#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
STM32 ST7735S LCD CJK 字模生成工具（Pillow 版）
生成 hz16_data.c + hz16_data.h（含二分查找函数）

用法:
  # CJK 常用汉字（GB2312 一级，约 3755 字）
  python font_gen.py --cjk --font simsun --size 32 --threshold 64

  # CJK + 标点
  python font_gen.py --cjk --punct
"""

import os
import sys
import argparse

try:
    from PIL import Image, ImageFont, ImageDraw
except ImportError:
    print("请安装 Pillow: pip install Pillow", file=sys.stderr)
    sys.exit(1)

# ── Otsu 自适应阈值 ──

def otsu_threshold(pixels: list) -> int:
    """Otsu 算法自动计算最佳二值化阈值"""
    hist = [0] * 256
    for v in pixels:
        hist[v] += 1
    total = len(pixels)
    sum_all = sum(i * hist[i] for i in range(256))
    sum_bg = 0
    w_bg = 0
    w_fg = 0
    max_var = 0
    best = 128
    for t in range(256):
        w_bg += hist[t]
        if w_bg == 0:
            continue
        w_fg = total - w_bg
        if w_fg == 0:
            break
        sum_bg += t * hist[t]
        mean_bg = sum_bg / w_bg
        mean_fg = (sum_all - sum_bg) / w_fg
        between_var = w_bg * w_fg * (mean_bg - mean_fg) * (mean_bg - mean_fg)
        if between_var > max_var:
            max_var = between_var
            best = t
    return best


# ── 通用渲染 ──

def render_char(font, char: str, w: int, h: int, scale: int = 1,
                threshold: int = None) -> list:
    """
    在 w*scale × h*scale 画布上渲染字符，缩放到 w×h
    threshold=None 时自动使用 Otsu 算法
    返回逐行字节列表（MSB 优先）
    """
    bw, bh = w * scale, h * scale
    img = Image.new("L", (bw, bh), 0)
    draw = ImageDraw.Draw(img)

    bbox = draw.textbbox((0, 0), char, font=font)
    tw = bbox[2] - bbox[0]
    th = bbox[3] - bbox[1]

    # 防止 bbox[0]/bbox[1] 为负导致偏移错误
    left = bbox[0] if bbox[0] > 0 else 0
    top = bbox[1] if bbox[1] > 0 else 0
    cx = (bw - tw) // 2 - left
    cy = (bh - th) // 2 - top
    draw.text((cx, cy), char, font=font, fill=255)

    small = img.resize((w, h), Image.LANCZOS)
    pixels = list(small.getdata())

    if threshold is None:
        threshold = otsu_threshold(pixels)

    bytes_per_row = (w + 7) // 8
    result = []
    for r in range(h):
        for br in range(bytes_per_row):
            byte_val = 0
            for c in range(8):
                col = br * 8 + c
                if col < w and pixels[r * w + col] >= threshold:
                    byte_val |= (1 << (7 - c))
            result.append(byte_val)
    return result


# ── CJK 16×16 ──

CJK_W, CJK_H = 16, 16
CJK_BYTES = CJK_H * ((CJK_W + 7) // 8)  # 32


def get_gb2312_level1_unicodes():
    """GB2312 一级汉字（常用 ~3755 字）的 Unicode 码点"""
    out = []
    for u in range(0x4E00, 0x9FA6):
        try:
            b = chr(u).encode("gb2312")
            if len(b) == 2:
                qu = b[0] - 0xA1
                if 16 <= qu <= 55:
                    out.append(u)
        except (UnicodeEncodeError, LookupError):
            pass
    return out


def get_gb2312_all_unicodes():
    """GB2312 全部汉字（一级+二级 ~6763 字）"""
    out = []
    for u in range(0x4E00, 0x9FA6):
        try:
            b = chr(u).encode("gb2312")
            if len(b) == 2:
                out.append(u)
        except (UnicodeEncodeError, LookupError):
            pass
    return out


