Metadata hook: enhance authors and keywords

README.md

@@ -38,6 +38,16 @@ dynamic = ["authors", "classifiers", "keywords", "license", "version", "urls"]
 additional-keywords = ["awesome"] # some additional keywords
 kind = "scraper" # indicate this is a scraper, so that additional keywords are added
 
+# Additional author #1
+[[tool.hatch.metadata.hooks.openzim-metadata.additional-authors]]
+name="Bob"
+email="[email protected]"
+
+# Additional author #2
+[[tool.hatch.metadata.hooks.openzim-metadata.additional-authors]]
+name="Alice"
+email="[email protected]"
+
 # Enable the hatch-openzim build hook to install files (e.g. JS libs) at build time.
 [tool.hatch.build.hooks.openzim-build]
 toml-config = "openzim.toml" # optional location of the configuration file
@@ -52,13 +62,15 @@ NOTA: the `dependencies` attribute is not specific to our hook(s), it is a gener
 
 | Variable | Required | Description |
 |---|---|---|
+| `additional-authors` | N | List of authors that will be appended to the automatic one |
+| `additional-keywords` | N | List of keywords that will be appended to the automatic ones |
+| `kind` | N | If set to `scraper`, scrapers keywords will be automatically added as well |
+| `organization` | N | Override organization (otherwise detected from Github repository to set author and keyword appropriately). Case-insentive. Supported values are `openzim`, `kiwix` and `offspot` |
 | `preserve-authors` | N | Boolean indicating that we do not want to set `authors` metadata but use the ones of `pyproject.toml` |
 | `preserve-classifiers` | N | Boolean indicating that we do not want to set `classifiers` metadata but use the ones of `pyproject.toml` |
 | `preserve-keywords` | N | Boolean indicating that we do not want to set `keywords` metadata but use the ones of `pyproject.toml` |
 | `preserve-license` | N | Boolean indicating that we do not want to set `license` metadata but use the one of `pyproject.toml` |
 | `preserve-urls` | N | Boolean indicating that we do not want to set `urls` metadata but use the ones of `pyproject.toml` |
-| `additional_keywords` | N | List of keywords that will be appended to the automatically added ones |
-| `kind` | N | If set to `scraper`, scrapers keywords will be automatically added as well |
 
 ### Behavior
 

src/hatch_openzim/files_install.py

@@ -112,7 +112,7 @@ def _process_execute_after(base_target_dir: Path, actions: List[str]):
         logger.info(f"    Executing '{action}'")
         process = subprocess.run(
             action,
-            shell=True,  # noqa: S602
+            shell=True,  # noqa: S602 # nosec: B602
             cwd=base_target_dir,
             text=True,
             check=True,

src/hatch_openzim/metadata.py

@@ -1,6 +1,6 @@
 from pathlib import Path
 
-from hatch_openzim.utils import get_github_project_homepage, get_python_versions
+from hatch_openzim.utils import get_github_info, get_python_versions
 
 
 def update(root: str, config: dict, metadata: dict):
@@ -23,19 +23,29 @@ def update(root: str, config: dict, metadata: dict):
                 "openzim metadata hook."
             )
 
+    github_info = get_github_info(git_config_path=Path(root) / ".git/config")
+
+    organization = config.get("organization", github_info.organization)
+
     if not config.get("preserve-urls", False):
         metadata["urls"] = {
             "Donate": "https://www.kiwix.org/en/support-us/",
-            "Homepage": get_github_project_homepage(
-                git_config_path=Path(root) / ".git/config"
-            ),
+            "Homepage": github_info.homepage,
         }
 
     if not config.get("preserve-authors", False):
-        metadata["authors"] = [{"name": "Kiwix", "email": "[email protected]"}]
+        if str(organization).lower() in ("kiwix", "offspot"):
+            authors = [{"name": "Kiwix", "email": "[email protected]"}]
+        else:
+            authors = [{"name": "openZIM", "email": "[email protected]"}]
+        authors.extend(config.get("additional-authors", []))
+        metadata["authors"] = authors
 
     if not config.get("preserve-keywords", False):
-        keywords = ["kiwix"]
+        if str(organization).lower() in ("kiwix", "offspot"):
+            keywords = ["kiwix"]
+        else:
+            keywords = ["openzim"]
         if config.get("kind", "") == "scraper":
             keywords.extend(["zim", "offline"])
         keywords.extend(config.get("additional-keywords", []))

