feat: caching

scrapers/codeforces.py

@@ -60,51 +60,135 @@ def parse_problem_url(contest_id: str, problem_letter: str) -> str:
     )
 
 
+def scrape_contest_problems(contest_id: str):
+    try:
+        contest_url = f"https://codeforces.com/contest/{contest_id}"
+        scraper = cloudscraper.create_scraper()
+        response = scraper.get(contest_url, timeout=10)
+        response.raise_for_status()
+
+        soup = BeautifulSoup(response.text, "html.parser")
+        problems = []
+
+        problem_links = soup.find_all("a", href=lambda x: x and f"/contest/{contest_id}/problem/" in x)
+
+        for link in problem_links:
+            href = link.get("href", "")
+            if f"/contest/{contest_id}/problem/" in href:
+                problem_letter = href.split("/")[-1].lower()
+                problem_name = link.get_text(strip=True)
+
+                if problem_letter and problem_name and len(problem_letter) == 1:
+                    problems.append({
+                        "id": problem_letter,
+                        "name": problem_name
+                    })
+
+        problems.sort(key=lambda x: x["id"])
+
+        seen = set()
+        unique_problems = []
+        for p in problems:
+            if p["id"] not in seen:
+                seen.add(p["id"])
+                unique_problems.append(p)
+
+        return unique_problems
+
+    except Exception as e:
+        print(f"Failed to scrape contest problems: {e}", file=sys.stderr)
+        return []
+
+
 def scrape_sample_tests(url: str):
     print(f"Scraping: {url}", file=sys.stderr)
     return scrape(url)
 
 
 def main():
-    if len(sys.argv) != 3:
+    if len(sys.argv) < 2:
         result = {
             "success": False,
-            "error": "Usage: codeforces.py <contest_id> <problem_letter>",
-            "problem_id": None,
+            "error": "Usage: codeforces.py metadata <contest_id> OR codeforces.py tests <contest_id> <problem_letter>",
         }
         print(json.dumps(result))
         sys.exit(1)
 
-    contest_id = sys.argv[1]
-    problem_letter = sys.argv[2]
-    problem_id = contest_id + problem_letter.lower()
+    mode = sys.argv[1]
 
-    url = parse_problem_url(contest_id, problem_letter)
-    tests = scrape_sample_tests(url)
+    if mode == "metadata":
+        if len(sys.argv) != 3:
+            result = {
+                "success": False,
+                "error": "Usage: codeforces.py metadata <contest_id>",
+            }
+            print(json.dumps(result))
+            sys.exit(1)
+
+        contest_id = sys.argv[2]
+        problems = scrape_contest_problems(contest_id)
+
+        if not problems:
+            result = {
+                "success": False,
+                "error": f"No problems found for contest {contest_id}",
+            }
+            print(json.dumps(result))
+            sys.exit(1)
 
-    if not tests:
         result = {
-            "success": False,
-            "error": f"No tests found for {contest_id} {problem_letter}",
+            "success": True,
+            "contest_id": contest_id,
+            "problems": problems,
+        }
+        print(json.dumps(result))
+
+    elif mode == "tests":
+        if len(sys.argv) != 4:
+            result = {
+                "success": False,
+                "error": "Usage: codeforces.py tests <contest_id> <problem_letter>",
+            }
+            print(json.dumps(result))
+            sys.exit(1)
+
+        contest_id = sys.argv[2]
+        problem_letter = sys.argv[3]
+        problem_id = contest_id + problem_letter.lower()
+
+        url = parse_problem_url(contest_id, problem_letter)
+        tests = scrape_sample_tests(url)
+
+        if not tests:
+            result = {
+                "success": False,
+                "error": f"No tests found for {contest_id} {problem_letter}",
+                "problem_id": problem_id,
+                "url": url,
+            }
+            print(json.dumps(result))
+            sys.exit(1)
+
+        test_cases = []
+        for input_data, output_data in tests:
+            test_cases.append({"input": input_data, "output": output_data})
+
+        result = {
+            "success": True,
             "problem_id": problem_id,
             "url": url,
+            "test_cases": test_cases,
+        }
+        print(json.dumps(result))
+
+    else:
+        result = {
+            "success": False,
+            "error": f"Unknown mode: {mode}. Use 'metadata' or 'tests'",
         }
         print(json.dumps(result))
         sys.exit(1)
 
-    test_cases = []
-    for input_data, output_data in tests:
-        test_cases.append({"input": input_data, "output": output_data})
-
-    result = {
-        "success": True,
-        "problem_id": problem_id,
-        "url": url,
-        "test_cases": test_cases,
-    }
-
-    print(json.dumps(result))
-
 
 if __name__ == "__main__":
     main()