CJK_PUNCTUATION = [
    0x00B7, 0x2014, 0x2018, 0x2019, 0x201C, 0x201D, 0x2026,
    0x3000, 0x3001, 0x3002, 0x300A, 0x300B, 0x300C, 0x300D,
    0x300E, 0x300F, 0x3010, 0x3011, 0x3014, 0x3015,
    0xFF01, 0xFF02, 0xFF07, 0xFF08, 0xFF09, 0xFF0C, 0xFF0E,
    0xFF1A, 0xFF1B, 0xFF1F, 0xFF3B, 0xFF3D, 0xFF5B, 0xFF5D,
]

CJK_FULLWIDTH_ASCII = list(range(0xFF01, 0xFF5F))  # 全角 ASCII


def generate_cjk(base_path: str, font_path: str, size: int,
                 threshold: int, scale: int = 2,
                 mode: str = 'common', punct: bool = False):
    """生成 CJK 16×16 字库 .c + .h"""

    font = ImageFont.truetype(font_path, size)

    if mode == 'common':
        unicodes = get_gb2312_level1_unicodes()
        print(f"CJK 范围: GB2312 一级汉字 ({len(unicodes)} 字)")
    elif mode == 'gb2312':
        unicodes = get_gb2312_all_unicodes()
        print(f"CJK 范围: GB2312 全部汉字 ({len(unicodes)} 字)")
    else:
        unicodes = []
        print("CJK 范围: 无")

    if punct:
        extra = set(CJK_PUNCTUATION) | set(CJK_FULLWIDTH_ASCII)
        unicodes = sorted(set(unicodes) | extra)
        print(f"已加入 {len(extra)} 个中文标点/全角字符")

    if not unicodes:
        print("未指定 CJK 范围，使用默认字符")
        unicodes = [ord(c) for c in "你好世界一二三"]

    unicodes = sorted(set(unicodes))
    N = len(unicodes)

    # 渲染
    bitmaps = []
    for i, u in enumerate(unicodes):
        ch = chr(u)
        bm = render_char(font, ch, CJK_W, CJK_H, scale, threshold)
        bitmaps.append(bm)
        if (i + 1) % 500 == 0:
            print(f"  CJK 进度: {i+1}/{N}")

    # ── 写 .c ──
    c_path = base_path + ".c"
    h_path = base_path + ".h"
    name_base = os.path.basename(base_path)
    guard = name_base.upper().replace(".", "_").replace("-", "_") + "_H"

    with open(c_path, 'w', encoding='utf-8', newline='\n') as f:
        f.write(f'/* 16×16 CJK 字库，由 font_gen.py 生成 */\n\n')
        f.write(f'#include "{name_base}.h"\n\n')
        f.write(f'static const uint8_t cjk_bitmaps[][{CJK_BYTES}] = {{\n')
        for u, bm in zip(unicodes, bitmaps):
            hex_str = ', '.join(f'0x{x:02X}' for x in bm)
            try:
                comment = f' /* U+{u:04X} {chr(u)} */'
            except:
                comment = f' /* U+{u:04X} */'
            f.write(f'    {{ {hex_str} }},{comment}\n')
        f.write('};\n\n')

        f.write(f'static const uint16_t cjk_unicodes[{N}] = {{\n')
        for i in range(0, N, 16):
            chunk = unicodes[i:i+16]
            f.write('    ' + ', '.join(f'0x{u:04X}' for u in chunk) + ',\n')
        f.write('};\n\n')

        f.write(f'const uint8_t* cjk_get_bitmap(uint16_t unicode)\n')
        f.write('{\n')
        f.write(f'    int lo = 0, hi = {N};\n')
        f.write('    while (lo < hi) {\n')
        f.write('        int mid = (lo + hi) >> 1;\n')
        f.write('        if (cjk_unicodes[mid] < unicode) lo = mid + 1;\n')
        f.write('        else if (cjk_unicodes[mid] > unicode) hi = mid;\n')
        f.write('        else return cjk_bitmaps[mid];\n')
        f.write('    }\n')
        f.write('    return (const uint8_t*)0;\n')
        f.write('}\n')

    print(f'已生成: {c_path}')

