# -*- coding: utf-8 -*-
"""
为网页发布目录生成 STEP 压缩副档 (.step.gz)，供浏览器端按需解压下载。

背景: 网页版购物车打包是逐个 fetch 原始 STEP (翻转缸单个 7-13MB), 选 20 个型号
      要传 40MB+。预生成 gz 副档后前端优先取 .gz 并在浏览器解压 (DecompressionStream),
      传输量降至约 1/4.5。

用法:
  PYTHONUTF8=1 python gen-step-gz.py <web目录>           # 全量生成(跳过已存在)
  PYTHONUTF8=1 python gen-step-gz.py <web目录> --force   # 强制重建
  PYTHONUTF8=1 python gen-step-gz.py <web目录> --clean   # 删除已有 .gz 后重建
"""
import os, sys, gzip, shutil, time
from concurrent.futures import ThreadPoolExecutor, as_completed

args = [a for a in sys.argv[1:] if not a.startswith('--')]
FORCE = '--force' in sys.argv
CLEAN = '--clean' in sys.argv
if not args:
    sys.exit('用法: python gen-step-gz.py <web目录> [--force|--clean]')
WEB = args[0]
STEP = os.path.join(WEB, 'step')
if not os.path.isdir(STEP):
    sys.exit('找不到 ' + STEP)

if CLEAN:
    n = 0
    for f in os.listdir(STEP):
        if f.lower().endswith('.gz'):
            os.remove(os.path.join(STEP, f))
            n += 1
    print(f'已清理 {n} 个 .gz')

files = [f for f in os.listdir(STEP)
         if f.lower().endswith(('.step', '.stp'))]
todo = []
for f in files:
    src = os.path.join(STEP, f)
    dst = src + '.gz'
    if FORCE or not os.path.exists(dst):
        todo.append((src, dst, f))
print(f'step 文件 {len(files)} 个, 待压缩 {len(todo)} 个')


def gz_one(item):
    src, dst, name = item
    tmp = dst + '.part'
    try:
        with open(src, 'rb') as fi, gzip.open(tmp, 'wb', compresslevel=6) as fo:
            shutil.copyfileobj(fi, fo, 1024 * 1024)
        os.replace(tmp, dst)
        return (name, os.path.getsize(src), os.path.getsize(dst), None)
    except Exception as e:
        try:
            os.remove(tmp)
        except OSError:
            pass
        return (name, 0, 0, str(e))


t0 = time.time()
raw = comp = done = err = 0
errors = []
with ThreadPoolExecutor(max_workers=4) as ex:
    futs = [ex.submit(gz_one, it) for it in todo]
    for i, fut in enumerate(as_completed(futs), 1):
        name, r, c, e = fut.result()
        if e:
            err += 1
            errors.append(f'{name}: {e}')
        else:
            done += 1
            raw += r
            comp += c
        if i % 200 == 0:
            print(f'  {i}/{len(todo)}  ({time.time()-t0:.0f}s)')

print(f'\n完成: {done} 个, 失败 {err} 个, 用时 {time.time()-t0:.0f}s')
if raw:
    print(f'  原始 {raw/1073741824:.2f} GB -> gz {comp/1073741824:.2f} GB '
          f'(省 {(1-comp/raw)*100:.1f}%)')
for e in errors[:10]:
    print('  !', e)
print(f'step 目录现有 .gz {sum(1 for f in os.listdir(STEP) if f.endswith(".gz"))} 个')