feat(sites): structure release detail pages and rework the detail drawer (DETAIL-1)
Adapters (pure functions + tests): - detail-description: strip HTML/BBCode residue, drop site template blocks and ad/signature lines (configurable rules), parse PT info blocks (◎ / 【】 / 『』 / 「键:」) into infoBlock and body paragraphs - mediainfo: MediaInfo and BDInfo (quick summary + full report) into a summary of container/duration/bitrate/size, video (HDR/DV), audio and subtitle tracks - file-tree: file list into a flattened directory tree with sizes and main/sample/subtitle/extra roles - formatDetailPage combines them; the server computes it when reading the cached detail page, so existing caches benefit without a refetch Contracts: SiteReleasePage gains infoBlock, paragraphs, media, fileTree. Web: drawer now reads top-down: meta row (promo/H&R, size, peers, upload time, IMDb/TMDB/豆瓣 tags) → 资料 → 简介 (collapsible) → 技术规格 (raw folded) → 截图 grid with keyboard lightbox → 文件 tree; inventory overlay moved into the fixed footer next to 下载. Fixtures: 6 desensitized samples from the local cache (site.detail and RSS descriptions) plus synthetic multi-track MediaInfo, BDInfo, bracket blocks and directory file lists. Co-Authored-By: Claude Opus 5.5 <[email protected]>
This commit is contained in:
rootandClaude Opus 5.5 committed
1 parent
c846e91cc4
commit
9be9813021
16 files changed
+1992
-175
No files matched your search
@@ -0,0 +1,283 @@
|
||||
/**
|
||||
* 详情页简介清洗与结构化(DETAIL-1):纯函数,输入是抓取时得到的简介文本(可能残留 HTML / BBCode),
|
||||
* 输出「信息块」键值对(◎片名 / 【年代】/ 『导演』/ 片名: 等)与正文段落。
|
||||
* 清洗规则可配置:站点模板段、广告签名按正则整段或整行去掉,调用方可在默认规则上追加。
|
||||
*/
|
||||
|
||||
export interface DetailInfoItem {
|
||||
key: string;
|
||||
value: string;
|
||||
}
|
||||
|
||||
export interface StructuredDescription {
|
||||
infoBlock: DetailInfoItem[];
|
||||
paragraphs: string[];
|
||||
}
|
||||
|
||||
export interface DescriptionRules {
|
||||
/** 整段去掉(对整篇文本匹配,可跨行):站点模板段、更新提示等 */
|
||||
dropBlocks: RegExp[];
|
||||
/** 整行去掉:广告、群号、转载声明、感谢签名、分隔线 */
|
||||
dropLines: RegExp[];
|
||||
}
|
||||
|
||||
export const DEFAULT_DESCRIPTION_RULES: DescriptionRules = {
|
||||
dropBlocks: [
|
||||
// 憨憨豆瓣信息块末尾:「注意：该信息最后更新于…点击 这里 更新信息」
|
||||
/注意[:：]\s*该信息最后更新于[\s\S]{0,80}?更新信息/g,
|
||||
],
|
||||
dropLines: [
|
||||
/^[-=_*~—─━·.]{5,}$/,
|
||||
/^(?:QQ|qq|微信|TG|Telegram|电报|Discord)\s*群?(?:号)?\s*[:：]/,
|
||||
/(?:禁止|严禁|请勿|不得).{0,12}转载|转载.{0,8}(?:请注明|须注明)/,
|
||||
/^(?:感谢|鸣谢|Thanks?\s+to)\s*.{0,20}(?:分享|发布|制作|压制|上传)/i,
|
||||
/^(?:本资源|本种子?|资源)?由\s*\S{1,24}\s*(?:小组|组)?\s*(?:制作|压制|发布|分享)\S{0,4}$/,
|
||||
/^(?:点击|更多).{0,10}(?:下载|资源).{0,20}(?:访问|请到|见)/,
|
||||
],
|
||||
};
|
||||
|
||||
/** 常见信息块键(去掉空白后比较);冒号式 / 【】/『』 只认这些,避免把正文里的「【第一版】」当键 */
|
||||
const KNOWN_KEYS = [
|
||||
"译名",
|
||||
"片名",
|
||||
"原名",
|
||||
"又名",
|
||||
"别名",
|
||||
"中文名",
|
||||
"英文名",
|
||||
"年代",
|
||||
"年份",
|
||||
"产地",
|
||||
"国家",
|
||||
"地区",
|
||||
"国家地区",
|
||||
"制片国家/地区",
|
||||
"类别",
|
||||
"类型",
|
||||
"语言",
|
||||
"对白语言",
|
||||
"字幕",
|
||||
"上映日期",
|
||||
"首播",
|
||||
"首播日期",
|
||||
"IMDb评分",
|
||||
"IMDB评分",
|
||||
"IMDb链接",
|
||||
"IMDB链接",
|
||||
"豆瓣评分",
|
||||
"豆瓣链接",
|
||||
"TMDB",
|
||||
"TMDb链接",
|
||||
"评分",
|
||||
"票数",
|
||||
"集数",
|
||||
"季数",
|
||||
"片长",
|
||||
"单集片长",
|
||||
"视频尺寸",
|
||||
"文件格式",
|
||||
"文件大小",
|
||||
"导演",
|
||||
"编剧",
|
||||
"主演",
|
||||
"演员",
|
||||
"配音",
|
||||
"制片人",
|
||||
"标签",
|
||||
"标语",
|
||||
"简介",
|
||||
"剧情简介",
|
||||
"内容简介",
|
||||
"剧情概要",
|
||||
"剧情介绍",
|
||||
"获奖情况",
|
||||
];
|
||||
const KNOWN = new Set(KNOWN_KEYS.map((k) => k.toLowerCase()));
|
||||
/** 值就是正文的键:内容进段落 */
|
||||
const DESC_KEYS = new Set(["简介", "剧情简介", "内容简介", "剧情概要", "剧情介绍"]);
|
||||
/** 人名列表:续行合并,修复「Kai Xu周也」这类元素拼接 */
|
||||
const PEOPLE_KEYS = new Set(["导演", "编剧", "主演", "演员", "配音", "制片人"]);
|
||||
|
||||
const WS = "[\\s\\u3000]";
|
||||
const KEY_ALTERNATION = [...KNOWN_KEYS]
|
||||
.sort((a, b) => b.length - a.length)
|
||||
.map((k) => [...k].map((c) => c.replace(/[.*+?^${}()|[\]\\/]/g, "\\$&")).join(`${WS}*`))
|
||||
.join("|");
|
||||
/** ◎ 标记:认已知键,否则取到第一个空白 / 冒号为止的短词 */
|
||||
const MARK_KNOWN = new RegExp(`^◎${WS}*(${KEY_ALTERNATION})${WS}*[:：]?${WS}*(.*)$`, "i");
|
||||
const MARK_ANY = /^◎[\s ]*([^\s :：]{1,12})[\s ]*[:：]?[\s ]*(.*)$/;
|
||||
const BRACKET = new RegExp(`^(?:【|『|\\[)${WS}*(${KEY_ALTERNATION})${WS}*(?:】|』|\\])${WS}*[:：]?${WS}*(.*)$`, "i");
|
||||
const COLON = new RegExp(`^(${KEY_ALTERNATION})${WS}*[:：]${WS}*(.*)$`, "i");
|
||||
|
||||
const ENTITIES: Record<string, string> = {
|
||||
nbsp: " ",
|
||||
amp: "&",
|
||||
lt: "<",
|
||||
gt: ">",
|
||||
quot: '"',
|
||||
apos: "'",
|
||||
ensp: " ",
|
||||
emsp: " ",
|
||||
};
|
||||
|
||||
function decodeEntities(s: string): string {
|
||||
return s.replace(/&(#x[0-9a-f]+|#\d+|[a-z]+);/gi, (all, e: string) => {
|
||||
if (e[0] === "#") {
|
||||
const code = e[1] === "x" || e[1] === "X" ? Number.parseInt(e.slice(2), 16) : Number.parseInt(e.slice(1), 10);
|
||||
return Number.isFinite(code) && code > 0 && code < 0x110000 ? String.fromCodePoint(code) : all;
|
||||
}
|
||||
return ENTITIES[e.toLowerCase()] ?? all;
|
||||
});
|
||||
}
|
||||
|
||||
/** 去 HTML / BBCode 残留:图片整段去掉,链接留文字,块级标签换行,实体解码 */
|
||||
export function stripMarkup(text: string): string {
|
||||
return decodeEntities(
|
||||
text
|
||||
.replace(/\r/g, "")
|
||||
.replace(/<(script|style)\b[\s\S]*?<\/\1>/gi, "")
|
||||
// 站点截断在标签中间:「…<br…」
|
