0.2.6-dev: the maintainer's decisions of 2026-10-12 — the 玩法说明 pages' WebP copies are lossy at quality 90 (25.6 MB of PNG → 4.2 MB; lossless stays a GUIDE_WEBP_MODE away) and /healthz?traffic=1 answers the counters to a loopback or private-network client only (a public visitor, also one a local tunnel or proxy forwards, gets the plain report; TRUST_PROXY applies as for the WebSocket limits); golden: unchanged

This commit is contained in:
sganggs committed 2026-10-12 19:09:54 +08:00
1 parent 35f4895219
commit 79bb26b317
10 files changed
+71 -26

No files matched your search

+2 -2
View File
@@ -5,8 +5,8 @@
- 1 KB 以上的 WebSocket 消息用 permessage-deflate 压缩,一局单人对局的服务器消息实测从 173 KB 降到 60 KB。
- 素材地址带内容哈希:浏览器把图片、模型和音频缓存一年,服务器更新、重新解压或 Docker 重建后只重新下载内容变了的文件;哈希存在 `.cache/asset-hashes.json`,`SP_ASSET_INDEX=0` 可关闭。
- 支持 brotli 压缩(浏览器只在 HTTPS 页面和 localhost 上使用),`.skel` 和数据文件比 gzip 再小约四分之一。
- 本地提取的「玩法说明」教程图另存无损 WebP(25.6 MB → 17.4 MB),看说明时只预载下一页。
- 流量统计:`/healthz?traffic=1` 显示服务器发出的各类流量;「问题反馈」的诊断信息增加本页下载量。
- 本地提取的「玩法说明」教程图另存 WebP(质量 90,25.6 MB → 4.2 MB),看说明时只预载下一页。
- 流量统计:`/healthz?traffic=1` 显示服务器发出的各类流量(只对本机和局域网的访问显示);「问题反馈」的诊断信息增加本页下载量。
- 部署指南:固定访问地址、反代开 HTTP/2、Cloudflare 缓存规则。
## 0.2.5 — 2026-10-12
+1 -1
View File
@@ -211,7 +211,7 @@ The `stem` of a Spine model is the upstream file name. Two examples: `char_107_l
- **Fonts:** OTF/TTF files are converted to WOFF2 by a built-in encoder (`tools/assets/woff2.mjs`: Brotli with null transforms).
- Its output was verified lossless against Google's reference `woff2` decoder.
- `fonts.css` lists WOFF2 first and falls back to the original file.
- Images stay PNG. WebP conversion is not done: it would need a native dependency. The local-client board textures are the exception: `tools/local-extract/extract.py` (Python, where Pillow is already a dependency) writes WebP copies of the 12 textures the 3D board downloads (`WEBP`: colour maps lossy at quality 95 with the alpha and the RGB under transparent texels kept, normal and data maps lossless) and `data/local-assets.json` lists the copies; the PNGs stay beside them for `tools/crop-board-atlas.mjs`. The 19 local 玩法说明 pages get a WebP copy too (`GUIDE_WEBP_MODE`, lossless: 25.6 MB → 17.4 MB on the CN client's pages; lossy quality 90 would be 4.2 MB and waits for the maintainer's side-by-side look), listed in place of the PNG, which stays beside it; the mirror copies under `ui/guide/` stay PNG. `extract.py --webp` adds the copies to an existing extraction without the client.
- Images stay PNG. WebP conversion is not done: it would need a native dependency. The local-client board textures are the exception: `tools/local-extract/extract.py` (Python, where Pillow is already a dependency) writes WebP copies of the 12 textures the 3D board downloads (`WEBP`: colour maps lossy at quality 95 with the alpha and the RGB under transparent texels kept, normal and data maps lossless) and `data/local-assets.json` lists the copies; the PNGs stay beside them for `tools/crop-board-atlas.mjs`. The 19 local 玩法说明 pages get a WebP copy too (`GUIDE_WEBP_MODE`, lossy quality 90 since the maintainer's side-by-side look of 2026-10-12: 25.6 MB → 4.2 MB on the CN client's pages; lossless would be 17.4 MB), listed in place of the PNG, which stays beside it; the mirror copies under `ui/guide/` stay PNG. `extract.py --webp` adds the copies to an existing extraction without the client.
