import sys
import csv
import argparse

try:
    import pymupdf  # noqa: F401  (probed here for a friendly error)
except ImportError:
    sys.exit("Missing dependency. Run:  pip install pymupdf")

from .measure import measure
from .convert import convert


def report(label, value, ok=None, bad=""):
    """One line of the results table. ok=None means purely informational."""
    tail = "" if ok is None else ("   ok" if ok else f"   <-- {bad}")
    print(f"  {label + ' ':.<28} {value}{tail}")


def main():
    ap = argparse.ArgumentParser(
        prog="pinion",
        description="Rebuild a CSV from a spreadsheet that was printed to PDF.")
    ap.add_argument("pdf")
    ap.add_argument("out_csv")
    ap.add_argument("--limit", type=int, default=0,
                    help="only convert the first N batches, for a trial run")
    args = ap.parse_args()

    M = measure(args.pdf)
    n_batches = min(args.limit, M["n_batches"]) if args.limit else M["n_batches"]

    print("CONVERTING  (takes a few minutes)")
    print("-" * 62)
    doc = pymupdf.open(args.pdf)
    with open(args.out_csv, "w", newline="", encoding="utf-8") as fh:
        stats = convert(doc, M, csv.writer(fh), n_batches)
    doc.close()

    expected = n_batches * M["rows_per_page"] - 1
    misplaced = stats["text_no_column"] + stats["text_no_row"]
    tally = {}
    for _row, letter, _width in stats["blackouts"]:
        tally[letter] = tally.get(letter, 0) + 1

    print(f"\nWritten to {args.out_csv}\n")
    print("-" * 62)
    report("rows written", f"{stats['n_rows']:,}")
    report("rows expected", f"{expected:,}", stats["n_rows"] == expected, "MISMATCH")
    report("row numbers 1..N unbroken", "yes" if stats["chain_ok"] else "NO",
           stats["chain_ok"], "GAPS OR DUPLICATES")
    report("unplaced text",
           f"{stats['text_no_column']:,} by column, {stats['text_no_row']:,} by row",
           misplaced == 0, "SOME TEXT LOST")
    report("unplaced bars", f"{stats['bars_no_row']:,}",
           stats["bars_no_row"] == 0, "REDACTIONS LOST")
    report("disagreed numbers", f"{stats['gutter_disagreed']:,}",
           stats["gutter_disagreed"] == 0, "LAYOUT WRONG")
    report("row numbers hidden by bars",
           f"{stats['gutter_unreadable']:,} (placed by spacing instead)")
    report("cells drawn in pieces",
           f"{stats['pieces']:,} (rejoined; nothing lost)")
    print()
    report("blackout bars found", f"{len(stats['blackouts']):,}")
    report("rows affected", f"{len(set(b[0] for b in stats['blackouts'])):,}")
    report("which columns", tally or "none")
    if stats["readable"]:
        report("TEXT READABLE UNDER BARS", f"{len(stats['readable']):,}",
               False, "REDACTION IS COSMETIC, HAHA! TEXT RECOVERABLE")
        for row, letter, text in stats["readable"][:10]:
            print(f"       row {row:,} column {letter}: {text!r}")
    else:
        report("text readable under bars", "none (deleted)")

    if n_batches * M["rows_per_page"] == 1048576:
        print("\n  WARNING: exactly 1,048,576 rows. That's an Excel magic number")
        print("  The source was almost certainly truncated.")
        print("  Go tell the source they owe you more data.")