#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
聲請人歸屬標註　第 1 步：建工作檔
=================================
把每一件釋憲案做成一張標註卡：官網原文（解釋文／主文＋理由書／理由）
加上該案所有待歸屬的確定終局裁判案號。

    python _pipeline\\聲請人_建工作檔.py

輸出於 50_聲請人標註/：
    工作檔.json      一件釋憲案一筆，含原文全文與待歸屬案號清單
    工作檔清單.csv   一列一筆待歸屬裁判，供進度追蹤與併檔
    卡片/<釋憲案>.txt 逐件的可讀卡片（標註時實際閱讀的東西）

鐵律（見 00_索引/碼表_聲請人歸屬.md）：
  * 標註依據是判決全文，不是關鍵詞窗格。
  * 每筆必須附逐字證據句，且證據句必須能在原文中以子字串還原。
  * 一次看一整件釋憲案的全部待歸屬裁判，不得一次看一筆。
"""

import csv, json, re, sys, collections
from pathlib import Path

ROOT = Path(__file__).resolve().parent.parent
PIPE = ROOT / "_pipeline"
SRC = ROOT / "40_案件歷程"
OUT = ROOT / "50_聲請人標註"
CARD = OUT / "卡片"

csv.field_size_limit(10 ** 8)

# 原文中屬於「本文」的段落標籤。判決摘要與意見書不列入標註依據。
BODY = ("解釋文", "主文", "理由書", "理由")


def main():
    try:
        sys.stdout.reconfigure(encoding="utf-8")
    except Exception:
        pass

    dbp = PIPE / "資料庫.json"
    fjp = SRC / "確定終局裁判.csv"
    for p in (dbp, fjp):
        if not p.exists():
            print("找不到", p)
            sys.exit(1)

    recs = {r["k"]: r for r in json.load(open(dbp, encoding="utf-8"))}
    fj = list(csv.DictReader(open(fjp, encoding="utf-8-sig")))

    tgt = collections.OrderedDict()
    nohit = collections.Counter()
    for r in fj:
        k = r["釋憲案"]
        if not r.get("法院"):
            nohit[k] += 1
            continue
        tgt.setdefault(k, []).append({
            "案號": f'{r["法院"]} {r["年"]}年度{r["字別"]}字第{r["號"]}號',
            "法院": r["法院"], "年": r["年"], "字別": r["字別"], "號": r["號"],
            "官網原文法院": r.get("官網原文法院", ""),
        })

    OUT.mkdir(exist_ok=True)
    CARD.mkdir(exist_ok=True)

    work, rows, nochar = [], [], []
    for k, cases in tgt.items():
        rec = recs.get(k)
        if not rec:
            nochar.append(k)
            continue
        secs = [(lab, "\n".join((f"（{n}）" if n else "") + t for n, t in ps))
                for lab, ps in rec["s"] if lab in BODY]
        text = "\n\n".join(f"【{lab}】\n{body}" for lab, body in secs)
        work.append({"k": k, "z": rec.get("z", k), "n": rec.get("n", ""),
                     "u": rec.get("u", ""), "ty": rec.get("ty", ""),
                     "cases": cases, "text": text, "chars": len(text)})
        for c in cases:
            rows.append([k, rec.get("z", k), c["案號"], len(text), "", "", "", "", ""])

        head = (f"釋憲案　{k}　{rec.get('z','')}"
                + (f"【{rec.get('n')}】" if rec.get("n") else "")
                + f"\n官網原文　{rec.get('u','')}\n"
                + f"待歸屬之確定終局裁判　{len(cases)} 筆：\n"
                + "".join(f"  [{i+1:02d}] {c['案號']}\n" for i, c in enumerate(cases))
                + "=" * 78 + "\n")
        (CARD / f"{k}.txt").write_text(head + text + "\n", encoding="utf-8")

    json.dump(work, open(OUT / "工作檔.json", "w", encoding="utf-8"), ensure_ascii=False)
    with open(OUT / "工作檔清單.csv", "w", encoding="utf-8-sig", newline="") as fh:
        w = csv.writer(fh)
        w.writerow(["釋憲案", "釋憲字號", "確定終局裁判", "原文字數",
                    "P碼", "聲請人序號", "聲請人稱謂原文", "證據句", "註記"])
        w.writerows(rows)

    ch = [x["chars"] for x in work]
    ch.sort()
    print(f"待標註釋憲案　{len(work)} 件")
    print(f"待歸屬裁判　　{len(rows)} 筆")
    print(f"原文字數　中位 {ch[len(ch)//2]:,}／最長 {ch[-1]:,}／最短 {ch[0]:,}")
    print(f"合計字數　　{sum(ch):,}")
    print(f"每件待歸屬裁判數　{collections.Counter(len(x['cases']) for x in work).most_common()}")
    if nochar:
        print("！有裁判但找不到原文：", nochar)
    print(f"另有 {sum(nohit.values())} 筆未取得裁判字號（分屬 {len(nohit)} 件），不列入標註母體")
    print("卡片在", CARD)


if __name__ == "__main__":
    main()