blindforge/src/api_client.rs

use std::{collections::HashSet, io::Write, path::PathBuf};

use crate::{
	config::Config,
	repo::{RepoMetadata, WriteRepoMetadataError},
};

use rand::RngExt;
use serde::Deserialize;
use sha2::Digest;
use trillium_client::Client;

pub const USER_AGENT: &str = "Blindforge";
pub const MAX_PAGE: u32 = 256;
pub const MAX_ENTRIES: u32 = 256;
pub const MAX_FILE_SIZE: u64 = 8 * 1024 * 1024;
pub const MAX_PATH_SIZE: usize = 1024;
pub const MAX_URL_SIZE: usize = 256;

#[derive(Deserialize)]
struct RepoGetTagResponse {
	tarball_url: String,
}

#[derive(Debug)]
pub enum ClientError {
	Client(trillium_client::Error),
	Http(trillium_client::UnexpectedStatusError),
	Parse(trillium_client::ClientSerdeError),
}

impl From<trillium_client::Error> for ClientError {
	fn from(value: trillium_client::Error) -> Self {
		Self::Client(value)
	}
}

impl From<trillium_client::UnexpectedStatusError> for ClientError {
	fn from(value: trillium_client::UnexpectedStatusError) -> Self {
		Self::Http(value)
	}
}

impl From<trillium_client::ClientSerdeError> for ClientError {
	fn from(value: trillium_client::ClientSerdeError) -> Self {
		Self::Parse(value)
	}
}

pub fn make_client() -> Client {
	Client::new(trillium_native_tls::NativeTlsConfig::<
		trillium_smol::ClientConfig,
	>::default())
	.with_default_header("User-Agent", USER_AGENT)
}

pub async fn fetch_repo_archive_url(
	client: &mut Client,
	url: &str,
	owner: &str,
	repo: &str,
	tag: &str,
	token: Option<&str>,
) -> Result<String, ClientError> {
	let mut req = client.get(format!("{url}/api/v1/repos/{owner}/{repo}/tags/{tag}"));
	if let Some(token) = token {
		req = req.with_request_header("Authorization", format!("token {token}"));
	}
	let res: RepoGetTagResponse = req.await?.success()?.response_json().await?;
	Ok(res.tarball_url)
}

#[derive(Deserialize)]
struct GitTreeResponse {
	tree: Vec<GitEntry>,
	truncated: bool,
	total_count: u32,
}

#[derive(Deserialize)]
struct GitEntry {
	path: String,
	#[serde(rename = "type")]
	entry_type: String,
	size: u64,
	url: String,
}

async fn fetch_repo_tree_at_commit_page(
	client: &mut Client,
	url: &str,
	owner: &str,
	repo: &str,
	commit_hash: &str,
	token: Option<&str>,
	page: u32,
) -> Result<GitTreeResponse, ClientError> {
	let mut req = client.get(format!(
		"{url}/api/v1/repos/{owner}/{repo}/git/trees/{commit_hash}?recursive=true&page={page}"
	));
	if let Some(token) = token {
		req = req.with_request_header("Authorization", format!("token {token}"));
	}
	// TODO ensure we don't use too much memory
	Ok(req.await?.success()?.response_json().await?)
}

#[derive(Debug)]
pub enum FetchRepoError {
	CannotCreateDir(std::io::Error),
	TooManyEntries,
	Client(ClientError),
	UrlParsing,
	WriteRepoMetadata(WriteRepoMetadataError),
}

impl From<ClientError> for FetchRepoError {
	fn from(value: ClientError) -> Self {
		Self::Client(value)
	}
}

impl From<trillium_client::Error> for FetchRepoError {
	fn from(value: trillium_client::Error) -> Self {
		Self::Client(value.into())
	}
}

impl From<trillium_client::UnexpectedStatusError> for FetchRepoError {
	fn from(value: trillium_client::UnexpectedStatusError) -> Self {
		Self::Client(value.into())
	}
}

impl From<trillium_client::ClientSerdeError> for FetchRepoError {
	fn from(value: trillium_client::ClientSerdeError) -> Self {
		Self::Client(value.into())
	}
}

impl From<url::ParseError> for FetchRepoError {
	fn from(_value: url::ParseError) -> Self {
		Self::UrlParsing
	}
}

impl From<WriteRepoMetadataError> for FetchRepoError {
	fn from(value: WriteRepoMetadataError) -> Self {
		Self::WriteRepoMetadata(value)
	}
}

