Add support for generating commit and patch URLs for all supported Git repositories.

Treat GitLab subdomains as GitLab when using the repository URL instead of a generic type.

Add support for salsa.debian.org, gitlab.eclipse.org, forge.fedoraproject.org domains

Signed-off-by: ziad hany <ziadhany2016@gmail.com>

Author: ziadhany

src/packageurl/contrib/purl2url.py

@@ -53,6 +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):
@@ -70,6 +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.
@@ -160,18 +186,29 @@ 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 name and namespace:
-        return f"https://gitlab.com/{namespace}/{name}"
+    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 = {
@@ -182,8 +219,9 @@ def build_gitlab_repo_url(purl):
         "cgit\.git\.savannah\.gnu\.org",
         "web\.git\.kernel\.org",
     ): {
-        "commit_url": "https://{namespace}/{name}.git/commit/?id={version}",
         "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
     (
@@ -192,13 +230,15 @@ def build_gitlab_repo_url(purl):
         r"chromium\.googlesource\.com",
         r"gerrit\.googlesource\.com",
     ): {
-        "commit_url": "https://{namespace}/{name}/+/{version}",
         "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"): {
-        "commit_url": "https://{namespace}/{name}/ci/{version}",
         "repo_url": "https://{namespace}/{name}",
+        "commit_url": "https://{namespace}/{name}/ci/{version}",
+        "patch_url": "https://{namespace}/{name}/ci/{version}",  # FIXME
     },
     # gitweb
     (
@@ -210,35 +250,29 @@ def build_gitlab_repo_url(purl):
     ): {
         "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}",
-    },
-    # sub gitlab ( excludes gitlab.com )
-    (
-        r"git\.codelinaro\.org.*",
-        r"gitlab\.(?!com\b)[^/]+",
-    ): {
-        "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 Commit URL from the `purl` string.
+    Return a Repo 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):
         return
@@ -248,14 +282,109 @@ def build_generic_repo_url(purl):
             if not re.match(pattern, namespace):
                 continue
 
-            if version:
-                return template_url["commit_url"].format(
-                    namespace=namespace, name=name, version=version
-                )
             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 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
+            )
+    return
+
+
+@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
+
+            return template_url["patch_url"].format(namespace=namespace, name=name, version=version)
+    return
+
+
+@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)/.*")
 def build_rubygems_repo_url(purl):
     """

src/packageurl/contrib/url2purl.py

@@ -675,7 +675,7 @@ def build_route_regex(domain_patterns, path_suffix="/.*"):
     return rf"https?://({domain_pattern}){path_suffix}"
 
 
-SUB_GITLAB_DOMAINS = [r"git\.codelinaro\.org", r"gitlab\.(?!com\b)[^/]+"]
+SUB_GITLAB_DOMAINS = [r"git\.codelinaro\.org", r"salsa\.debian\.org", r"gitlab\.(?!com\b)[^/]+"]
 SUB_GITLAB_ROUTE_REGEX = build_route_regex(SUB_GITLAB_DOMAINS)
 
 
@@ -686,29 +686,46 @@ def build_gitlab_sub_purl(url):
     For example:
     https://gitlab.gnome.org/GNOME/gimp
     https://git.codelinaro.org/clo/qsdk/oss/kernel/linux-msm
+    https://gitlab.eclipse.org/eclipse/asciidoc-lang/asciidoc-lang
     https://gitlab.gnome.org/GNOME/gimp/-/commit/112a5e038f0646eae5ae314988ec074433d2b365
     https://git.codelinaro.org/linaro/qcom/project/-/commit/a40a9732c840e5a324fba78b0ff7980b497c3831
     """
+    archive_pattern = (
+        r"^https?://" r"(?P<namespace>[^/]+)/" r".+?/-/archive/[^/]+/" r"(?P<name>[^/]+)$"
+    )
 
-    gitlab_sub_commit_pattern = (
+    commit_pattern = (
         r"^https?://"
         r"(?P<namespace>.+?)/"
         r"(?P<name>[^/]+)"
         r"(?:/-/commit/(?P<version>[0-9a-fA-F]{7,64}))?"
         r"/?$"
     )
 
-    commit_match = re.search(gitlab_sub_commit_pattern, url)
+    archive_match = re.search(archive_pattern, url)
+    if archive_match:
+        namespace = archive_match.group("namespace")
+        name = archive_match.group("name")
+        return PackageURL(
+            type="gitlab", namespace=namespace, name=name, qualifiers={"download_url": url}
+        )
+
+    commit_match = re.search(commit_pattern, url)
     if commit_match:
+        namespace = commit_match.group("namespace")
+        name = commit_match.group("name")
+        version = commit_match.group("version")
+
         return PackageURL(
-            type="generic",
-            namespace=commit_match.group("namespace"),
-            name=commit_match.group("name"),
-            version=commit_match.group("version"),
+            type="gitlab",
+            namespace=namespace,
+            name=name,
+            version=version,
+            qualifiers={"repository_url": f"https://{namespace}/{name}"},
         )
 
 
-GITEA_DOMAINS = ["codeberg\.org", "gitea\.com"]
+GITEA_DOMAINS = [r"codeberg\.org", r"gitea\.com", r"forge\.fedoraproject\.org"]
 GITEA_ROUTE_REGEX = build_route_regex(GITEA_DOMAINS)
 
 
@@ -722,7 +739,6 @@ def build_gitea_purl(url):
     https://codeberg.org/alpinelinux/aports/commit/a40a9732c840e5a324fba78b0ff7980b497c3831
     https://gitea.com/htc47/entur/commit/271b852cfb761a1fe257aa0f0a12ff38bd8bfd1c
     """
-
     gitea_commit_pattern = (
         r"^https?://"
         r"(?P<namespace>.+?)/"
@@ -771,7 +787,7 @@ def build_cgit_purl(url):
     """
 
     # https://git.kernel.org/stable/c/<hash>
-    kernel_shorthand = r"^https?://git\.kernel\.org/stable/c/" r"(?P<version>[0-9a-fA-F]{7,64})/?$"
+    kernel_shorthand = r"^https?://git\.kernel\.org/stable/c/(?P<version>[0-9a-fA-F]{7,64})/?$"
 
     cgit_project_pattern = (
         r"^https?://"