||||
.replace(/<[a-z/][^>\n]*…?$/i, "…")
|
||||
.replace(/<img\b[^>]*>/gi, "")
|
||||
.replace(/<br\s*\/?>[ \t]*\n?/gi, "\n")
|
||||
.replace(/<\/(?:p|div|li|tr|h[1-6]|blockquote|fieldset|pre)>[ \t]*\n?/gi, "\n")
|
||||
.replace(/<[a-z/!][^>]*>/gi, "")
|
||||
.replace(/\[img(?:=[^\]]*)?\][^[]*\[\/img\]|\[img=[^\]]+\]/gi, "")
|
||||
.replace(/\[url=[^\]]*\]([\s\S]*?)\[\/url\]/gi, "$1")
|
||||
.replace(
|
||||
/\[\/?(?:b|i|u|s|color|size|font|align|center|left|right|url|quote|code|spoiler|hide|box|list|\*|pre|table|tr|td)(?:=[^\]]*)?\]/gi,
|
||||
"",
|
||||
),
|
||||
).replace(/ /g, " ");
|
||||
}
|
||||
|
||||
/** 清洗:去标记、按规则删段删行、行尾空白、连续空行压成一行 */
|
||||
export function cleanDescription(text: string, rules: DescriptionRules = DEFAULT_DESCRIPTION_RULES): string {
|
||||
let t = stripMarkup(text);
|
||||
for (const re of rules.dropBlocks) t = t.replace(re, "");
|
||||
return t
|
||||
.split("\n")
|
||||
.map((l) => l.replace(/[\s ]+$/, ""))
|
||||
.filter((l) => {
|
||||
const s = l.trim().replace(/^ +/, "");
|
||||
return !s || !rules.dropLines.some((re) => re.test(s));
|
||||
})
|
||||
.join("\n")
|
||||
.replace(/\n{3,}/g, "\n\n")
|
||||
.trim();
|
||||
}
|
||||
|
||||
interface KeyHit {
|
||||
key: string;
|
||||
value: string;
|
||||
}
|
||||
|
||||
const normKey = (k: string) => k.replace(/[\s ]+/g, "");
|
||||
|
||||
function matchKey(line: string): KeyHit | null {
|
||||
const m = MARK_KNOWN.exec(line) ?? MARK_ANY.exec(line) ?? BRACKET.exec(line) ?? COLON.exec(line);
|
||||
if (!m) return null;
|
||||
const key = normKey(m[1]!);
|
||||
if (!key) return null;
|
||||
// 冒号式:键后面要么没东西(值在下一行),要么是冒号分隔的值;「片名」开头的正文句子不当键
|
||||
if (!line.startsWith("◎") && !KNOWN.has(key.toLowerCase())) return null;
|
||||
return { key, value: (m[2] ?? "").trim() };
|
||||
}
|
||||
|
||||
const isSeparator = (l: string) => /^[-=_*~—─━·.]{5,}$/.test(l);
|
||||
const SENTENCE_END = /[。！？!?.…」』"”)）]$/;
|
||||
|
||||
/** 相邻行合并成段:英文硬换行(下一行小写开头且上一行没有句末标点)接成一行,其余每行一段 */
|
||||
function toParagraphs(lines: string[]): string[] {
|
||||
const out: string[] = [];
|
||||
let prevJoinable = false;
|
||||
for (const raw of lines) {
|
||||
const l = raw.replace(/^[\s ]+/, "").trim();
|
||||
if (!l) {
|
||||
prevJoinable = false;
|
||||
continue;
|
||||
}
|
||||
if (prevJoinable && /^[a-z]/.test(l) && out.length) out[out.length - 1] += ` ${l}`;
|
||||
else out.push(l);
|
||||
prevJoinable = /[A-Za-z,]$/.test(l) && !SENTENCE_END.test(l);
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
function tidyValue(key: string, lines: string[]): string {
|
||||
const parts = lines.map((l) => l.replace(/[\s ]+/g, " ").trim()).filter(Boolean);
|
||||
if (PEOPLE_KEYS.has(key)) {
|
||||
return parts
|
||||
.flatMap((p) => p.replace(/([a-z])(?=[一-鿿])/g, "$1\u0000").split("\u0000"))
|
||||
.map((p) => p.trim())
|
||||
.filter(Boolean)
|
||||
.join(" / ");
|
||||
}
|
||||
return parts.every((p) => p.length <= 40) ? parts.join(" / ") : parts.join("\n");
|
||||
}
|
||||
|
||||
const URL_ONLY = /^(?:https?:\/\/\S+|www\.\S+)$/i;
|
||||
const isLinkKey = (k: string) => /(?:链接|网址|link)$/i.test(k) || /^tmdb$/i.test(k);
|
||||
|
||||
/**
|
||||
* 信息块 + 段落。信息块的键:◎ 开头任意短键;【】『』[] 包裹或「键:」开头只认常见键。
|
||||
* 键后的值可在同一行,也可在下一个非空行(憨憨豆瓣块);续行(主演多行)并入上一键,空行结束。
|
||||
* 简介类键的内容进段落;信息块之前的零碎短行(海报卡片的片名 / 年份 / 评分)在有信息块时丢弃。
|
||||
*/
|
||||
export function structureDescription(
|
||||
text: string | null | undefined,
|
||||
rules: DescriptionRules = DEFAULT_DESCRIPTION_RULES,
|
||||
): StructuredDescription {
|
||||
if (!text?.trim()) return { infoBlock: [], paragraphs: [] };
|
||||
const lines = cleanDescription(text, rules).split("\n");
|
||||
const info: { key: string; lines: string[] }[] = [];
|
||||
const lead: string[] = [];
|
||||
const body: string[] = [];
|
||||
let cur: { key: string; lines: string[] } | "desc" | null = null;
|
||||
const pushBody = (l: string) => (info.length ? body : lead).push(l);
|
||||
|
||||
for (const line of lines) {
|
||||
const trimmed = line.trim();
|
||||
if (isSeparator(trimmed)) {
|
||||
if (cur && cur !== "desc" && cur.lines.length) cur = null;
|
||||
if (cur === "desc") body.push("");
|
||||
continue;
|
||||
}
|
||||
const hit = trimmed ? matchKey(trimmed) : null;
|
||||
if (hit) {
|
||||
if (DESC_KEYS.has(hit.key)) {
|
||||
cur = "desc";
|
||||
body.push("");
|
||||
if (hit.value) body.push(hit.value);
|
||||
} else {
|
||||
const entry = { key: hit.key, lines: hit.value ? [hit.value] : [] };
|
||||
info.push(entry);
|
||||
cur = entry;
|
||||
}
|
||||
continue;
|
||||
}
|
||||
if (!trimmed) {
|
||||
if (cur === "desc") body.push("");
|
||||
else if (cur?.lines.length) cur = null;
|
||||
else if (!cur) pushBody("");
|
||||
continue;
|
||||
}
|
||||
if (cur && cur !== "desc") cur.lines.push(line);
|
||||
else pushBody(line);
|
||||
}
|
||||
|
||||
const seen = new Set<string>();
|
||||
const infoBlock: DetailInfoItem[] = [];
|
||||
for (const e of info) {
|
||||
const value = tidyValue(e.key, e.lines);
|
||||
if (!value || seen.has(e.key)) continue;
|
||||
if (isLinkKey(e.key) && value.split(/\s+/).every((v) => URL_ONLY.test(v) || !v)) continue;
|
||||
seen.add(e.key);
|
||||
infoBlock.push({ key: e.key, value });
|
||||
}
|
||||
// 信息块前面只有零碎短行(卡片摘要)就丢掉;有像样的句子则保留为正文
|
||||
const leadParas = toParagraphs(lead);
|
||||
const keepLead = !infoBlock.length || leadParas.some((p) => p.length > 24 || /[。！？!?]$/.test(p));
|
||||
