Fix/reuse 2.0 implementation (#17)

Co-authored-by: Roberto Pastor Muela <[email protected]>

.gitignore

@@ -157,3 +157,6 @@ cython_debug/
 #.idea/
 
 # End of https://www.toptal.com/developers/gitignore/api/python
+
+# IDE
+.vscode

doc/source/_autoapi_templates/index.rst

@@ -0,0 +1,13 @@
+API reference
+=============
+
+This page contains the ``ansys-pre-commit-hooks`` API reference.
+
+.. toctree::
+   :titlesonly:
+
+   {% for page in pages %}
+   {% if page.top_level_object and page.display %}
+   {{ page.include_path }}
+   {% endif %}
+   {% endfor %}

doc/source/conf.py

@@ -38,6 +38,8 @@
 extensions = [
     "sphinx.ext.autodoc",
     "sphinx.ext.autosummary",
+    "autoapi.extension",
+    "sphinx_autodoc_typehints",
     "numpydoc",
     "sphinx.ext.intersphinx",
     "sphinx_copybutton",
@@ -65,7 +67,7 @@
 numpydoc_validation_checks = {
     "GL06",  # Found unknown section
     "GL07",  # Sections are in the wrong order.
-    "GL08",  # The object does not have a docstring
+    # "GL08",  # The object does not have a docstring
     "GL09",  # Deprecation warning should precede extended summary
     "GL10",  # reST directives {directives} must be followed by two colons
     "SS01",  # No summary found
@@ -90,7 +92,22 @@
 # The master toctree document.
 master_doc = "index"
 
+# TODO: Once the repo goes public... remove these links
 linkcheck_ignore = [
     r"https://github.com/ansys/pre-commit-hooks/*",
     r"https://pypi.org/project/ansys-pre-commit-hooks",
 ]
+
+# Configuration for Sphinx autoapi
+autoapi_type = "python"
+autoapi_dirs = ["../../src/ansys/pre_commit_hooks/"]
+autoapi_options = [
+    "members",
+    "undoc-members",
+    "show-inheritance",
+    "show-module-summary",
+    "imported-members",
+]
+autoapi_template_dir = "_autoapi_templates"
+autoapi_python_use_implicit_namespaces = True
+exclude_patterns = ["_autoapi_templates/index.rst"]

doc/source/index.rst

@@ -4,3 +4,9 @@
    here.
 
 .. include:: ../../README.rst
+
+.. toctree::
+   :hidden:
+   :maxdepth: 3
+
+   autoapi/index

pyproject.toml

@@ -25,19 +25,23 @@ classifiers = [
 ]
 dependencies = [
     "importlib-metadata >=4.0",
-    "reuse <2",
+    "reuse>=2",
+    "GitPython>=3",
 ]
 
 [project.optional-dependencies]
 tests = [
     "pytest==7.3.0",
     "pytest-cov==4.0.0",
-    "reuse <2",
+    "reuse==2.1.0",
+    "GitPython==3.1.32",
 ]
 doc = [
     "ansys-sphinx-theme==0.10.0",
     "numpydoc==1.5.0",
     "sphinx==7.1.2",
+    "sphinx-autoapi==2.1.1",
+    "sphinx-autodoc-typehints==1.22.0",
     "sphinx-copybutton==0.5.1",
 ]
 

src/ansys/pre_commit_hooks/__init__.py

@@ -24,7 +24,7 @@
 
 try:
     import importlib.metadata as importlib_metadata
-except ModuleNotFoundError:
+except ModuleNotFoundError:  # pragma: no cover
     import importlib_metadata
 
 __version__ = importlib_metadata.version(__name__.replace(".", "-"))

src/ansys/pre_commit_hooks/add_license_headers.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python
-
 # Copyright (C) 2023 ANSYS, Inc. and/or its affiliates.
 # SPDX-License-Identifier: MIT
 #
@@ -24,98 +23,183 @@
 
 """Run reuse for files missing license headers."""
 import argparse
-import datetime
+from datetime import date as dt
+import json
 import os
-import subprocess
-import sys
 from tempfile import NamedTemporaryFile
 
-from reuse import lint
-from reuse.project import Project
-from reuse.report import ProjectReport
+import git
+from reuse import header, lint, project
 
 
-def get_args(args):
-    """Retrieve value of --loc argument."""
-    parser = argparse.ArgumentParser(description="Get repository location.")
-    parser.add_argument(
-        "--loc", type=str, required=True, help="Path to repository location", default="src"
-    )
-    return parser.parse_args(args).loc
+def set_lint_args(parser):
+    """
+    Add arguments to parser for reuse lint.
+
+    Parameters
+    ----------
+    parser: argparse.ArgumentParser
+        Parser without any arguments.
+
+    Returns
+    -------
+    argparse.Namespace
+        Parser Namespace containing lint arguments.
+    """
+    parser.add_argument("--loc", type=str, help="Path to repository location", default="src")
+    parser.add_argument("--parser")
+    parser.add_argument("--no_multiprocessing", action="store_true")
+    lint.add_arguments(parser)
+
+    return parser.parse_args()
 
 
-def get_files(loc):
+def list_noncompliant_files(args, proj):
     """
-    Generate report containing files without the license header.
+    Retrieve list of noncompliant files.
 
