Add support for GitHub annotations in GitHub actions

AUTHORS.rst

@@ -150,3 +150,4 @@ Contributors
 - rajivsunar07 <[email protected]>
 - Сергій <[email protected]>
 - Mersho <[email protected]>
+- Michal Čihař <[email protected]>

changelog.d/added/format-env.md

@@ -0,0 +1,2 @@
+- Added `REUSE_OUTPUT_FORMAT` environment variable to configure output for
+  `lint`.

changelog.d/added/github-format.md

@@ -0,0 +1 @@
+- Added `--github` output option for `lint`.

docs/man/reuse-lint.rst

@@ -96,6 +96,19 @@ Options
 
   Output one line per error, prefixed by the file path.
 
+.. option:: -g, --github
+
+  Output one line per error in GitHub workflow command syntax.
+
 .. option:: -h, --help
 
   Display help and exit.
+
+Environment
+-----------
+
+.. envvar:: REUSE_OUTPUT_FORMAT
+
+   Specifies output format, one of ``plain``, ``lines``, ``github``, ``json``
+
+   It behaves same as corresponding command line options.

src/reuse/lint.py

@@ -2,21 +2,25 @@
 # SPDX-FileCopyrightText: 2022 Florian Snow <[email protected]>
 # SPDX-FileCopyrightText: 2023 DB Systel GmbH
 # SPDX-FileCopyrightText: 2024 Nico Rikken <[email protected]>
+# SPDX-FileCopyrightText: 2024 Michal Čihař <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
 """All linting happens here. The linting here is nothing more than reading
 the reports and printing some conclusions.
 """
 
+from __future__ import annotations
+
 import json
+import os
 import sys
 from argparse import ArgumentParser, Namespace
 from gettext import gettext as _
 from io import StringIO
 from pathlib import Path
 from textwrap import TextWrapper
-from typing import IO, Any, Optional
+from typing import IO, Any, Iterable, Optional
 
 from . import __REUSE_version__
 from .project import Project
@@ -44,6 +48,12 @@ def add_arguments(parser: ArgumentParser) -> None:
         action="store_true",
         help=_("formats output as errors per line"),
     )
+    mutex_group.add_argument(
+        "-g",
+        "--github",
+        action="store_true",
+        help=_("formats output as GitHub workflow commands per line"),
+    )
 
 
 # pylint: disable=too-many-branches,too-many-statements,too-many-locals
@@ -264,18 +274,19 @@ def custom_serializer(obj: Any) -> Any:
     )
 
 
-def format_lines(report: ProjectReport) -> str:
-    """Formats data dictionary as plaintext strings to be printed to sys.stdout
+def get_errors(
+    report: ProjectReport,
+) -> Iterable[tuple[Path | str | None, str]]:
+    """Returns data dictionary iterable of paths and errors.
     Sorting of output is not guaranteed.
     Symbolic links can result in multiple entries per file.
 
     Args:
         report: ProjectReport data
 
     Returns:
-        String (in plaintext) that can be output to sys.stdout
+        Iterable of tuples containing path and error message.
     """
-    output = StringIO()
 
     def license_path(lic: str) -> Optional[Path]:
         """Resolve a license identifier to a license path."""
@@ -285,55 +296,79 @@ def license_path(lic: str) -> Optional[Path]:
         # Bad licenses
         for lic, files in sorted(report.bad_licenses.items()):
             for path in sorted(files):
-                output.write(
-                    _("{path}: bad license {lic}\n").format(path=path, lic=lic)
-                )
+                yield (path, _("bad license {lic}").format(lic=lic))
 
         # Deprecated licenses
         for lic in sorted(report.deprecated_licenses):
             lic_path = license_path(lic)
-            output.write(
-                _("{lic_path}: deprecated license\n").format(lic_path=lic_path)
-            )
+            yield (lic_path, _("deprecated license"))
 
         # Licenses without extension
         for lic in sorted(report.licenses_without_extension):
             lic_path = license_path(lic)
-            output.write(
-                _("{lic_path}: license without file extension\n").format(
-                    lic_path=lic_path
-                )
-            )
+            yield (lic_path, _("license without file extension"))
 
         # Unused licenses
         for lic in sorted(report.unused_licenses):
             lic_path = license_path(lic)
-            output.write(
-                _("{lic_path}: unused license\n").format(lic_path=lic_path)
-            )
+            yield lic_path, _("unused license")
 
         # Missing licenses
         for lic, files in sorted(report.missing_licenses.items()):
             for path in sorted(files):
-                output.write(
-                    _("{path}: missing license {lic}\n").format(
-                        path=path, lic=lic
-                    )
-                )
+                yield (path, _("missing license {lic}").format(lic=lic))
 
         # Read errors
         for path in sorted(report.read_errors):
-            output.write(_("{path}: read error\n").format(path=path))
+            yield (path, _("read error"))
 
         # Without licenses
         for path in report.files_without_licenses:
-            output.write(_("{path}: no license identifier\n").format(path=path))
+            yield (path, _("no license identifier"))
 