const paragraphs = [...(keepLead ? leadParas : []), ...toParagraphs(body)];
|
||||
return { infoBlock, paragraphs };
|
||||
}
|
||||
@@ -0,0 +1,287 @@
|
||||
import {
|
||||
REAL_DETAIL_HH_DV_ONLY,
|
||||
REAL_DETAIL_HH_INFO_CARD,
|
||||
REAL_RSS_DESCR_AWARDS,
|
||||
REAL_RSS_DESCR_CAST,
|
||||
REAL_RSS_DESCR_INFO_BLOCK,
|
||||
REAL_RSS_DESCR_PLAIN,
|
||||
SYNTH_BDINFO_FULL,
|
||||
SYNTH_BDINFO_QUICK,
|
||||
SYNTH_DESCR_BRACKETS,
|
||||
SYNTH_FILES_BDMV,
|
||||
SYNTH_FILES_DIRECTORY,
|
||||
SYNTH_MEDIAINFO_MULTI_TRACK,
|
||||
} from "@umt/testkit";
|
||||
import { describe, expect, it } from "vitest";
|
||||
import {
|
||||
cleanDescription,
|
||||
DEFAULT_DESCRIPTION_RULES,
|
||||
stripMarkup,
|
||||
structureDescription,
|
||||
} from "./detail-description.ts";
|
||||
import { formatDetailPage } from "./detail-format.ts";
|
||||
import { buildFileTree } from "./file-tree.ts";
|
||||
import { parseMediaSummary } from "./mediainfo.ts";
|
||||
|
||||
const info = (text: string | null) =>
|
||||
Object.fromEntries(structureDescription(text).infoBlock.map((x) => [x.key, x.value]));
|
||||
|
||||
describe("简介结构化:真实缓存样本", () => {
|
||||
it("憨憨豆瓣信息块:「片名:」换行取值,卡片摘要与更新提示去掉,主演拆开", () => {
|
||||
const d = structureDescription(REAL_DETAIL_HH_INFO_CARD.descriptionText);
|
||||
expect(Object.fromEntries(d.infoBlock.map((x) => [x.key, x.value]))).toMatchObject({
|
||||
片名: "一瓯春",
|
||||
年代: "2026",
|
||||
片长: "45分钟",
|
||||
类别: "爱情,古装",
|
||||
导演: "潘安子 Anzi Pan",
|
||||
编剧: "尤四姐",
|
||||
主演: "许凯 Kai Xu / 周也 Ye Zhou / 白澍 Shu Bai",
|
||||
});
|
||||
// 空值键(别名 / 标语)不出现
|
||||
expect(d.infoBlock.map((x) => x.key)).not.toContain("别名");
|
||||
expect(d.infoBlock.map((x) => x.key)).not.toContain("标语");
|
||||
expect(d.paragraphs).toHaveLength(2);
|
||||
expect(d.paragraphs[0]).toMatch(/^“杀伐千面腹黑男”/);
|
||||
expect(d.paragraphs.join("\n")).not.toMatch(/该信息最后更新于|更新信息|^-{5,}|\(11053\)/m);
|
||||
});
|
||||
|
||||
it("◎ 信息块 + ◎简介:链接键去掉,简介进段落", () => {
|
||||
const d = structureDescription(REAL_RSS_DESCR_INFO_BLOCK);
|
||||
expect(d.infoBlock.map((x) => x.key)).toEqual([
|
||||
"译名",
|
||||
"片名",
|
||||
"年代",
|
||||
"产地",
|
||||
"类别",
|
||||
"语言",
|
||||
"上映日期",
|
||||
"集数",
|
||||
"片长",
|
||||
"主演",
|
||||
]);
|
||||
expect(info(REAL_RSS_DESCR_INFO_BLOCK).译名).toBe("Stay Low Profile, Sect Chief / 掌门低调点 3D动画版");
|
||||
expect(d.paragraphs).toHaveLength(1);
|
||||
expect(d.paragraphs[0]).toMatch(/^大雪覆门/);
|
||||
expect(JSON.stringify(d)).not.toMatch(/<| |douban\.com/);
|
||||
});
|
||||
|
||||
it("主演多行续行合并;◎获奖情况进信息块;站点截断的半个标签去掉", () => {
|
||||
const i = info(REAL_RSS_DESCR_AWARDS);
|
||||
expect(i.主演).toBe(
|
||||
"张维为 Weiwei Zhang / 亚历山大·杜金 Alexander Dugin / 李世默 Eric Li / 范勇鹏 Yongpeng Fan / 余亮 Liang Yu",
|
||||
);
|
||||
expect(i.获奖情况).toBe("第28届上海电视节白玉兰奖 (2023) / 综艺节目 最佳电视综艺节目(提名)…");
|
||||
const d = structureDescription(REAL_RSS_DESCR_AWARDS);
|
||||
expect(d.paragraphs).toHaveLength(2);
|
||||
expect(d.paragraphs[1]).toMatch(/^《这就是中国》/);
|
||||
});
|
||||
|
||||
it("豆瓣评分 / 导演 / 编剧,长主演列表", () => {
|
||||
const i = info(REAL_RSS_DESCR_CAST);
|
||||
expect(i).toMatchObject({ 豆瓣评分: "5.4/10 from 11075 users", 导演: "潘安子 Anzi Pan", 编剧: "尤四姐" });
|
||||
expect(i.主演!.split(" / ")).toHaveLength(16);
|
||||
expect(structureDescription(REAL_RSS_DESCR_CAST).paragraphs).toEqual([]);
|
||||
});
|
||||
|
||||
it("没有信息块:整段是正文,海报图去掉", () => {
|
||||
const d = structureDescription(REAL_RSS_DESCR_PLAIN);
|
||||
expect(d.infoBlock).toEqual([]);
|
||||
expect(d.paragraphs).toHaveLength(1);
|
||||
expect(d.paragraphs[0]).toMatch(/^21岁的玉雕师方雾/);
|
||||
});
|
||||
|
||||
it("只有 MediaInfo 的详情页:信息块和段落都为空", () => {
|
||||
expect(structureDescription(REAL_DETAIL_HH_DV_ONLY.descriptionText)).toEqual({ infoBlock: [], paragraphs: [] });
|
||||
});
|
||||
});
|
||||
|
||||
describe("简介清洗:【】『』、BBCode / HTML 残留、广告签名", () => {
|
||||
const d = structureDescription(SYNTH_DESCR_BRACKETS);
|
||||
|
||||
it("【】『』键认得,正文里的「【第一版】」不当键", () => {
|
||||
expect(Object.fromEntries(d.infoBlock.map((x) => [x.key, x.value]))).toMatchObject({
|
||||
译名: "示例电影 / Example Movie",
|
||||
字幕: "中英双语",
|
||||
IMDb评分: "8.1/10 from 12,345 users",
|
||||
导演: "某导演 Some Director",
|
||||
主演: "演员甲 Actor A / 演员乙 Actor B",
|
||||
});
|
||||
expect(d.infoBlock.map((x) => x.key)).not.toContain("IMDb链接");
|
||||
expect(structureDescription("◎简 介\n\n 【第一版】仙界至暗，众生皆材。").paragraphs).toEqual([
|
||||
"【第一版】仙界至暗，众生皆材。",
|
||||
]);
|
||||
});
|
||||
|
||||
it("段落:标记、实体、多余空行、转载声明 / 群号 / 感谢签名 / 分隔线都去掉", () => {
|
||||
expect(d.paragraphs).toEqual([
|
||||
"一位科学家在沙漠深处发现了改变世界的秘密。<他必须>在时间耗尽前做出选择。",
|
||||
"影片获得多项提名,查看获奖名单。",
|
||||
]);
|
||||
const cleaned = cleanDescription(SYNTH_DESCR_BRACKETS);
|
||||
expect(cleaned).not.toMatch(/\[\/?\w+|<\/?\w+| |QQ群|转载|感谢|\n{3,}|-{5,}/);
|
||||
});
|
||||
|
||||
it("规则可配置:追加整行规则", () => {
|
||||
const rules = { ...DEFAULT_DESCRIPTION_RULES, dropLines: [...DEFAULT_DESCRIPTION_RULES.dropLines, /^影片获得/] };
|
||||
expect(structureDescription(SYNTH_DESCR_BRACKETS, rules).paragraphs).toHaveLength(1);
|
||||
});
|
||||
|
||||
it("英文硬换行接回一段", () => {
|
||||
expect(
|
||||
structureDescription("A scientist discovers a secret\nthat changes the world.\n\nSecond one.").paragraphs,
|
||||
).toEqual(["A scientist discovers a secret that changes the world.", "Second one."]);
|
||||
});
|
||||
|
||||
it("stripMarkup:实体与数字引用", () => {
|
||||
expect(stripMarkup("a&bAB <b>c</b>[b]d[/b]")).toBe("a&bAB cd");
