Add docx template variable for headers

Author: jonathansick

README.rst

@@ -136,3 +136,14 @@ This is useful for explaining that the file is autogenerated::
            args:
              - "--header"
              - "THIS FILE IS AUTOGENERATED"
+
+You can also insert the name of the source docx file using Python format string syntax and the ``docx`` template variable::
+
+   repos:
+     - repo: https://github.com/jsickcodes/pre-commit-docx-plain
+       rev: 0.2.0
+       hooks:
+         - id: docxplain
+           args:
+             - "--header"
+             - "This file is autogenerated from {docx}. Do not edit."

src/docxplain/converter.py

@@ -44,7 +44,7 @@ def convert_file(
     pypandoc.convert_file(str(docx_path), "plain", outputfile=str(plain_path))
 
     if header:
-        insert_header(plain_path, header)
+        insert_header(plain_path, header, docx_path.name)
 
     if exists:
         final_hash = get_hash(plain_path)
@@ -53,10 +53,11 @@ def convert_file(
         return True
 
 
-def insert_header(path: Path, header: str) -> None:
+def insert_header(path: Path, header: str, docx_name: str) -> None:
     """Add a header to the beginning of a plain text file."""
     content = path.read_text()
-    content = "\n\n".join((header, content))
+    context = {"docx": docx_name}
+    content = "\n\n".join((header.format(**context), content))
     path.write_text(content)
 
 

tests/converter_test.py

@@ -56,3 +56,19 @@ def test_header(tmp_path: Path) -> None:
     assert plain_path.is_file()
     content = plain_path.read_text().splitlines()
     assert content[0] == "This file is autogenerated."
+
+
+def test_header_templating(tmp_path: Path) -> None:
+    """Test the case of a templated custom file header that includes the
+    document name.
+    """
+    repo_data = Path(__file__).parent.joinpath("data/new")
+    work_dir = tmp_path / "header"
+    shutil.copytree(repo_data, work_dir)
+    docxpath = work_dir.joinpath("test_doc.docx")
+    header = "This file is autogenerated from {docx}."
+    assert convert_file(str(docxpath), header=header) is True
+    plain_path = docxpath.with_suffix(".txt")
+    assert plain_path.is_file()
+    content = plain_path.read_text().splitlines()
+    assert content[0] == "This file is autogenerated from test_doc.docx."