infer bug_tracking_url and code_view_url from vcs_url

Signed-off-by: Kaushik <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, version=self.version)
+        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 +116,62 @@ def normalize_vcs_url(repo_url, vcs_tool=None):
     return repo_url
 
 
+BUG_TRACKING_SUFFIXES = {
+    'github.com':    '/issues',
+    'gitlab.com':    '/-/issues',
+    'codeberg.org':  '/issues',
+    'bitbucket.org': '/issues',
+}
+
+# each platform has a different path scheme for browsing a tagged tree
+CODE_VIEW_SUFFIXES = {
+    'github.com':    '/tree/{version}',
+    'gitlab.com':    '/-/tree/{version}',
+    'codeberg.org':  '/src/tag/{version}',
+    'bitbucket.org': '/src/{version}',
+}
+
+
+def parse_vcs_urls(vcs_url, version=None):
+    """
+    Given a ``vcs_url`` and an optional ``version``, return a
+    (code_view_url, bug_tracking_url) tuple for recognized hosting platforms,
+    or (None, None) otherwise.
+
+    ``code_view_url`` points to the source tree at a specific version tag and
+    is only populated when a concrete ``version`` is provided.
+    """
+    cleaned = normalize_vcs_url(vcs_url)
+    if not cleaned:
+        return None, None
+
+    if cleaned.endswith('.git'):
+        cleaned = cleaned[:-4]
+
+    # urlsplit can't parse git@ SSH form like 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}'
+
+    bug_suffix = BUG_TRACKING_SUFFIXES.get(host)
+    if not bug_suffix:
+        return None, None
+
+    bug_tracking_url = base + bug_suffix
+
+    code_view_url = None
+    if version:
+        code_suffix = CODE_VIEW_SUFFIXES.get(host)
+        if code_suffix:
+            code_view_url = base + code_suffix.format(version=version)
+
+    return code_view_url, bug_tracking_url
+
+
 def build_description(summary, description):
     """
     Return a description string from a summary and description

tests/packagedcode/data/cargo/cargo_toml/clap/Cargo.toml.expected

@@ -33,8 +33,8 @@
     "md5": null,
     "sha256": null,
     "sha512": null,
-    "bug_tracking_url": null,
-    "code_view_url": null,
+    "bug_tracking_url": "https://github.com/clap-rs/clap/issues",
+    "code_view_url": "https://github.com/clap-rs/clap/tree/2.32.0",
     "vcs_url": "https://github.com/clap-rs/clap",
     "copyright": null,
     "holder": null,

tests/packagedcode/data/cargo/cargo_toml/clippy/Cargo.toml.expected

@@ -60,8 +60,8 @@
     "md5": null,
     "sha256": null,
     "sha512": null,
-    "bug_tracking_url": null,
-    "code_view_url": null,
+    "bug_tracking_url": "https://github.com/rust-lang/rust-clippy/issues",
+    "code_view_url": "https://github.com/rust-lang/rust-clippy/tree/0.0.212",
     "vcs_url": "https://github.com/rust-lang/rust-clippy",
     "copyright": null,
     "holder": null,

tests/packagedcode/data/cargo/cargo_toml/mdbook/Cargo.toml.expected

@@ -45,8 +45,8 @@
     "md5": null,
     "sha256": null,
     "sha512": null,
-    "bug_tracking_url": null,
-    "code_view_url": null,
+    "bug_tracking_url": "https://github.com/rust-lang-nursery/mdBook/issues",
+    "code_view_url": "https://github.com/rust-lang-nursery/mdBook/tree/0.2.4-alpha.0",
     "vcs_url": "https://github.com/rust-lang-nursery/mdBook",
     "copyright": null,
     "holder": null,

tests/packagedcode/data/cargo/cargo_toml/rustfmt/Cargo.toml.expected

@@ -35,8 +35,8 @@
     "md5": null,
     "sha256": null,
     "sha512": null,
-    "bug_tracking_url": null,
-    "code_view_url": null,
+    "bug_tracking_url": "https://github.com/rust-lang/rustfmt/issues",
+    "code_view_url": "https://github.com/rust-lang/rustfmt/tree/1.0.3",
     "vcs_url": "https://github.com/rust-lang/rustfmt",
     "copyright": null,
     "holder": null,