    # ── 写 .h ──
    with open(h_path, 'w', encoding='utf-8', newline='\n') as f:
        f.write(f'/* 16×16 CJK 字库，由 font_gen.py 生成 */\n\n')
        f.write(f'#ifndef {guard}\n#define {guard}\n\n')
        f.write(f'#include <stdint.h>\n\n')
        f.write(f'#define CJK_BYTES_PER_CHAR {CJK_BYTES}\n')
        f.write(f'#define CJK_WIDTH  {CJK_W}\n')
        f.write(f'#define CJK_HEIGHT {CJK_H}\n')
        f.write(f'#define CJK_NUM_CHARS {N}\n\n')
        f.write(f'/* 按 Unicode 码点查找 16×16 点阵，返回 {CJK_BYTES} 字节，未找到返回 NULL */\n')
        f.write(f'const uint8_t* cjk_get_bitmap(uint16_t unicode);\n\n')
        f.write(f'#endif /* {guard} */\n')

    print(f'已生成: {h_path}')


def generate_cjk_direct(base_path: str, font_path: str, size: int,
                        threshold: int, scale: int,
                        unicodes: list):
    """用指定的 Unicode 列表生成 CJK 字库"""
    font = ImageFont.truetype(font_path, size)
    N = len(unicodes)

    print(f"CJK 自定义字符: {N} 字: {''.join(chr(u) for u in unicodes[:50])}{'...' if N > 50 else ''}")

    bitmaps = []
    for i, u in enumerate(unicodes):
        bm = render_char(font, chr(u), CJK_W, CJK_H, scale, threshold)
        bitmaps.append(bm)

    # ── 写 .c ──
    c_path = base_path + ".c"
    h_path = base_path + ".h"
    name_base = os.path.basename(base_path)
    guard = name_base.upper().replace(".", "_").replace("-", "_") + "_H"

    with open(c_path, 'w', encoding='utf-8', newline='\n') as f:
        f.write(f'/* 16×16 CJK 字库，由 font_gen.py 生成（自定义） */\n\n')
        f.write(f'#include "{name_base}.h"\n\n')
        f.write(f'static const uint8_t cjk_bitmaps[][{CJK_BYTES}] = {{\n')
        for u, bm in zip(unicodes, bitmaps):
            hex_str = ', '.join(f'0x{x:02X}' for x in bm)
            f.write(f'    {{ {hex_str} }}, /* U+{u:04X} {chr(u)} */\n')
        f.write('};\n\n')

        f.write(f'static const uint16_t cjk_unicodes[{N}] = {{\n')
        for i in range(0, N, 16):
            chunk = unicodes[i:i+16]
            f.write('    ' + ', '.join(f'0x{u:04X}' for u in chunk) + ',\n')
        f.write('};\n\n')

        f.write(f'const uint8_t* cjk_get_bitmap(uint16_t unicode)\n')
        f.write('{\n')
        f.write(f'    int lo = 0, hi = {N};\n')
        f.write('    while (lo < hi) {\n')
        f.write('        int mid = (lo + hi) >> 1;\n')
        f.write('        if (cjk_unicodes[mid] < unicode) lo = mid + 1;\n')
        f.write('        else if (cjk_unicodes[mid] > unicode) hi = mid;\n')
        f.write('        else return cjk_bitmaps[mid];\n')
        f.write('    }\n')
        f.write('    return (const uint8_t*)0;\n')
        f.write('}\n')

    print(f'已生成: {c_path}')

