Merge pull request #309 from sash2104/fix/list_embedded_urls

Fix broken URL detection

Author: kmyk

onlinejudge_verify/languages/special_comments.py

@@ -59,9 +59,16 @@ def list_embedded_urls(path: pathlib.Path) -> List[str]:
     urls = []
     for url in pattern.findall(content):
         # The URL may be written like `"https://atcoder.jp/"`. In this case, we need to remove `"`s around the URL.
+        # We also need to remove trailing superfluous chars in a case like `{"url":"https://atcoder.jp/"}`.
         for quote in ("'", '"', '`'):
-            if len(url) >= 2 and url.startswith(quote) and url.endswith(quote):
-                url = url[1:-1]
+            if url.startswith(quote):
+                end_quote_pos = url.rfind(quote)
+                if end_quote_pos == 0:
+                    # Remove opening quote from the URL like `"https://atcoder.jp/`
+                    url = url[1:]
+                else:
+                    # Remove quotes and trailing superfluous chars around the URL
+                    url = url[1:end_quote_pos]
                 break
         urls.append(url)
     return sorted(set(urls))

tests/test_special_comments.py

@@ -0,0 +1,45 @@
+import textwrap
+import unittest
+from typing import *
+
+import onlinejudge_verify.languages.special_comments as special_comments
+import tests.utils
+
+
+class TestSpecialComments(unittest.TestCase):
+    """Unit tests for languages/special_comments.py
+    """
+    def test_list_embedded_urls(self) -> None:
+        files = {
+            'main.cpp': textwrap.dedent("""\
+                // URL with quotes
+                #define PROBLEM "http://judge.u-aizu.ac.jp/onlinejudge/description.jsp?id=ITP1_1_A"
+                // url='https://atcoder.jp/'
+                // `https://atcoder.jp/contests/abc001`
+                //
+                // URL without quotes
+                // @see https://atcoder.jp/contests/abc002
+                //
+                // URL with quotes and extra characters
+                // {"url": "https://atcoder.jp/contests/abc003"}
+                // ('https://atcoder.jp/contests/abc004')
+                //
+                // URL with opening quote and without closing quote
+                // "https://atcoder.jp/contests/abc005
+                """).encode(),
+        }
+        expected = [
+            'http://judge.u-aizu.ac.jp/onlinejudge/description.jsp?id=ITP1_1_A',
+            'https://atcoder.jp/',
+            'https://atcoder.jp/contests/abc001',
+            'https://atcoder.jp/contests/abc002',
+            'https://atcoder.jp/contests/abc003',
+            'https://atcoder.jp/contests/abc004',
+            'https://atcoder.jp/contests/abc005',
+        ]
+
+        with tests.utils.load_files(files) as tempdir:
+            with tests.utils.chdir(tempdir):
+                file_path = tempdir / 'main.cpp'
+                actual = sorted(special_comments.list_embedded_urls(file_path))
+                self.assertEqual(actual, expected)