## Manifest schema (`data/assets.json`)
+2 -2
View File
@@ -296,7 +296,7 @@ services:
| 本地提取失败 | 游戏照常运行,只是第 6 节表格里的几样换成替代样式。确认客户端已下载全部资源;Python 版本太新导致依赖安装失败时,安装 Python 3.12 后删除 `.venv-extract` 再运行 `node tools/setup.mjs --local` |
| 3D 棋盘没出现 | 需要本地提取的棋盘贴图(`node tools/doctor.mjs` 会显示「3D 棋盘可用」),以及支持 WebGL2 的浏览器。没有客户端的服务器可以从同一版本的整合包复制本地素材(第 6 节) |
| 素材地址里的 `~` 加一串字母数字 | `/assets/~<哈希>/…` 是 0.2.6 起的带版本地址(见第 0 节「网络」),浏览器可以长期缓存,属于正常现象。怀疑它导致问题时,可以用 `SP_ASSET_INDEX=0` 启动服务器临时关闭(清单改回普通地址),并请在 issue 里说明 |
| 想知道开服电脑上传了多少流量 | 浏览器打开 `http://localhost:3000/healthz?traffic=1`(换成你的端口):`traffic.http` 是服务器启动以来各类内容(`assets` 图片和模型、`audio` 音频、`code` 页面和脚本、`data` 游戏数据和素材清单)发出的字节数和请求数(压缩后的实际字节,`notModified` 是只回了「没变」的 304),`traffic.ws` 是 WebSocket 的线上字节(`wireOut` / `wireIn`,压缩后)和压缩前的消息字节(`payloadOut` / `payloadIn`)。数字只在内存里,重启后从 0 开始。玩家那边:设置 →「问题反馈」复制的诊断里有一行 `traffic:`,是这个页面从打开起下载的字节数(`cached` 是浏览器缓存直接给的,不占流量;`external` 是其他网站,例如 Google 字体) |
| 想知道开服电脑上传了多少流量 | 浏览器打开 `http://localhost:3000/healthz?traffic=1`(换成你的端口):`traffic.http` 是服务器启动以来各类内容(`assets` 图片和模型、`audio` 音频、`code` 页面和脚本、`data` 游戏数据和素材清单)发出的字节数和请求数(压缩后的实际字节,`notModified` 是只回了「没变」的 304),`traffic.ws` 是 WebSocket 的线上字节(`wireOut` / `wireIn`,压缩后)和压缩前的消息字节(`payloadOut` / `payloadIn`)。数字只在内存里,重启后从 0 开始;只有本机和局域网(含虚拟局域网地址)的访问能看到,公网访问者(包括经隧道 / 反代转发来的)只拿到普通的 `/healthz`。玩家那边:设置 →「问题反馈」复制的诊断里有一行 `traffic:`,是这个页面从打开起下载的字节数(`cached` 是浏览器缓存直接给的,不占流量;`external` 是其他网站,例如 Google 字体) |
| 断线 | 同盟模拟 10 分钟内、独立模拟 24 小时内(`config.constants.singleReconnectTime`)用同一浏览器重新打开页面,自动回到原座位。同盟掉线期间按原阵容自动作战、到时自动准备(不会代为购买;想让 AI 代打请用「离开模拟 → 暂离(AI 托管)」);独立模拟不计时,等你回来 |
## 6. 本地客户端素材(可选)
@@ -323,7 +323,7 @@ services:
**3D 棋盘贴图的下载量**:每位玩家进入对局时都要从开服的电脑下载 3D 棋盘的 12 张贴图。提取时会给这 12 张各写一份 WebP(颜色贴图有损、质量 95,法线和数据贴图无损),清单里列的是 WebP,同名 PNG 留在旁边给裁切工具和 setup 用。这部分下载量从约 6.7 MB 降到约 2 MB,网速慢的远程联机最明显。只有 PNG 的本地素材(例如在这一改动之前提取的)可以用提取时的 Python 环境运行 `tools/local-extract/extract.py --webp` 就地补上,只需要 Pillow,不需要客户端。
**「玩法说明」教程图的下载量**(0.2.6 起):本地提取的 19 页教程图每页 1.2–1.5 MB,玩家每翻一页都要从开服的电脑下载。提取时同样给每页写一份无损 WebP,清单列 WebP,PNG 留在旁边;19 页合计从约 25.6 MB 降到约 17.4 MB,画面完全不变。看说明时只预载下一页(以前前后两页都预载)。已经提取过的同样用 `tools/local-extract/extract.py --webp` 补上。
**「玩法说明」教程图的下载量**(0.2.6 起):本地提取的 19 页教程图每页 1.2–1.5 MB,玩家每翻一页都要从开服的电脑下载。提取时同样给每页写一份 WebP(有损,质量 90,放大两倍对比看不出差别),清单列 WebP,PNG 留在旁边;19 页合计从约 25.6 MB 降到约 4.2 MB。看说明时只预载下一页(以前前后两页都预载)。已经提取过的同样用 `tools/local-extract/extract.py --webp` 补上。
## 7. 打包发布(维护者)
+1 -1
View File
@@ -75,7 +75,7 @@ Audio: autochess BGM per phase, UI SFX (buy/sell/refresh/level/merge/ready/timer
- `scene/<stageId>/` and `scene/tex/` — the official scene of a battlefield (§15 "Official battlefield scenes", DESIGN §30.21): `scene.json` (materials, key light, chunk records), `scene.bin` (the meshes in board space), the lightmap pages `lm<i>`, and the materials' albedo textures shared by the scenes. Read from the scene bundles `scenes/activities/<season>/level_<stage>/…`, which the client carries only while the season runs: `extract.py --only scene` takes them from the client or from `--scenes <dir>` / `.cache/official-scenes/*` (`tools/local-extract/fetch_scenes.py` downloads them from the official CDN's in-season version). The 0.2.5 full release zip includes the extracted files when the maintainer extracted them before packaging; a lite zip and a source install may fetch and extract them (`node tools/setup.mjs --scenes`). The repository never stores the bundles or the extracted files.
