Extend url2purl/purl2url coverage for Git-based source hosts

src/packageurl/contrib/purl2url.py

@@ -24,6 +24,8 @@
 # Visit https://github.com/package-url/packageurl-python for support and
 # download.
 
+import re
+
 from packageurl import PackageURL
 from packageurl.contrib.route import NoRouteAvailable
 from packageurl.contrib.route import Router
@@ -51,6 +53,8 @@ def get_repo_download_url_by_package_type(
 
 repo_router = Router()
 download_router = Router()
+commit_router = Router()
+patch_router = Router()
 
 
 def _get_url_from_router(router, purl):
@@ -68,6 +72,30 @@ def get_repo_url(purl):
     return _get_url_from_router(repo_router, purl)
 
 
+def get_commit_url(purl):
+    """
+    Return a Commit URL inferred from the `purl` string.
+    """
+    commit_url = _get_url_from_router(commit_router, purl)
+    if commit_url:
+        return commit_url
+
+    purl_data = PackageURL.from_string(purl)
+    return purl_data.qualifiers.get("commit_url", None)
+
+
+def get_patch_url(purl):
+    """
+    Return a Patch URL inferred from the `purl` string.
+    """
+    patch_url = _get_url_from_router(patch_router, purl)
+    if patch_url:
+        return patch_url
+
+    purl_data = PackageURL.from_string(purl)
+    return purl_data.qualifiers.get("patch_url", None)
+
+
 def get_download_url(purl):
     """
     Return a download URL inferred from the `purl` string.
@@ -158,18 +186,203 @@ def build_github_repo_url(purl):
     return repo_url
 
 
+SUB_GITLAB_DOMAINS = [r"^git\.codelinaro\.org", r"^salsa\.debian\.org", r"^gitlab\.(?!com\b)[^/]+"]
+
+
 @repo_router.route("pkg:gitlab/.*")
 def build_gitlab_repo_url(purl):
     """
     Return a gitlab repo URL from the `purl` string.
     """
     purl_data = PackageURL.from_string(purl)
+    namespace = purl_data.namespace
+    name = purl_data.name
+    qualifiers = purl_data.qualifiers
+    repository_url = qualifiers.get("repository_url")
+    if not (namespace and name):
+        return
 
+    if repository_url:
+        clean_url = re.sub(r"^https?://", "", repository_url)
+        for pattern in SUB_GITLAB_DOMAINS:
+            if re.match(pattern, clean_url):
+                return f"https://{namespace}/{name}"
+
+    return f"https://gitlab.com/{namespace}/{name}"
+
+
+GIT_REPO_GENERIC = {
+    # cgit
+    (
+        r"git\.kernel\.org",
+        r"gitweb\.gentoo\.org",
+        "cgit\.git\.savannah\.gnu\.org",
+        "web\.git\.kernel\.org",
+    ): {
+        "repo_url": "https://{namespace}/{name}.git",
+        "commit_url": "https://{namespace}/{name}.git/commit/?id={version}",
+        "patch_url": "https://{namespace}/{name}.git/patch/?id={version}",
+    },
+    # gitiles
+    (
+        r"android\.googlesource\.com",
+        r"aomedia\.googlesource\.com",
+        r"chromium\.googlesource\.com",
+        r"gerrit\.googlesource\.com",
+    ): {
+        "repo_url": "https://{namespace}/{name}",
+        "commit_url": "https://{namespace}/{name}/+/{version}",
+        "patch_url": "https://{namespace}/{name}/+/{version}^!?format=TEXT",  # base64 encoded
+    },
+    # allura
+    (r"sourceforge\.net", r"forge-allura\.apache\.org"): {
+        "repo_url": "https://{namespace}/{name}",
+        "commit_url": "https://{namespace}/{name}/ci/{version}",
+        "patch_url": None,  # Can't find a direct patch URL
+    },
+    # gitweb
+    (
+        r"gcc\.gnu\.org/git",
+        r"git\.postgresql\.org",
+        "sourceware\.org",
+        "git\.openssl\.org",
+        "gitbox\.apache\.org",
+    ): {
+        "commit_url": "https://{namespace}/?p={name}.git;a=commit;h={version}",
+        "repo_url": "https://{namespace}/?p={name}.git",
+        "patch_url": "https://{namespace}/?p={name}.git;a=patch;h={version}",
+    },
+    # gitea / forgejo
+    (
+        r"codeberg\.org",
+        r"gitea\.com",
+        r"forge\.fedoraproject\.org",
+    ): {
+        "commit_url": "https://{namespace}/{name}/commit/{version}",
+        "repo_url": "https://{namespace}/{name}",
+        "patch_url": "https://{namespace}/{name}/commit/{version}.patch",
+    },
+}
+
+
+@repo_router.route("pkg:generic/.*")
+def build_generic_repo_url(purl):
+    """
+    Return a Repo URL from the `purl` string.
+    """
+    purl_data = PackageURL.from_string(purl)
+    name = purl_data.name
     namespace = purl_data.namespace
+
+    if not (namespace and name):
+        return
+
+    for patterns, template_url in GIT_REPO_GENERIC.items():
+        for pattern in patterns:
+            if not re.match(pattern, namespace):
+                continue
+
+            return template_url["repo_url"].format(namespace=namespace, name=name)
+    return
+
+
+@commit_router.route("pkg:generic/.*")
+def build_generic_commit_url(purl):
+    """
+    Return a Commit URL from the `purl` string.
+    """
+    purl_data = PackageURL.from_string(purl)
     name = purl_data.name
+    namespace = purl_data.namespace
+    version = purl_data.version
 