|
||||
});
|
||||
});
|
||||
|
||||
describe("MediaInfo / BDInfo 概要", () => {
|
||||
it("真实样本:中文摘要头 + General / Video / Audio", () => {
|
||||
expect(parseMediaSummary(REAL_DETAIL_HH_INFO_CARD.mediaInfo)).toEqual({
|
||||
kind: "mediainfo",
|
||||
general: {
|
||||
fileName: "Spring.of.the.Blade.S01E18.2026.2160p.YK.WEB-DL.H.265.HFR.HQ.AAC2.0-HHWEB.mkv",
|
||||
container: "Matroska",
|
||||
duration: "43 min 55 s",
|
||||
overallBitrate: "15.6 Mb/s",
|
||||
size: "4.79 GiB",
|
||||
},
|
||||
video: [
|
||||
{
|
||||
codec: "HEVC",
|
||||
resolution: "3840×1608",
|
||||
frameRate: "60 fps",
|
||||
bitrate: "15.3 Mb/s",
|
||||
bitDepth: "10 bit",
|
||||
hdr: [],
|
||||
},
|
||||
],
|
||||
audio: [
|
||||
{
|
||||
language: "中文",
|
||||
codec: "AAC",
|
||||
channels: "2.0",
|
||||
bitrate: "256 kb/s",
|
||||
title: "中文 [AAC 2.0]",
|
||||
isDefault: true,
|
||||
},
|
||||
],
|
||||
subtitles: [],
|
||||
});
|
||||
});
|
||||
|
||||
it("真实样本:杜比视界 Profile 5", () => {
|
||||
const m = parseMediaSummary(REAL_DETAIL_HH_DV_ONLY.mediaInfo)!;
|
||||
expect(m.video[0]).toMatchObject({ hdr: ["Dolby Vision P5"], frameRate: "25 fps" });
|
||||
});
|
||||
|
||||
it("多音轨多字幕:编码短名、声道、默认 / 强制、语言译名", () => {
|
||||
const m = parseMediaSummary(SYNTH_MEDIAINFO_MULTI_TRACK)!;
|
||||
expect(m.video[0]).toMatchObject({
|
||||
resolution: "3840×2160",
|
||||
frameRate: "23.976 fps",
|
||||
hdr: ["Dolby Vision P7.6", "HDR10"],
|
||||
});
|
||||
expect(m.audio.map((a) => [a.language, a.codec, a.channels, a.bitrate, a.isDefault])).toEqual([
|
||||
["英语", "TrueHD Atmos", "7.1", "4561 kb/s", true],
|
||||
["中文", "AC-3", "5.1", "640 kb/s", false],
|
||||
["英语", "AAC", "2.0", "192 kb/s", false],
|
||||
]);
|
||||
expect(m.subtitles.map((s) => [s.language, s.format, s.isDefault, s.forced])).toEqual([
|
||||
["英语", "PGS", false, true],
|
||||
["中文", "UTF-8", true, false],
|
||||
["繁体中文", "ASS", false, false],
|
||||
["英语", "PGS", false, false],
|
||||
]);
|
||||
});
|
||||
|
||||
it("BDInfo 简版", () => {
|
||||
const m = parseMediaSummary(SYNTH_BDINFO_QUICK)!;
|
||||
expect(m.kind).toBe("bdinfo");
|
||||
expect(m.general).toMatchObject({
|
||||
container: "Blu-ray",
|
||||
duration: "2 h 22 min",
|
||||
size: "57.57 GiB",
|
||||
overallBitrate: "55.20 Mbps",
|
||||
});
|
||||
expect(m.video.map((v) => [v.resolution, v.hdr])).toEqual([
|
||||
["2160p", ["HDR10"]],
|
||||
["1080p", ["Dolby Vision FEL"]],
|
||||
]);
|
||||
expect(m.audio[0]).toMatchObject({
|
||||
language: "英语",
|
||||
codec: "TrueHD Atmos",
|
||||
channels: "7.1",
|
||||
bitrate: "4561 kbps",
|
||||
});
|
||||
expect(m.subtitles.map((s) => s.language)).toEqual(["英语", "中文", "中文"]);
|
||||
});
|
||||
|
||||
it("BDInfo 完整报告表格", () => {
|
||||
const m = parseMediaSummary(SYNTH_BDINFO_FULL)!;
|
||||
expect(m.general.duration).toBe("52 min 10 s");
|
||||
expect(m.video).toEqual([
|
||||
{
|
||||
codec: "MPEG-4 AVC",
|
||||
resolution: "1080p",
|
||||
frameRate: "23.976 fps",
|
||||
bitrate: "31456 kbps",
|
||||
bitDepth: null,
|
||||
hdr: [],
|
||||
},
|
||||
]);
|
||||
expect(m.audio.map((a) => [a.language, a.codec, a.channels])).toEqual([
|
||||
["日语", "DTS-HD MA", "5.1"],
|
||||
["日语", "LPCM", "2.0"],
|
||||
]);
|
||||
expect(m.subtitles).toHaveLength(2);
|
||||
});
|
||||
|
||||
it("认不出返回 null", () => {
|
||||
expect(parseMediaSummary("just some text")).toBeNull();
|
||||
expect(parseMediaSummary(null)).toBeNull();
|
||||
});
|
||||
});
|
||||
|
||||
describe("文件清单 → 目录树", () => {
|
||||
it("真实样本:8 集平铺,全是主视频", () => {
|
||||
const t = buildFileTree(REAL_DETAIL_HH_INFO_CARD.files);
|
||||
expect(t).toHaveLength(REAL_DETAIL_HH_INFO_CARD.files.length);
|
||||
expect(t.every((r) => !r.dir && r.depth === 0 && r.role === "main")).toBe(true);
|
||||
});
|
||||
|
||||
it("目录型:目录在前、大小求和,主视频 / sample / 字幕 / 附件", () => {
|
||||
const t = buildFileTree(SYNTH_FILES_DIRECTORY);
|
||||
expect(t[0]).toMatchObject({ dir: true, depth: 0, fileCount: 6, sizeBytes: 15_085_574_144 });
|
||||
const role = (name: string) => t.find((r) => r.name === name)?.role;
|
||||
expect(role("Example.Movie.2023.1080p.BluRay.x264-GROUP.mkv")).toBe("main");
|
||||
expect(role("example.movie.sample.mkv")).toBe("sample");
|
||||
expect(role("Example.Movie.2023.chs.srt")).toBe("subtitle");
|
||||
expect(role("poster.jpg")).toBe("extra");
|
||||
expect(t.map((r) => `${" ".repeat(r.depth)}${r.name}`).slice(0, 4)).toEqual([
|
||||
"Example.Movie.2023.1080p.BluRay.x264-GROUP",
|
||||
" Sample",
|
||||
" example.movie.sample.mkv",
|
||||
" Subs",
|
||||
]);
|
||||
});
|
||||
|
||||
it("原盘:最大的 m2ts 是正片,小片段不是", () => {
|
||||
const t = buildFileTree(SYNTH_FILES_BDMV);
|
||||
expect(t.find((r) => r.name === "00001.m2ts")?.role).toBe("main");
|
||||
expect(t.find((r) => r.name === "00002.m2ts")?.role).toBe("video");
|
||||
});
|
||||
});
|
||||
|
||||
describe("formatDetailPage", () => {
|
||||
it("组合四部分", () => {
|
||||
const f = formatDetailPage(REAL_DETAIL_HH_INFO_CARD);
|
||||
expect(f.infoBlock.length).toBeGreaterThan(5);
|
||||
expect(f.paragraphs).toHaveLength(2);
|
||||
expect(f.media?.kind).toBe("mediainfo");
|
||||
expect(f.fileTree).toHaveLength(8);
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,33 @@
|
||||
/**
|
||||
* 详情页扩展内容的展示结构(DETAIL-1):简介 → 信息块 + 段落,MediaInfo / BDInfo → 技术规格概要,文件清单 → 目录树。
|
||||
* 纯函数;服务端读缓存时现算(旧缓存也能用),原文(descriptionText / mediaInfo / files)照旧保留。
|
||||
*/
|
||||
import {
|
||||
