add wikidata api bindings (untested)

1 files changed, 120 insertions, 0 deletions

diff --git a/import/src/wikidata.rs b/import/src/wikidata.rs
new file mode 100644
index 0000000..267899f
--- /dev/null
+++ b/import/src/wikidata.rs
@@ -0,0 +1,120 @@
+/*
+    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::{bail, Result};
+use jellybase::cache::async_cache_memory;
+use log::info;
+use reqwest::{
+    header::{HeaderMap, HeaderName, HeaderValue},
+    Client, ClientBuilder,
+};
+use serde::Deserialize;
+use serde_json::Value;
+use std::collections::BTreeMap;
+
+pub struct Wikidata {
+    client: Client,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+pub struct WikidataResponse {
+    entities: BTreeMap<String, WikidataEntity>,
+}
+
+#[derive(Debug, Deserialize, 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, Clone)]
+pub struct WikidataClaim {
+    pub r#type: String,
+    pub id: String,
+    pub rank: String,
+    pub mainsnak: WikidataSnak,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+pub struct WikidataSnak {
+    pub snaktype: String,
+    pub property: String,
+    pub hash: String,
+    pub datavalue: WikidataValue,
+    pub datatype: String,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+pub struct WikidataValue {
+    pub value: Value,
+    pub r#type: String,
+}
+
+pub mod properties {
+    pub static IMAGE: &str = "P18";
+}
+
+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 async fn query_image_path(&self, id: String) -> Result<Option<String>> {
+        let response = self.query(id.clone()).await?;
+        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 Value::String(filename) = &image.mainsnak.datavalue.value {
+                        return Ok(Some(filename.to_owned()));
+                    }
+                }
+            }
+        }
+        Ok(None)
+    }
+
+    pub async fn query(&self, id: String) -> Result<WikidataResponse> {
+        let json = async_cache_memory("api-wikidata", id.clone(), || async move {
+            info!("entity query: {id}");
+
+            let resp = self
+                .client
+                .get(format!("https://www.wikidata.org/entity/{id}"))
+                .send()
+                .await?
+                .error_for_status()?
+                .text()
+                .await?;
+
+            Ok(resp)
+        })
+        .await?;
+
+        Ok(serde_json::from_str(&json)?)
+    }
+}