- `ui/{common,battle,outer}/*` — every sprite of the three official autochess UI bundles (tier chips I–VI, bond disc rings, frames, banners, HUD pieces, badges, backgrounds, mode art).
- `emoticon/<dir>/<picId>.png` — the 36 official battle emotes, 6 themes (`shared/constants.js → EMOTE_THEMES / EMOTES`, `emoteArtPath`; data/emotes.json); also downloaded from the mirror (§22.5).
- `guide/autochess_{home,shop,handbook}_N.png` — the 19 official tutorial pages; stored squashed to 1024² — **display at 16:9** (the mirror copies too, §22.5). `extract.py` writes a lossless WebP copy beside each page and the manifest lists the copy (`GUIDE_WEBP_MODE`; 25.6 MB → 17.4 MB); the viewer preloads only the next page (`ui/guide.js guidePreloadUrl`), not the previous one too.
- `guide/autochess_{home,shop,handbook}_N.png` — the 19 official tutorial pages; stored squashed to 1024² — **display at 16:9** (the mirror copies too, §22.5). `extract.py` writes a WebP copy (lossy, quality 90) beside each page and the manifest lists the copy (`GUIDE_WEBP_MODE`; 25.6 MB → 4.2 MB); the viewer preloads only the next page (`ui/guide.js guidePreloadUrl`), not the previous one too.
- `projectiles/*` — battle projectile sprites.
- `spine/enemy/<id>/` — enemy models no web dump carries (灼热 / 炽焰源石虫, `extract.py ENEMY_SPINES`): manifest `enemies[id].spineLocal` (metadata from the committed `tools/assets/local-enemy-spines.json`); drawn instead of the tinted web alias when every file is listed (§21.14).
- `spine/token/<id>/` — token (summon) models no web dump carries (40: most 自选 summons, 凯瑟琳's 爬行号·防护单元, 凛御银灰's 风雪之眼; `extract.py TOKEN_SPINES`, the Front renderer's skeleton of each battle token prefab): manifest `tokens[id].spineLocal` (metadata from the committed `tools/assets/local-token-spines.json`); drawn instead of the avatar diamond when every file is listed (docs/ASSETS.md "Token models from the local client").
+8 -5
View File
@@ -24,14 +24,17 @@ committed data.
line: design/network.md §8 Transport.
- **玩法说明 pages as WebP.** `tools/local-extract/extract.py` writes a WebP copy beside each of the 19 local pages and
the manifest lists it (`GUIDE_WEBP_MODE`); the guide preloads only the next page instead of both neighbours. The
copies are lossless (25.6 MB → 17.4 MB on the CN client's pages). Lossy quality 90 would make them 4.2 MB; per the
decision that tutorial pages may go lossy only after a side-by-side comparison, that switch (one constant, then
`extract.py --webp`) is left to the maintainer. Normative lines: design/client.md §13, ASSETS.md Post-processing.
copies are lossy at quality 90: 25.6 MB → 4.2 MB on the CN client's pages (lossless would be 17.4 MB). Tutorial pages
could go lossy only after a side-by-side comparison; the maintainer looked at 2× crops of four pages on 2026-10-12 and
chose quality 90 (`GUIDE_WEBP_MODE`; `extract.py --webp` rewrites existing copies). Normative lines: design/client.md
§13, ASSETS.md Post-processing.
