Extract used wikidata properties from infobox template

2025-08-03 13:40:55 +02:00
parent 9d4136c993
commit e9cfb0293b
5 changed files with 1015 additions and 0 deletions
@@ -0,0 +1,56 @@
+#[doc = include_str!("../README.md")]
+pub mod wikipedia_infobox_analyzer {
+    use regex::Regex;
+
+    /// Looks for a used template that does lists wikidata properties listing
+    pub fn extract_used_properties_from_template(template: String) -> Vec<String> {
+        let lines = template.lines();
+        let used = lines
+            .filter(|line| line.starts_with("{{Gebruikt Wikidata"))
+            .nth(0)
+            .expect("Template should have a line declaring which properties are used");
+
+        // The properties are listed inside the template with their number: P1|P2|P...
+        let re = Regex::new(r"P\d+").unwrap();
+
+        // Find all matches and collect into vector
+        let properties = re
+            .find_iter(used)
+            .map(|m| m.as_str().to_owned())
+            .collect::<Vec<String>>();
+
+        return properties;
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::wikipedia_infobox_analyzer::*;
+    use std::fs;
+
+    #[test]
+    fn test_extract_used_properties_from_template() {
+        let contents = fs::read_to_string("examples/templates/nl_infobox_bedrijf")
+            .expect("Template file should exist");
+
+        assert_eq!(
+            extract_used_properties_from_template(contents),
+            vec![
+                "P18", "P154", "P1128", "P4103", "P2139", "P2295", "P2226", "P856", "P169",
+                "P1448", "P159", "P749", "P355", "P17", "P1056", "P452", "P576", "P112", "P127",
+                "P856", "P2096"
+            ]
+        );
+
+        let contents = fs::read_to_string("examples/templates/nl_infobox_software")
+            .expect("Template file should exist");
+
+        assert_eq!(
+            extract_used_properties_from_template(contents),
+            vec![
+                "P18", "P154", "P170", "P178", "P275", "P277", "P306", "P348", "P400", "P548",
+                "P571", "P577", "P856", "P1324", "P2096"
+            ]
+        );
+    }
+}