-    if name and namespace:
-        return f"https://gitlab.com/{namespace}/{name}"
+    if not (namespace and name and version):
+        return
+
+    for patterns, template_url in GIT_REPO_GENERIC.items():
+        for pattern in patterns:
+            if not re.match(pattern, namespace):
+                continue
+
+            return template_url["commit_url"].format(
+                namespace=namespace, name=name, version=version
+            )
+
+
+@patch_router.route("pkg:generic/.*")
+def build_generic_patch_url(purl):
+    """
+    Return a Patch URL from the `purl` string.
+    """
+    purl_data = PackageURL.from_string(purl)
+    name = purl_data.name
+    namespace = purl_data.namespace
+    version = purl_data.version
+
+    if not (namespace and name and version):
+        return
+
+    for patterns, template_url in GIT_REPO_GENERIC.items():
+        for pattern in patterns:
+            if not re.match(pattern, namespace):
+                continue
+
+            patch_template = template_url.get("patch_url")
+            if patch_template:
+                return patch_template.format(namespace=namespace, name=name, version=version)
+
+
+@commit_router.route("pkg:gitlab/.*", "pkg:bitbucket/.*", "pkg:github/.*")
+def build_main_commit_url(purl):
+    """
+    Return a github/gitlab/bitbucket Commit URL from the `purl` string.
+    """
+    purl_data = PackageURL.from_string(purl)
+    purl_type = purl_data.type
+    name = purl_data.name
+    namespace = purl_data.namespace
+    version = purl_data.version
+    if not (namespace and name and version):
+        return
+
+    commit_url_template = {
+        "github": f"https://github.com/{namespace}/{name}/commit/{version}",
+        "gitlab": f"https://gitlab.com/{namespace}/{name}/-/commit/{version}",
+        "sub-gitlab": f"https://{namespace}/{name}/-/commit/{version}",
+        "bitbucket": f"https://bitbucket.org/{namespace}/{name}/commits/{version}",
+    }
+
+    if purl_type == "gitlab" and purl_data.qualifiers.get("repository_url"):
+        purl_type = "sub-gitlab"
+
+    return commit_url_template[purl_type].format(namespace=namespace, name=name, version=version)
+
+
+@patch_router.route("pkg:gitlab/.*", "pkg:bitbucket/.*", "pkg:github/.*")
+def build_main_patch_url(purl):
+    """
+    Return a github/gitlab/bitbucket Patch URL from the `purl` string.
+    """
+    purl_data = PackageURL.from_string(purl)
+    purl_type = purl_data.type
+    name = purl_data.name
+    namespace = purl_data.namespace
+    version = purl_data.version
+
+    if not (namespace and name and version):
+        return
+
+    patch_url_templates = {
+        "github": f"https://github.com/{namespace}/{name}/commit/{version}.patch",
+        "gitlab": f"https://gitlab.com/{namespace}/{name}/-/commit/{version}.patch",
+        "sub-gitlab": f"https://{namespace}/{name}/-/commit/{version}.patch",
+        "bitbucket": f"https://bitbucket.org/{namespace}/{name}/commits/{version}/raw",
+    }
+
+    if purl_type == "gitlab" and purl_data.qualifiers.get("repository_url"):
+        purl_type = "sub-gitlab"
+
+    return patch_url_templates[purl_type].format(namespace=namespace, name=name, version=version)
 
 
 @repo_router.route("pkg:(gem|rubygems)/.*")