- **Traffic counters.** `GET /healthz?traffic=1` adds `traffic` (HTTP body bytes and requests by category — assets,
audio, code, data, other — and 304s; WebSocket wire bytes after compression and payload bytes before it) and
`assetHashes` (the index state). The 问题反馈 diagnostics gain a `traffic:` line: the page's Resource Timing
`transferSize` by the same categories (`shared/assetPaths.js trafficCategory`), other sites apart. Nothing is stored
or uploaded.
or uploaded. The maintainer's decision of 2026-10-12: only a loopback or private-network client gets the counters
(`server/http/routes.js`, the address from `net.js clientAddress`, so TRUST_PROXY applies as for the WebSocket
limits); any other client — also a visitor that a tunnel or proxy on this machine forwards — gets the plain report.
- **DEPLOY.md**: a stable address keeps every player's cache, settings and reconnect token (a quick tunnel's new domain
is a new site to the browser); HTTP/2 through the proxy saves round trips, not bytes; browsers offer brotli on HTTPS
pages and localhost only; a Cloudflare cache rule for `/assets/`, `/fonts/`, `/media/` (with the risk note).
@@ -82,7 +85,7 @@ changed, while a hashed URL is never asked for again until its bytes change.
### 31.4 Open items and tests
Open: `/fonts/fonts.css` (linked by index.html) and the woff2 files it names keep plain URLs (content ETags only); the
optional versioned vendor URLs (layer 1.5) are not done; lossy tutorial pages wait for the comparison.
optional versioned vendor URLs (layer 1.5) are not done.
Tests: `test/ui/audio.test.js` (BGM gating; hashed audio through `/media/~hash/`), `test/ws-compression.test.js`,
`test/local-extract.test.js` (guide WebP), `test/ui/localArt.test.js` (next-page preload), `test/traffic.test.js`,
+9 -4
View File
@@ -5,7 +5,9 @@
// * any method but GET / HEAD → 405 with `Allow: GET, HEAD`;
// * GET /healthz → JSON status (protocol `version`, release `app`, uptime, the served `build`, sockets, sessions,
// rooms, matches), never cached; `?traffic=1` adds the traffic counters (traffic.js: bytes by category since the
// start, WebSocket wire and payload bytes);
// start, WebSocket wire and payload bytes) for a loopback or private-network client only — the maintainer's
// decision of 2026-10-12: the figures are the host's, so any other client, also one a tunnel or proxy on this
// machine forwards, gets the plain report (net.js clientAddress, honouring TRUST_PROXY like the WebSocket limits);
// * everything else → the static files (static.js).
// Every response's body bytes count in the traffic counters (traffic.js meter). A route that throws is logged and
// answers 500.
@@ -13,6 +15,7 @@
import { PROTOCOL_VERSION, APP_VERSION } from '../../shared/constants.js';
import { buildTag } from './buildTag.js';
import { setSecurityHeaders, sendError, sendJson, splitUrl } from './common.js';
import { clientAddress, isLocalIp } from '../net.js';
const MAX_URL_LENGTH = 4096;
@@ -45,10 +48,11 @@ const wantsTraffic = (query) => /(?:^|&)traffic(?:=(?:1|true|yes))?(?:&|$)/i.tes
* The request listener for `http.createServer`.
