Merge pull request #787 from carmenbianca/info-correct-source

Make sure the sources are correct when information is aggregated

CHANGELOG.md

@@ -73,6 +73,8 @@ CLI command and its behaviour. There are no guarantees of stability for the
 - Added `--contributor` option to `annotate`. (#669)
 - Added `--json` flag to `lint` command (#654).
 - `reuse.ReuseInfo` now has `copy` and `union` methods. (#759)
+- `reuse.ReuseInfo` now stores information about the source from which the
+  information was gathered. (#654, #787)
 - Added Ukrainian and Czech translations (#767)
 - Added `--suppress-deprecation` to hide (verbose) deprecation warnings. (#778)
 
@@ -81,12 +83,16 @@ CLI command and its behaviour. There are no guarantees of stability for the
 - Bumped SPDX license list to v3.20. (#692)
 - `reuse.SpdxInfo` was renamed to `reuse.ReuseInfo`. It is now a (frozen)
   dataclass instead of a namedtuple. This is only relevant if you're using reuse
-  as a library in Python. (#669)
+  as a library in Python. Other functions and methods were similarly renamed.
+  (#669)
 - Sphinx documentation: Switched from RTD theme to Furo. (#673, #716)
 - Removed dependency on setuptools' `pkg_resources` to determine the installed
   version of reuse. (#724)
 - Bumped SPDX license list to v3.21. (#763)
 - Bumped REUSE Spec version to 3.1. (#768)
+- `Project.reuse_info_of` now returns a list of `ReuseInfo` objects instead of a
+  single one. This is because the source information is now stored alongside the
+  REUSE information. (#787)
 
 ### Deprecated
 

src/reuse/__init__.py

@@ -1,5 +1,6 @@
 # SPDX-FileCopyrightText: 2017 Free Software Foundation Europe e.V. <https://fsfe.org>
 # SPDX-FileCopyrightText: 2021 Alliander N.V.
+# SPDX-FileCopyrightText: 2023 Carmen Bianca BAKKER <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -91,11 +92,11 @@ class SourceType(Enum):
     """
 
     #: A .license file containing license information.
-    DOT_LICENSE_FILE = ".license file"
+    DOT_LICENSE = "dot-license"
     #: A file header containing license information.
-    FILE_HEADER = "file header"
+    FILE_HEADER = "file-header"
     #: A .reuse/dep5 file containing license information.
-    DEP5_FILE = ".reuse/dep5 file"
+    DEP5 = "dep5"
 
 
 # TODO: In Python 3.10+, add kw_only=True
@@ -106,6 +107,7 @@ class ReuseInfo:
     spdx_expressions: Set[Expression] = field(default_factory=set)
     copyright_lines: Set[str] = field(default_factory=set)
     contributor_lines: Set[str] = field(default_factory=set)
+    path: Optional[str] = None
     source_path: Optional[str] = None
     source_type: Optional[SourceType] = None
 
@@ -153,6 +155,17 @@ 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 contains_info(self) -> bool:
+        """Any field except *path*, *source_path* and *source_type* is
+        non-empty.
+        """
+        keys = {
+            key
+            for key in self.__dict__
+            if key not in ("path", "source_path", "source_type")
+        }
+        return any(self.__dict__[key] for key in keys)
+
     def __bool__(self) -> bool:
         return any(self.__dict__.values())
 

src/reuse/_util.py

@@ -26,7 +26,7 @@
 from hashlib import sha1
 from itertools import chain
 from os import PathLike
-from pathlib import Path
+from pathlib import Path, PurePath
 from typing import IO, Any, BinaryIO, Dict, Iterator, List, Optional, Set, Union
 
 from boolean.boolean import Expression, ParseError
@@ -225,7 +225,8 @@ def _determine_license_suffix_path(path: StrPath) -> Path:
 
 def _copyright_from_dep5(path: StrPath, 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())
+    path = PurePath(path).as_posix()
+    result = dep5_copyright.find_files_paragraph(path)
 
     if result is None:
         return ReuseInfo()
@@ -237,7 +238,9 @@ def _copyright_from_dep5(path: StrPath, dep5_copyright: Copyright) -> ReuseInfo:
         copyright_lines=set(
             map(str.strip, result.copyright.splitlines())  # type: ignore
         ),
-        source_type=SourceType.DEP5_FILE,
+        path=path,
+        source_type=SourceType.DEP5,
+        source_path=".reuse/dep5",
     )
 
