dsh-plugin-lookatstudy 0.11.0 → 0.12.0

package/lib/docx-parser-BhyqPImb.mjs

@@ -0,0 +1,55 @@
+import { n as readZipText, t as readZip } from "./zip-reader-KnRrq0av.mjs";
+//#region src/vendor/docx-parser.ts
+function decodeEntities(s) {
+	return s.replace(/<[^>]+>/g, "").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, "\"").replace(/&apos;/g, "'").replace(/&amp;/g, "&");
+}
+function parseParagraphs(documentXml) {
+	const paras = [];
+	const pRe = /<w:p\b[^>]*\/>|<w:p\b[^>]*>[\s\S]*?<\/w:p>/g;
+	let pm;
+	while ((pm = pRe.exec(documentXml)) !== null) {
+		const p = pm[0];
+		const style = p.match(/<w:pStyle w:val="([^"]+)"/i)?.[1] ?? "";
+		let text = "";
+		const tRe = /<w:t(?:\s[^>]*)?>([\s\S]*?)<\/w:t>|<w:tab\b[^>]*\/>|<w:br\b[^>]*\/>/g;
+		let tm;
+		while ((tm = tRe.exec(p)) !== null) if (tm[1] !== void 0) text += decodeEntities(tm[1]);
+		else if (/w:tab/.test(tm[0])) text += "	";
+		else text += "\n";
+		paras.push({
+			style: style.trim(),
+			text
+		});
+	}
+	return paras;
+}
+function headingLevel(style) {
+	const s = style.toLowerCase().replace(/\s+/g, "");
+	const m = s.match(/heading(\d)/) ?? s.match(/titre(\d)/) ?? s.match(/berschrift(\d)/);
+	if (m) return Math.min(6, Math.max(1, Number(m[1])));
+	if (/^(title|titre)$/.test(s)) return 1;
+	return 0;
+}
+/** .docx → markdown。Word Heading 样式保真为 # 级标题,代码样式段落围栏。 */
+function parseDocx(buf) {
+	const entries = readZip(buf);
+	const doc = readZipText(entries, "word/document.xml");
+	if (!doc) throw new Error("docx 结构异常:缺 word/document.xml");
+	const lines = [];
+	for (const para of parseParagraphs(doc)) {
+		const t = para.text.trim();
+		if (!t) {
+			lines.push("");
+			continue;
+		}
+		const lvl = headingLevel(para.style);
+		if (lvl > 0) lines.push("", "#".repeat(lvl) + " " + t, "");
+		else if (/code|sourcecode/i.test(para.style)) lines.push("```", t, "```", "");
+		else lines.push(t, "");
+	}
+	const md = lines.join("\n").replace(/\n{3,}/g, "\n\n").trim();
+	if (!md) throw new Error("docx 里没有可识别的文本");
+	return md;
+}
+//#endregion
+export { parseDocx };

package/lib/epub-parser-oH96guBW.mjs

@@ -0,0 +1,306 @@
+import { n as htmlToMarkdown } from "./html-article-Da8ksU0i.mjs";
+import { n as readZipText, t as readZip } from "./zip-reader-KnRrq0av.mjs";
+//#region src/vendor/epub-parser.ts
+/** 提取某标签的全部出现(OPF 的属性顺序不定,先抓整标签再逐个提属性)。 */
+function tags(xml, tagName) {
+	const out = [];
+	const re = new RegExp(`<${tagName}\\b[^>]*>`, "g");
+	let m;
+	while ((m = re.exec(xml)) !== null) {
+		const raw = m[0];
+		const attrs = {};
+		const attrRe = /([\w:-]+)\s*=\s*"([^"]*)"/g;
+		let a;
+		while ((a = attrRe.exec(raw)) !== null) attrs[a[1]] = a[2];
+		out.push({
+			raw,
+			attrs
+		});
+	}
+	return out;
+}
+function firstTagText(xml, tagName) {
+	return xml.match(new RegExp(`<${tagName}[^>]*>([^<]*)</${tagName}>`, "i"))?.[1]?.trim() ?? "";
+}
+/** zip 内路径归一:posix 分隔 + href 相对 OPF 目录解析。 */
+function resolveZipPath(opfPath, href) {
+	const cleanHref = decodeURIComponent(href.split("#")[0] ?? href).replace(/\\/g, "/");
+	if (!opfPath.includes("/")) return cleanHref.replace(/^\.\//, "");
+	const parts = `${opfPath.slice(0, opfPath.lastIndexOf("/"))}/${cleanHref}`.split("/");
+	const resolved = [];
+	for (const p of parts) if (p === "..") resolved.pop();
+	else if (p !== "." && p !== "") resolved.push(p);
+	return resolved.join("/");
+}
+/** EPUB3 nav.xhtml 或 EPUB2 toc.ncx → href(去 fragment)→ 章节标题。 */
+function parseTocLabels(tocXml, isNcx) {
+	const labels = /* @__PURE__ */ new Map();
+	if (isNcx) {
