024-03-08 14:44:45 +00:00
use tokio_tar::Archive;
pub struct OciImageFetcher {
2024-03-13 14:20:22 +00:00
seed: Option<PathBuf>,
platform: OciPlatform,
progress: OciBoundProgress,
2024-01-18 10:16:59 -08:00
}
2024-03-08 14:44:45 +00:00
#[derive(Clone, Debug, PartialEq, Eq)]
pub enum OciImageLayerCompression {
None,
Gzip,
Zstd,
}
#[derive(Clone, Debug)]
pub struct OciImageLayer {
pub metadata: Descriptor,
2024-03-08 14:44:45 +00:00
pub path: PathBuf,
pub digest: String,
pub compression: OciImageLayerCompression,
}
2024-01-18 10:16:59 -08:00
#[async_trait::async_trait]
pub trait OciImageLayerReader: AsyncRead + Sync {
async fn position(&mut self) -> Result<u64>;
}
#[async_trait::async_trait]
impl OciImageLayerReader for BufReader<File> {
async fn position(&mut self) -> Result<u64> {
Ok(self.seek(SeekFrom::Current(0)).await?)
}
}
#[async_trait::async_trait]
impl OciImageLayerReader for GzipDecoder<BufReader<File>> {
async fn position(&mut self) -> Result<u64> {
self.get_mut().position().await
}
}
#[async_trait::async_trait]
impl OciImageLayerReader for ZstdDecoder<BufReader<File>> {
async fn position(&mut self) -> Result<u64> {
self.get_mut().position().await
}
}
2024-03-08 14:44:45 +00:00
impl OciImageLayer {
pub async fn decompress(&self) -> Result<Pin<Box<dyn OciImageLayerReader + Send>>> {
2024-03-08 14:44:45 +00:00
let file = File::open(&self.path).await?;
let reader = BufReader::new(file);
let reader: Pin<Box<dyn OciImageLayerReader + Send>> = match self.compression {
2024-03-08 14:44:45 +00:00
OciImageLayerCompression::None => Box::pin(reader),
OciImageLayerCompression::Gzip => Box::pin(GzipDecoder::new(reader)),
OciImageLayerCompression::Zstd => Box::pin(ZstdDecoder::new(reader)),
};
Ok(reader)
2024-01-18 10:16:59 -08:00
}
pub async fn archive(&self) -> Result<Archive<Pin<Box<dyn OciImageLayerReader + Send>>>> {
2024-03-08 14:44:45 +00:00
let decompress = self.decompress().await?;
Ok(Archive::new(decompress))
2024-01-18 10:16:59 -08:00
}
2024-03-08 14:44:45 +00:00
}
2024-01-18 10:16:59 -08:00
2024-03-08 14:44:45 +00:00
#[derive(Clone, Debug)]
pub struct OciResolvedImage {
pub name: ImageName,
pub digest: String,
pub descriptor: Descriptor,
pub manifest: OciSchema<ImageManifest>,
2024-03-08 14:44:45 +00:00
}
#[derive(Clone, Debug)]
pub struct OciLocalImage {
pub image: OciResolvedImage,
pub config: OciSchema<ImageConfiguration>,
2024-03-08 14:44:45 +00:00
pub layers: Vec<OciImageLayer>,
}
impl OciImageFetcher {
2024-03-13 14:20:22 +00:00
pub fn new(
seed: Option<PathBuf>,
platform: OciPlatform,
progress: OciBoundProgress,
) -> OciImageFetcher {
OciImageFetcher {
2024-03-13 14:20:22 +00:00
seed,
platform,
progress,
2024-03-13 14:20:22 +00:00
}
}
async fn load_seed_json_blob<T: Clone + Debug + DeserializeOwned>(
2024-03-13 14:20:22 +00:00
&self,
descriptor: &Descriptor,
) -> Result<Option<OciSchema<T>>> {
2024-03-13 14:20:22 +00:00
let digest = descriptor.digest();
let Some((digest_type, digest_content)) = digest.split_once(':') else {
return Err(anyhow!("digest content was not properly formatted"));
};
let want = format!("blobs/{}/{}", digest_type, digest_content);
self.load_seed_json(&want).await
}
async fn load_seed_json<T: Clone + Debug + DeserializeOwned>(
&self,
want: &str,
) -> Result<Option<OciSchema<T>>> {
2024-03-13 14:20:22 +00:00
let Some(ref seed) = self.seed else {
return Ok(None);
};
let file = File::open(seed).await?;
let mut archive = Archive::new(file);
let mut entries = archive.entries()?;
while let Some(entry) = entries.next().await {
let mut entry = entry?;