src/hatch_openzim/shared.py

@@ -1,6 +1,16 @@
 import logging
 import os
+import sys
 
+DEFAULT_FORMAT = "%(name)s:%(levelname)s:%(message)s"
+
+# create logger
 logger = logging.getLogger("hatch_openzim")
+logger.setLevel(logging.DEBUG)
+
+# setup console logging
+console_handler = logging.StreamHandler(sys.stdout)
+console_handler.setFormatter(logging.Formatter(DEFAULT_FORMAT))
 log_level = logging.getLevelName(os.getenv("HATCH_OPENZIM_LOG_LEVEL", "INFO"))
-logger.setLevel(log_level)
+console_handler.setLevel(log_level)
+logger.addHandler(console_handler)

src/hatch_openzim/utils.py

@@ -1,5 +1,6 @@
 import configparser
 import re
+from collections import namedtuple
 from pathlib import Path
 from typing import List
 
@@ -13,12 +14,16 @@
     r"""(?P<repository>.*?)(?:.git)?$"""
 )
 
-DEFAULT_GITHUB_PROJECT_HOMEPAGE = "https://www.kiwix.org"
+GithubInfo = namedtuple("GithubInfo", ["homepage", "organization", "repository"])
+
+DEFAULT_GITHUB_INFO = GithubInfo(
+    homepage="https://www.kiwix.org", organization=None, repository=None
+)
 
 
-def get_github_project_homepage(git_config_path: Path, remote: str = "origin") -> str:
+def get_github_info(git_config_path: Path, remote: str = "origin") -> GithubInfo:
     if not git_config_path.exists() or not git_config_path.is_file():
-        return DEFAULT_GITHUB_PROJECT_HOMEPAGE
+        return DEFAULT_GITHUB_INFO
 
     try:
         config = configparser.ConfigParser()
@@ -27,13 +32,15 @@ def get_github_project_homepage(git_config_path: Path, remote: str = "origin") -
         match = REMOTE_REGEXP.match(git_remote_url)
         if not match:
             raise Exception(f"Unexpected remote url: {git_remote_url}")
-        return (
-            f"https://github.com/{match.group('organization')}/"
-            f"{match.group('repository')}"
+        return GithubInfo(
+            homepage=f"https://github.com/{match.group('organization')}/"
+            f"{match.group('repository')}",
+            organization=match.group("organization"),
+            repository=match.group("repository"),
         )
     except Exception as exc:
         logger.error("Failed to read Github URL", exc_info=exc)
-        return DEFAULT_GITHUB_PROJECT_HOMEPAGE
+        return DEFAULT_GITHUB_INFO
 
 
 def get_python_versions(requires_python: str) -> List[str]:

tests/test_metadata.py

@@ -32,14 +32,14 @@ def test_metadata_nominal(metadata):
         metadata=metadata,
     )
 
-    assert metadata["authors"] == [{"email": "dev@kiwix.org", "name": "Kiwix"}]
+    assert metadata["authors"] == [{"email": "dev@openzim.org", "name": "openZIM"}]
     assert metadata["classifiers"] == [
         "License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)",
         "Programming Language :: Python :: 3",
         "Programming Language :: Python :: 3.10",
         "Programming Language :: Python :: 3.11",
     ]
-    assert metadata["keywords"] == ["kiwix"]
+    assert metadata["keywords"] == ["openzim"]
     assert metadata["license"] == {"text": "GPL-3.0-or-later"}
     assert metadata["urls"] == {
         "Donate": "https://www.kiwix.org/en/support-us/",
@@ -126,7 +126,53 @@ def test_metadata_additional_keywords(metadata):
         metadata=metadata,
     )
     # we compare sets because order is not relevant
-    assert set(metadata["keywords"]) == {"kiwix", "keyword1", "keyword2"}
+    assert set(metadata["keywords"]) == {"openzim", "keyword1", "keyword2"}
+
+
+def test_metadata_additional_authors(metadata):
+    config = {}
+    config["additional-authors"] = [{"email": "[email protected]", "name": "Some One"}]
+    update(
+        root=str(Path(os.path.dirname(os.path.abspath(__file__))).parent),
+        config=config,
+        metadata=metadata,
+    )
+    # we compare sets because order is not relevant
+    assert metadata["authors"] == [
+        {"email": "[email protected]", "name": "openZIM"},
+        {"email": "[email protected]", "name": "Some One"},
+    ]
+
+
+@pytest.mark.parametrize(
+    "organization, expected_result",
+    [
+        ("kiwix", "kiwix"),
+        ("Kiwix", "kiwix"),
+        ("openzim", "openzim"),
+        ("openZIM", "openzim"),
+        ("offspot", "kiwix"),
+        ("unknown", "openzim"),
+        (None, "openzim"),
+    ],
+)
+def test_metadata_organization(organization, expected_result, metadata):
+    config = {}
+    if organization:
+        config["organization"] = organization
+    update(
+        root=str(Path(os.path.dirname(os.path.abspath(__file__))).parent),
+        config=config,
+        metadata=metadata,
+    )
+    if expected_result == "kiwix":
+        assert metadata["authors"] == [{"email": "[email protected]", "name": "Kiwix"}]
+        assert metadata["keywords"] == ["kiwix"]
+    elif expected_result == "openzim":
+        assert metadata["authors"] == [{"email": "[email protected]", "name": "openZIM"}]
+        assert metadata["keywords"] == ["openzim"]
+    else:
+        raise Exception(f"Unexpected expected result: {expected_result}")
 
