1 files changed, 129 insertions, 0 deletions
diff --git a/import/src/plugins/wikidata.rs b/import/src/plugins/wikidata.rs
new file mode 100644
index 0000000..358996e
--- /dev/null
+++ b/import/src/plugins/wikidata.rs
@@ -0,0 +1,129 @@
+/*
+    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::{Context, Result, bail};
+use jellycache::{EscapeKey, cache_memory};
+use log::info;
+use reqwest::{
+    Client, ClientBuilder,
+    header::{HeaderMap, HeaderName, HeaderValue},
+};
+use serde::{Deserialize, Serialize};
+use serde_json::Value;
+use std::{collections::BTreeMap, sync::Arc};
+use tokio::runtime::Handle;
+
+pub struct Wikidata {
+    client: Client,
+}
+
+#[derive(Debug, Deserialize, Serialize, Clone)]
+pub struct WikidataResponse {
+    entities: BTreeMap<String, WikidataEntity>,
+}
+
+#[derive(Debug, Deserialize, Serialize, Clone)]
+pub struct WikidataEntity {
+    pub pageid: u64,
+    pub ns: u64,
+    pub title: String,
+    pub lastrevid: u64,
+    pub modified: String,
+    pub r#type: String,
+    pub id: String,
+    pub claims: BTreeMap<String, Vec<WikidataClaim>>,
+}
+#[derive(Debug, Deserialize, Serialize, Clone)]
+pub struct WikidataClaim {
+    pub r#type: String,
+    pub id: String,
+    pub rank: String,
+    pub mainsnak: WikidataSnak,
+}
+
+#[derive(Debug, Deserialize, Serialize, Clone)]
+pub struct WikidataSnak {
+    pub snaktype: String,
+    pub property: String,
+    pub hash: String,
+    pub datavalue: Option<WikidataValue>,
+    pub datatype: String,
+}
+
+#[derive(Debug, Deserialize, Serialize, Clone)]
+pub struct WikidataValue {
+    pub value: Value,
+    pub r#type: String,
+}
+
+pub mod properties {
+    pub static IMAGE: &str = "P18";
+}
+
+impl Default for Wikidata {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl Wikidata {
+    pub fn new() -> Self {
+        let client = ClientBuilder::new()
+            .default_headers(HeaderMap::from_iter([
+                (
+                    HeaderName::from_static("accept"),
+                    HeaderValue::from_static("application/json"),
+                ),
+                (
+                    HeaderName::from_static("user-agent"),
+                    HeaderValue::from_static(USER_AGENT),
+                ),
+            ]))
+            .build()
+            .unwrap();
+        Self { client }
+    }
+
+    pub fn query_image_path(&self, id: String, rt: &Handle) -> Result<Option<String>> {
+        let response = self.query(id.clone(), rt)?;
+        if let Some(entity) = response.entities.get(&id) {
+            if let Some(images) = entity.claims.get(properties::IMAGE) {
+                for image in images {
+                    if image.mainsnak.datatype != "commonsMedia" {
+                        bail!("image is of type {:?}", image.mainsnak.datatype);
+                    }
+                    if let Some(dv) = &image.mainsnak.datavalue {
+                        if let Value::String(filename) = &dv.value {
+                            return Ok(Some(filename.to_owned()));
+                        }
+                    }
+                }
+            }
+        }
+        Ok(None)
+    }
+
+    pub fn query(&self, id: String, rt: &Handle) -> Result<Arc<WikidataResponse>> {
+        cache_memory(
+            &format!("ext/wikidata/{}.json", EscapeKey(&id)),
+            move || {
+                rt.block_on(async {
+                    info!("entity query: {id}");
+                    Ok(self
+                        .client
+                        .get(format!("https://www.wikidata.org/entity/{id}"))
+                        .send()
+                        .await?
+                        .error_for_status()?
+                        .json()
+                        .await?)
+                })
+            },
+        )
+        .context("wikidata entity")
+    }
+}