"""只读诊断：同一个会话的档案指纹在多次采样之间是否稳定。

档案键 = 头像感知哈希(8B) + 名称哈希(32B)，而 store.has_record() 是精确匹配。
只要指纹漂移，同一个联系人就会被当成“首次遇到”，从而丢弃复制到的聊天文字、
只靠截图回复。这里量化漂移幅度。
"""

import os
import sys
import time

sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

import numpy as np

import wechat_bot as bot_module


def hamming(a: bytes, b: bytes) -> int:
    return (int.from_bytes(a, "big") ^ int.from_bytes(b, "big")).bit_count()


def main():
    rounds = int(sys.argv[1]) if len(sys.argv) > 1 else 8
    bot = bot_module.WeChatBot()
    bot.safe_window_mode = True
    bot.auto_activate_window = True
    if not bot.connect(activate=False, wait_if_missing=False):
        print("[-] 未挂载到企业微信窗口")
        return
    if not bot._ensure_visible():
        print("[-] 企业微信未能切到前台")
        return
    print(
        "容差: 头像 %d 位以内算同一人；名称使用模糊匹配"
        % bot_module.WeChatBot._FP_HAMMING_TOL
    )

    samples = []
    for index in range(rounds):
        img = bot.capture_session_list()
        selected_y = bot.detect_selected_row(img)
        if selected_y < 0:
            print("  第 %d 次采样：未检测到选中行" % (index + 1))
            time.sleep(0.4)
            continue
        avatar = bot._raw_session_fingerprint(img, selected_y, row_center=True)
        name = bot._session_name_fingerprint(img, selected_y, row_center=True)
        full = bot._session_fingerprint(img, selected_y, row_center=True)
        samples.append((selected_y, avatar, name, full))
        print(
            "  第 %d 次采样：选中行 y=%-4d 头像=%s 名称=%s 完整键=%s"
            % (index + 1, selected_y, avatar.hex(), name.hex()[:16] + "…", full.hex()[:16] + "…")
        )
        time.sleep(0.4)

    if len(samples) < 2:
        print("样本不足，无法比较。")
        return

    print("\n[行中心] y 取值: %s" % sorted({s[0] for s in samples}))
    print("[头像哈希] 去重后 %d 种" % len({s[1] for s in samples}))
    print("[名称哈希] 去重后 %d 种" % len({s[2] for s in samples}))
    print("[完整档案键] 去重后 %d 种  <- 大于 1 就意味着同一会话会被反复当成新会话"
          % len({s[3] for s in samples}))

    base_y, base_avatar, base_name, _ = samples[0]
    worst = 0
    for y, avatar, name, _full in samples[1:]:
        distance = hamming(base_avatar, avatar)
        worst = max(worst, distance)
        if distance:
            print(
                "  头像相对第 1 次漂移 %2d 位（y %d→%d）%s"
                % (distance, base_y, y, "  超出容差！" if distance > bot_module.WeChatBot._FP_HAMMING_TOL else "")
            )
    print("[结论] 头像最大漂移 %d 位，容差 %d 位" % (worst, bot_module.WeChatBot._FP_HAMMING_TOL))

    # 行中心偏移对指纹的影响：模拟 ±1~6 像素的行中心估算误差。
    img = bot.capture_session_list()
    selected_y = bot.detect_selected_row(img)
    if selected_y >= 0:
        anchor = bot._raw_session_fingerprint(img, selected_y, row_center=True)
        print("\n[敏感度] 同一张图，仅把行中心挪动若干像素：")
        for offset in (1, 2, 3, 4, 6, 8):
            for sign in (-1, 1):
                shifted = bot._raw_session_fingerprint(
                    img,
                    selected_y + sign * offset,
                    row_center=True,
                )
                print(
                    "  行中心 %+d px -> 头像漂移 %2d 位%s"
                    % (
                        sign * offset,
                        hamming(anchor, shifted),
                        "  超出容差！"
                        if hamming(anchor, shifted) > bot_module.WeChatBot._FP_HAMMING_TOL
                        else "",
                    )
                )


if __name__ == "__main__":
    main()