Corriger et améliorer la recherches des versions des paquets NPM

Arsène Volte · fwininger · commit 0800cd8ebea2 · 2025-10-07T21:10:27.000+02:00
diff --git a/hashtheplanet/resources/npm_resource.py b/hashtheplanet/resources/npm_resource.py
@@ -6,11 +6,11 @@
 import tarfile
 import tempfile
 from typing import Dict, List, Set, Tuple
-import requests
+import json
 
 # third party imports
 from loguru import logger
-from bs4 import BeautifulSoup
+import requests
 
 # project imports
 from hashtheplanet.sql.db_connector import Hash
@@ -34,22 +34,8 @@ def retrieve_versions(npm_module_name: str) -> Set[VersionName]:
         """
         This method retrieves npm module versions.
         """
-        page = requests.get(f"https://www.npmjs.com/package/{npm_module_name}?activeTab=versions", timeout=10)
-        soup = BeautifulSoup(page.content, "html.parser")
-        versions = set()
-
-        table = soup.find('div', attrs={'id':'tabpanel-versions'})
-
-        table_body_version_elements = list(table.find_all())[0]
-
-        version_rows = table_body_version_elements.find_all("li")
-
-        for row in list(version_rows):
-            version = row.find("a")
-            if version is None:
-                continue
-            versions.add(version.get_text())
-        return versions
+        page = requests.get(f"https://registry.npmjs.org/{npm_module_name}", timeout=10)
+        return set(json.loads(page.content)["versions"])
 
     @staticmethod
     def save_tar_to_disk(file_path: str, npm_module_name: str, version: str):
diff --git a/tests/resources/test_npm_resource.py b/tests/resources/test_npm_resource.py
@@ -19,27 +19,12 @@ def test_retrieve_versions():
     class MockedPage():
         def __init__(self) -> None:
             self.content = """
-            <html>
-            <head><title></title></head>
-            <body>
-                <div id="tabpanel-versions">
-                    <div>
-                        <div></div>
-                        <h3></h3>
-                        <ul><li></li></ul>
-                        <ul>
-                            <li></li>
-                            <li><a>1.0.1</a></li>
-                            <li><a>1.0.2</a></li>
-                            <li><a>1.0.3</a></li>
-                        </ul>
-                    </div>
-                    <div></div>
-                </div>
-            </body></html>
+            {
+              "versions": ["1.0.1", "1.0.2", "1.0.3"]
+            }
             """
     def mocked_get_request(url: str, *args, **kwargs):
-        assert url == f"https://www.npmjs.com/package/{npm_module_name}?activeTab=versions"
+        assert url == f"https://registry.npmjs.org/{npm_module_name}"
         return MockedPage()
 
     with mock.patch("requests.get", MagicMock(side_effect=mocked_get_request)):
