Automatically detect LICENSE files and mark them with %license macro

README.md

@@ -184,11 +184,10 @@ For example, if a package provides the modules `requests` and `_requests`, write
     %pyproject_save_files requests _requests
 
 To add listed files to the `%files` section, use `%files -f %{pyproject_files}`.
-Note that you still need to add any documentation and license manually (for now).
+Note that you still need to add any documentation manually (for now).
 
     %files -n python3-requests -f %{pyproject_files}
     %doc README.rst
-    %license LICENSE
 
 You can use globs in the module names if listing them explicitly would be too tedious:
 
@@ -214,10 +213,12 @@ However, in Fedora packages, always list executables explicitly to avoid uninten
     
     %files -n python3-requests -f %{pyproject_files}
     %doc README.rst
-    %license LICENSE
     %{_bindir}/downloader
 
-`%pyproject_save_files` also automatically recognizes language (`*.mo`) files and marks them with `%lang` macro and appropriate language code.
+`%pyproject_save_files` can automatically mark license files with `%license` macro
+and  language (`*.mo`) files with `%lang` macro and appropriate language code.
+Only license files declared via [PEP 639] `License-Field` field are detected.
+[PEP 639] is still a draft and can be changed in the future.
 
 Note that `%pyproject_save_files` uses data from the [RECORD file](https://www.python.org/dev/peps/pep-0627/).
 If you wish to rename, remove or otherwise change the installed files of a package
@@ -303,6 +304,7 @@ so be prepared for problems.
 [PEP 508]: https://www.python.org/dev/peps/pep-0508/
 [PEP 517]: https://www.python.org/dev/peps/pep-0517/
 [PEP 518]: https://www.python.org/dev/peps/pep-0518/
+[PEP 639]: https://www.python.org/dev/peps/pep-0639/
 [pip's documentation]: https://pip.pypa.io/en/stable/cli/pip_install/#vcs-support
 
 

pyproject-rpm-macros.spec

@@ -117,6 +117,7 @@ export HOSTNAME="rpmbuild"  # to speedup tox in network-less mock, see rhbz#1856
 - %%pyproject_buildrequires now fails when it encounters an invalid requirement
 - Fixes: rhbz#1983053
 - Rename %%_pyproject_ghost_distinfo and %%_pyproject_record to indicate they are private
+- Automatically detect LICENSE files and mark them with %%license macro
 
 * Fri Jul 23 2021 Fedora Release Engineering <releng@fedoraproject.org> - 0-45
 - Rebuilt for https://fedoraproject.org/wiki/Fedora_35_Mass_Rebuild

pyproject_save_files.py

@@ -5,6 +5,7 @@ import os
 
 from collections import defaultdict
 from pathlib import PosixPath, PurePosixPath
+from importlib.metadata import Distribution
 
 
 # From RPM's build/files.c strtokWithQuotes delim argument
@@ -144,7 +145,7 @@ def add_lang_to_module(paths, module_name, path):
 
 
 def classify_paths(
-    record_path, parsed_record_content, sitedirs, python_version
+    record_path, parsed_record_content, metadata, sitedirs, python_version
 ):
     """
     For each BuildrootPath in parsed_record_content classify it to a dict structure
@@ -160,7 +161,7 @@ def classify_paths(
             "files": [],  # regular %file entries with dist-info content
             "dirs": [distinfo],  # %dir %file entries with dist-info directory
             "docs": [],  # to be used once there is upstream way to recognize READMEs
-            "licenses": [],  # to be used once there is upstream way to recognize LICENSEs
+            "licenses": [],  # %license entries parsed from dist-info METADATA file
         },
         "lang": {}, # %lang entries: [module_name or None][language_code] lists of .mo files
         "modules": defaultdict(list),  # each importable module (directory, .py, .so)
@@ -170,6 +171,7 @@ def classify_paths(
     # In RECORDs generated by pip, there are no directories, only files.
     # The example RECORD from PEP 376 does not contain directories either.
     # Hence, we'll only assume files, but TODO get it officially documented.
+    license_files = metadata.get_all('License-File')
     for path in parsed_record_content:
         if path.suffix == ".pyc":
             # we handle bytecode separately
@@ -180,8 +182,10 @@ def classify_paths(
                 # RECORD and REQUESTED files are removed in %pyproject_install
                 # See PEP 627
                 continue
-            # TODO is this a license/documentation?
+            if license_files and path.name in license_files:
-            paths["metadata"]["files"].append(path)
+                paths["metadata"]["licenses"].append(path)
+            else:
+                paths["metadata"]["files"].append(path)
             continue
 
         for sitedir in sitedirs:
@@ -423,6 +427,14 @@ def load_parsed_record(pyproject_record):
     return parsed_record
 
