Improve scheme check in repo url

mypy.ini

@@ -38,6 +38,9 @@ ignore_missing_imports = True
 ; https://github.com/pypa/readme_renderer/issues/166
 ignore_missing_imports = True
 
+[mypy-rfc3986]
+ignore_missing_imports = True
+
 [mypy-setuptools]
 ; https://github.com/python/typeshed/issues/2171
 ignore_missing_imports = True

setup.cfg

@@ -42,6 +42,7 @@ install_requires=
     tqdm >= 4.14
     importlib_metadata; python_version < "3.8"
     keyring >= 15.1
+    rfc3986 >= 1.4.0
 setup_requires =
     setuptools_scm >= 1.15
 

tests/test_utils.py

@@ -184,15 +184,44 @@ def test_get_repository_config_missing(tmpdir):
     assert utils.get_repository_from_config(pypirc, "pypi") == exp
 
 
-def test_get_repository_config_invalid_url(tmpdir):
+def test_get_repository_config_invalid_scheme(tmpdir):
     """
-    Test if we get an URL without a protocol
+    Test if we get an URL with a invalid scheme
+    """
+
+    pypirc = os.path.join(str(tmpdir), ".pypirc")
+
+    with pytest.raises(
+        exceptions.UnreachableRepositoryURLDetected,
+        match=r"Invalid repository URL: "
+        r"scheme was required to be one of \['http', 'https'\] but was 'ftp'.",
+    ):
+        utils.get_repository_from_config(pypirc, "foo.bar", "ftp://test.pypi.org")
+
+
+def test_get_repository_config_missing_components(tmpdir):
+    """
+    Test if we get an URL with missing components
     """
     pypirc = os.path.join(str(tmpdir), ".pypirc")
 
-    repository_url = "foo.bar"
-    with pytest.raises(exceptions.UnreachableRepositoryURLDetected):
-        utils.get_repository_from_config(pypirc, "foo.bar", repository_url)
+    with pytest.raises(
+        exceptions.UnreachableRepositoryURLDetected,
+        match="Invalid repository URL: host was required but missing.",
+    ):
+        utils.get_repository_from_config(pypirc, "foo.bar", "https:/")
+
+    with pytest.raises(
+        exceptions.UnreachableRepositoryURLDetected,
+        match="Invalid repository URL: scheme was required but missing.",
+    ):
+        utils.get_repository_from_config(pypirc, "foo.bar", "//test.pypi.org")
+
+    with pytest.raises(
+        exceptions.UnreachableRepositoryURLDetected,
+        match="Invalid repository URL: host, scheme were required but missing.",
+    ):
+        utils.get_repository_from_config(pypirc, "foo.bar", "foo.bar")
 
 
 def test_get_repository_config_missing_config(tmpdir):

twine/utils.py

@@ -28,6 +28,7 @@
 from urllib.parse import urlunparse
 
 import requests
+import rfc3986
 
 from twine import exceptions
 
@@ -99,23 +100,38 @@ def get_config(path: str = "~/.pypirc") -> Dict[str, RepositoryConfig]:
     return dict(config)
 
 
+def _validate_repository_url(repository_url: str) -> None:
+    """Validate the given url for allowed schemes and components"""
+
+    # Allowed schemes are http and https, based on whether the repository
+    # supports TLS or not, and scheme and host must be present in the URL
+    validator = (
+        rfc3986.validators.Validator()
+        .allow_schemes("http", "https")
+        .require_presence_of("scheme", "host")
+    )
+    try:
+        validator.validate(rfc3986.uri_reference(repository_url))
+    except rfc3986.exceptions.RFC3986Exception as exc:
+        raise exceptions.UnreachableRepositoryURLDetected(
+            f"Invalid repository URL: {exc.args[0]}."
+        )
+
+
 def get_repository_from_config(
     config_file: str, repository: str, repository_url: Optional[str] = None
 ) -> RepositoryConfig:
     # Get our config from, if provided, command-line values for the
     # repository name and URL, or the .pypirc file
-    if repository_url and "://" in repository_url:
+
+    if repository_url:
+        _validate_repository_url(repository_url)
         # prefer CLI `repository_url` over `repository` or .pypirc
         return {
             "repository": repository_url,
             "username": None,
             "password": None,
         }
-    if repository_url and "://" not in repository_url:
-        raise exceptions.UnreachableRepositoryURLDetected(
-            "Repository URL {} has no protocol. Please add "
-            "'https://'. \n".format(repository_url)
-        )
     try:
         return get_config(config_file)[repository]
     except KeyError: