tools: add md-to-ipynb converter script
Browse files- scripts/md_to_ipynb.py +139 -0
scripts/md_to_ipynb.py
ADDED
|
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Convert grpo_vertex_v3.md → grpo_vertex_v3.ipynb
|
| 3 |
+
|
| 4 |
+
Parses markdown with ```python fenced code blocks into Jupyter notebook cells.
|
| 5 |
+
- ```python blocks → code cells
|
| 6 |
+
- Everything else → markdown cells
|
| 7 |
+
- Consecutive markdown sections are merged into single cells
|
| 8 |
+
- Empty cells are skipped
|
| 9 |
+
"""
|
| 10 |
+
|
| 11 |
+
import json
|
| 12 |
+
import re
|
| 13 |
+
import sys
|
| 14 |
+
from pathlib import Path
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
def md_to_notebook(md_text: str) -> dict:
|
| 18 |
+
"""Parse markdown into notebook cells."""
|
| 19 |
+
cells = []
|
| 20 |
+
|
| 21 |
+
# Split on ```python and ``` boundaries
|
| 22 |
+
# We need to track whether we're inside a code block
|
| 23 |
+
lines = md_text.split('\n')
|
| 24 |
+
|
| 25 |
+
current_type = 'markdown' # or 'code'
|
| 26 |
+
current_lines = []
|
| 27 |
+
|
| 28 |
+
for line in lines:
|
| 29 |
+
if line.strip() == '```python':
|
| 30 |
+
# Flush current markdown
|
| 31 |
+
if current_lines and current_type == 'markdown':
|
| 32 |
+
text = '\n'.join(current_lines).strip()
|
| 33 |
+
if text:
|
| 34 |
+
cells.append(make_markdown_cell(text))
|
| 35 |
+
current_lines = []
|
| 36 |
+
current_type = 'code'
|
| 37 |
+
elif line.strip() == '```' and current_type == 'code':
|
| 38 |
+
# Flush current code
|
| 39 |
+
if current_lines:
|
| 40 |
+
code = '\n'.join(current_lines)
|
| 41 |
+
# Remove trailing whitespace but keep structure
|
| 42 |
+
code = code.rstrip()
|
| 43 |
+
if code:
|
| 44 |
+
cells.append(make_code_cell(code))
|
| 45 |
+
current_lines = []
|
| 46 |
+
current_type = 'markdown'
|
| 47 |
+
else:
|
| 48 |
+
current_lines.append(line)
|
| 49 |
+
|
| 50 |
+
# Flush remaining
|
| 51 |
+
if current_lines:
|
| 52 |
+
text = '\n'.join(current_lines).strip()
|
| 53 |
+
if text:
|
| 54 |
+
if current_type == 'code':
|
| 55 |
+
cells.append(make_code_cell(text))
|
| 56 |
+
else:
|
| 57 |
+
cells.append(make_markdown_cell(text))
|
| 58 |
+
|
| 59 |
+
# Build notebook
|
| 60 |
+
notebook = {
|
| 61 |
+
"nbformat": 4,
|
| 62 |
+
"nbformat_minor": 5,
|
| 63 |
+
"metadata": {
|
| 64 |
+
"kernelspec": {
|
| 65 |
+
"display_name": "Python 3 (ipykernel)",
|
| 66 |
+
"language": "python",
|
| 67 |
+
"name": "python3"
|
| 68 |
+
},
|
| 69 |
+
"language_info": {
|
| 70 |
+
"name": "python",
|
| 71 |
+
"version": "3.10.0",
|
| 72 |
+
"mimetype": "text/x-python",
|
| 73 |
+
"file_extension": ".py"
|
| 74 |
+
}
|
| 75 |
+
},
|
| 76 |
+
"cells": cells
|
| 77 |
+
}
|
| 78 |
+
|
| 79 |
+
return notebook
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
def make_code_cell(source: str) -> dict:
|
| 83 |
+
"""Create a code cell."""
|
| 84 |
+
return {
|
| 85 |
+
"cell_type": "code",
|
| 86 |
+
"execution_count": None,
|
| 87 |
+
"metadata": {},
|
| 88 |
+
"outputs": [],
|
| 89 |
+
"source": source.split('\n') # Will be joined with \n later
|
| 90 |
+
}
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
def make_markdown_cell(source: str) -> dict:
|
| 94 |
+
"""Create a markdown cell."""
|
| 95 |
+
return {
|
| 96 |
+
"cell_type": "markdown",
|
| 97 |
+
"metadata": {},
|
| 98 |
+
"source": source.split('\n')
|
| 99 |
+
}
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
def format_notebook(notebook: dict) -> str:
|
| 103 |
+
"""
|
| 104 |
+
Format notebook JSON with proper source line handling.
|
| 105 |
+
Each line in source needs a trailing \n except the last.
|
| 106 |
+
"""
|
| 107 |
+
for cell in notebook["cells"]:
|
| 108 |
+
lines = cell["source"]
|
| 109 |
+
if lines:
|
| 110 |
+
# Add \n to all lines except the last
|
| 111 |
+
cell["source"] = [line + '\n' for line in lines[:-1]] + [lines[-1]]
|
| 112 |
+
|
| 113 |
+
return json.dumps(notebook, indent=1, ensure_ascii=False)
|
| 114 |
+
|
| 115 |
+
|
| 116 |
+
def main():
|
| 117 |
+
input_path = Path(sys.argv[1]) if len(sys.argv) > 1 else Path("grpo_vertex_v3.md")
|
| 118 |
+
output_path = input_path.with_suffix('.ipynb')
|
| 119 |
+
|
| 120 |
+
if not input_path.exists():
|
| 121 |
+
print(f"Error: {input_path} not found")
|
| 122 |
+
sys.exit(1)
|
| 123 |
+
|
| 124 |
+
md_text = input_path.read_text(encoding='utf-8')
|
| 125 |
+
notebook = md_to_notebook(md_text)
|
| 126 |
+
|
| 127 |
+
# Stats
|
| 128 |
+
code_cells = sum(1 for c in notebook["cells"] if c["cell_type"] == "code")
|
| 129 |
+
md_cells = sum(1 for c in notebook["cells"] if c["cell_type"] == "markdown")
|
| 130 |
+
|
| 131 |
+
output_path.write_text(format_notebook(notebook), encoding='utf-8')
|
| 132 |
+
|
| 133 |
+
print(f"✓ Converted {input_path} → {output_path}")
|
| 134 |
+
print(f" {code_cells} code cells, {md_cells} markdown cells")
|
| 135 |
+
print(f" Size: {output_path.stat().st_size / 1024:.1f} KB")
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
if __name__ == "__main__":
|
| 139 |
+
main()
|