normalize source vcs urls (#701)

This change ensures when referencing source url for package
specifications of source type "git" or for VCS dependencies, a
normalized version returned at all times.

Relates-to: python-poetry/poetry#8999

src/poetry/core/packages/specification.py

@@ -35,7 +35,7 @@ def __init__(
         self._pretty_name = name
         self._name = canonicalize_name(name)
         self._source_type = source_type
-        self._source_url = source_url
+        self._source_url = self._normalize_source_url(source_type, source_url)
         self._source_reference = source_reference
         self._source_resolved_reference = source_resolved_reference
         self._source_subdirectory = source_subdirectory
@@ -45,6 +45,17 @@ def __init__(
 
         self._features = frozenset(canonicalize_name(feature) for feature in features)
 
+    @staticmethod
+    def _normalize_source_url(
+        source_type: str | None, source_url: str | None
+    ) -> str | None:
+        if source_type and source_url and source_type == "git":
+            from poetry.core.vcs.git import ParsedUrl
+
+            return ParsedUrl.parse(source_url).url
+
+        return source_url
+
     @property
     def name(self) -> NormalizedName:
         return self._name

src/poetry/core/packages/vcs_dependency.py

@@ -32,7 +32,6 @@ def __init__(
         # Attributes must be immutable for clone() to be safe!
         # (For performance reasons, clone only creates a copy instead of a deep copy).
         self._vcs = vcs
-        self._source = source
 
         self._branch = branch
         self._tag = tag
@@ -47,13 +46,15 @@ def __init__(
             optional=optional,
             allows_prereleases=True,
             source_type=self._vcs.lower(),
-            source_url=self._source,
+            source_url=source,
             source_reference=branch or tag or rev or "HEAD",
             source_resolved_reference=resolved_rev,
             source_subdirectory=directory,
             extras=extras,
         )
 
+        self._source = self.source_url or source
+
     @property
     def vcs(self) -> str:
         return self._vcs

src/poetry/core/vcs/git.py

@@ -119,7 +119,7 @@ def parse(cls, url: str) -> ParsedUrl:
             if m:
                 groups = m.groupdict()
                 return ParsedUrl(
-                    groups.get("protocol"),
+                    groups.get("protocol", "ssh"),
                     groups.get("resource"),
                     groups.get("pathname"),
                     groups.get("user"),

tests/packages/test_specification.py

@@ -115,3 +115,41 @@ def test_equal_specifications_have_same_hash(
     assert spec1 == spec2
     assert spec2 == spec1
     assert hash(spec1) == hash(spec2)
+
+
+@pytest.mark.parametrize(
+    "source_url,normalized_url",
+    [
+        ("https://github.com/demo/demo.git", "https://github.com/demo/demo.git"),
+        ("git@github.com:demo/demo.git", "ssh://git@github.com/demo/demo.git"),
+    ],
+)
+def test_specification_normalize_source_url_method(
+    source_url: str, normalized_url: str
+) -> None:
+    assert (
+        PackageSpecification._normalize_source_url("git", source_url) == normalized_url
+    )
+    assert (
+        PackageSpecification._normalize_source_url("notgit", source_url) == source_url
+    )
+
+
+@pytest.mark.parametrize(
+    "source_url,normalized_url",
+    [
+        ("https://github.com/demo/demo.git", "https://github.com/demo/demo.git"),
+        ("git@github.com:demo/demo.git", "ssh://git@github.com/demo/demo.git"),
+    ],
+)
+def test_specification_uses_normalize_source_url_for_git(
+    source_url: str, normalized_url: str
+) -> None:
+    assert (
+        PackageSpecification(
+            name="demo",
+            source_type="git",
+            source_url=source_url,
+        ).source_url
+        == normalized_url
+    )

tests/packages/test_vcs_dependency.py

@@ -229,3 +229,21 @@ def test_vcs_dependencies_are_equal_if_resolved_references_match() -> None:
     )
 
     assert dependency1 == dependency2
+
+
+@pytest.mark.parametrize(
+    "source_url,normalized_url",
+    [
+        ("https://github.com/demo/demo.git", "https://github.com/demo/demo.git"),
+        ("git@github.com:demo/demo.git", "ssh://git@github.com/demo/demo.git"),
+    ],
+)
+def test_vcs_source_is_normalized(source_url: str, normalized_url: str) -> None:
+    dependency = VCSDependency(
+        name="demo",
+        vcs="git",
+        source=source_url,
+        branch="main",
+    )
+    assert dependency.source == normalized_url
+    assert dependency.source_url == normalized_url

tests/vcs/test_vcs.py

@@ -258,7 +258,7 @@ def test_normalize_url(url: str, normalized: GitUrl) -> None:
         ),
         (
             "git@github.com:org/repo",
-            ParsedUrl(None, "github.com", ":org/repo", "git", None, "repo", None),
+            ParsedUrl("ssh", "github.com", ":org/repo", "git", None, "repo", None),
         ),
         (
             "git+https://github.com/sdispater/pendulum",