Files
proxy-scraper-checker/src/ipdb.rs
T

192 lines
5.8 KiB
Rust
Raw Normal View History

2025-05-24 02:31:23 +03:00
use std::{io, path::PathBuf};
use color_eyre::eyre::{WrapErr as _, eyre};
use tokio::io::AsyncWriteExt as _;
#[cfg(feature = "tui")]
use crate::event::{AppEvent, Event};
use crate::{fs::get_cache_path, utils::is_docker};
#[derive(Clone, Copy)]
2025-05-24 02:31:23 +03:00
pub enum DbType {
Asn,
Geo,
}
impl DbType {
const fn name(self) -> &'static str {
2025-05-24 13:51:22 +03:00
match self {
Self::Asn => "ASN",
Self::Geo => "geolocation",
}
}
const fn url(self) -> &'static str {
2025-05-24 02:31:23 +03:00
match self {
Self::Asn => {
2025-09-18 14:20:57 +03:00
"https://raw.githubusercontent.com/P3TERX/GeoLite.mmdb/refs/heads/download/GeoLite2-ASN.mmdb"
}
Self::Geo => {
2025-09-18 14:20:57 +03:00
"https://raw.githubusercontent.com/P3TERX/GeoLite.mmdb/refs/heads/download/GeoLite2-City.mmdb"
}
2025-05-24 02:31:23 +03:00
}
}
async fn db_path(self) -> crate::Result<PathBuf> {
2025-09-26 17:08:57 +03:00
let mut cache_path = get_cache_path().await?;
2025-05-24 02:31:23 +03:00
match self {
Self::Asn => cache_path.push("asn_database.mmdb"),
Self::Geo => cache_path.push("geolocation_database.mmdb"),
}
Ok(cache_path)
}
async fn etag_path(self) -> crate::Result<PathBuf> {
2025-09-26 17:08:57 +03:00
let mut db_path = self.db_path().await?;
2025-05-24 02:31:23 +03:00
db_path.set_extension("mmdb.etag");
Ok(db_path)
}
async fn save_db(
self,
2025-05-24 02:31:23 +03:00
mut response: reqwest::Response,
#[cfg(feature = "tui")] tx: tokio::sync::mpsc::UnboundedSender<Event>,
) -> crate::Result<()> {
2025-05-24 02:31:23 +03:00
#[cfg(feature = "tui")]
2025-07-15 15:13:55 +03:00
drop(tx.send(Event::App(AppEvent::IpDbTotal(
self,
2025-05-24 02:31:23 +03:00
response.content_length(),
2025-07-15 15:13:55 +03:00
))));
2025-05-24 02:31:23 +03:00
let db_path = self.db_path().await?;
let mut file =
tokio::fs::File::create(&db_path).await.wrap_err_with(|| {
2025-09-30 07:53:39 +03:00
format!("failed to create file: {}", db_path.display())
2025-05-24 02:31:23 +03:00
})?;
2025-09-26 17:08:57 +03:00
while let Some(chunk) = response.chunk().await? {
2025-05-24 02:31:23 +03:00
file.write_all(&chunk).await.wrap_err_with(|| {
2025-09-30 07:53:39 +03:00
format!("failed to write to file: {}", db_path.display())
2025-05-24 02:31:23 +03:00
})?;
#[cfg(feature = "tui")]
drop(
tx.send(Event::App(AppEvent::IpDbDownloaded(
self,
chunk.len(),
))),
);
2025-05-24 02:31:23 +03:00
}
Ok(())
}
async fn save_etag(self, etag: impl AsRef<[u8]>) -> crate::Result<()> {
2025-05-24 13:51:22 +03:00
let path = self.etag_path().await?;
tokio::fs::write(&path, etag).await.wrap_err_with(move || {
2025-09-30 07:53:39 +03:00
format!("failed to write to file: {}", path.display())
2025-05-24 13:51:22 +03:00
})
}
async fn read_etag(
self,
) -> crate::Result<Option<reqwest::header::HeaderValue>> {
2025-05-24 13:51:22 +03:00
let path = self.etag_path().await?;
match tokio::fs::read_to_string(&path).await {
Ok(text) => Ok(text.parse().ok()),
