"""Regenerate all REVIEW.md files from GT JSONs.
Uses HTML <table> for any table containing multi-line cells (equations,
structure descriptions, numbered lists). Markdown tables used for simple
single-line content. VS Code markdown preview renders HTML blocks correctly.
Usage::
"./.venv/Scripts/python.exe" tests/regenerate_review_md.py
"""
from __future__ import annotations
import html
import json
from pathlib import Path
def _has_newlines(headers: list[str], rows: list[list[str]]) -> bool:
"""Check if any cell in the table contains newlines."""
for h in headers:
if "\n" in h:
return True
for row in rows:
for cell in row:
if "\n" in cell:
return True
return False
def _escape_pipe(s: str) -> str:
"""Escape literal pipe characters for markdown tables."""
return s.replace("|", "\\|")
def _html_cell(s: str) -> str:
"""Convert cell content to HTML-safe text with <br> for newlines."""
s = html.escape(s)
s = s.replace("\n", "<br>")
return s
def _render_html_table(headers: list[str], rows: list[list[str]]) -> list[str]:
"""Render a table as an HTML <table> block."""
ncols = len(headers)
lines = []
lines.append("<table>")
lines.append("<thead><tr>")
for h in headers:
lines.append(f"<th>{_html_cell(h)}</th>")
lines.append("</tr></thead>")
lines.append("<tbody>")
for row in rows:
padded = list(row) + [""] * (ncols - len(row))
lines.append("<tr>")
for cell in padded[:ncols]:
lines.append(f"<td>{_html_cell(cell)}</td>")
lines.append("</tr>")
lines.append("</tbody>")
lines.append("</table>")
return lines
def _render_md_table(headers: list[str], rows: list[list[str]]) -> list[str]:
"""Render a table as a markdown table (no newlines in cells)."""
ncols = len(headers)
lines = []
header_cells = [_escape_pipe(h) for h in headers]
lines.append("| " + " | ".join(header_cells) + " |")
lines.append("|" + "|".join(["---"] * ncols) + "|")
for row in rows:
padded = list(row) + [""] * (ncols - len(row))
cells = [_escape_pipe(c) for c in padded[:ncols]]
lines.append("| " + " | ".join(cells) + " |")
return lines
def render_paper(paper_dir: Path) -> str:
"""Render a complete REVIEW.md for one paper directory."""
manifest = json.loads((paper_dir / "manifest.json").read_text(encoding="utf-8"))
paper_name = manifest["paper"]
item_key = manifest["item_key"]
gt_files = sorted(paper_dir.glob("table_*_gt.json"))
tables_info = []
for gt_path in gt_files:
data = json.loads(gt_path.read_text(encoding="utf-8"))
tables_info.append(data)
num_total = len(tables_info)
num_artifacts = sum(
1 for t in tables_info
if not t.get("headers") and not t.get("rows")
)
num_data = num_total - num_artifacts
lines = []
lines.append(f"# {paper_name} — Ground Truth Review")
lines.append("")
lines.append(f"Paper: {paper_name}")
lines.append(f"Item key: {item_key}")
artifact_note = f" ({num_artifacts} artifact + {num_data} data)" if num_artifacts else ""
lines.append(f"Total tables: {num_total}{artifact_note}")
lines.append("")
for i, data in enumerate(tables_info):
idx = data.get("table_index", i)
page = data.get("page_num", "?")
caption = data.get("caption", "")
headers = data.get("headers", [])
rows = data.get("rows", [])
notes = data.get("notes", "")
verified = data.get("verified", False)
artifact = data.get("artifact_type", "")
ncols = len(headers)
nrows = len(rows)
lines.append("---")
lines.append("")
# Section heading
if not headers and not rows:
atype = artifact if artifact else "unknown"
lines.append(f"## table_{idx} (p{page}) — ARTIFACT: {atype} [{ncols} cols × {nrows} rows]")
else:
short_caption = caption[:80] + ("..." if len(caption) > 80 else "")
lines.append(f"## table_{idx} (p{page}) — {short_caption} [{ncols} cols × {nrows} rows]")
lines.append("")
# Metadata
lines.append(f"- **Caption:** {caption}")
if artifact:
lines.append(f"- **Type:** {artifact}")
if notes:
lines.append(f"- **Notes:** {notes}")
lines.append(f"- **Verified:** {str(verified).lower()}")
lines.append("")
if not headers and not rows:
lines.append("(No table to render — headers and rows are empty.)")
lines.append("")
continue
# Choose HTML or markdown based on whether cells have newlines
if _has_newlines(headers, rows):
lines.extend(_render_html_table(headers, rows))
else:
lines.extend(_render_md_table(headers, rows))
lines.append("")
# Review checklist
lines.append("---")
lines.append("")
lines.append("## Review Checklist")
lines.append("")
for data in tables_info:
idx = data.get("table_index", "?")
headers = data.get("headers", [])
rows = data.get("rows", [])
ncols = len(headers)
nrows = len(rows)
if not headers and not rows:
lines.append(f"- [ ] table_{idx} — artifact confirmed")
else:
lines.append(f"- [ ] table_{idx} — {nrows} rows, {ncols} cols verified")
lines.append("")
return "\n".join(lines)
def main():
workspace = Path(__file__).resolve().parent.parent / "tests" / "ground_truth_workspace"
paper_dirs = sorted(
p for p in workspace.iterdir()
if p.is_dir() and (p / "manifest.json").exists()
)
for paper_dir in paper_dirs:
content = render_paper(paper_dir)
review_path = paper_dir / "REVIEW.md"
review_path.write_text(content, encoding="utf-8")
print(f" [{paper_dir.name}] REVIEW.md written")
print(f"\nDone. {len(paper_dirs)} REVIEW.md files regenerated.")
if __name__ == "__main__":
main()