 
 def test_metadata_is_scraper(metadata):
@@ -138,4 +184,4 @@ def test_metadata_is_scraper(metadata):
         metadata=metadata,
     )
     # we compare sets because order is not relevant
-    assert set(metadata["keywords"]) == {"kiwix", "offline", "zim"}
+    assert set(metadata["keywords"]) == {"openzim", "offline", "zim"}

tests/test_utils.py

@@ -5,7 +5,7 @@
 
 import pytest
 
-from hatch_openzim.utils import get_github_project_homepage, get_python_versions
+from hatch_openzim.utils import GithubInfo, get_github_info, get_python_versions
 
 
 @pytest.fixture
@@ -32,46 +32,55 @@ def _mock_git_config(git_origin_url: str, remote_name: str = "origin"):
 
 
 @pytest.mark.parametrize(
-    "git_url, expected_homepage_url",
+    "git_url, expected_homepage_url, expected_organization, expected_repository",
     [
         (
             "https://github.com/oneuser/onerepo.git",
             "https://github.com/oneuser/onerepo",
+            "oneuser",
+            "onerepo",
         ),
         (
             "https://github.com/oneuser/onerepo",
             "https://github.com/oneuser/onerepo",
+            "oneuser",
+            "onerepo",
         ),
         (
             "[email protected]:oneuser/one-repo.git",
             "https://github.com/oneuser/one-repo",
+            "oneuser",
+            "one-repo",
         ),
     ],
 )
 def test_get_github_project_homepage_valid_url(
-    mock_git_config, git_url, expected_homepage_url
+    mock_git_config,
+    git_url,
+    expected_homepage_url,
+    expected_organization,
+    expected_repository,
 ):
     with mock_git_config(git_url) as git_config_path:
-        assert (
-            get_github_project_homepage(git_config_path=git_config_path)
-            == expected_homepage_url
+        assert get_github_info(git_config_path=git_config_path) == GithubInfo(
+            homepage=expected_homepage_url,
+            organization=expected_organization,
+            repository=expected_repository,
         )
 
 
 def test_get_github_project_homepage_invalid_url(mock_git_config):
     # Test the function with an invalid URL
     with mock_git_config("http://github.com/oneuser/onerepo.git") as git_config_path:
-        assert (
-            get_github_project_homepage(git_config_path=git_config_path)
-            == "https://www.kiwix.org"
+        assert get_github_info(git_config_path=git_config_path) == GithubInfo(
+            homepage="https://www.kiwix.org", organization=None, repository=None
         )
 
 
 def test_get_github_project_missing_git_config():
     # Test the function with an invalid URL
-    assert (
-        get_github_project_homepage(git_config_path=Path("i_m_not_here.config"))
-        == "https://www.kiwix.org"
+    assert get_github_info(git_config_path=Path("i_m_not_here.config")) == GithubInfo(
+        homepage="https://www.kiwix.org", organization=None, repository=None
     )
 
 
@@ -80,9 +89,8 @@ def test_get_github_project_homepage_invalid_remote(mock_git_config):
     with mock_git_config(
         "https://github.com/oneuser/onerepo.git", remote_name="origin2"
     ) as git_config_path:
-        assert (
-            get_github_project_homepage(git_config_path=git_config_path)
-            == "https://www.kiwix.org"
+        assert get_github_info(git_config_path=git_config_path) == GithubInfo(
+            homepage="https://www.kiwix.org", organization=None, repository=None
         )