* @param {{ serveStatic: (req: import('node:http').IncomingMessage, res: import('node:http').ServerResponse,
* rawPath: string, query: string) => Promise<void>,
* health: Parameters<typeof healthReport>[0], log: object }} deps
* health: Parameters<typeof healthReport>[0], log: object, trustProxy?: 'auto' | boolean }} deps
* trustProxy: whose forwarding headers name the client (net.js clientAddress; the server's TRUST_PROXY)
* @returns {(req: import('node:http').IncomingMessage, res: import('node:http').ServerResponse) => void}
*/
export function createRequestHandler({ serveStatic, health, log }) {
export function createRequestHandler({ serveStatic, health, log, trustProxy = 'auto' }) {
async function handleRequest(req, res) {
const url = req.url || '/';
if (url.length > MAX_URL_LENGTH) { sendError(req, res, 414, '请求地址过长 · URI too long'); return; }
@@ -60,7 +64,8 @@ export function createRequestHandler({ serveStatic, health, log }) {
return;
}
if (parts.rawPath === '/healthz') {
sendJson(req, res, 200, healthReport(health, wantsTraffic(parts.query)));
const withTraffic = wantsTraffic(parts.query) && isLocalIp(clientAddress(req, trustProxy).ip);
sendJson(req, res, 200, healthReport(health, withTraffic));
return;
}
await serveStatic(req, res, parts.rawPath, parts.query);
+1 -1
View File
@@ -90,7 +90,7 @@ export async function startServer(opts = {}) {
buildTag();
const traffic = createTrafficCounter();
const server = http.createServer(createRequestHandler({ serveStatic, health: { startedAt, network, registry, lobby, traffic, assetIndex, views }, log }));
const server = http.createServer(createRequestHandler({ serveStatic, health: { startedAt, network, registry, lobby, traffic, assetIndex, views }, log, trustProxy: network.opts.trustProxy }));
server.on('clientError', answerClientError);
const wss = attachWebSocket(server, { network, log, traffic });
+11 -6
View File
@@ -210,11 +210,11 @@ print(json.dumps(e.prefab_node(go)))`);
assert.equal(raw.stdout.trim(), '[0.0, 0, 0.0]');
});
test('guide pages (host bandwidth layer 0): --webp writes a lossless WebP copy of every page PNG, the manifest lists it, the PNG stays',
test('guide pages (host bandwidth layer 0): --webp writes a WebP copy of every page PNG (lossy q90 by default, lossless on request), the manifest lists it, the PNG stays',
{ skip: spawnSync(PY, ['-c', "import sys; from PIL import features; sys.exit(0 if features.check('webp') else 1)"], { env: ENV }).status !== 0 && 'no Pillow with WebP' }, () => {
const jobs = JSON.parse(spawnSync(PY, [path.join(TOOL, 'extract.py'), '--print-jobs'], { encoding: 'utf8', env: ENV }).stdout);
assert.deepEqual(jobs.guideWebp.sub, 'guide');
assert.equal(jobs.guideWebp.mode, 'lossless', 'lossy pages wait for the maintainer\'s comparison');
assert.equal(jobs.guideWebp.mode, 'lossy', 'quality 90, the maintainer\'s choice of 2026-10-12');
assert.ok(!jobs.webp.some((w) => w.sub === 'guide'), 'the board texture table is unchanged');
const out = py(`
import tempfile, pathlib
@@ -235,20 +235,25 @@ mf.write_text(json.dumps({'version': 1, 'source': 'local-client', 'count': 4, 'g
code = e.webp_only(root, mf, lambda *a: None)
doc = json.loads(mf.read_text())
g = doc['groups']['guide']
modes = {k: v.get('webp') for k, v in g.items()}
px = lambda p: list(Image.open(p).convert('RGBA').getdata())
print(json.dumps({'code': code, 'paths': {k: v['path'] for k, v in g.items()}, 'modes': {k: v.get('webp') for k, v in g.items()},
e.GUIDE_WEBP_MODE = 'lossless'
again = e.run_webp(root, 'guide', doc['groups'], lambda *a: None)
print(json.dumps({'code': code, 'paths': {k: v['path'] for k, v in g.items()}, 'modes': modes,
'relisted': {k: v.get('webp') for k, v in g.items()},
'other': doc['groups']['ui/battle']['x']['path'],
'same': all(px(sub / f'{n}.png') == px(sub / f'{n}.webp') for n in ['autochess_home_1', 'autochess_shop_2']),
'png': all((sub / f'{n}.png').exists() for n in ['autochess_home_1', 'autochess_shop_2']),
'again': e.run_webp(root, 'guide', doc['groups'], lambda *a: None)}))`);
'again': again}))`);
assert.equal(out.code, 0);
const at = (n, ext) => `/assets/local/guide/${n}.${ext}`;