tests/packagedcode/data/cargo/cargo_toml/rustup/Cargo.toml.expected

@@ -31,8 +31,8 @@
     "md5": null,
     "sha256": null,
     "sha512": null,
-    "bug_tracking_url": null,
-    "code_view_url": null,
+    "bug_tracking_url": "https://github.com/rust-lang/rustup.rs/issues",
+    "code_view_url": "https://github.com/rust-lang/rustup.rs/tree/1.17.0",
     "vcs_url": "https://github.com/rust-lang/rustup.rs",
     "copyright": null,
     "holder": null,

tests/packagedcode/data/cargo/cargo_toml/single-file-scan/Cargo.toml.expected

@@ -37,8 +37,8 @@
           "md5": null,
           "sha256": null,
           "sha512": null,
-          "bug_tracking_url": null,
-          "code_view_url": null,
+          "bug_tracking_url": "https://github.com/cesarb/constant_time_eq/issues",
+          "code_view_url": "https://github.com/cesarb/constant_time_eq/tree/0.4.2",
           "vcs_url": "https://github.com/cesarb/constant_time_eq",
           "copyright": null,
           "holder": null,
@@ -123,4 +123,4 @@
       "scan_errors": []
     }
   ]
-}
+}

tests/packagedcode/data/cargo/scan-package-only.expected.json

@@ -327,8 +327,8 @@
           "md5": null,
           "sha256": null,
           "sha512": null,
-          "bug_tracking_url": null,
-          "code_view_url": null,
+          "bug_tracking_url": "https://github.com/indygreg/PyOxidizer/issues",
+          "code_view_url": "https://github.com/indygreg/PyOxidizer/tree/0.4.0-pre",
           "vcs_url": "https://github.com/indygreg/PyOxidizer.git",
           "copyright": null,
           "holder": null,
@@ -669,8 +669,8 @@
           "md5": null,
           "sha256": null,
           "sha512": null,
-          "bug_tracking_url": null,
-          "code_view_url": null,
+          "bug_tracking_url": "https://github.com/daac-tools/daachorse/issues",
+          "code_view_url": "https://github.com/daac-tools/daachorse/tree/0.4.1",
           "vcs_url": "https://github.com/daac-tools/daachorse",
           "copyright": null,
           "holder": null,

tests/packagedcode/data/cargo/scan.expected.json

@@ -30,8 +30,8 @@
       "md5": null,
       "sha256": null,
       "sha512": null,
-      "bug_tracking_url": null,
-      "code_view_url": null,
+      "bug_tracking_url": "https://github.com/indygreg/PyOxidizer/issues",
+      "code_view_url": "https://github.com/indygreg/PyOxidizer/tree/0.4.0-pre",
       "vcs_url": "https://github.com/indygreg/PyOxidizer.git",
       "copyright": null,
       "holder": null,
@@ -128,8 +128,8 @@
       "md5": null,
       "sha256": null,
       "sha512": null,
-      "bug_tracking_url": null,
-      "code_view_url": null,
+      "bug_tracking_url": "https://github.com/daac-tools/daachorse/issues",
+      "code_view_url": "https://github.com/daac-tools/daachorse/tree/0.4.1",
       "vcs_url": "https://github.com/daac-tools/daachorse",
       "copyright": null,
       "holder": null,
@@ -1173,8 +1173,8 @@
           "md5": null,
           "sha256": null,
           "sha512": null,
-          "bug_tracking_url": null,
-          "code_view_url": null,
+          "bug_tracking_url": "https://github.com/indygreg/PyOxidizer/issues",
+          "code_view_url": "https://github.com/indygreg/PyOxidizer/tree/0.4.0-pre",
           "vcs_url": "https://github.com/indygreg/PyOxidizer.git",
           "copyright": null,
           "holder": null,
@@ -1540,8 +1540,8 @@
           "md5": null,
           "sha256": null,
           "sha512": null,
-          "bug_tracking_url": null,
-          "code_view_url": null,
+          "bug_tracking_url": "https://github.com/daac-tools/daachorse/issues",
+          "code_view_url": "https://github.com/daac-tools/daachorse/tree/0.4.1",
           "vcs_url": "https://github.com/daac-tools/daachorse",
           "copyright": null,
           "holder": null,