DEFAULT_DESCRIPTION_RULES,
|
||||
type DescriptionRules,
|
||||
type DetailInfoItem,
|
||||
structureDescription,
|
||||
} from "./detail-description.ts";
|
||||
import { buildFileTree, type FileTreeRow } from "./file-tree.ts";
|
||||
import { type MediaSummary, parseMediaSummary } from "./mediainfo.ts";
|
||||
import type { SiteDetailPage } from "./types.ts";
|
||||
|
||||
export interface FormattedDetailPage {
|
||||
infoBlock: DetailInfoItem[];
|
||||
paragraphs: string[];
|
||||
media: MediaSummary | null;
|
||||
fileTree: FileTreeRow[];
|
||||
}
|
||||
|
||||
export function formatDetailPage(
|
||||
page: Pick<SiteDetailPage, "descriptionText" | "mediaInfo" | "files">,
|
||||
rules: DescriptionRules = DEFAULT_DESCRIPTION_RULES,
|
||||
): FormattedDetailPage {
|
||||
const { infoBlock, paragraphs } = structureDescription(page.descriptionText, rules);
|
||||
return {
|
||||
infoBlock,
|
||||
paragraphs,
|
||||
media: parseMediaSummary(page.mediaInfo),
|
||||
fileTree: buildFileTree(page.files ?? []),
|
||||
};
|
||||
}
|
||||
@@ -0,0 +1,103 @@
|
||||
/**
|
||||
* 详情页文件清单 → 目录树(DETAIL-1,纯函数)。按「/」或「\」拆路径,目录大小为子项之和;
|
||||
* 标出主视频(最大视频的 30% 以上,排除 sample)、sample、字幕与附件。输出按先序展平,前端按 depth 缩进。
|
||||
*/
|
||||
import type { SiteDetailFile } from "./types.ts";
|
||||
|
||||
export type FileRole = "main" | "sample" | "subtitle" | "extra" | "video" | "other";
|
||||
|
||||
export interface FileTreeRow {
|
||||
/** 完整路径(目录也有),作 key 与折叠状态 */
|
||||
path: string;
|
||||
name: string;
|
||||
depth: number;
|
||||
dir: boolean;
|
||||
/** 目录:子项之和;有未知大小的子项时仍给已知部分之和 */
|
||||
sizeBytes: number | null;
|
||||
/** 目录为 null */
|
||||
role: FileRole | null;
|
||||
/** 目录下的文件数(递归);文件为 null */
|
||||
fileCount: number | null;
|
||||
}
|
||||
|
||||
const VIDEO = /\.(mkv|mp4|m2ts|ts|avi|wmv|mov|m4v|iso|rmvb|flv|webm|vob|mpg|mpeg)$/i;
|
||||
const SUB = /\.(srt|ass|ssa|sup|sub|idx|vtt|smi)$/i;
|
||||
const EXTRA = /\.(nfo|txt|jpe?g|png|gif|webp|bmp|md5|sfv|sha1|sha256|url|html?|pdf|cue|log|torrent)$/i;
|
||||
const SAMPLE = /(^|[\\/._\-\s])sample([\\/._\-\s]|$)/i;
|
||||
|
||||
interface Node {
|
||||
name: string;
|
||||
path: string;
|
||||
children: Map<string, Node> | null;
|
||||
sizeBytes: number | null;
|
||||
}
|
||||
|
||||
export function buildFileTree(files: SiteDetailFile[]): FileTreeRow[] {
|
||||
const root: Node = { name: "", path: "", children: new Map(), sizeBytes: null };
|
||||
for (const f of files) {
|
||||
const parts = f.name.split(/[\\/]+/).filter(Boolean);
|
||||
if (!parts.length) continue;
|
||||
let node = root;
|
||||
parts.forEach((part, i) => {
|
||||
const last = i === parts.length - 1;
|
||||
const path = node.path ? `${node.path}/${part}` : part;
|
||||
let child = node.children!.get(part);
|
||||
if (!child) {
|
||||
child = { name: part, path, children: last ? null : new Map(), sizeBytes: last ? f.sizeBytes : null };
|
||||
node.children!.set(part, child);
|
||||
} else if (!last && !child.children) {
|
||||
child.children = new Map();
|
||||
}
|
||||
node = child;
|
||||
});
|
||||
}
|
||||
|
||||
const videoSizes = files.filter((f) => VIDEO.test(f.name) && !SAMPLE.test(f.name)).map((f) => f.sizeBytes ?? 0);
|
||||
const maxVideo = Math.max(0, ...videoSizes);
|
||||
const roleOf = (n: Node): FileRole => {
|
||||
if (SUB.test(n.name)) return "subtitle";
|
||||
if (VIDEO.test(n.name)) {
|
||||
if (SAMPLE.test(n.path)) return "sample";
|
||||
// 大小未知时所有视频都当主视频;已知时按最大视频的 30% 划线(剧集各集、原盘正片)
|
||||
if (!maxVideo || n.sizeBytes === null || n.sizeBytes >= maxVideo * 0.3) return "main";
|
||||
return "video";
|
||||
}
|
||||
if (EXTRA.test(n.name)) return "extra";
|
||||
return "other";
|
||||
};
|
||||
|
||||
const out: FileTreeRow[] = [];
|
||||
const walk = (n: Node, depth: number): { size: number | null; count: number } => {
|
||||
const kids = [...n.children!.values()].sort((a, b) =>
|
||||
!!a.children === !!b.children ? a.name.localeCompare(b.name, "zh-CN", { numeric: true }) : a.children ? -1 : 1,
|
||||
);
|
||||
let size: number | null = null;
|
||||
let count = 0;
|
||||
for (const k of kids) {
|
||||
if (k.children) {
|
||||
const idx = out.length;
|
||||
out.push({ path: k.path, name: k.name, depth, dir: true, sizeBytes: null, role: null, fileCount: null });
|
||||
const sub = walk(k, depth + 1);
|
||||
out[idx]!.sizeBytes = sub.size;
|
||||
out[idx]!.fileCount = sub.count;
|
||||
if (sub.size !== null) size = (size ?? 0) + sub.size;
|
||||
count += sub.count;
|
||||
} else {
|
||||
out.push({
|
||||
path: k.path,
|
||||
name: k.name,
|
||||
depth,
|
||||
dir: false,
|
||||
sizeBytes: k.sizeBytes,
|
||||
role: roleOf(k),
|
||||
fileCount: null,
|
||||
});
|
||||
if (k.sizeBytes !== null) size = (size ?? 0) + k.sizeBytes;
|
||||
count += 1;
|
||||
}
|
||||
}
|
||||
return { size, count };
|
||||
};
|
||||
walk(root, 0);
|
||||
return out;
|
||||
}
|
||||
@@ -16,12 +16,23 @@ export {
|
||||
SpiderFilterSchema,
|
||||
} from "./definitions/index.ts";
|
||||
export type { SiteAdapterDeps } from "./deps.ts";
|
||||
export {
|
||||
cleanDescription,
|
||||
DEFAULT_DESCRIPTION_RULES,
|
||||
type DescriptionRules,
|
||||
type DetailInfoItem,
|
||||
type StructuredDescription,
|
||||
stripMarkup,
|
||||
structureDescription,
|
||||
} from "./detail-description.ts";
|
||||
export { type FormattedDetailPage, formatDetailPage } from "./detail-format.ts";
|
||||
export {
|
||||
type FeedMediaType,