+		const blockRe = /<navPoint\b[\s\S]*?<\/navPoint>/g;
+		let m;
+		while ((m = blockRe.exec(tocXml)) !== null) {
+			const label = firstTagText(m[0], "text");
+			const src = m[0].match(/<content[^>]+src="([^"]+)"/i)?.[1];
+			if (label && src) labels.set(decodeURIComponent(src.split("#")[0].replace(/\\/g, "/")), label);
+		}
+	} else {
+		const aRe = /<a\b[^>]*href="([^"]+)"[^>]*>([\s\S]*?)<\/a>/gi;
+		let m;
+		while ((m = aRe.exec(tocXml)) !== null) {
+			const label = (m[2] ?? "").replace(/<[^>]+>/g, "").trim();
+			if (label) labels.set(decodeURIComponent((m[1] ?? "").split("#")[0].replace(/\\/g, "/")), label);
+		}
+	}
+	return labels;
+}
+function sanitizeFileName(title) {
+	return (title || "chapter").replace(/[\\/:*?"<>|#\s]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 50) || "chapter";
+}
+/** 文件内章标记。两类形态:heading 行(# 前缀)与**裸行**(Gutenberg 常见——章号在
+* 原书 html 里是段落不是标题,转 markdown 后成裸文本 "CHAPTER XX.")。
+* 裸行限行长(≤60)防正文中引用句误切;罗马数字/裸序号低置信需序列验证。 */
+const CH_HEADING = /^(?:#{1,3}\s*)?(?:CHAPTER|Chapter|chap\.)\s+[IVXLC0-9]{1,7}\b[.:)]?\s*(.*)$/;
+const LETTER_HEADING = /^(?:#{1,3}\s*)?(?:Letter|LETTER)\s+\d{1,3}\b[.:)]?\s*(.*)$/;
+const ROMAN_HEADING = /^(?:#{1,3}\s*)?(?:[—[]\s*)?([IVXLC]{1,7})\s*[—\]]?\.?\s*$/;
+const NUM_HEADING = /^(?:#{1,3}\s*)?[—[]?\s*(\d{1,3})\s*[—\]]?\.?\s*$/;
+const romanValue = (s) => {
+	const vals = {
+		I: 1,
+		V: 5,
+		X: 10,
+		L: 50,
+		C: 100
+	};
+	let v = 0;
+	for (let i = 0; i < s.length; i++) {
+		const cur = vals[s[i]] ?? 0;
+		const next = vals[s[i + 1]] ?? 0;
+		v += cur < next ? -cur : cur;
+	}
+	return v;
+};
+/**
+* 文件内按章标记切分。返回 null = 无高置信切分(保持单章)。
+* 置信规则:CHAPTER/Letter 标记 ≥2 即切;罗马数字/裸序号须形成连续递增
+* 序列(≥3 个)才切——正文中孤立的 "I" / "2" 不当章号。
+* 首标记前的引言/卷头(<800 字符)并入首章;够长则独立成"引言"章。
+*/
+function splitChaptersInBody(body, opts) {
+	const lines = body.split("\n");
+	const marks = [];
+	for (let i = 0; i < lines.length; i++) {
+		const ln = lines[i].trimEnd();
+		if (!ln || ln.length > 60) continue;
+		const strip = (s) => s.replace(/^#{1,3}\s*/, "");
+		let m = ln.match(CH_HEADING);
+		if (m) {
+			marks.push({
+				line: i,
+				title: strip(ln),
+				kind: "ch",
+				seq: 0
+			});
+			continue;
+		}
+		m = ln.match(LETTER_HEADING);
+		if (m) {
+			marks.push({
+				line: i,
+				title: strip(ln),
+				kind: "letter",
+				seq: 0
+			});
+			continue;
+		}
+		m = ln.match(ROMAN_HEADING);
+		if (m && (ln.startsWith("#") || /[—[]/.test(ln))) {
+			marks.push({
+				line: i,
+				title: strip(ln),
+				kind: "roman",
+				seq: romanValue(m[1])
+			});
+			continue;
+		}
+		m = ln.match(NUM_HEADING);
+		if (m && (ln.startsWith("#") || /[—[]/.test(ln))) {
+			marks.push({
+				line: i,
+				title: strip(ln),
+				kind: "num",
+				seq: parseInt(m[1], 10)
+			});
+			continue;
+		}
+	}
+	const minHigh = opts?.minHighConf ?? 2;
+	const highConf = marks.filter((mk) => mk.kind === "ch" || mk.kind === "letter");
+	let splitMarks;
+	if (highConf.length >= minHigh) splitMarks = highConf;
+	else {
+		splitMarks = [];
+		for (const kind of ["roman", "num"]) {
+			const seq = marks.filter((mk) => mk.kind === kind);
+			const chain = [];
+			let expect = 0;
+			for (const mk of seq) if (expect === 0) {
+				if (mk.seq === 1 || mk.seq === 2) {
+					chain.push(mk);
+					expect = mk.seq + 1;
+				}
+			} else if (mk.seq === expect) {
+				chain.push(mk);
+				expect++;
+			}
+			if (chain.length >= 3) splitMarks.push(...chain);
+		}
+		if (splitMarks.length < 3) return null;
+	}
+	if (splitMarks.length < 1) return null;
+	const out = [];