         # Without copyright
         for path in report.files_without_copyright:
-            output.write(_("{path}: no copyright notice\n").format(path=path))
+            yield (path, _("no copyright notice"))
 
-    return output.getvalue()
+
+def format_lines(report: ProjectReport) -> str:
+    """Formats data dictionary as plaintext strings to be printed to sys.stdout
+    Sorting of output is not guaranteed.
+    Symbolic links can result in multiple entries per file.
+
+    Args:
+        report: ProjectReport data
+
+    Returns:
+        String (in plaintext) that can be output to sys.stdout
+    """
+    if not report.is_compliant:
+        return "".join(
+            f"{path}: {error}\n" for path, error in get_errors(report)
+        )
+
+    return ""
+
+
+def format_github(report: ProjectReport) -> str:
+    """Formats data dictionary as GitHub workflow commands
+    to be printed to sys.stdout
+    Sorting of output is not guaranteed.
+    Symbolic links can result in multiple entries per file.
+
+    Args:
+        report: ProjectReport data
+
+    Returns:
+        String (in plaintext) that can be output to sys.stdout
+    """
+    if not report.is_compliant:
+        return "".join(
+            f"::error file={path}::{error}\n"
+            for path, error in get_errors(report)
+        )
+
+    return ""
 
 
 def run(args: Namespace, project: Project, out: IO[str] = sys.stdout) -> int:
@@ -342,13 +377,26 @@ def run(args: Namespace, project: Project, out: IO[str] = sys.stdout) -> int:
         project, do_checksum=False, multiprocessing=not args.no_multiprocessing
     )
 
-    if args.quiet:
-        pass
-    elif args.json:
-        out.write(format_json(report))
-    elif args.lines:
-        out.write(format_lines(report))
-    else:
-        out.write(format_plain(report))
+    formatters = {
+        "json": format_json,
+        "lines": format_lines,
+        "github": format_github,
+        "plain": format_plain,
+    }
+
+    if not args.quiet:
+        output_format = os.environ.get("REUSE_OUTPUT_FORMAT")
+
+        if output_format is not None and output_format in formatters:
+            formatter = formatters[output_format]
+            out.write(formatter(report))
+        elif args.json:
+            out.write(format_json(report))
+        elif args.lines:
+            out.write(format_lines(report))
+        elif args.github:
+            out.write(format_github(report))
+        else:
+            out.write(format_plain(report))
 
     return 0 if report.is_compliant else 1

tests/test_lint.py

@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: 2019 Free Software Foundation Europe e.V. <https://fsfe.org>
 # SPDX-FileCopyrightText: 2022 Florian Snow <[email protected]>
 # SPDX-FileCopyrightText: 2024 Nico Rikken <[email protected]>
+# SPDX-FileCopyrightText: 2024 Michal Čihař <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -11,7 +12,7 @@
 
 from conftest import cpython, posix
 
-from reuse.lint import format_lines, format_plain
+from reuse.lint import format_github, format_lines, format_plain
 from reuse.project import Project
 from reuse.report import ProjectReport
 
@@ -271,4 +272,45 @@ def test_lint_lines_read_errors(fake_repository):
     assert "read error" in result
 
 
+def test_lint_github_output(fake_repository):
+    """Complete test for lint with github  output."""
+    # Prepare a repository that includes all types of situations:
+    # missing_licenses, unused_licenses, bad_licenses, deprecated_licenses,
+    # licenses_without_extension, files_without_copyright,
+    # files_without_licenses, read_errors
+    (fake_repository / "invalid-license.py").write_text(
+        "SPDX-License-Identifier: invalid"
+    )
+    (fake_repository / "no-license.py").write_text(
+        "SPDX-FileCopyrightText: Jane Doe"
+    )
+    (fake_repository / "LICENSES" / "invalid-license-text").write_text(
+        "An invalid license text"
+    )
+    (fake_repository / "LICENSES" / "Nokia-Qt-exception-1.1.txt").write_text(
+        "Deprecated"
+    )
+    (fake_repository / "LICENSES" / "MIT").write_text("foo")
+    (fake_repository / "file with spaces.py").write_text("foo")
+
+    project = Project.from_directory(fake_repository)
+    report = ProjectReport.generate(project)
+
+    lines_result = format_github(report)
+    lines_result_lines = lines_result.splitlines()
+
+    assert len(lines_result_lines) == 12
+
+    for line in lines_result_lines:
+        assert re.match("::error file=.+::[^:]+", line)
+
+    assert lines_result.count("invalid-license.py") == 3
+    assert lines_result.count("no-license.py") == 1
+    assert lines_result.count("LICENSES") == 6
+    assert lines_result.count("invalid-license-text") == 3
+    assert lines_result.count("Nokia-Qt-exception-1.1.txt") == 2
+    assert lines_result.count("MIT") == 2
+    assert lines_result.count("file with spaces.py") == 2
+
+
 # REUSE-IgnoreEnd