assert.deepEqual(out.paths, { autochess_home_1: at('autochess_home_1', 'webp'), autochess_shop_2: at('autochess_shop_2', 'webp'), autochess_handbook_9: at('autochess_handbook_9', 'png') });
assert.deepEqual(out.modes, { autochess_home_1: 'lossless', autochess_shop_2: 'lossless', autochess_handbook_9: null });
assert.deepEqual(out.modes, { autochess_home_1: 'lossy', autochess_shop_2: 'lossy', autochess_handbook_9: null });
assert.equal(out.other, '/assets/local/ui/battle/x.png', 'other groups keep their PNGs');
assert.ok(out.same, 'lossless: the same pixels, alpha included');
assert.ok(out.png, 'the PNG stays beside the copy');
assert.equal(out.again, 2, 'a second run (another GUIDE_WEBP_MODE) rewrites the copies the manifest already lists');
assert.deepEqual(out.relisted, { autochess_home_1: 'lossless', autochess_shop_2: 'lossless', autochess_handbook_9: null });
assert.ok(out.same, 'lossless: the same pixels, alpha included');
});
});
+32
View File
@@ -103,3 +103,35 @@ test('the counter on its own: a string body in another encoding, an aborted resp
assert.equal(ws.payloadIn, 7);
assert.equal(tc.snapshot().seconds, 0);
});
test('/healthz?traffic=1 answers the counters to a loopback / LAN client only (maintainer, 2026-10-12)', async () => {
const tmp = fs.mkdtempSync(path.join(os.tmpdir(), 'sp-traffic-lan-'));
fs.mkdirSync(path.join(tmp, 'public'), { recursive: true });
fs.mkdirSync(path.join(tmp, 'data'), { recursive: true });
fs.mkdirSync(path.join(tmp, 'shared'), { recursive: true });
fs.writeFileSync(path.join(tmp, 'public/index.html'), '<!doctype html><p>hi</p>');
const dirs = { publicDir: path.join(tmp, 'public'), dataDir: path.join(tmp, 'data'), sharedDir: path.join(tmp, 'shared') };
const has = async (port, headers) => {
const r = await get(port, '/healthz?traffic=1', headers);
assert.equal(r.status, 200);
const body = JSON.parse(r.body);
assert.equal(body.ok, true, 'the plain report still answers');
return 'traffic' in body;
};
const auto = await startServer({ port: 0, host: '127.0.0.1', quiet: true, ...dirs });
try {
assert.equal(await has(auto.port), true, 'the host itself (loopback)');
assert.equal(await has(auto.port, { 'x-forwarded-for': '192.168.1.20' }), true, 'a LAN client behind a local proxy');
assert.equal(await has(auto.port, { 'x-forwarded-for': '100.101.102.103' }), true, 'a virtual-LAN (CGNAT range) address');
assert.equal(await has(auto.port, { 'x-forwarded-for': '203.0.113.9' }), false, 'a public visitor through a tunnel / proxy on this machine');
assert.equal(await has(auto.port, { 'cf-connecting-ip': '2001:db8::7' }), false, 'a public IPv6 visitor through cloudflared');
} finally {
await auto.close();
}
const ignore = await startServer({ port: 0, host: '127.0.0.1', quiet: true, trustProxy: false, ...dirs });
try {
assert.equal(await has(ignore.port, { 'x-forwarded-for': '203.0.113.9' }), true, 'TRUST_PROXY=0: the loopback peer itself counts');
} finally {
await ignore.close();
}
});
+4 -4
View File
@@ -191,12 +191,12 @@ WEBP_OPTIONS = {
# The 19 玩法说明 pages (output subdir guide: 1024², opaque) get a WebP copy too — every page PNG of the group, each a
# 1.2–1.5 MB download whenever a player opens the guide (public/js/ui/guide.js shows the local page first and preloads
# the next one). Lossless by default: 25.6 MB → 17.4 MB on the CN client's pages. Lossy quality 90 would make them
# 4.2 MB; the maintainer decides on that after a side-by-side comparison (DESIGN §31.1, docs/history/0.2.6.md),
# and then only GUIDE_WEBP_MODE changes (`extract.py --webp` rewrites the copies). Method 4: method 6 spends about 8 s
# the next one). Lossy quality 90, the maintainer's choice of 2026-10-12 after a side-by-side comparison at 2× (DESIGN
# §31.1, docs/history/0.2.6.md): 25.6 MB of PNG → 4.2 MB on the CN client's pages (lossless would be 17.4 MB);
# `extract.py --webp` rewrites the copies when GUIDE_WEBP_MODE changes. Method 4: method 6 spends about 8 s
# a page for 0.4 % less.
GUIDE_WEBP_SUB = 'guide'
GUIDE_WEBP_MODE = 'lossless'
GUIDE_WEBP_MODE = 'lossy'
GUIDE_WEBP_OPTIONS = {
'lossless': {'lossless': True, 'quality': 100, 'method': 4, 'exact': True},
'lossy': {'quality': 90, 'method': 4, 'alpha_quality': 100},