let path = String::from_utf8(entry.path_bytes().to_vec())?;
if path == want {
let mut content = Vec::new();
entry.read_to_end(&mut content).await?;
let item = serde_json::from_slice::<T>(&content)?;
return Ok(Some(OciSchema::new(content, item)));
2024-03-13 14:20:22 +00:00
}
}
Ok(None)
}
async fn extract_seed_blob(&self, descriptor: &Descriptor, to: &Path) -> Result<bool> {
let Some(ref seed) = self.seed else {
return Ok(false);
};
let digest = descriptor.digest();
let Some((digest_type, digest_content)) = digest.split_once(':') else {
return Err(anyhow!("digest content was not properly formatted"));
};
let want = format!("blobs/{}/{}", digest_type, digest_content);
let seed = File::open(seed).await?;
let mut archive = Archive::new(seed);
let mut entries = archive.entries()?;
while let Some(entry) = entries.next().await {
let mut entry = entry?;
let path = String::from_utf8(entry.path_bytes().to_vec())?;
if path == want {
let file = File::create(to).await?;
let mut bufwrite = BufWriter::new(file);
tokio::io::copy(&mut entry, &mut bufwrite).await?;
2024-03-13 14:20:22 +00:00
return Ok(true);
}
}
Ok(false)
2024-02-25 05:38:23 +00:00
}
2024-03-08 14:44:45 +00:00
pub async fn resolve(&self, image: ImageName) -> Result<OciResolvedImage> {
2024-03-13 14:20:22 +00:00
debug!("resolve manifest image={}", image);
if let Some(index) = self.load_seed_json::<ImageIndex>("index.json").await? {
let mut found: Option<&Descriptor> = None;
for manifest in index.item().manifests() {
2024-03-13 14:20:22 +00:00
let Some(annotations) = manifest.annotations() else {
continue;
};
let mut image_name = annotations.get("io.containerd.image.name");
if image_name.is_none() {
image_name = annotations.get("org.opencontainers.image.ref.name");
}
let Some(image_name) = image_name else {
2024-03-13 14:20:22 +00:00
continue;
};
if *image_name != image.to_string() {
continue;
}
if let Some(platform) = manifest.platform() {
if *platform.architecture() != self.platform.arch
|| *platform.os() != self.platform.os
{
continue;
}
}
found = Some(manifest);
break;
}
if let Some(found) = found {
if let Some(manifest) = self.load_seed_json_blob(found).await? {
debug!(
"found seeded manifest image={} manifest={}",
image,
found.digest()
);
return Ok(OciResolvedImage {
name: image,
descriptor: found.clone(),
2024-03-13 14:20:22 +00:00
digest: found.digest().clone(),
manifest,
});
}
}
}
2024-03-08 14:44:45 +00:00
let mut client = OciRegistryClient::new(image.registry_url()?, self.platform.clone())?;
let (manifest, descriptor, digest) = client
2024-03-08 14:44:45 +00:00
.get_manifest_with_digest(&image.name, &image.reference)
2024-02-25 05:38:23 +00:00
.await?;
let descriptor = descriptor.unwrap_or_else(|| {
DescriptorBuilder::default()
.media_type(MediaType::ImageManifest)
.size(manifest.raw().len() as i64)
.digest(digest.clone())
.build()
.unwrap()
});
2024-03-08 14:44:45 +00:00
Ok(OciResolvedImage {
name: image,
descriptor,
2024-03-08 14:44:45 +00:00
digest,
manifest,
})
2024-01-20 02:41:49 -08:00
}
pub async fn download(
&self,
image: &OciResolvedImage,
layer_dir: &Path,
) -> Result<OciLocalImage> {
let config: OciSchema<ImageConfiguration>;
self.progress
.update(|progress| {
progress.phase = OciProgressPhase::ConfigDownload;
})
.await;
2024-03-08 14:44:45 +00:00
let mut client = OciRegistryClient::new(image.name.registry_url()?, self.platform.clone())?;
2024-03-13 14:20:22 +00:00
if let Some(seeded) = self
.load_seed_json_blob::<ImageConfiguration>(image.manifest.item().config())
2024-03-13 14:20:22 +00:00
.await?
