mod archive;
mod error;
mod manifest;
mod metadata;
pub use error::Error;
pub use manifest::{Executable, Facility, File, Identity, Limits, Manifest, PackageMetadata};
use std::collections::{BTreeMap, BTreeSet};
use std::io::Write;
use std::ops::Range;
use std::path::Path;
use harmos_domain::inventory::Declaration;
pub struct Source<'a> {
pub path: &'a str,
pub bytes: &'a [u8],
pub executable: Option<Executable>,
}
pub struct Bundle {
manifest: Manifest,
bytes: Vec<u8>,
contents: BTreeMap<String, Range<usize>>,
declarations: BTreeMap<String, Declaration>,
}
impl Bundle {
#[tracing::instrument(level = "debug", skip_all)]
pub fn build(identity: Identity, sources: &[Source<'_>], limits: Limits) -> Result<Self, Error> {
if sources.len() > limits.files {
return Err(Error::Limit("file count"));
}
let mut files = Vec::with_capacity(sources.len());
let mut contents = BTreeMap::new();
let mut total = 0usize;
let mut metadata_bytes = metadata::identity_size(&identity, limits.manifest_bytes)?;
for source in sources {
archive::path(source.path)?;
total = total.checked_add(source.bytes.len()).ok_or(Error::Limit("total file bytes"))?;
if source.bytes.len() > limits.file_bytes || total > limits.total_file_bytes {
return Err(Error::Limit("file bytes"));
}
if contents.insert(source.path.to_owned(), source.bytes).is_some() {
return Err(invalid("duplicate file path"));
}
if source.executable.as_ref().is_some_and(|value| value.registrations.len() > limits.registrations) {
return Err(Error::Limit("registration count"));
}
let sha256 = crate::digest(source.bytes);
metadata::file_size(source, &sha256, !files.is_empty(), &mut metadata_bytes, limits.manifest_bytes)?;
let mut executable = source.executable.clone();
if let Some(executable) = &mut executable {
executable.registrations.sort_by(|left, right| left.namespace.cmp(&right.namespace));
}
files.push(File {
path: source.path.to_owned(),
size: source.bytes.len() as u64,
sha256,
executable,
});
}
files.sort_by(|left, right| left.path.cmp(&right.path));
let manifest = Manifest { schema: 1, identity, files };
validate(&manifest, limits)?;
let bytes = archive::encode(&manifest, &contents, limits)?;
Self::inspect(&bytes, limits)
}
#[tracing::instrument(level = "debug", skip_all)]
pub fn inspect(bytes: &[u8], limits: Limits) -> Result<Self, Error> {
let (manifest, contents) = archive::decode(bytes, limits)?;
let mut declarations = BTreeMap::new();
let mut owners = BTreeSet::new();
for file in &manifest.files {
let content = &bytes[contents[&file.path].clone()];
if let Some(executable) = &file.executable {
let facility = match executable.facility {
Facility::Sidecar => harmos_domain::inventory::Facility::Sidecar,
Facility::Guest => harmos_domain::inventory::Facility::Guest,
};
let declaration = Declaration::inspect(content, facility).map_err(|error| invalid(error.to_string()))?;
if !owners.insert((executable.facility, declaration.id.clone(), &executable.target)) {
return Err(invalid("duplicate executable facility, owner, and target"));
}
declarations.insert(file.path.clone(), declaration);
}
}
Ok(Self {
manifest,
bytes: bytes.to_vec(),
contents,
declarations,
})
}
pub fn manifest(&self) -> &Manifest {
&self.manifest
}
pub fn bytes(&self) -> &[u8] {
&self.bytes
}
pub fn declaration(&self, path: &str) -> Option<&Declaration> {
self.declarations.get(path)
}
pub fn file(&self, path: &str) -> Option<&[u8]> {
self.contents.get(path).map(|range| &self.bytes[range.clone()])
}
pub fn extract(&self, parent: &Path) -> Result<tempfile::TempDir, Error> {
let directory = archive::staging(parent)?;
for file in &self.manifest.files {
let path = directory.path().join(&file.path);
std::fs::create_dir_all(path.parent().expect("a file has a staging parent"))?;
let mut output = std::fs::OpenOptions::new().write(true).create_new(true).open(path)?;
output.write_all(self.file(&file.path).expect("verified file exists"))?;
archive::permissions(&output, file.executable.as_ref().is_some_and(|value| value.facility == Facility::Sidecar))?;
}
let path = directory.path().join("manifest.json");
let mut output = std::fs::OpenOptions::new().write(true).create_new(true).open(path)?;
serde_json::to_writer(&mut output, &self.manifest)?;
archive::permissions(&output, false)?;
Ok(directory)
}
}
fn invalid(message: impl Into<String>) -> Error {
Error::Invalid(message.into())
}
fn validate(manifest: &Manifest, limits: Limits) -> Result<(), Error> {
if manifest.schema != 1 {
return Err(invalid("unsupported schema"));
}
let name = &manifest.identity.name;
if name.is_empty() || !name.bytes().all(|byte| byte.is_ascii_alphanumeric() || matches!(byte, b'-' | b'_')) {
return Err(invalid("invalid Cargo package name"));
}
if manifest.files.len() > limits.files {
return Err(Error::Limit("file count"));
}
let mut paths = BTreeSet::new();
let mut previous = None;
for file in &manifest.files {
archive::path(&file.path)?;
if previous.is_some_and(|value: &str| value >= file.path.as_str()) {
return Err(invalid("files are not in unique path order"));
}
previous = Some(&file.path);
if !paths.insert(file.path.to_ascii_lowercase()) {
return Err(invalid("case-insensitive file collision"));
}
if let Some(executable) = &file.executable {
if executable.target.is_empty()
|| executable.target.len() > 128
|| !executable.target.bytes().all(|byte| byte.is_ascii_alphanumeric() || matches!(byte, b'-' | b'_' | b'.'))
{
return Err(invalid("invalid executable target"));
}
if executable.registrations.len() > limits.registrations {
return Err(Error::Limit("registration count"));
}
let mut previous = None;
for registration in &executable.registrations {
if !harmos_domain::artifact_subset(®istration.namespace)
|| registration.schema == 0
|| previous.is_some_and(|value: &str| value >= registration.namespace.as_str())
{
return Err(invalid("invalid, duplicate, or unordered registration"));
}
previous = Some(®istration.namespace);
serde_json::from_slice::<&serde_json::value::RawValue>(®istration.payload)?;
}
}
}
for path in &paths {
for (offset, _) in path.match_indices('/') {
if paths.contains(&path[..offset]) {
return Err(invalid("file path is another file's parent"));
}
}
}
Ok(())
}