Merge pull request #1438 from roumcha/fix-language-name-normalization

Fix language name normalization
kenkoooo · Oct 15, 2023 · bfcd349 · bfcd349
2 parents dc92d7b + 09b6617
commit bfcd349
Show file tree

Hide file tree

Showing 3 changed files with 88 additions and 12 deletions.
diff --git a/atcoder-problems-backend/sql-client/src/language_count.rs b/atcoder-problems-backend/sql-client/src/language_count.rs
@@ -192,12 +192,33 @@ impl LanguageCountClient for PgPool {
     }
 }
 
+const MAPPING: [(&str, &str); 9] = [
+    ("PyPy", "Python"),
+    ("Python (Cython", "Cython"),
+    ("Assembly x64", "Assembly x64"),
+    ("Awk", "AWK"),
+    ("IOI-Style", "C++"),
+    ("LuaJIT", "Lua"),
+    ("Seed7", "Seed7"),
+    ("Perl6", "Raku"),
+    ("Objective-C", "Objective-C"),
+];
+
 fn simplify_language(lang: &str) -> String {
-    let re = Regex::new(r"\d*\s*\(.*\)").unwrap();
-    if lang.starts_with("Perl6") {
-        "Raku".to_string()
-    } else {
-        re.replace(lang, "").to_string()
+    for (beginning, simplified) in MAPPING {
+        if lang.starts_with(beginning) {
+            return simplified.to_string();
+        }
+    }
+
+    let simplified = Regex::new(r"\s*[\d(\-].*")
+        .unwrap()
+        .replace(lang, "")
+        .to_string();
+
+    match simplified.len() {
+        0 => String::from("Unknown"),
+        _ => simplified,
     }
 }
 
@@ -207,12 +228,31 @@ mod tests {
 
     #[test]
     fn test_simplify_language() {
-        assert_eq!(simplify_language("language1"), "language1");
+        assert_eq!(simplify_language("language1"), "language");
         assert_eq!(simplify_language("Perl (5)"), "Perl");
         assert_eq!(simplify_language("Perl6"), "Raku");
         assert_eq!(simplify_language("Fortran(GNU Fortran 9.2.1)"), "Fortran");
         assert_eq!(simplify_language("Ada2012 (GNAT 9.2.1)"), "Ada");
-        assert_eq!(simplify_language("PyPy2 (7.3.0)"), "PyPy");
-        assert_eq!(simplify_language("Haxe (4.0.3); js"), "Haxe; js");
+        assert_eq!(simplify_language("Haxe (4.0.3); js"), "Haxe");
+        assert_eq!(simplify_language("C++11 (Clang++ 3.4)"), "C++");
+        assert_eq!(simplify_language("C++ 20 (gcc 12.2)"), "C++");
+        assert_eq!(simplify_language("C# 11.0 (.NET 7.0.7)"), "C#");
+        assert_eq!(simplify_language("C# 11.0 AOT (.NET 7.0.7)"), "C#");
+        assert_eq!(simplify_language("Visual Basic 16.9 (...)"), "Visual Basic");
+        assert_eq!(simplify_language("><> (fishr 0.1.0)"), "><>");
+        assert_eq!(simplify_language("プロデル (...)"), "プロデル");
+
+        // mapped individually
+        assert_eq!(simplify_language("Assembly x64"), "Assembly x64");
+        assert_eq!(simplify_language("Awk (GNU Awk 4.1.4)"), "AWK");
+        assert_eq!(simplify_language("IOI-Style C++ (GCC 5.4.1)"), "C++");
+        assert_eq!(simplify_language("LuaJIT (2.0.4)"), "Lua");
+        assert_eq!(simplify_language("Objective-C (Clang3.8.0)"), "Objective-C");
+        assert_eq!(simplify_language("PyPy2 (7.3.0)"), "Python");
+        assert_eq!(simplify_language("Python (Cython 0.29.34)"), "Cython");
+        assert_eq!(simplify_language("Cython (0.29.16)"), "Cython");
+        assert_eq!(simplify_language("Seed7 (Seed7 3.2.1)"), "Seed7");
+
+        assert_eq!(simplify_language("1234"), "Unknown");
     }
 }
diff --git a/atcoder-problems-frontend/src/utils/LanguageNormalizer.test.ts b/atcoder-problems-frontend/src/utils/LanguageNormalizer.test.ts
@@ -7,4 +7,26 @@ test("normalize language", () => {
   expect(normalizeLanguage("C++11 (Clang++ 3.4)")).toBe("C++");
   expect(normalizeLanguage("Scala (2.11.7)")).toBe("Scala");
   expect(normalizeLanguage("Fortran(GNU Fortran 9.2.1)")).toBe("Fortran");
+  expect(normalizeLanguage("Ada2012 (GNAT 9.2.1)")).toBe("Ada");
+  expect(normalizeLanguage("Haxe (4.0.3); js")).toBe("Haxe");
+  expect(normalizeLanguage("C++11 (Clang++ 3.4)")).toBe("C++");
+  expect(normalizeLanguage("C++ 20 (gcc 12.2)")).toBe("C++");
+  expect(normalizeLanguage("C# 11.0 (.NET 7.0.7)")).toBe("C#");
+  expect(normalizeLanguage("C# 11.0 AOT (.NET 7.0.7)")).toBe("C#");
+  expect(normalizeLanguage("Visual Basic 16.9 (...)")).toBe("Visual Basic");
+  expect(normalizeLanguage("><> (fishr 0.1.0)")).toBe("><>");
+  expect(normalizeLanguage("プロデル (...)")).toBe("プロデル");
+
+  // mapped individually
+  expect(normalizeLanguage("Assembly x64")).toBe("Assembly x64");
+  expect(normalizeLanguage("Awk (GNU Awk 4.1.4)")).toBe("AWK");
+  expect(normalizeLanguage("IOI-Style C++ (GCC 5.4.1)")).toBe("C++");
+  expect(normalizeLanguage("LuaJIT (2.0.4)")).toBe("Lua");
+  expect(normalizeLanguage("Objective-C (Clang3.8.0)")).toBe("Objective-C");
+  expect(normalizeLanguage("PyPy2 (7.3.0)")).toBe("Python");
+  expect(normalizeLanguage("Python (Cython 0.29.34)")).toBe("Cython");
+  expect(normalizeLanguage("Cython (0.29.16)")).toBe("Cython");
+  expect(normalizeLanguage("Seed7 (Seed7 3.2.1)")).toBe("Seed7");
+
+  expect(normalizeLanguage("1234")).toBe("Unknown");
 });
diff --git a/atcoder-problems-frontend/src/utils/LanguageNormalizer.ts b/atcoder-problems-frontend/src/utils/LanguageNormalizer.ts
@@ -1,7 +1,21 @@
+const mapping: [beginning: string, normalized: string][] = [
+  ["PyPy", "Python"],
+  ["Python (Cython", "Cython"],
+  ["Assembly x64", "Assembly x64"],
+  ["Awk", "AWK"],
+  ["IOI-Style", "C++"],
+  ["LuaJIT", "Lua"],
+  ["Seed7", "Seed7"],
+  ["Perl6", "Raku"],
+  ["Objective-C", "Objective-C"],
+];
+
 export const normalizeLanguage = (language: string): string => {
-  if (language.startsWith("Perl6")) {
-    return "Raku";
-  } else {
-    return language.replace(/\d*\s*\(.*\)$/, "");
+  for (const [beginning, normalized] of mapping) {
+    if (language.startsWith(beginning)) {
+      return normalized;
+    }
   }
+
+  return language.replace(/\s*[\d(\-].*/, "") || "Unknown";
 };