Merge pull request #198 from vladh/master

GAP: improve filename format
lovasoa · Mar 12, 2023 · ec8a0ff · ec8a0ff
2 parents 17ecc72 + 8b8b3f3
commit ec8a0ff
Showing 1 changed file with 22 additions and 5 deletions.
diff --git a/src/google_arts_and_culture/tile_info.rs b/src/google_arts_and_culture/tile_info.rs
@@ -39,6 +39,27 @@ impl PageInfo {
     }
 }
 
+fn get_name_from_gap_html(html: &str) -> String {
+    let name = Regex::new(r#"<h1 class="[^"]+">([^<]+)</h1><h2 class="[^"]+"><span class="[^"]+"><a href="[^"]+">([^"]+) ([^"]+)</a></span><span class="[^"]+">([^<]+)</span></h2>"#)
+        .unwrap()
+        .captures(html)
+        .map(|c| format!("{}, {}; {}; {}",
+            &(c[3]),
+            &(c[2]),
+            &(c[1]),
+            &(c[4])));
+
+    if let Some(result) = name {
+        return result
+    }
+
+    Regex::new(r#""name":"([^"]+)"#)
+        .unwrap()
+        .captures(html)
+        .map(|c| (c[1]).to_string())
+        .unwrap_or_else(||"Google Arts and Culture Image".into())
+}
+
 impl FromStr for PageInfo {
     type Err = PageParseError;
 
@@ -51,11 +72,7 @@ impl FromStr for PageInfo {
             .get(2)
             .map_or_else(Default::default, |s| s.as_str().into());
 
-        let name = Regex::new(r#""name":"([^"]+)"#)
-            .unwrap()
-            .captures(s)
-            .map(|c| c[1].to_string())
-            .unwrap_or_else(|| "Google Arts and culture image".into());
+        let name = get_name_from_gap_html(s);
 
         Ok(PageInfo {
             base_url,