Fetch identifiers by article title

src/lib.rs

@@ -85,6 +85,22 @@ pub mod wikipedia_infobox_analyzer {
            .unique()
            .collect::<Vec<u64>>()
     }
+
+    /// Fetch the wikidata identifier associated with the wikipedia article
+    pub fn fetch_wiki_item_by_article_title(title: String, language_code: String) -> u64 {
+        let uri = format!("https://{}.wikipedia.org/w/api.php?action=query&prop=pageprops&format=json&titles={}", &language_code, &title);
+
+        let res = reqwest::blocking::get(uri).unwrap();
+        let text = res.text().unwrap();
+        if text.contains("-1") {
+            return 0;
+        }
+
+        let re = Regex::new("\"wikibase_item\":\"Q(\\d+)\"").unwrap();
+        let Some(qid) = re.captures(&text) else { return 0 };
+
+        qid[1].parse().expect("Should be a numeric value")
+    }
 }
 
 #[cfg(test)]
@@ -158,4 +174,11 @@ mod tests {
         assert!(!properties.contains(&19)); // property: place of birth
         assert!(!properties.contains(&30)); // property: continent
     }
+
+    #[test]
+    fn test_fetch_wiki_item_by_article_title() {
+        assert_eq!(fetch_wiki_item_by_article_title("Earth".to_string(), "en".to_string()), 2);
+        assert_eq!(fetch_wiki_item_by_article_title("Train".to_string(), "en".to_string()), 870);
+        assert_eq!(fetch_wiki_item_by_article_title("Baum".to_string(),  "de".to_string()), 10884);
+    }
 }