Merge pull request #9 from jsickcodes/trim-trailing-whitespace

Trim trailing whitespace

Author: jonathansick

.pre-commit-config.yaml

@@ -6,6 +6,7 @@ repos:
       - id: check-toml
       - id: check-json
       - id: trailing-whitespace
+        exclude: tests/data/trailing-whitespace/input\.txt
 
   - repo: https://github.com/Lucas-C/pre-commit-hooks-markup
     rev: v1.0.1

CHANGELOG.rst

@@ -1,6 +1,11 @@
 Change log
 ==========
 
+Unreleased
+----------
+
+- Trim trailing whitespace from the output.
+
 0.2.0 (2021-03-23)
 ------------------
 

src/docxplain/converter.py

@@ -46,6 +46,8 @@ def convert_file(
     if header:
         insert_header(plain_path, header, docx_path.name)
 
+    trim_trailing_whitespace(plain_path)
+
     if exists:
         final_hash = get_hash(plain_path)
         return final_hash != initial_hash
@@ -61,6 +63,16 @@ def insert_header(path: Path, header: str, docx_name: str) -> None:
     path.write_text(content)
 
 
+def trim_trailing_whitespace(path: Path) -> None:
+    """Trim trailing whitespace from the plain text file, updating it
+    in place.
+    """
+    content = path.read_text()
+    formatted_lines = [line.rstrip() for line in content.splitlines()]
+    new_content = "\n".join(formatted_lines) + "\n"
+    path.write_text(new_content)
+
+
 def get_hash(path: Path) -> str:
     """Get the SHA256 hash digest of a file."""
     m = hashlib.sha256()

tests/converter_test.py

@@ -3,7 +3,7 @@
 import shutil
 from pathlib import Path
 
-from docxplain.converter import convert_file
+from docxplain.converter import convert_file, trim_trailing_whitespace
 
 
 def test_unchanged(tmp_path: Path) -> None:
@@ -72,3 +72,21 @@ def test_header_templating(tmp_path: Path) -> None:
     assert plain_path.is_file()
     content = plain_path.read_text().splitlines()
     assert content[0] == "This file is autogenerated from test_doc.docx."
+
+
+def test_trim_trailing_whitespace(tmp_path: Path) -> None:
+    data_file = Path(__file__).parent.joinpath(
+        "data/trailing-whitespace/input.txt"
+    )
+    working_path = tmp_path / "trailing_whitespace.txt"
+    shutil.copyfile(data_file, working_path)
+    trim_trailing_whitespace(working_path)
+
+    output = working_path.read_text()
+    expected = (
+        Path(__file__)
+        .parent.joinpath("data/trailing-whitespace/output.txt")
+        .read_text()
+    )
+
+    assert output == expected

tests/data/trailing-whitespace/input.txt

@@ -0,0 +1,4 @@
+   hello world    
+   hello
+
+

tests/data/trailing-whitespace/output.txt

@@ -0,0 +1,4 @@
+   hello world
+   hello
+
+