Err(e) if e.kind() == io::ErrorKind::NotFound => Ok(None),
Err(e) => Err(e).wrap_err_with(move || {
2025-09-30 07:53:39 +03:00
format!("failed to read file to string: {}", path.display())
2025-05-24 13:51:22 +03:00
}),
}
}
async fn remove_etag(self) -> crate::Result<()> {
2025-05-24 13:51:22 +03:00
let path = self.etag_path().await?;
match tokio::fs::remove_file(&path).await {
Ok(()) => Ok(()),
Err(e) if e.kind() == io::ErrorKind::NotFound => Ok(()),
Err(e) => Err(e).wrap_err_with(move || {
2025-09-30 07:53:39 +03:00
format!("failed to remove file: {}", path.display())
2025-05-24 13:51:22 +03:00
}),
}
}
pub async fn download(
2025-05-24 02:31:23 +03:00
self,
http_client: reqwest_middleware::ClientWithMiddleware,
2025-05-24 02:31:23 +03:00
#[cfg(feature = "tui")] tx: tokio::sync::mpsc::UnboundedSender<Event>,
) -> crate::Result<()> {
2025-05-24 02:31:23 +03:00
let db_path = self.db_path().await?;
let mut headers = reqwest::header::HeaderMap::new();
#[expect(clippy::collapsible_if)]
2025-05-24 02:31:23 +03:00
if tokio::fs::metadata(&db_path).await.is_ok_and(|m| m.is_file()) {
2025-09-26 17:08:57 +03:00
if let Some(etag) = self.read_etag().await? {
2025-05-24 02:31:23 +03:00
headers.insert(reqwest::header::IF_NONE_MATCH, etag);
}
}
let response = http_client
.get(self.url())
.headers(headers)
.send()
2025-09-26 17:08:57 +03:00
.await?
.error_for_status()?;
2025-05-24 02:31:23 +03:00
if response.status() == reqwest::StatusCode::NOT_MODIFIED {
2025-05-30 21:58:07 +00:00
tracing::info!(
2025-05-24 02:31:23 +03:00
"Latest {} database is already cached at {}",
self.name(),
db_path.display()
);
return Ok(());
}
if response.status() != reqwest::StatusCode::OK {
return Err(eyre!(
"HTTP status error ({}) for url ({})",
response.status(),
response.url()
));
}
let etag = response.headers().get(reqwest::header::ETAG).cloned();
self.save_db(
response,
#[cfg(feature = "tui")]
tx.clone(),
)
2025-09-26 17:08:57 +03:00
.await?;
2025-05-24 02:31:23 +03:00
if is_docker().await {
2025-05-30 21:58:07 +00:00
tracing::info!(
2025-05-24 02:31:23 +03:00
"Downloaded {} database to Docker volume ({} in container)",
self.name(),
db_path.display()
);
} else {
2025-05-30 21:58:07 +00:00
tracing::info!(
2025-05-24 02:31:23 +03:00
"Downloaded {} database to {}",
self.name(),
db_path.display()
);
}
2025-08-15 17:19:50 +03:00
drop(db_path);
2025-05-24 02:31:23 +03:00
if let Some(etag_value) = etag {
2025-09-26 17:08:57 +03:00
self.save_etag(etag_value).await
2025-05-24 02:31:23 +03:00
} else {
2025-09-26 17:08:57 +03:00
self.remove_etag().await
2025-05-24 02:31:23 +03:00
}
}
2025-05-24 13:51:22 +03:00
pub async fn open_mmap(
self,
) -> crate::Result<maxminddb::Reader<maxminddb::Mmap>> {
2025-05-24 13:51:22 +03:00
let path = self.db_path().await?;
tokio::task::spawn_blocking(move || maxminddb::Reader::open_mmap(path))
2025-09-26 17:08:57 +03:00
.await?
2025-05-24 13:51:22 +03:00
.wrap_err_with(move || {
2025-09-30 07:53:39 +03:00
format!("failed to open IP database: {}", self.name())
2025-05-24 13:51:22 +03:00
})
}
2025-05-24 02:31:23 +03:00
}