"""Table-only companion route; preserve raw cell arrays separately from Markdown.

Run with the repository's preinstalled tools/lab-venv/bin/python.
The output deliberately excludes ordinary prose; it is not a document converter.
"""
import hashlib
import importlib.metadata
import json
import platform
import time
from datetime import datetime
from pathlib import Path

import pdfplumber

LAB = Path(__file__).resolve().parent
OUT = LAB / "pdfplumber-tables"
OUT.mkdir(exist_ok=True)
runs_path = LAB / "runs.json"
runs = json.loads(runs_path.read_text())
results = []
for source in sorted((LAB / "corpus").glob("*.pdf")):
    started = time.perf_counter()
    blocks, pages = [], []
    with pdfplumber.open(source) as pdf:
        for page in pdf.pages:
            tables = page.extract_tables()
            pages.append({"page": page.page_number, "tables": tables})
            for table in tables:
                rows = [[(cell or "").replace("\n", "<br>").replace("|", "\\|")
                         for cell in row] for row in table]
                if not rows:
                    continue
                lines = ["| " + " | ".join(rows[0]) + " |",
                         "| " + " | ".join("---" for _ in rows[0]) + " |"]
                lines.extend("| " + " | ".join(row) + " |" for row in rows[1:])
                blocks.append("\n".join(lines))
    elapsed = round(time.perf_counter() - started, 4)
    (OUT / f"{source.stem}.md").write_text("\n\n".join(blocks), encoding="utf-8")
    (OUT / f"{source.stem}.json").write_text(json.dumps({
        "source": source.name, "source_sha256": hashlib.sha256(source.read_bytes()).hexdigest(),
        "pages": pages, "route": "page.extract_tables() defaults; table-only",
    }, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
    results.append({"tool": "pdfplumber-tables", "sample": source.stem,
                    "output": f"pdfplumber-tables/{source.stem}.md",
                    "elapsed_s": elapsed, "error": None})
    print(f"pdfplumber-tables x {source.stem}: {elapsed}s")

runs["results"] = [r for r in runs["results"] if r["tool"] != "pdfplumber-tables"] + results
runs["tools"]["pdfplumber-tables"] = {
    "version": importlib.metadata.version("pdfplumber"), "device": "cpu",
    "elapsed_s": round(sum(r["elapsed_s"] for r in results), 4), "error": None,
    "scope": "table-only; prose and non-table pages intentionally excluded",
}
runs_path.write_text(json.dumps(runs, ensure_ascii=False, indent=1) + "\n")
print("python", platform.python_version(), "executed", datetime.now().astimezone().isoformat())