infer bug_tracking_url and code_view_url from vcs_url

Signed-off-by: Kaushik Kumar <kaushikrjpm10@gmail.com>

Author: Kaushik-Kumar-CEG

src/packagedcode/models.py

@@ -768,6 +768,8 @@ def from_data(cls, package_data, package_only=False):
         else:
             package_data.normalize_extracted_license_statement()
 
+        package_data.infer_vcs_urls()
+
         return package_data
 
     @property
@@ -857,6 +859,16 @@ def populate_license_fields(self):
 
         self.normalize_extracted_license_statement()
 
+    def infer_vcs_urls(self):
+        from packagedcode.utils import parse_vcs_urls
+        if not self.vcs_url:
+            return
+        code_view, bug_track = parse_vcs_urls(self.vcs_url)
+        if not self.code_view_url and code_view:
+            self.code_view_url = code_view
+        if not self.bug_tracking_url and bug_track:
+            self.bug_tracking_url = bug_track
+
     def update_purl_fields(self, package_data, replace=False):
 
         if not self.type == package_data.type:

src/packagedcode/utils.py

@@ -7,6 +7,8 @@
 # See https://aboutcode.org for more information about nexB OSS projects.
 #
 
+from urllib.parse import urlsplit
+
 from packageurl import PackageURL
 
 try:
@@ -114,6 +116,42 @@ def normalize_vcs_url(repo_url, vcs_tool=None):
     return repo_url
 
 
+# map known VCS host -> bug_tracking_url suffix to append
+BUG_TRACKING_SUFFIXES = {
+    'github.com':    '/issues',
+    'gitlab.com':    '/-/issues',
+    'codeberg.org':  '/issues',
+    'bitbucket.org': '/issues',
+}
+
+
+def parse_vcs_urls(vcs_url):
+    '''
+    given a vcs_url, return (code_view_url, bug_tracking_url) when the
+    host platform is recognized, otherwise (None, None)
+    '''
+    cleaned = normalize_vcs_url(vcs_url)
+    if not cleaned:
+        return None, None
+
+    if cleaned.endswith('.git'):
+        cleaned = cleaned[:-4]
+
+    # urlsplit does not handle git@ SSH form, e.g. git@github.com:owner/repo
+    if cleaned.startswith('git@'):
+        cleaned = 'https://' + cleaned[4:].replace(':', '/')
+
+    parsed = urlsplit(cleaned)
+    host = parsed.netloc.lower()
+    path = parsed.path.rstrip('/')
+    base = f'https://{host}{path}'
+
+    suffix = BUG_TRACKING_SUFFIXES.get(host)
+    if suffix:
+        return base, base + suffix
+    return None, None
+
+
 def build_description(summary, description):
     """
     Return a description string from a summary and description

tests/packagedcode/test_utils.py

@@ -10,6 +10,7 @@
 from unittest import TestCase
 
 from packagedcode.utils import normalize_vcs_url
+from packagedcode.utils import parse_vcs_urls
 
 
 class TestPackageUtils(TestCase):
@@ -128,3 +129,45 @@ def test_normalize_vcs_url_does_not_fail_on_empty(self):
         assert normalize_vcs_url(None) == None
         assert normalize_vcs_url('') == None
         assert normalize_vcs_url(' ') == None
+
+    # parse_vcs_urls tests
+
+    def test_parse_vcs_urls_github_git_ssh(self):
+        code, bugs = parse_vcs_urls('git@github.com:owner/repo.git')
+        assert code == 'https://github.com/owner/repo'
+        assert bugs == 'https://github.com/owner/repo/issues'
+
+    def test_parse_vcs_urls_github_https(self):
+        code, bugs = parse_vcs_urls('https://github.com/owner/repo.git')
+        assert code == 'https://github.com/owner/repo'
+        assert bugs == 'https://github.com/owner/repo/issues'
+
+    def test_parse_vcs_urls_github_git_plus_https(self):
+        code, bugs = parse_vcs_urls('git+https://github.com/owner/repo.git')
+        assert code == 'https://github.com/owner/repo'
+        assert bugs == 'https://github.com/owner/repo/issues'
+
+    def test_parse_vcs_urls_gitlab(self):
+        code, bugs = parse_vcs_urls('git+https://gitlab.com/owner/repo')
+        assert code == 'https://gitlab.com/owner/repo'
+        assert bugs == 'https://gitlab.com/owner/repo/-/issues'
+
+    def test_parse_vcs_urls_bitbucket_git_ssh(self):
+        code, bugs = parse_vcs_urls('git@bitbucket.org:owner/repo')
+        assert code == 'https://bitbucket.org/owner/repo'
+        assert bugs == 'https://bitbucket.org/owner/repo/issues'
+
+    def test_parse_vcs_urls_codeberg(self):
+        code, bugs = parse_vcs_urls('https://codeberg.org/owner/repo.git')
+        assert code == 'https://codeberg.org/owner/repo'
+        assert bugs == 'https://codeberg.org/owner/repo/issues'
+
+    def test_parse_vcs_urls_unknown_host(self):
+        code, bugs = parse_vcs_urls('https://gitea.example.com/owner/repo.git')
+        assert code is None
+        assert bugs is None
+
+    def test_parse_vcs_urls_empty(self):
+        assert parse_vcs_urls(None) == (None, None)
+        assert parse_vcs_urls('') == (None, None)
+        assert parse_vcs_urls(' ') == (None, None)