Merge pull request #654 from fsfe/lint-json

Adding `--json` to lint

CHANGELOG.md

@@ -65,6 +65,7 @@ CLI command and its behaviour. There are no guarantees of stability for the
   SPDX license expression remains the same:
   `Apache-2.0 AND CC0-1.0 AND CC-BY-SA-4.0 AND GPL-3.0-or-later`. (#733)
 - Added `--contributor` option to `annotate`. (#669)
+- Added `--json` flag to `lint` command (#654).
 
 ### Changed
 

src/reuse/__init__.py

@@ -20,7 +20,8 @@
 import os
 import re
 from dataclasses import dataclass, field
-from typing import NamedTuple, Set
+from enum import Enum, auto
+from typing import NamedTuple, Optional, Set
 
 try:
     from importlib.metadata import PackageNotFoundError, version
@@ -88,13 +89,34 @@
 _IGNORE_FILE_PATTERNS.extend(_IGNORE_SPDX_PATTERNS)
 
 
+class SourceType(Enum):
+    """
+    An enumeration representing the types of sources for license information.
+
+    Potential values:
+        DOT_LICENSE_FILE: A .license file containing license information.
+        FILE_HEADER: A file header containing license information.
+        DEP5_FILE: A .reuse/dep5 file containing license information.
+    """
+
+    DOT_LICENSE_FILE = ".license file"
+    FILE_HEADER = "file header"
+    DEP5_FILE = ".reuse/dep5 file"
+
+
 @dataclass(frozen=True)
-class SpdxInfo:
-    """Simple class holding SPDX information"""
+class ReuseInfo:
+    """Simple dataclass holding licensing and copyright information"""
 
     spdx_expressions: Set[Expression] = field(default_factory=set)
     copyright_lines: Set[str] = field(default_factory=set)
     contributor_lines: Set[str] = field(default_factory=set)
+    source_path: Optional[str] = None
+    source_type: Optional[SourceType] = None
+
+    def contains_copyright_or_licensing(self) -> bool:
+        """Either *spdx_expressions* or *copyright_lines* is non-empty."""
+        return bool(self.spdx_expressions or self.copyright_lines)
 
     def __bool__(self):
         return any(self.__dict__.values())

src/reuse/_util.py

@@ -31,7 +31,7 @@
 from debian.copyright import Copyright
 from license_expression import ExpressionError, Licensing
 
-from . import SpdxInfo
+from . import ReuseInfo
 from ._licenses import ALL_NON_DEPRECATED_MAP
 from .comment import _all_style_classes
 
@@ -203,16 +203,19 @@ def _determine_license_suffix_path(path: PathLike) -> Path:
     return Path(f"{path}.license")
 
 
-def _copyright_from_dep5(path: PathLike, dep5_copyright: Copyright) -> SpdxInfo:
+def _copyright_from_dep5(
+    path: PathLike, dep5_copyright: Copyright
+) -> ReuseInfo:
     """Find the reuse information of *path* in the dep5 Copyright object."""
     result = dep5_copyright.find_files_paragraph(Path(path).as_posix())
 
     if result is None:
-        return SpdxInfo(set(), set())
+        return ReuseInfo(set(), set(), source_path=str(path))
 
-    return SpdxInfo(
+    return ReuseInfo(
         set(map(_LICENSING.parse, [result.license.synopsis])),
         set(map(str.strip, result.copyright.splitlines())),
+        source_path=str(path),
     )
 
 
@@ -289,7 +292,7 @@ def merge_copyright_lines(copyright_lines: Set[str]) -> Set[str]:
     return copyright_out
 
 
-def extract_spdx_info(text: str) -> SpdxInfo:
+def extract_spdx_info(text: str) -> ReuseInfo:
     """Extract SPDX information from comments in a string.
 
     :raises ExpressionError: if an SPDX expression could not be parsed
@@ -316,7 +319,7 @@ def extract_spdx_info(text: str) -> SpdxInfo:
                 copyright_matches.add(match.groupdict()["copyright"].strip())
                 break
 
-    return SpdxInfo(expressions, copyright_matches)
+    return ReuseInfo(expressions, copyright_matches, "")
 
 
 def find_license_identifiers(text: str) -> Iterator[str]:

src/reuse/header.py

@@ -33,7 +33,7 @@
 from jinja2.exceptions import TemplateNotFound
 from license_expression import ExpressionError
 
-from . import SpdxInfo
+from . import ReuseInfo
 from ._util import (
     _COPYRIGHT_STYLES,
     PathType,
@@ -81,7 +81,7 @@ class MissingSpdxInfo(Exception):
 
 # TODO: Add a template here maybe.
 def _create_new_header(
-    spdx_info: SpdxInfo,
+    spdx_info: ReuseInfo,
     template: Template = None,
     template_is_commented: bool = False,
     style: CommentStyle = None,
@@ -131,7 +131,7 @@ def _create_new_header(
 
 # pylint: disable=too-many-arguments
 def create_header(
-    spdx_info: SpdxInfo,
+    spdx_info: ReuseInfo,
     header: str = None,
     template: Template = None,
     template_is_commented: bool = False,
@@ -174,10 +174,11 @@ def create_header(
             )
 
         # TODO: This behaviour does not match the docstring.
-        spdx_info = SpdxInfo(
+        spdx_info = ReuseInfo(
             spdx_info.spdx_expressions.union(existing_spdx.spdx_expressions),
             spdx_copyrights,
             spdx_info.contributor_lines.union(existing_spdx.contributor_lines),
+            "",
         )
 
     new_header += _create_new_header(
@@ -249,7 +250,7 @@ def _extract_shebang(prefix: str, text: str) -> Tuple[str, str]:
 # pylint: disable=too-many-arguments
 def find_and_replace_header(
     text: str,
-    spdx_info: SpdxInfo,
+    spdx_info: ReuseInfo,
     template: Template = None,
     template_is_commented: bool = False,
     style: CommentStyle = None,
@@ -326,7 +327,7 @@ def find_and_replace_header(
 # pylint: disable=too-many-arguments
 def add_new_header(
     text: str,
-    spdx_info: SpdxInfo,
+    spdx_info: ReuseInfo,
     template: Template = None,
     template_is_commented: bool = False,
     style: CommentStyle = None,
@@ -463,7 +464,7 @@ def _find_template(project: Project, name: str) -> Template:
 
 def _add_header_to_file(
     path: PathLike,
-    spdx_info: SpdxInfo,
+    spdx_info: ReuseInfo,
     template: Template,
     template_is_commented: bool,
     style: Optional[str],
@@ -784,7 +785,7 @@ def run(args, project: Project, out=sys.stdout) -> int:
         set(args.contributor) if args.contributor is not None else set()
     )
 
-    spdx_info = SpdxInfo(expressions, copyright_lines, contributors)
+    spdx_info = ReuseInfo(expressions, copyright_lines, contributors, "")
 
     result = 0
     for path in paths: