✨ Add --unix-newline flag to change newline behavior

README.md

@@ -310,6 +310,13 @@ To drop all cells tagged with "solution" run:
 
 The option accepts a list of tags separated by whitespace.
 
+### Forcing UNIX newlines
+
+Force UNIX (LF) newlines in the output (useful on Windows to keep consistent
+line endings in filtered output or textconv diffs):
+
+    nbstripout --unix-newlines FILE.ipynb
+
 ### Keeping some output
 
 Do not strip the execution count/prompt number:

nbstripout/_nbstripout.py

@@ -531,6 +531,12 @@ def main():
 
     parser.add_argument('--textconv', '-t', action='store_true', help='Prints stripped files to STDOUT')
 
+    parser.add_argument(
+        '--unix-newlines',
+        action='store_true',
+        help='Force UNIX line endings in output (if unset, normalize to os.linesep)',
+    )
+
     parser.add_argument('files', nargs='*', help='Files to strip output from')
     args = parser.parse_args()
     git_config = ['git', 'config']
@@ -600,10 +606,14 @@ def main():
     keep_metadata_keys.extend(args.keep_metadata_keys.split())
     extra_keys = [i for i in extra_keys if i not in keep_metadata_keys]
 
+    # Note that we can't actually preserve newlines from the input file: nbformat implicitly converts all newlines to \n
+    # and setting newline='' disables normalization of newlines on output, so the output will always use \n as newlines.
+    newline = '' if args.unix_newlines else None
+
     # Wrap input/output stream in UTF-8 encoded text wrapper
     # https://stackoverflow.com/a/16549381
     input_stream = io.TextIOWrapper(sys.stdin.buffer, encoding='utf-8') if sys.stdin else None
-    output_stream = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8', newline='')
+    output_stream = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8', newline=newline)
 
     process_notebook = {'jupyter': process_jupyter_notebook, 'zeppelin': process_zeppelin_notebook}[args.mode]
     any_change = False
@@ -612,7 +622,7 @@ def main():
             continue
 
         try:
-            with io.open(filename, 'r+', encoding='utf8') as f:
+            with io.open(filename, 'r+', encoding='utf8', newline=newline) as f:
                 out = output_stream if args.textconv or args.dry_run else f
                 if process_notebook(
                     input_stream=f, output_stream=out, args=args, extra_keys=extra_keys, filename=filename

tests/test_end_to_end.py

@@ -1,4 +1,5 @@
 import os
+import sys
 from pathlib import Path
 import re
 from subprocess import run, PIPE
@@ -202,3 +203,31 @@ def test_nochange_notebook_unchanged():
     zpln_mtime_after = zpln_file.stat().st_mtime_ns
 
     assert zpln_mtime_after == zpln_mtime_before
+
+
+def test_newline_behavior(tmp_path: Path):
+    input_content = (NOTEBOOKS_FOLDER / 'test_drop_empty_cells.ipynb').read_bytes().replace(b'\n', b'\r\n')
+
+    to_os_eol = tmp_path / 'should-have-os-eol.ipynb'
+    to_os_eol.write_bytes(input_content)
+
+    run([nbstripout_exe(), to_os_eol])
+    if sys.platform == 'win32':
+        assert b'\r\n' in to_os_eol.read_bytes()
+    else:
+        assert b'\r\n' not in to_os_eol.read_bytes()
+
+    pc = run([nbstripout_exe(), '--textconv', to_os_eol], stdout=PIPE)
+    if sys.platform == 'win32':
+        assert b'\r\n' in pc.stdout
+    else:
+        assert b'\r\n' not in pc.stdout
+
+    to_lf_eol = tmp_path / 'should-have-lf-eol.ipynb'
+    to_lf_eol.write_bytes(input_content)
+
+    run([nbstripout_exe(), '--unix-newlines', to_lf_eol])
+    assert b'\r\n' not in to_lf_eol.read_bytes()
+
+    pc = run([nbstripout_exe(), '--unix-newlines', '--textconv', to_lf_eol], stdout=PIPE)
+    assert b'\r\n' not in pc.stdout