    # ── 写 .h ──
    with open(h_path, 'w', encoding='utf-8', newline='\n') as f:
        f.write(f'/* 16×16 CJK 字库，由 font_gen.py 生成（自定义） */\n\n')
        f.write(f'#ifndef {guard}\n#define {guard}\n\n')
        f.write(f'#include <stdint.h>\n\n')
        f.write(f'#define CJK_BYTES_PER_CHAR {CJK_BYTES}\n')
        f.write(f'#define CJK_WIDTH  {CJK_W}\n')
        f.write(f'#define CJK_HEIGHT {CJK_H}\n')
        f.write(f'#define CJK_NUM_CHARS {N}\n\n')
        f.write(f'const uint8_t* cjk_get_bitmap(uint16_t unicode);\n\n')
        f.write(f'#endif /* {guard} */\n')

    print(f'已生成: {h_path}')


# ── 主入口 ──

def _resolve_font(name_or_path: str) -> str:
    if os.path.isfile(name_or_path):
        return name_or_path
    font_dir = r'C:\Windows\Fonts'
    key = name_or_path.lower()
    known = {
        'consolas': 'consola.ttf', 'arial': 'arial.ttf',
        'tahoma': 'tahoma.ttf', 'simsun': 'simsun.ttc',
        'songti': 'simsun.ttc', '宋体': 'simsun.ttc',
        'simhei': 'simhei.ttf', '黑体': 'simhei.ttf',
        'yahei': 'msyh.ttc', '微软雅黑': 'msyh.ttc',
    }
    if key in known:
        full = os.path.join(font_dir, known[key])
        if os.path.isfile(full):
            return full
    for f in os.listdir(font_dir):
        if f.lower().startswith(key) and f.lower().endswith(('.ttf', '.ttc')):
            return os.path.join(font_dir, f)
    # fallback
    fallback = os.path.join(font_dir, 'simsun.ttc')
    if os.path.isfile(fallback):
        return fallback
    return name_or_path


def main():
    p = argparse.ArgumentParser(description='CJK 字模生成工具')
    p.add_argument('--font', default='simsun', help='字体名称或路径')
    p.add_argument('--size', type=int, default=None, help='渲染字号')
    p.add_argument('--threshold', type=int, default=None, help='二值化阈值（默认 Otsu 自适应）')
    p.add_argument('--scale', type=int, default=None, help='缩放倍率（越大边缘越平滑）')
    p.add_argument('--cjk', action='store_true', help='生成 CJK 汉字字模（必须指定）')
    p.add_argument('--common', action='store_true', help='CJK：仅常用汉字（GB2312 一级）')
    p.add_argument('--gb2312', action='store_true', help='CJK：GB2312 全部汉字')
    p.add_argument('--punct', action='store_true', help='CJK：加入中文标点/全角字符')
    p.add_argument('--string', type=str, default=None, help='CJK：仅生成指定字符串中的字符（节省空间）')
    p.add_argument('--output', '-o', default=None, help='CJK 输出基名（不含扩展名）')

    args = p.parse_args()

    font_path = _resolve_font(args.font)
    print(f"字体: {font_path}")

    if not args.cjk:
        print("请指定 --cjk 来生成 CJK 字库，ASCII 字库已不再由此工具生成")
        return

    proj_dir = os.path.abspath(os.path.join(
        os.path.dirname(__file__),
        '..', 'STM32F103C8T6', 'Drivers', 'BSP', 'LCD'))

    size = args.size if args.size is not None else 48
    scale = args.scale if args.scale is not None else 3
    out_base = args.output or os.path.join(proj_dir, 'hz16_data')

    if args.string is not None:
        unicodes = sorted(set(ord(c) for c in args.string))
        generate_cjk_direct(out_base, font_path, size, args.threshold, scale, unicodes)
    else:
        generate_cjk(out_base, font_path, size, args.threshold, scale,
                     mode='common' if args.common else ('gb2312' if args.gb2312 else 'common'),
                     punct=args.punct)


if __name__ == '__main__':
    main()