Add support for Markdown files with frontmatter

Author: mschoettle

src/reuse/comment.py

@@ -20,6 +20,7 @@
 # SPDX-FileCopyrightText: 2024 Anthony Loiseau <[email protected]>
 # SPDX-FileCopyrightText: 2025 Raphael Schlarb <[email protected]>
 # SPDX-FileCopyrightText: 2025 Kiko Fernandez-Reyes <[email protected]>
+# SPDX-FileCopyrightText: 2025 Matthias Schoettle <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -386,6 +387,16 @@ class FortranCommentStyle(CommentStyle):
     INDENT_AFTER_SINGLE = " "
 
 
+class FrontmatterCommentStyle(CommentStyle):
+    """Frontmatter comment style."""
+
+    SHORTHAND = "frontmatter"
+
+    SINGLE_LINE = "#"
+    INDENT_AFTER_SINGLE = " "
+    SHEBANGS = ["---"]
+
+
 class ModernFortranCommentStyle(CommentStyle):
     """Fortran (free form) comment style."""
 

src/reuse/header.py

@@ -10,6 +10,7 @@
 # SPDX-FileCopyrightText: 2022 Yaman Qalieh
 # SPDX-FileCopyrightText: 2022 Carmen Bianca Bakker <[email protected]>
 # SPDX-FileCopyrightText: 2025 Rivos Inc.
+# SPDX-FileCopyrightText: 2025 Matthias Schoettle <[email protected]>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -24,7 +25,7 @@
 from license_expression import ExpressionError
 
 from . import ReuseInfo
-from .comment import CommentStyle, EmptyCommentStyle, PythonCommentStyle
+from .comment import CommentStyle, EmptyCommentStyle, FrontmatterCommentStyle, HtmlCommentStyle, PythonCommentStyle
 from .copyright import merge_copyright_lines
 from .exceptions import (
     CommentCreateError,
@@ -279,6 +280,9 @@ def find_and_replace_header(
     if style is EmptyCommentStyle:
         after = ""
 
+    if style is HtmlCommentStyle and text.startswith(FrontmatterCommentStyle.SHEBANGS[0]):
+        style = FrontmatterCommentStyle
+
     _LOGGER.debug(f"before = {repr(before)}")
     _LOGGER.debug(f"header = {repr(header)}")
     _LOGGER.debug(f"after = {repr(after)}")

tests/test_cli_annotate.py

@@ -393,6 +393,191 @@ def test_shebang_wrong_comment_style(self, fake_repository):
         assert result.exit_code == 0
         assert simple_file.read_text() == expected
 
+    def test_xml(self, fake_repository):
+        """Keep the XML header when annotating."""
+        simple_file = fake_repository / "foo.xml"
+        simple_file.write_text(
+            cleandoc(
+                """
+                <?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+                <root>
+                </root>
+                """
+            )
+        )
+        expected = cleandoc(
+            """
+            <?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+
+            <!--
+            SPDX-License-Identifier: MIT
+            -->
+
+            <root>
+            </root>
+            """
+        )
+
+        result = CliRunner().invoke(
+            main,
+            [
+                "annotate",
+                "--license",
+                "MIT",
+                "foo.xml",
+            ],
+        )
+
+        assert result.exit_code == 0
+        assert simple_file.read_text() == expected
+
+    def test_xml_wrong_comment_style(self, fake_repository):
+        """If a comment style does not support the XML header at the top, don't
+        treat the shebang as special.
+        """
+        simple_file = fake_repository / "foo.sh"
+        simple_file.write_text(
+            cleandoc(
+                """
+                <?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+                <root>
+                </root>
+                """
+            )
+        )
+        expected = cleandoc(
+            """
+            # SPDX-License-Identifier: MIT
+
+            <?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+            <root>
+            </root>
+            """
+        )
+
+        result = CliRunner().invoke(
+            main,
+            [
+                "annotate",
+                "--license",
+                "MIT",
+                "foo.sh",
+            ],
+        )
+
+        assert result.exit_code == 0
+        assert simple_file.read_text() == expected
+
+    def test_markdown(self, fake_repository):
+        """Markdown uses the HTML comment style."""
+        simple_file = fake_repository / "foo.md"
+        simple_file.write_text(
+            cleandoc(
+                """
+                # Heading
+                """
+            )
+        )
+        expected = cleandoc(
+            """
+            <!--
+            SPDX-License-Identifier: MIT
+            -->
+
+            # Heading
+            """
+        )
+
+        result = CliRunner().invoke(
+            main,
+            [
+                "annotate",
+                "--license",
+                "MIT",
+                "foo.md",
+            ],
+        )
+
+        assert result.exit_code == 0
+        assert simple_file.read_text() == expected
+
+    def test_frontmatter(self, fake_repository):
+        """Keep the frontmatter when annotating."""
+        simple_file = fake_repository / "foo.md"
+        simple_file.write_text(
+            cleandoc(
+                """
+                ---
+                description: test
+                ---
+                # Heading
+                """
+            )
+        )
+        expected = cleandoc(
+            """
+            ---
+
+            # SPDX-License-Identifier: MIT
+
+            description: test
+            ---
+            # Heading
+            """
+        )
+
+        result = CliRunner().invoke(
+            main,
+            [
+                "annotate",
+                "--license",
+                "MIT",
+                "foo.md",
+            ],
+        )
+
+        assert result.exit_code == 0
+        assert simple_file.read_text() == expected
+
+    def test_frontmatter_wrong_comment_style(self, fake_repository):
+        """If a comment style does not support the frontmatter header at the top, don't
+        treat the shebang as special.
+        """
+        simple_file = fake_repository / "foo.sh"
+        simple_file.write_text(
+            cleandoc(
+                """
+                ---
+                description: test
+                ---
+                # Heading
+                """
+            )
+        )
+        expected = cleandoc(
+            """
+            # SPDX-License-Identifier: MIT
+
+            ---
+            description: test
+            ---
+            # Heading
+            """
+        )
+
+        result = CliRunner().invoke(
+            main,
+            [
+                "annotate",
+                "--license",
+                "MIT",
+                "foo.sh",
+            ],
+        )
+
+        assert result.exit_code == 0
+        assert simple_file.read_text() == expected
+
     def test_contributors_only(
         self, fake_repository, mock_date_today, contributors
     ):