     Parameters
     ----------
-    loc : str
-        Path to repository location.
+    args: argparse.Namespace
+        Namespace of arguments with their values.
+    proj: project.Project
+        Project reuse runs on.
+
+    Returns
+    -------
+    list
+        List of files without license headers.
     """
-    project = Project(loc)
-    report = ProjectReport.generate(project)
-
-    with NamedTemporaryFile(mode="w") as tmp:
-        output = lint.lint_files_without_copyright_and_licensing(report, tmp)
-
-    return output
-
-
-def get_files_missing_header(loc):
-    """Retrieve files without license header."""
-    try:
-        # Run reuse lint command to find files without license header
-        return get_files(loc)
-    except NotADirectoryError as e:  # When loc is an invalid path
-        return str(e) + " for a repository"
-
-
-def run_reuse_cmd(file):
-    """Run the reuse command for files missing the license header."""
-    year = datetime.date.today().year
-    cmd = [
-        "reuse",
-        "annotate",
-        "--year",
-        rf"{year}",
-        "--copyright-style",
-        "string-c",
-        "--merge-copyright",
-        "--template=ansys",
-        "-l",
-        "MIT",
-        "-c",
-        "ANSYS, Inc. and/or its affiliates.",
-        "--skip-unrecognised",
-        rf"{file}",
-    ]
-    subprocess.check_call(cmd, stdout=subprocess.PIPE)
-
-
-def run_reuse_on_files(loc):
-    """Run the reuse annotate command on all files without license header."""
-    missing_header = get_files_missing_header(loc)
-    if missing_header:
-        # Run reuse command for files without license header
-        for file in missing_header:
-            print(f"License header added to {file}")
-            if os.path.isfile(file):
-                run_reuse_cmd(file)
-        return 1
+    # Create a temporary file containing lint.run json output
+    filename = None
+    with NamedTemporaryFile(mode="w", delete=False) as tmp:
+        args.json = True
+        lint.run(args, proj, tmp)
+        filename = tmp.name
+
+    # Open the temporary file, load the json, and find files missing copyright & licensing info.
+    lint_json = None
+    with open(filename, "rb") as file:
+        lint_json = json.load(file)
+
+    missing_headers = set(
+        lint_json["non_compliant"]["missing_copyright_info"]
+        + lint_json["non_compliant"]["missing_licensing_info"]
+    )
+
+    # Remove temporary file
+    os.remove(filename)
+
+    return missing_headers
+
+
+def find_files_missing_header():
+    """
+    Retrieve files without license header.
+
+    Returns
+    -------
+    int
+        Returns non-zero int if one or more files are noncompliant,
+        or if the .reuse directory is missing.
+    """
+    # Set up argparse for location, parser, and lint
+    # Lint contains 4 args: quiet, json, plain, and no_multiprocessing
+    parser = argparse.ArgumentParser()
+    args = set_lint_args(parser)
+    proj = project.Project(rf"{args.loc}")
+
+    missing_headers = list_noncompliant_files(args, proj)
+
+    # Get current year for license file
+    year = dt.today().year
+
+    # If there are files missing headers, run reuse and return 1
+    if missing_headers:
+        # Returns 1 if reuse changes all noncompliant files
+        # Returns 2 if .reuse directory does not exist
+        return run_reuse(parser, year, args.loc, missing_headers)
 
-    # Return zero if all files have license header
     return 0
 
 
+def check_reuse_dir():
+    """
+    Check .reuse directory exists in root of git repository.
+
+    Returns
+    -------
+    str
+        Root path of git repository.
+    int
+        If .reuse directory does not exist at root path of git repository.
+    """
+    # Get root directory of current git repository
+    git_repo = git.Repo(os.getcwd(), search_parent_directories=True)
+    git_root = git_repo.git.rev_parse("--show-toplevel")
+
+    # If .reuse folder does not exist in git repository root, return 1
+    if not os.path.isdir(os.path.join(git_root, ".reuse")):
+        print(f"Please ensure the .reuse directory is in {git_root}")
+        return 1
+
+    return git_root
+
+
+def run_reuse(parser, year, loc, missing_headers):
+    """
+    Run reuse command on files without license headers.
+
+    Parameters
+    ----------
+    parser: argparse.ArgumentParser
+        Parser containing previously set arguments.
+    year: int
+        Current year for license header.
+    loc: str
+        Location to search for files missing headers.
+    missing_headers: list
+        List of files that are missing headers.
+
+    Returns
+    -------
+    int
+        Fails pre-commit hook on return 1
+    """
+    # Add header arguments to parser which include: copyright, license, contributor, year,
+    # style, copyright-style, template, exclude-year, merge-copyrights, single-line,
+    # multi-line, explicit-license, force-dot-license, recursive, no-replace,
+    # skip-unrecognized, skip-existing
+    header.add_arguments(parser)
+
+    git_root = check_reuse_dir()
+    if git_root == 1:
+        # Previous check_reuse_dir() function returned error because .reuse
+        # directory does not exist... returning 2
+        return 2
+
+    # Add missing license header to each file in the list
+    for file in missing_headers:
+        args = parser.parse_args([rf"--loc={loc}", file])
+        args.year = [str(year)]
+        args.copyright_style = "string-c"
+        args.copyright = ["ANSYS, Inc. and/or its affiliates."]
+        args.merge_copyrights = True
+        args.template = "ansys"
+        args.license = ["MIT"]
+        args.skip_unrecognised = True
+        args.parser = parser
+
+        # Requires .reuse directory to be in git_root directory
+        proj = project.Project(git_root)
+        header.run(args, proj)
+
+    return 1
+
+
 def main():
-    """Run reuse on files without license headers."""
-    try:
-        # Get repository location argument
-        loc = get_args([sys.argv[1]])
-        return run_reuse_on_files(loc)
-    except IndexError as e:
-        print("Please provide the --loc argument. IndexError: " + str(e))
+    """Find files missing license header & run reuse on them."""
+    return find_files_missing_header()
 
 
 if __name__ == "__main__":