Skip to content

Commit

Permalink
Merge pull request #1440 from kenkoooo/staging
Browse files Browse the repository at this point in the history 
言語をなんとかするやつ
  • Loading branch information
@kenkoooo
kenkoooo committed Oct 15, 2023
2 parents a6bdfad + 5a0a454 commit 163bb96
Show file tree
Hide file tree
Showing 5 changed files with 111 additions and 36 deletions.
57 changes: 48 additions & 9 deletions atcoder-problems-backend/sql-client/src/language_count.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,6 @@ use crate::models::{Submission, UserLanguageCount, UserLanguageCountRank, UserPr
use crate::{PgPool, MAX_INSERT_ROWS};
use anyhow::Result;
use async_trait::async_trait;
use regex::Regex;
use sqlx::postgres::PgRow;
use sqlx::Row;
use std::collections::{BTreeMap, BTreeSet};
Expand Down Expand Up @@ -192,12 +191,33 @@ 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 = lang
.chars()
.take_while(|&c| !c.is_numeric() && c != '(' && c != '-')
.collect::<String>();

match simplified.len() {
0 => lang.to_string(),
_ => simplified.trim().to_string(),
}
}

Expand All @@ -207,12 +227,31 @@ 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"), "1234");
}
}
42 changes: 21 additions & 21 deletions atcoder-problems-backend/sql-client/tests/test_language_count.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,35 +11,35 @@ async fn test_language_count() {
id: 1,
problem_id: "problem1".to_owned(),
user_id: "user1".to_owned(),
language: "language1".to_owned(),
language: "language x".to_owned(),
..Default::default()
},
Submission {
id: 2,
problem_id: "problem2".to_owned(),
user_id: "user1".to_owned(),
language: "language1".to_owned(),
language: "language x".to_owned(),
..Default::default()
},
Submission {
id: 3,
problem_id: "problem1".to_owned(),
user_id: "user1".to_owned(),
language: "language1".to_owned(),
language: "language x".to_owned(),
..Default::default()
},
Submission {
id: 4,
problem_id: "problem1".to_owned(),
user_id: "user1".to_owned(),
language: "language2".to_owned(),
language: "language y".to_owned(),
..Default::default()
},
Submission {
id: 5,
problem_id: "problem1".to_owned(),
user_id: "user2".to_owned(),
language: "language1".to_owned(),
language: "language x".to_owned(),
..Default::default()
},
Submission {
Expand All @@ -65,17 +65,17 @@ async fn test_language_count() {
vec![
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 2
},
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language2".to_owned(),
simplified_language: "language y".to_owned(),
problem_count: 1
},
UserLanguageCount {
user_id: "user2".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 1
},
UserLanguageCount {
Expand Down Expand Up @@ -106,17 +106,17 @@ async fn test_language_count() {
vec![
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 2
},
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language2".to_owned(),
simplified_language: "language y".to_owned(),
problem_count: 1
},
UserLanguageCount {
user_id: "user2".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 1
},
UserLanguageCount {
Expand All @@ -133,7 +133,7 @@ async fn test_language_count() {
);

let language_count_1st_to_2nd = pool
.load_language_count_in_range("language1", 0..2)
.load_language_count_in_range("language x", 0..2)
.await
.unwrap();
assert_eq!(
Expand All @@ -151,7 +151,7 @@ async fn test_language_count() {
);

let language_count_2nd_to_2nd = pool
.load_language_count_in_range("language1", 1..2)
.load_language_count_in_range("language x", 1..2)
.await
.unwrap();
assert_eq!(
Expand All @@ -163,7 +163,7 @@ async fn test_language_count() {
);

assert_eq!(
pool.load_language_count_in_range("language1", 0..10)
pool.load_language_count_in_range("language x", 0..10)
.await
.unwrap()
.len(),
Expand All @@ -176,12 +176,12 @@ async fn test_language_count() {
vec![
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 2,
},
UserLanguageCount {
user_id: "user1".to_owned(),
simplified_language: "language2".to_owned(),
simplified_language: "language y".to_owned(),
problem_count: 1,
},
]
Expand All @@ -192,12 +192,12 @@ async fn test_language_count() {
vec![
UserLanguageCountRank {
user_id: "user1".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
rank: 1,
},
UserLanguageCountRank {
user_id: "user1".to_owned(),
simplified_language: "language2".to_owned(),
simplified_language: "language y".to_owned(),
rank: 1,
},
]
Expand All @@ -208,7 +208,7 @@ async fn test_language_count() {
language_count,
vec![UserLanguageCount {
user_id: "user2".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
problem_count: 1,
},]
);
Expand All @@ -217,7 +217,7 @@ async fn test_language_count() {
language_count_rank,
vec![UserLanguageCountRank {
user_id: "user2".to_owned(),
simplified_language: "language1".to_owned(),
simplified_language: "language x".to_owned(),
rank: 2,
},]
);
Expand Down Expand Up @@ -255,5 +255,5 @@ async fn test_language_count() {
]
);
let languages = pool.load_languages().await.unwrap();
assert_eq!(languages, ["language1", "language2", "Perl", "Raku"]);
assert_eq!(languages, ["language x", "language y", "Perl", "Raku"]);
}
4 changes: 2 additions & 2 deletions atcoder-problems-frontend/package.json
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,9 +11,9 @@
"format": "yarn run sort-package-json && yarn run prettier && yarn run lint:fix",
"lint": "eslint --ignore-path .gitignore \"./src/**/*.{js,jsx,ts,tsx}\"",
"lint:fix": "yarn run lint --fix",
"md:serve": "mdbook serve ../guide/",
"md:clean": "mdbook clean ../guide/ -d=../atcoder-problems-frontend/build/book/",
"md:build": "mkdir -p build/book && yarn run md:clean && mdbook build ../guide/ -d=../atcoder-problems-frontend/build/book/",
"md:clean": "mdbook clean ../guide/ -d=../atcoder-problems-frontend/build/book/",
"md:serve": "mdbook serve ../guide/",
"prepare-ci": "rm -rf serve && mkdir serve && cp -r build serve/atcoder",
"prettier": "prettier --write --ignore-path .gitignore \"**/*.{css,scss,html,js,json,jsx,md,ts,tsx}\"",
"sort-package-json": "sort-package-json",
Expand Down
22 changes: 22 additions & 0 deletions atcoder-problems-frontend/src/utils/LanguageNormalizer.test.ts
View file
Original file line number Diff line number Diff line change
Expand Up @@ -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");
});
22 changes: 18 additions & 4 deletions atcoder-problems-frontend/src/utils/LanguageNormalizer.ts
View file
Original file line number Diff line number Diff line change
@@ -1,7 +1,21 @@
const mapping: [string, 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";
};

0 comments on commit 163bb96

Please sign in to comment.