#[derive(Default)]
pub struct RepoIndex {
	pub files: Vec<RepoIndexFile>,
}

pub struct RepoIndexFile {
	pub path: String,
	pub url: String,
}

pub async fn fetch_repo_tree_index_at_commit(
	client: &mut Client,
	repo_url: &str,
	commit_hash: &str,
	token: Option<&str>,
) -> Result<(RepoIndex, RepoMetadata), FetchRepoError> {
	let parsed = url::Url::parse(repo_url)?;
	let mut base = parsed.clone();
	base.set_fragment(None);
	base.set_path("");
	let base_url = base.as_str();
	// Is the URL always /owner/repo?
	let mut segments = parsed.path_segments().ok_or(FetchRepoError::UrlParsing)?;
	let owner = segments.next().ok_or(FetchRepoError::UrlParsing)?;
	let repo = segments.next().ok_or(FetchRepoError::UrlParsing)?;
	let repo = repo.strip_suffix(".git").unwrap_or(repo);

	let mut repo_index = RepoIndex::default();
	let mut count: u32 = 0;
	for page in 1..MAX_PAGE {
		let res =
			fetch_repo_tree_at_commit_page(client, base_url, owner, repo, commit_hash, token, page)
				.await?;
		count = count.saturating_add(res.total_count);
		if count > MAX_ENTRIES || res.total_count > MAX_ENTRIES {
			return Err(FetchRepoError::TooManyEntries);
		}
		if res.tree.is_empty() {
			break;
		}
		for entry in res.tree {
			if entry.entry_type.as_str() == "blob"
				&& entry.size <= MAX_FILE_SIZE
				&& entry.path.len() <= MAX_PATH_SIZE
				&& entry.url.len() <= MAX_URL_SIZE
			{
				repo_index.files.push(RepoIndexFile {
					path: entry.path,
					url: entry.url,
				});
			}
		}
		if !res.truncated || count >= res.total_count {
			break;
		}
	}

	Ok((
		repo_index,
		RepoMetadata {
			date: std::time::UNIX_EPOCH.elapsed().unwrap().as_secs(),
			repo_url: repo_url.to_string(),
			commit_hash: commit_hash.to_string(),
			content: Vec::new(),
		},
	))
}

#[derive(Deserialize)]
struct GitBlob {
	content: String,
}

pub async fn fetch_repo_files(
	config: &Config,
	client: &mut Client,
	repo_index: &RepoIndex,
	repo_metadata: &mut RepoMetadata,
	token: Option<&str>,
) -> Result<(), FetchRepoError> {
	let mut hasher = sha2::Sha256::default();

	let repo_key: [u8; 32] = rand::rng().random();
	hasher.update(repo_key);
	let repo_id: [u8; 32] = hasher.finalize_reset().into();
	let mut repo_id_str = [0; 24];
	base64_turbo::URL_SAFE
		.encode_into(&repo_id[0..16], &mut repo_id_str)
		.expect("unreachable");
	let repo_id_str = str::from_utf8(&repo_id_str).expect("unreachable");

	let repo_dir = PathBuf::from(&config.data_dir)
		.join(crate::SUBDIR_REPOS)
		.join(repo_id_str);
	std::fs::create_dir(&repo_dir).map_err(FetchRepoError::CannotCreateDir)?;

	let mut file_index = HashSet::new();

	for file in repo_index.files.iter() {
		let mut req = client.get(file.url.clone());
		if let Some(token) = token {
			req = req.with_request_header("Authorization", format!("token {token}"));
		}
		// TODO ensure we don't use too much memory
		let blob: GitBlob = req.await?.success()?.response_json().await?;
		if base64_turbo::STANDARD.estimate_decoded_len(blob.content.len()) as u64 > MAX_FILE_SIZE {
			continue;
		}
		if let Ok(content) = base64_turbo::STANDARD.decode(&blob.content) {
			hasher.update(&content);
			let file_name: [u8; 32] = hasher.finalize_reset().into();
			if file_index.insert(file_name) {
				let mut file_name_str = [0; 44];
				base64_turbo::URL_SAFE
					.encode_into(file_name, &mut file_name_str)
					.expect("unreachable");

				repo_metadata
					.content
					.extend_from_slice(&(file.path.len() as u32).to_be_bytes());
				repo_metadata
					.content
					.extend_from_slice(file.path.as_bytes());
				repo_metadata.content.extend_from_slice(&file_name_str);

				let file_name_str = str::from_utf8(&file_name_str).expect("unreachable");
				if let Ok(mut file) = std::fs::OpenOptions::new()
					.create_new(true)
					.write(true)
					.open(repo_dir.join(file_name_str))
				{
					file.write_all(&content).unwrap();
				}
			}
		}
	}

	repo_metadata.write_to_file(config, &repo_dir)?;

	Ok(())
}