 
+def dist_metadata(buildroot, record_path):
+    """
+    Returns distribution metadata (email.message.EmailMessage), possibly empty
+    """
+    real_dist_path = record_path.parent.to_real(buildroot)
+    dist = Distribution.at(real_dist_path)
+    return dist.metadata
+
 def pyproject_save_files(buildroot, sitelib, sitearch, python_version, pyproject_record, varargs):
     """
     Takes arguments from the %{pyproject_save_files} macro
@@ -439,8 +451,9 @@ def pyproject_save_files(buildroot, sitelib, sitearch, python_version, pyproject
     final_file_list = []
 
     for record_path, files in parsed_records.items():
+        metadata = dist_metadata(buildroot, record_path)
         paths_dict = classify_paths(
-            record_path, files, sitedirs, python_version
+            record_path, files, metadata, sitedirs, python_version
         )
 
         final_file_list.extend(

pyproject_save_files_test_data.yaml

@@ -50,11 +50,11 @@ classified:
       docs: []
       files:
       - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/INSTALLER
-      - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/LICENSE
       - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/METADATA
       - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/WHEEL
       - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/top_level.txt
-      licenses: []
+      licenses:
+      - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/LICENSE
     modules:
       requests:
       - files:
@@ -415,13 +415,13 @@ classified:
       files:
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/AUTHORS
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/INSTALLER
-      - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE
-      - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE.python
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/METADATA
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/WHEEL
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/entry_points.txt
       - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/top_level.txt
-      licenses: []
+      licenses:
+      - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE
+      - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE.python
     lang:
       django:
         af:
@@ -7434,8 +7434,8 @@ dumped:
   - - '%dir /usr/lib/python3.7/site-packages/requests'
     - '%dir /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info'
     - '%dir /usr/lib/python3.7/site-packages/requests/__pycache__'
+    - '%license /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/LICENSE'
     - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/INSTALLER
-    - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/LICENSE
     - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/METADATA
     - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/WHEEL
     - /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/top_level.txt
@@ -11252,12 +11252,12 @@ dumped:
     - '%lang(zh) /usr/lib/python3.7/site-packages/django/contrib/sessions/locale/zh_Hant/LC_MESSAGES/django.mo'
     - '%lang(zh) /usr/lib/python3.7/site-packages/django/contrib/sites/locale/zh_Hans/LC_MESSAGES/django.mo'
     - '%lang(zh) /usr/lib/python3.7/site-packages/django/contrib/sites/locale/zh_Hant/LC_MESSAGES/django.mo'
+    - '%license /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE'
+    - '%license /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE.python'
     - /usr/bin/django-admin
     - /usr/bin/django-admin.py
     - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/AUTHORS
     - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/INSTALLER
-    - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE
-    - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/LICENSE.python
     - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/METADATA
     - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/WHEEL
     - /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/entry_points.txt
@@ -14487,6 +14487,23 @@ dumped:
     - /usr/share/pronterface/zoom_in.png
     - /usr/share/pronterface/zoom_out.png
 
+metadata:
+  requests:
+    path: /usr/lib/python3.7/site-packages/requests-2.22.0.dist-info/METADATA
+    content: |
+      Name: requests
+      Version: 2.22.0
+      License-File: LICENSE
+      Whatever: False data
+  django:
+    path: /usr/lib/python3.7/site-packages/Django-3.0.7.dist-info/METADATA
+    content: |
+      Name: Django
+      Version: 3.0.7
+      License-File: LICENSE
+      License-File: LICENSE.python
+      Whatever: False data
+
 records:
   kerberos:
     path: /usr/lib64/python3.7/site-packages/kerberos-1.3.0.dist-info/RECORD

test_pyproject_save_files.py

@@ -20,6 +20,7 @@ yaml_data = yaml.safe_load(yaml_file.read_text())
 EXPECTED_DICT = yaml_data["classified"]
 EXPECTED_FILES = yaml_data["dumped"]
 TEST_RECORDS = yaml_data["records"]
+TEST_METADATAS = yaml_data["metadata"]
 
 
 @pytest.fixture
@@ -47,6 +48,10 @@ def prepare_pyproject_record(tmp_path, package=None, content=None):
         # Get test data and write dist-info/RECORD file
         record_path = BuildrootPath(TEST_RECORDS[package]["path"])
         record_file.write_text(TEST_RECORDS[package]["content"])
+        if package in TEST_METADATAS:
+            metadata_path = BuildrootPath(TEST_METADATAS[package]["path"]).to_real(tmp_path)
+            metadata_path.parent.mkdir(parents=True, exist_ok=True)
+            metadata_path.write_text(TEST_METADATAS[package]["content"])
         # Parse RECORD file
         parsed_record = parse_record(record_path, read_record(record_file))
         # Save JSON content to pyproject-record

tests/python-setuptools.spec

@@ -67,8 +67,12 @@ rm pyproject.toml
 # We only run a subset of tests to speed things up and be less fragile
 PYTHONPATH=$(pwd) %pytest --ignore=pavement.py -k "sdist"
 
+# Internal check that license file was recognized correctly
+grep '^%%license' %{pyproject_files} > tested.license
+echo '%%license %{python3_sitelib}/setuptools-%{version}.dist-info/LICENSE' > expected.license
+diff tested.license expected.license
+
 
 %files -n python3-setuptools -f %{pyproject_files}
-%license LICENSE
 %doc docs/* CHANGES.rst README.rst
 %{python3_sitelib}/distutils-precedence.pth