|
| 1 | +import codecs |
| 2 | +import json |
| 3 | +import argparse |
| 4 | +from typing import List, Dict, Any, Optional |
| 5 | + |
| 6 | + |
| 7 | +def get_default_css(css_filename: str) -> str: |
| 8 | + """기본 CSS 스타일을 반환합니다.""" |
| 9 | + try: |
| 10 | + with open(css_filename, "r", encoding="utf-8") as f: |
| 11 | + css_content = f.read() |
| 12 | + return f"<style>\n{css_content}\n</style>" |
| 13 | + except FileNotFoundError: |
| 14 | + print("Could not find styles.css file.") |
| 15 | + return "" |
| 16 | + |
| 17 | + |
| 18 | +def _read_notebook_file(filename: str) -> Dict[str, Any]: |
| 19 | + """노트북 파일을 읽어서 JSON으로 파싱합니다.""" |
| 20 | + try: |
| 21 | + with codecs.open(filename, "r") as f: |
| 22 | + source = f.read() |
| 23 | + except UnicodeDecodeError: |
| 24 | + with codecs.open(filename, "r", encoding="utf-8") as f: |
| 25 | + source = f.read() |
| 26 | + except Exception as e: |
| 27 | + raise Exception(f"파일 변환에 실패했습니다. 에러 메시지: {str(e)}") |
| 28 | + |
| 29 | + return json.loads(source) |
| 30 | + |
| 31 | + |
| 32 | +def _process_code_output(output: Dict[str, Any], cells: List[str]) -> None: |
| 33 | + """코드 셀의 출력을 처리합니다.""" |
| 34 | + if "data" in output: |
| 35 | + outputs_data = output["data"] |
| 36 | + for key, value in outputs_data.items(): |
| 37 | + if key == "text/html": |
| 38 | + v = [v_.replace("\n", "") for v_ in value] |
| 39 | + cells.extend(v) |
| 40 | + cells.append("\n") |
| 41 | + break |
| 42 | + elif key == "text/plain": |
| 43 | + v = [v_.replace("\n", "") for v_ in value] |
| 44 | + v.insert(0, '<pre class="custom">') |
| 45 | + v.append("</pre>") |
| 46 | + cells.extend(v) |
| 47 | + cells.append("\n\n") |
| 48 | + break |
| 49 | + elif key == "image/png": |
| 50 | + plain_image = '<img src="data:image/png;base64,{}"/>\n'.format( |
| 51 | + value.replace("\n", "") |
| 52 | + ) |
| 53 | + |
| 54 | + cells.append(plain_image) |
| 55 | + cells.append("\n\n") |
| 56 | + break |
| 57 | + elif output.get("output_type") == "stream": |
| 58 | + v = output["text"] |
| 59 | + v.insert(0, '<pre class="custom">') |
| 60 | + v.append("</pre>\n\n") |
| 61 | + cells.extend(v) |
| 62 | + |
| 63 | + |
| 64 | +def _process_code_cell(cell: Dict[str, Any], cells: List[str]) -> None: |
| 65 | + """코드 셀을 처리합니다.""" |
| 66 | + work_flag = True |
| 67 | + |
| 68 | + if "source" in cell: |
| 69 | + cells.append("\n```python\n") |
| 70 | + cells.extend(cell["source"]) |
| 71 | + cells.append("\n```\n") |
| 72 | + work_flag = False |
| 73 | + |
| 74 | + outputs = cell["outputs"] |
| 75 | + if outputs: |
| 76 | + for output in outputs: |
| 77 | + _process_code_output(output, cells) |
| 78 | + elif work_flag: |
| 79 | + cells.append("\n```python") |
| 80 | + code = [c.replace("\n", "") for c in cell["source"]] |
| 81 | + cells.extend(code) |
| 82 | + cells.append("```\n\n") |
| 83 | + |
| 84 | + |
| 85 | +def _process_converter(filename: str, output_filename: str, css_filename: str) -> str: |
| 86 | + """노트북을 마크다운으로 변환합니다.""" |
| 87 | + notebook = _read_notebook_file(filename) |
| 88 | + cells: List[str] = [] |
| 89 | + |
| 90 | + for cell in notebook["cells"]: |
| 91 | + if cell["cell_type"] == "code": |
| 92 | + _process_code_cell(cell, cells) |
| 93 | + elif cell["cell_type"] == "markdown": |
| 94 | + cells.extend(cell["source"]) |
| 95 | + cells.append("\n") |
| 96 | + else: |
| 97 | + print(f"Unknown cell type. Verification Needed. file : {filename}") |
| 98 | + |
| 99 | + final_output = f"{get_default_css(css_filename)}\n\n{''.join(cells)}" |
| 100 | + |
| 101 | + with open(output_filename, "w") as f: |
| 102 | + f.write(final_output) |
| 103 | + |
| 104 | + return output_filename |
| 105 | + |
| 106 | + |
| 107 | +def convert_markdown_from_notebook( |
| 108 | + filename: str, |
| 109 | + css_filename: str = "css/styles.css", |
| 110 | + post_fix: str = "-(NEED-REVIEW)", |
| 111 | +) -> str: |
| 112 | + """노트북 파일을 마크다운으로 변환하는 메인 함수입니다.""" |
| 113 | + |
| 114 | + output_filename = filename.replace(".ipynb", f"{post_fix}.md") |
| 115 | + |
| 116 | + return _process_converter(filename, output_filename, css_filename) |
| 117 | + |
| 118 | + |
| 119 | +if __name__ == "__main__": |
| 120 | + parser = argparse.ArgumentParser( |
| 121 | + description="주피터 노트북을 마크다운으로 변환합니다." |
| 122 | + ) |
| 123 | + parser.add_argument( |
| 124 | + "--filename", required=True, help="변환할 주피터 노트북 파일 경로" |
| 125 | + ) |
| 126 | + parser.add_argument("--css", default="css/styles.css", help="css file path") |
| 127 | + args = parser.parse_args() |
| 128 | + |
| 129 | + convert_markdown_from_notebook(args.filename) |
0 commit comments