|
||||
normalizeSiteFeedItem,
|
||||
type SiteFeedItemFields,
|
||||
siteCategoryMediaType,
|
||||
} from "./feed-item.ts";
|
||||
export { buildFileTree, type FileRole, type FileTreeRow } from "./file-tree.ts";
|
||||
export {
|
||||
classifyLoginError,
|
||||
loginFailureText,
|
||||
@@ -33,6 +44,14 @@ export {
|
||||
type SiteLoginFailure,
|
||||
totpCode,
|
||||
} from "./login/nexusphp.ts";
|
||||
export {
|
||||
type MediaAudioTrack,
|
||||
type MediaGeneral,
|
||||
type MediaSummary,
|
||||
type MediaTextTrack,
|
||||
type MediaVideoTrack,
|
||||
parseMediaSummary,
|
||||
} from "./mediainfo.ts";
|
||||
export { createMteamAdapter, mteamTorrentToRelease } from "./mteam/adapter.ts";
|
||||
export { type MteamApiPath, mteamPost } from "./mteam/client.ts";
|
||||
export { type MteamDefinition, mteamApiBase, mteamDefinition } from "./mteam/definition.ts";
|
||||
|
||||
@@ -0,0 +1,405 @@
|
||||
/**
|
||||
* MediaInfo / BDInfo 文本 → 技术规格概要(DETAIL-1,纯函数)。原文另存,这里只取展示需要的字段:
|
||||
* 概要(容器、时长、总码率、大小)、视频轨、音频轨、字幕轨。认不出的字段留 null,不猜。
|
||||
*/
|
||||
|
||||
export interface MediaGeneral {
|
||||
fileName: string | null;
|
||||
container: string | null;
|
||||
duration: string | null;
|
||||
overallBitrate: string | null;
|
||||
size: string | null;
|
||||
}
|
||||
|
||||
export interface MediaVideoTrack {
|
||||
codec: string | null;
|
||||
resolution: string | null;
|
||||
frameRate: string | null;
|
||||
bitrate: string | null;
|
||||
bitDepth: string | null;
|
||||
/** HDR10 / HDR10+ / Dolby Vision P5 / HLG … 空数组 = SDR 或未知 */
|
||||
hdr: string[];
|
||||
}
|
||||
|
||||
export interface MediaAudioTrack {
|
||||
language: string | null;
|
||||
codec: string | null;
|
||||
channels: string | null;
|
||||
bitrate: string | null;
|
||||
title: string | null;
|
||||
isDefault: boolean | null;
|
||||
}
|
||||
|
||||
export interface MediaTextTrack {
|
||||
language: string | null;
|
||||
format: string | null;
|
||||
title: string | null;
|
||||
isDefault: boolean | null;
|
||||
forced: boolean | null;
|
||||
}
|
||||
|
||||
export interface MediaSummary {
|
||||
kind: "mediainfo" | "bdinfo";
|
||||
general: MediaGeneral;
|
||||
video: MediaVideoTrack[];
|
||||
audio: MediaAudioTrack[];
|
||||
subtitles: MediaTextTrack[];
|
||||
}
|
||||
|
||||
const LANG: Record<string, string> = {
|
||||
chinese: "中文",
|
||||
mandarin: "国语",
|
||||
cantonese: "粤语",
|
||||
zh: "中文",
|
||||
chi: "中文",
|
||||
zho: "中文",
|
||||
"zh-cn": "简体中文",
|
||||
"zh-hans": "简体中文",
|
||||
"zh-tw": "繁体中文",
|
||||
"zh-hk": "繁体中文",
|
||||
"zh-hant": "繁体中文",
|
||||
english: "英语",
|
||||
en: "英语",
|
||||
eng: "英语",
|
||||
japanese: "日语",
|
||||
ja: "日语",
|
||||
jpn: "日语",
|
||||
korean: "韩语",
|
||||
ko: "韩语",
|
||||
kor: "韩语",
|
||||
french: "法语",
|
||||
fr: "法语",
|
||||
german: "德语",
|
||||
de: "德语",
|
||||
spanish: "西班牙语",
|
||||
es: "西班牙语",
|
||||
italian: "意大利语",
|
||||
russian: "俄语",
|
||||
thai: "泰语",
|
||||
portuguese: "葡萄牙语",
|
||||
hindi: "印地语",
|
||||
};
|
||||
|
||||
function language(v: string | null | undefined): string | null {
|
||||
if (!v) return null;
|
||||
const t = v.trim();
|
||||
return LANG[t.toLowerCase()] ?? t;
|
||||
}
|
||||
|
||||
function yesNo(v: string | undefined): boolean | null {
|
||||
if (!v) return null;
|
||||
if (/^(yes|是)$/i.test(v.trim())) return true;
|
||||
if (/^(no|否)$/i.test(v.trim())) return false;
|
||||
return null;
|
||||
}
|
||||
|
||||
/** 「3 840 pixels」→ 3840 */
|
||||
const num = (v: string | undefined): number | null => {
|
||||
const m = v ? /([\d\s]+)/.exec(v.replace(/(\d)\s+(?=\d)/g, "$1")) : null;
|
||||
return m ? Number.parseInt(m[1]!.replace(/\s/g, ""), 10) || null : null;
|
||||
};
|
||||
|
||||
/** 码率:去掉数字里的千分位空格 「4 561 kb/s」→「4561 kb/s」 */
|
||||
const rate = (v: string | undefined): string | null => (v ? v.replace(/(\d)\s+(?=\d)/g, "$1").trim() : null);
|
||||
|
||||
function frameRate(v: string | undefined): string | null {
|
||||
const m = v ? /([\d.]+)/.exec(v) : null;
|
||||
if (!m) return null;
|
||||
const n = Number.parseFloat(m[1]!);
|
||||
return Number.isFinite(n) ? `${Number.isInteger(n) ? n : Number(n.toFixed(3))} fps` : null;
|
||||
}
|
||||
|
||||
/** 声道数 + 布局 → 2.0 / 5.1 / 7.1 */
|
||||
function channels(count: string | undefined, layout: string | undefined): string | null {
|
||||
const n = num(count);
|
||||
if (!n) return null;
|
||||
const lfe = layout ? /\bLFE\b/.test(layout) : n === 6 || n === 8;
|
||||
return lfe ? `${n - 1}.1` : `${n}.0`;
|
||||
}
|
||||
|
||||
function hdrOf(fields: Record<string, string>): string[] {
|
||||
const out: string[] = [];
|
||||
const f = `${fields["HDR format"] ?? ""} ${fields.HDR_Format ?? ""}`;
|
||||
const dv = /Dolby Vision[^,]*,[^,]*,\s*Profile\s*([\d.]+)/i.exec(f);
|
||||
if (dv) out.push(`Dolby Vision P${dv[1]}`);
|
||||
else if (/Dolby Vision/i.test(f)) out.push("Dolby Vision");
|
||||
if (/HDR10\+|SMPTE ST 2094/i.test(f)) out.push("HDR10+");
|
||||
if (
|
||||
/HDR10(?!\+)|SMPTE ST 2086/i.test(f) ||
|
||||
(!out.length && /PQ|SMPTE ST 2084/.test(fields["Transfer characteristics"] ?? ""))
|
||||
)
|
||||
out.push("HDR10");
|
||||
if (/HLG/i.test(`${f} ${fields["Transfer characteristics"] ?? ""}`)) out.push("HLG");
|
||||
return [...new Set(out)];
|
||||
}
|
||||
|
||||
/** 音频编码短名:MLP FBA → TrueHD,E-AC-3 → DDP,DTS XLL → DTS-HD MA;商业名带 Atmos 的加上 Atmos */
|
||||
function audioCodec(f: Record<string, string>): string | null {