{
config = seeded;
} else {
let config_bytes = client
.get_blob(&image.name.name, image.manifest.item().config())
2024-03-13 14:20:22 +00:00
.await?;
config = OciSchema::new(
config_bytes.to_vec(),
serde_json::from_slice(&config_bytes)?,
);
2024-03-13 14:20:22 +00:00
}
self.progress
.update(|progress| {
progress.phase = OciProgressPhase::LayerDownload;
for layer in image.manifest.item().layers() {
progress.add_layer(layer.digest());
}
})
.await;
2024-03-08 14:44:45 +00:00
let mut layers = Vec::new();
for layer in image.manifest.item().layers() {
self.progress
.update(|progress| {
progress.downloading_layer(layer.digest(), 0, layer.size() as u64);
})
.await;
layers.push(
self.acquire_layer(&image.name, layer, layer_dir, &mut client)
.await?,
);
self.progress
.update(|progress| {
progress.downloaded_layer(layer.digest(), layer.size() as u64);
})
.await;
2024-01-18 10:16:59 -08:00
}
2024-03-08 14:44:45 +00:00
Ok(OciLocalImage {
image: image.clone(),
2024-03-08 14:44:45 +00:00
config,
layers,
})
2024-01-18 10:16:59 -08:00
}
2024-03-13 14:20:22 +00:00
async fn acquire_layer(
2024-03-08 14:44:45 +00:00
&self,
image: &ImageName,
layer: &Descriptor,
layer_dir: &Path,
2024-03-08 14:44:45 +00:00
client: &mut OciRegistryClient,
) -> Result<OciImageLayer> {
debug!(
2024-03-13 14:20:22 +00:00
"acquire layer digest={} size={}",
2024-03-08 14:44:45 +00:00
layer.digest(),
layer.size()
);
let mut layer_path = layer_dir.to_path_buf();
2024-03-08 14:44:45 +00:00
layer_path.push(format!("{}.layer", layer.digest()));
2024-03-13 14:20:22 +00:00
let seeded = self.extract_seed_blob(layer, &layer_path).await?;
if !seeded {
let file = File::create(&layer_path).await?;
let size = client
.write_blob_to_file(&image.name, layer, file, Some(self.progress.clone()))
.await?;
2024-03-08 14:44:45 +00:00
if layer.size() as u64 != size {
return Err(anyhow!(
"downloaded layer size differs from size in manifest",
));
2024-01-18 10:16:59 -08:00
}
}
2024-03-08 14:44:45 +00:00
let metadata = fs::metadata(&layer_path).await?;
if layer.size() as u64 != metadata.size() {
return Err(anyhow!("layer size differs from size in manifest",));
}
2024-03-08 14:44:45 +00:00
let mut media_type = layer.media_type().clone();
// docker layer compatibility
if media_type.to_string() == MediaType::ImageLayerGzip.to_docker_v2s2()? {
media_type = MediaType::ImageLayerGzip;
}
let compression = match media_type {
MediaType::ImageLayer => OciImageLayerCompression::None,
MediaType::ImageLayerGzip => OciImageLayerCompression::Gzip,
MediaType::ImageLayerZstd => OciImageLayerCompression::Zstd,
other => return Err(anyhow!("found layer with unknown media type: {}", other)),
};
Ok(OciImageLayer {
metadata: layer.clone(),
2024-03-08 14:44:45 +00:00
path: layer_path,
digest: layer.digest().clone(),
compression,
})
2024-01-18 10:16:59 -08:00
}
}