pre-commit-hooks/pre_commit_hooks/string_fixer_for_jupyter_notebooks.py

66 lines
2.1 KiB
Python

from __future__ import annotations
import argparse
import json
from typing import Sequence
from pre_commit_hooks.util_string_fixer import fix_strings_in_file_contents
def fix_strings(filename: str) -> int:
if not filename.endswith('.ipynb'):
print(f'{filename}: not a Jupyter notebook file')
return 1
try:
with open(filename) as f:
notebook_contents = json.load(f)
except json.JSONDecodeError as exc:
print(f'{filename}: Failed to load ({exc})')
return 1
else:
cells = notebook_contents['cells']
return_value = 0
for cell in cells:
if cell.get('cell_type') == 'code' and 'source' in cell:
# Each element in cell['source'] is a string that ends with \n,
# except for the last element, which is why we don't join by \n
source_in_1_line = ''.join(cell['source'])
fixed = fix_strings_in_file_contents(source_in_1_line)
if fixed != source_in_1_line:
fixed_lines = fixed.split('\n')
cell['source'] = (
[_ + '\n' for _ in fixed_lines[:-1]] +
[fixed_lines[-1]]
)
return_value = 1
if return_value == 1:
notebook_contents['cells'] = cells
with open(filename, 'w') as f:
json.dump(notebook_contents, f, indent=1)
# Jupyter notebooks (.ipynb) always ends with a new line
# but json.dump does not.
f.write('\n')
return return_value
def main(argv: Sequence[str] | None = None) -> int:
parser = argparse.ArgumentParser()
parser.add_argument('filenames', nargs='*', help='Filenames to fix')
args = parser.parse_args(argv)
retv = 0
for filename in args.filenames:
return_value = fix_strings(filename)
if return_value != 0:
print(f'Fixing strings in {filename}')
retv |= return_value
return retv
if __name__ == '__main__':
raise SystemExit(main())