|
||||
const fmt = f.Format?.trim();
|
||||
if (!fmt) return f["Commercial name"] ?? null;
|
||||
const commercial = f["Commercial name"] ?? "";
|
||||
let name = fmt;
|
||||
if (/^MLP FBA/i.test(fmt)) name = "TrueHD";
|
||||
else if (/^E-AC-3/i.test(fmt)) name = "DDP";
|
||||
else if (/^AC-3/i.test(fmt)) name = "AC-3";
|
||||
else if (/^DTS/i.test(fmt))
|
||||
name = /XLL X|DTS:X/i.test(`${fmt} ${commercial}`)
|
||||
? "DTS:X"
|
||||
: /XLL|Master Audio/i.test(`${fmt} ${commercial}`)
|
||||
? "DTS-HD MA"
|
||||
: "DTS";
|
||||
else if (/^AAC/i.test(fmt)) name = "AAC";
|
||||
else if (/^PCM/i.test(fmt)) name = "LPCM";
|
||||
if (/Atmos/i.test(`${fmt} ${commercial}`) && !/Atmos/i.test(name)) name += " Atmos";
|
||||
return name;
|
||||
}
|
||||
|
||||
/** 中文版 MediaInfo 键 → 英文键 */
|
||||
const ZH_KEYS: Record<string, string> = {
|
||||
完整名称: "Complete name",
|
||||
格式: "Format",
|
||||
文件大小: "File size",
|
||||
时长: "Duration",
|
||||
总体码率: "Overall bit rate",
|
||||
码率: "Bit rate",
|
||||
宽度: "Width",
|
||||
高度: "Height",
|
||||
帧率: "Frame rate",
|
||||
位深: "Bit depth",
|
||||
声道数: "Channel(s)",
|
||||
声道布局: "Channel layout",
|
||||
语言: "Language",
|
||||
标题: "Title",
|
||||
默认: "Default",
|
||||
强制: "Forced",
|
||||
商业名称: "Commercial name",
|
||||
"HDR 格式": "HDR format",
|
||||
传输特质: "Transfer characteristics",
|
||||
};
|
||||
|
||||
type Section = { type: "general" | "video" | "audio" | "text" | "other"; fields: Record<string, string> };
|
||||
|
||||
const SECTION_HEAD =
|
||||
/^(General|Video|Audio|Text|Menu|Chapters|Image|Other|概览|常规|概要|视频|音频|文本|字幕|菜单|章节)(?:\s*#\s*\d+)?\s*$/i;
|
||||
|
||||
function sectionType(name: string): Section["type"] {
|
||||
const n = name.toLowerCase();
|
||||
if (/^(general|概览|常规|概要)/.test(n)) return "general";
|
||||
if (/^(video|视频)/.test(n)) return "video";
|
||||
if (/^(audio|音频)/.test(n)) return "audio";
|
||||
if (/^(text|文本|字幕)/.test(n)) return "text";
|
||||
return "other";
|
||||
}
|
||||
|
||||
function parseSections(text: string): Section[] {
|
||||
const out: Section[] = [];
|
||||
let cur: Section | null = null;
|
||||
for (const raw of text.replace(/\r/g, "").split("\n")) {
|
||||
const line = raw.trim();
|
||||
if (!line) continue;
|
||||
const head = SECTION_HEAD.exec(line);
|
||||
if (head) {
|
||||
cur = { type: sectionType(head[1]!), fields: {} };
|
||||
out.push(cur);
|
||||
continue;
|
||||
}
|
||||
if (!cur) continue;
|
||||
const m = /^(.+?)\s*[:：]\s(.*)$/.exec(line) ?? /^(.+?)\s{2,}[:：]\s*(.*)$/.exec(line);
|
||||
if (!m) continue;
|
||||
const key = ZH_KEYS[m[1]!.trim()] ?? m[1]!.trim();
|
||||
if (!(key in cur.fields)) cur.fields[key] = m[2]!.trim();
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
function baseName(p: string | undefined): string | null {
|
||||
return p ? (p.split(/[\\/]/).pop() ?? p).trim() || null : null;
|
||||
}
|
||||
|
||||
function parseMediaInfoText(text: string): MediaSummary | null {
|
||||
const sections = parseSections(text);
|
||||
const g = sections.find((s) => s.type === "general")?.fields;
|
||||
const videos = sections.filter((s) => s.type === "video");
|
||||
if (!g && !videos.length) return null;
|
||||
const general: MediaGeneral = {
|
||||
fileName: baseName(g?.["Complete name"] ?? g?.["Movie name"]),
|
||||
container: g?.Format ?? null,
|
||||
duration: g?.Duration ?? null,
|
||||
overallBitrate: rate(g?.["Overall bit rate"]),
|
||||
size: g?.["File size"] ?? null,
|
||||
};
|
||||
const video = videos.map(({ fields: f }): MediaVideoTrack => {
|
||||
const w = num(f.Width);
|
||||
const h = num(f.Height);
|
||||
return {
|
||||
codec: f.Format ?? null,
|
||||
resolution: w && h ? `${w}×${h}` : null,
|
||||
frameRate: frameRate(f["Frame rate"] ?? f["Original frame rate"]),
|
||||
bitrate: rate(f["Bit rate"] ?? f["Nominal bit rate"]),
|
||||
bitDepth: num(f["Bit depth"]) ? `${num(f["Bit depth"])} bit` : null,
|
||||
hdr: hdrOf(f),
|
||||
};
|
||||
});
|
||||
const audio = sections
|
||||
.filter((s) => s.type === "audio")
|
||||
.map(
|
||||
({ fields: f }): MediaAudioTrack => ({
|
||||
language: language(f.Language),
|
||||
codec: audioCodec(f),
|
||||
channels: channels(f["Channel(s)"], f["Channel layout"] ?? f["Channel positions"]),
|
||||
bitrate: rate(f["Bit rate"]),
|
||||
title: f.Title ?? null,
|
||||
isDefault: yesNo(f.Default),
|
||||
}),
|
||||
);
|
||||
const subtitles = sections
|
||||
.filter((s) => s.type === "text")
|
||||
.map(
|
||||
({ fields: f }): MediaTextTrack => ({
|
||||
language: language(f.Language),
|
||||
format: f.Format ?? null,
|
||||
title: f.Title ?? null,
|
||||
isDefault: yesNo(f.Default),
|
||||
forced: yesNo(f.Forced),
|
||||
}),
|
||||
);
|
||||
return { kind: "mediainfo", general, video, audio, subtitles };
|
||||
}
|
||||
|
||||
// ---------------- BDInfo ----------------
|
||||
|
||||
function bdVideo(codec: string, desc: string, bitrate: string | null): MediaVideoTrack {
|
||||
const parts = desc.split(/\s+\/\s+/).map((s) => s.trim());
|
||||
const res = parts.find((p) => /^\d{3,4}[pi]$/i.test(p)) ?? null;
|
||||
const hdr: string[] = [];
|
||||
for (const p of parts) {
|
||||
if (/^Dolby Vision/i.test(p)) hdr.push(p.replace(/^Dolby Vision\s*/i, "Dolby Vision ").trim());
|
||||
else if (/^HDR10\+?$/i.test(p) || /^HLG$/i.test(p)) hdr.push(p.toUpperCase());
|
||||
}
|
||||
return {
|
||||
codec: codec.replace(/\s*Video$/i, "").trim() || null,
|
||||
resolution: res,