 

src/reuse/project.py

@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: 2017 Free Software Foundation Europe e.V. <https://fsfe.org>
 # SPDX-FileCopyrightText: 2022 Florian Snow <[email protected]>
 # SPDX-FileCopyrightText: 2023 DB Systel GmbH
+# SPDX-FileCopyrightText: 2023 Carmen Bianca BAKKER <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -13,7 +14,7 @@
 import warnings
 from gettext import gettext as _
 from pathlib import Path
-from typing import Dict, Iterator, Optional, Union, cast
+from typing import Dict, Iterator, List, Optional, Union, cast
 
 from boolean.boolean import ParseError
 from debian.copyright import Copyright
@@ -148,28 +149,30 @@ def all_files(self, directory: Optional[StrPath] = None) -> Iterator[Path]:
                 _LOGGER.debug("yielding '%s'", the_file)
                 yield the_file
 
-    def reuse_info_of(self, path: StrPath) -> ReuseInfo:
+    def reuse_info_of(self, path: StrPath) -> List[ReuseInfo]:
         """Return REUSE info of *path*.
 
-        This function will return any REUSE information that it can find, both
-        from within the file, the .license file and from the .reuse/dep5 file.
+        This function will return any REUSE information that it can find: from
+        within the file, the .license file and/or from the .reuse/dep5 file.
 
-        It also returns a single primary source path of the license/copyright
-        information, where 'primary' means '.license file' > 'header' > 'dep5'
+        The presence of a .license file always means that the file itself will
+        not be parsed for REUSE information.
+
+        When the .reuse/dep5 file covers a file and there is also REUSE
+        information within that file (or within its .license file), then two
+        :class:`ReuseInfo` objects are returned in the set, each with respective
+        discovered REUSE information and information about the source.
         """
         original_path = path
         path = _determine_license_path(path)
-        dep5_path: Optional[str] = None
-        source_path = ""
-        source_type = None
 
         _LOGGER.debug(f"searching '{path}' for REUSE information")
 
         # This means that only one 'source' of licensing/copyright information
         # is captured in ReuseInfo
         dep5_result = ReuseInfo()
         file_result = ReuseInfo()
-        final_result = ReuseInfo()
+        result = []
 
         # Search the .reuse/dep5 file for REUSE information.
         if self._copyright:
@@ -180,8 +183,6 @@ def reuse_info_of(self, path: StrPath) -> ReuseInfo:
                 _LOGGER.info(
                     _("'{path}' covered by .reuse/dep5").format(path=path)
                 )
-                source_path = str(self.root / ".reuse/dep5")
-                dep5_path = source_path
 
         # Search the file for REUSE information.
         with path.open("rb") as fp:
@@ -199,12 +200,16 @@ def reuse_info_of(self, path: StrPath) -> ReuseInfo:
                 file_result = extract_reuse_info(
                     decoded_text_from_binary(fp, size=read_limit)
                 )
-                if file_result:
-                    source_path = str(path)
+                if file_result.contains_copyright_or_licensing():
                     if path.suffix == ".license":
-                        source_type = SourceType.DOT_LICENSE_FILE
+                        source_type = SourceType.DOT_LICENSE
                     else:
                         source_type = SourceType.FILE_HEADER
+                    file_result = file_result.copy(
+                        path=self.relative_from_root(original_path).as_posix(),
+                        source_path=self.relative_from_root(path).as_posix(),
+                        source_type=source_type,
+                    )
 
             except (ExpressionError, ParseError):
                 _LOGGER.error(
@@ -215,11 +220,7 @@ def reuse_info_of(self, path: StrPath) -> ReuseInfo:
                 )
 
         # There is both information in a .dep5 file and in the file header
-        if (
-            dep5_result.contains_copyright_or_licensing()
-            and file_result.contains_copyright_or_licensing()
-        ):
-            final_result = file_result.union(dep5_result)
+        if dep5_result.contains_info() and file_result.contains_info():
             warnings.warn(
                 _(
                     "Copyright and licensing information for"
@@ -233,22 +234,17 @@ def reuse_info_of(self, path: StrPath) -> ReuseInfo:
                     " need do nothing yet. Run with"
                     " `--suppress-deprecation` to hide this warning."
                 ).format(
-                    original_path=original_path, path=path, dep5_path=dep5_path
+                    original_path=original_path,
+                    path=path,
+                    dep5_path=dep5_result.source_path,
                 ),
                 PendingDeprecationWarning,
             )
-        # Information is only found in a DEP5 file
-        elif (
-            dep5_result.contains_copyright_or_licensing()
-            and not file_result.contains_copyright_or_licensing()
-        ):
-            final_result = dep5_result.copy(source_path=source_path)
-        # There is a file header or a .license file
-        else:
-            final_result = file_result.copy(
-                source_path=source_path, source_type=source_type
-            )
-        return final_result
+        if dep5_result.contains_info():
+            result.append(dep5_result)
+        if file_result.contains_info():
+            result.append(file_result)
+        return result
 
     def relative_from_root(self, path: StrPath) -> Path:
         """If the project root is /tmp/project, and *path* is