1 files changed, 108 insertions, 0 deletions
diff --git a/import/src/vgmdb.rs b/import/src/vgmdb.rs
new file mode 100644
index 0000000..9ac76d6
--- /dev/null
+++ b/import/src/vgmdb.rs
@@ -0,0 +1,108 @@
+/*
+    This file is part of jellything (https://codeberg.org/metamuffin/jellything)
+    which is licensed under the GNU Affero General Public License (version 3); see /COPYING.
+    Copyright (C) 2025 metamuffin <metamuffin.org>
+*/
+
+use crate::USER_AGENT;
+use anyhow::Result;
+use jellybase::cache::{async_cache_file, async_cache_memory, CachePath};
+use log::info;
+use regex::Regex;
+use reqwest::{
+    header::{HeaderMap, HeaderName, HeaderValue},
+    Client, ClientBuilder,
+};
+use std::{
+    sync::{Arc, LazyLock},
+    time::Duration,
+};
+use tokio::{
+    io::AsyncWriteExt,
+    sync::Semaphore,
+    time::{sleep_until, Instant},
+};
+
+pub struct Vgmdb {
+    client: Client,
+    rate_limit: Arc<Semaphore>,
+}
+
+static RE_IMAGE_URL_FROM_HTML: LazyLock<Regex> = LazyLock::new(|| {
+    Regex::new(r#"href='(?<url>https://media.vgm.io/artists/[-/\w\.]+)'"#).unwrap()
+});
+
+impl Vgmdb {
+    pub fn new() -> Self {
+        let client = ClientBuilder::new()
+            .default_headers(HeaderMap::from_iter([
+                (
+                    HeaderName::from_static("user-agent"),
+                    HeaderValue::from_static(USER_AGENT),
+                ),
+                (
+                    HeaderName::from_static("x-comment"),
+                    HeaderValue::from_static("Please add an API, thanks!"),
+                ),
+            ]))
+            .build()
+            .unwrap();
+        Self {
+            client,
+            rate_limit: Arc::new(Semaphore::new(3)),
+        }
+    }
+
+    pub async fn get_artist_image(&self, id: u64) -> Result<Option<CachePath>> {
+        if let Some(url) = self.get_artist_image_url(id).await? {
+            Ok(Some(
+                async_cache_file("api-vgmdb-media", url.clone(), |mut file| async move {
+                    info!("downloading image {url:?}");
+                    let mut res = self.client.get(url).send().await?.error_for_status()?;
+                    while let Some(chunk) = res.chunk().await? {
+                        file.write_all(&chunk).await?;
+                    }
+                    Ok(())
+                })
+                .await?,
+            ))
+        } else {
+            Ok(None)
+        }
+    }
+
+    pub async fn get_artist_image_url(&self, id: u64) -> Result<Option<String>> {
+        let html = self.scrape_artist_page(id).await?;
+        if let Some(cap) = RE_IMAGE_URL_FROM_HTML.captures(&html) {
+            if let Some(url) = cap.name("url").map(|m| m.as_str()) {
+                return Ok(Some(url.to_string()));
+            }
+        }
+        return Ok(None);
+    }
+
+    pub async fn scrape_artist_page(&self, id: u64) -> Result<Arc<String>> {
+        async_cache_memory("api-vgmdb-artist", id.clone(), || async move {
+            let _permit = self.rate_limit.clone().acquire_owned().await?;
+            let permit_drop_ts = Instant::now() + Duration::from_secs(1);
+            info!("scrape artist: {id}");
+
+            let resp = self
+                .client
+                .get(format!("https://vgmdb.net/artist/{id}"))
+                .send()
+                .await?
+                .error_for_status()?
+                .text()
+                .await?;
+
+            tokio::task::spawn(async move {
+                sleep_until(permit_drop_ts).await;
+                drop(_permit);
+            });
+
+            Ok(resp)
+        })
+        .await
+    }
+}