|
||||
frameRate: frameRate(parts.find((p) => /fps$/i.test(p))),
|
||||
bitrate: bitrate ?? parts.find((p) => /bps$/i.test(p)) ?? null,
|
||||
bitDepth: parts.find((p) => /^\d+ bits?$/i.test(p))?.replace(/bits?$/i, "bit") ?? null,
|
||||
hdr,
|
||||
};
|
||||
}
|
||||
|
||||
function bdAudioCodec(codec: string): string | null {
|
||||
const c = codec.replace(/\s*Audio$/i, "").trim();
|
||||
if (!c) return null;
|
||||
if (/TrueHD\/Atmos/i.test(c)) return "TrueHD Atmos";
|
||||
if (/TrueHD/i.test(c)) return "TrueHD";
|
||||
if (/Digital Plus/i.test(c)) return /Atmos/i.test(c) ? "DDP Atmos" : "DDP";
|
||||
if (/^Dolby Digital$/i.test(c)) return "AC-3";
|
||||
if (/DTS:X/i.test(c)) return "DTS:X";
|
||||
if (/DTS-HD Master/i.test(c)) return "DTS-HD MA";
|
||||
return c;
|
||||
}
|
||||
|
||||
function bdAudio(codec: string, lang: string | null, desc: string, bitrate: string | null): MediaAudioTrack {
|
||||
const parts = desc.split(/\s+\/\s+/).map((s) => s.trim());
|
||||
return {
|
||||
language: language(lang),
|
||||
codec: bdAudioCodec(codec),
|
||||
channels: parts.find((p) => /^\d\.\d(?:\s|$|-)/.test(p))?.slice(0, 3) ?? null,
|
||||
bitrate: bitrate ?? parts.find((p) => /kbps$/i.test(p)) ?? null,
|
||||
title: null,
|
||||
isDefault: null,
|
||||
};
|
||||
}
|
||||
|
||||
/** 「61,812,345,678 bytes」→「57.57 GiB」;认不出原样返回 */
|
||||
function bdSize(v: string | null): string | null {
|
||||
const m = v ? /^([\d,]+)\s*bytes$/i.exec(v.trim()) : null;
|
||||
if (!m) return v;
|
||||
const n = Number(m[1]!.replace(/,/g, ""));
|
||||
const units = ["B", "KiB", "MiB", "GiB", "TiB"];
|
||||
let x = n;
|
||||
let i = 0;
|
||||
while (x >= 1024 && i < units.length - 1) {
|
||||
x /= 1024;
|
||||
i++;
|
||||
}
|
||||
return `${i ? x.toFixed(2) : x} ${units[i]}`;
|
||||
}
|
||||
|
||||
/** 「2:22:48.521」→「2 h 22 min」,与 MediaInfo 写法一致 */
|
||||
function bdDuration(v: string | null): string | null {
|
||||
const m = v ? /^(\d+):(\d{2}):(\d{2})/.exec(v.trim()) : null;
|
||||
if (!m) return v;
|
||||
const h = Number(m[1]);
|
||||
const min = Number(m[2]);
|
||||
return h ? `${h} h ${min} min` : `${min} min ${Number(m[3])} s`;
|
||||
}
|
||||
|
||||
function parseBdInfoText(text: string): MediaSummary | null {
|
||||
const lines = text.replace(/\r/g, "").split("\n");
|
||||
const kv = (label: RegExp) => {
|
||||
for (const l of lines) {
|
||||
const m = new RegExp(`^\\s*${label.source}\\s*:\\s*(.+)$`, "i").exec(l);
|
||||
if (m) return m[1]!.trim();
|
||||
}
|
||||
return null;
|
||||
};
|
||||
const general: MediaGeneral = {
|
||||
fileName: kv(/(?:Disc Title|Disc Label)/),
|
||||
container: "Blu-ray",
|
||||
duration: bdDuration(kv(/Length/)),
|
||||
overallBitrate: kv(/Total Bitrate/),
|
||||
size: bdSize(kv(/(?:Disc Size|Size)/)),
|
||||
};
|
||||
const video: MediaVideoTrack[] = [];
|
||||
const audio: MediaAudioTrack[] = [];
|
||||
const subtitles: MediaTextTrack[] = [];
|
||||
|
||||
// 简版:「Video: codec / bitrate / 2160p / …」「Audio: lang / codec / 7.1 / …」「Subtitle: lang / bitrate」
|
||||
for (const l of lines) {
|
||||
const m = /^\s*(Video|Audio|Subtitle)\s*:\s*(.+)$/i.exec(l);
|
||||
if (!m) continue;
|
||||
const parts = m[2]!.split(/\s+\/\s+/).map((s) => s.trim());
|
||||
const kind = m[1]!.toLowerCase();
|
||||
if (kind === "video") video.push(bdVideo(parts[0] ?? "", parts.slice(1).join(" / "), parts[1] ?? null));
|
||||
else if (kind === "audio") audio.push(bdAudio(parts[1] ?? "", parts[0] ?? null, parts.slice(2).join(" / "), null));
|
||||
else subtitles.push({ language: language(parts[0]), format: "PGS", title: null, isDefault: null, forced: null });
|
||||
}
|
||||
|
||||
// 完整报告:VIDEO: / AUDIO: / SUBTITLES: 下的表格,列之间两个以上空格
|
||||
if (!video.length && !audio.length) {
|
||||
let section: "video" | "audio" | "subs" | null = null;
|
||||
for (const l of lines) {
|
||||
const head =
|
||||
/^\s*(VIDEO|AUDIO|SUBTITLES|FILES|CHAPTERS|STREAM DIAGNOSTICS|PLAYLIST REPORT|DISC INFO)\s*:\s*$/.exec(l);
|
||||
if (head) {
|
||||
section =
|
||||
head[1] === "VIDEO" ? "video" : head[1] === "AUDIO" ? "audio" : head[1] === "SUBTITLES" ? "subs" : null;
|
||||
continue;
|
||||
}
|
||||
if (!section || !l.trim() || /^\s*(Codec|-----)/.test(l)) continue;
|
||||
const cols = l.trim().split(/\s{2,}/);
|
||||
if (section === "video") video.push(bdVideo(cols[0] ?? "", cols[2] ?? "", cols[1] ?? null));
|
||||
else if (section === "audio") audio.push(bdAudio(cols[0] ?? "", cols[1] ?? null, cols[3] ?? "", cols[2] ?? null));
|
||||
else
|
||||
subtitles.push({
|
||||
language: language(cols[1]),
|
||||
format: /Presentation Graphics/i.test(cols[0] ?? "") ? "PGS" : (cols[0] ?? null),
|
||||
title: null,
|
||||
isDefault: null,
|
||||
forced: null,
|
||||
});
|
||||
}
|
||||
}
|
||||
if (!video.length && !audio.length) return null;
|
||||
return { kind: "bdinfo", general, video, audio, subtitles };
|
||||
}
|
||||
|
||||
const isBdInfo = (t: string) =>
|
||||
/(^|\n)\s*(DISC INFO:|Disc Title\s*:|Disc Label\s*:|BDInfo\s*:|PLAYLIST REPORT:)/i.test(t);
|
||||
|
||||
/** MediaInfo 或 BDInfo 原文 → 概要;两种都认不出返回 null(前端只显示原文) */
|
||||
export function parseMediaSummary(text: string | null | undefined): MediaSummary | null {
|
||||
if (!text?.trim()) return null;
|
||||
if (isBdInfo(text)) return parseBdInfoText(text) ?? parseMediaInfoText(text);
|
||||
return parseMediaInfoText(text);
|
||||
}
|
||||
Reference in new issue
Block a user