+	const preamble = lines.slice(0, splitMarks[0].line).join("\n").trim();
+	for (let k = 0; k < splitMarks.length; k++) {
+		const start = splitMarks[k].line;
+		const end = k + 1 < splitMarks.length ? splitMarks[k + 1].line : lines.length;
+		let content = lines.slice(start + 1, end).join("\n").trim();
+		if (k === 0 && preamble && preamble.length < 800) content = `${preamble}\n\n${content}`.trim();
+		out.push({
+			title: splitMarks[k].title,
+			content
+		});
+	}
+	if (preamble && preamble.length >= 800) out.unshift({
+		title: "引言",
+		content: preamble
+	});
+	return out.filter((c) => c.content.replace(/\s/g, "").length > 0);
+}
+/**
+* 清理电子书正文噪声(Gutenberg 头/尾块、1894 插图版装饰行)。
+* 只删确定性标志,不猜内容:
+* - 头:"The Project Gutenberg eBook of…" 段落(heading 形态)
+* - 尾:"THE FULL PROJECT GUTENBERG™ LICENSE" 起截断到文件尾;"End of the Project Gutenberg" 行起同理
+* - 行内:`[ _Copyright 1894 by …_ ]` 插图版装饰片段
+*/
+function sanitizeEpubBody(body) {
+	let md = body;
+	const lines0 = md.split("\n");
+	for (let i = 0; i < lines0.length; i++) {
+		const ln = lines0[i].trim();
+		if (ln.length <= 100 && (/(?:THE\s+)?FULL PROJECT GUTENBERG/i.test(ln) || /END OF THE PROJECT GUTENBERG/i.test(ln) || ln === "START: FULL LICENSE" || ln === "START: THE FULL LICENSE")) {
+			md = lines0.slice(0, i).join("\n");
+			break;
+		}
+	}
+	md = md.replace(/^#{1,3}\s*(?:The Project Gutenberg eBook[^*\n]*|Project Gutenberg[^*\n]*)\n[\s\S]*?\n(?:\n|$)/im, "");
+	md = md.replace(/^#{1,3}[ \t]*$/gm, "");
+	md = md.replace(/\\?\[?\s*_?Copyright 1[6-9]\d\d[^_\n]*_?\s*\\?]?/g, "");
+	return md.trim();
+}
+function parseEpub(buf) {
+	const entries = readZip(buf);
+	const read = (p) => readZipText(entries, p);
+	const opfPath = read("META-INF/container.xml").match(/full-path="([^"]+)"/i)?.[1];
+	if (!opfPath) throw new Error("epub 结构异常:找不到 container.xml 里的 OPF 路径");
+	const opf = read(opfPath);
+	if (!opf) throw new Error(`epub 结构异常:OPF 文件缺失(${opfPath})`);
+	const bookTitle = firstTagText(opf, "dc:title") || "未命名电子书";
+	const manifest = /* @__PURE__ */ new Map();
+	for (const t of tags(opf, "item")) manifest.set(t.attrs["id"] ?? "", {
+		href: t.attrs["href"] ?? "",
+		mediaType: (t.attrs["media-type"] ?? "").toLowerCase(),
+		properties: t.attrs["properties"] ?? ""
+	});
+	const spineIds = tags(opf, "itemref").map((t) => t.attrs["idref"] ?? "").filter(Boolean);
+	const spineTocId = opf.match(/<spine\b[^>]*\btoc="([^"]+)"/i)?.[1];
+	let tocLabels = /* @__PURE__ */ new Map();
+	const navItem = [...manifest.values()].find((it) => it.properties.split(/\s+/).includes("nav"));
+	if (navItem?.href) tocLabels = parseTocLabels(read(resolveZipPath(opfPath, navItem.href)), false);
+	if (tocLabels.size === 0 && spineTocId && manifest.has(spineTocId)) tocLabels = parseTocLabels(read(resolveZipPath(opfPath, manifest.get(spineTocId).href)), true);
+	const opfDirKey = (href) => decodeURIComponent(href.split("#")[0] ?? href).replace(/\\/g, "/");
+	/** 无标题标记(不再编造"第 N 章";配对/课程设计层的锚点) */
+	const UNTITLED_MARK = "未命名章节";
+	const chapters = [];
+	let n = 0;
+	for (const id of spineIds) {
+		const item = manifest.get(id);
+		if (!item?.href) continue;
+		if (!(item.mediaType === "application/xhtml+xml" || item.mediaType === "text/html" || /\.(xhtml|html|htm)$/i.test(item.href))) continue;
+		if (item.properties.split(/\s+/).includes("nav")) continue;
+		const xhtml = read(resolveZipPath(opfPath, item.href));
+		if (!xhtml) continue;
+		const md = htmlToMarkdown(xhtml, { stripImages: true });
+		if (!md || md.replace(/[#\s>*-]/g, "").length < 8) continue;
+		n++;
+		const body0 = md.startsWith("# ") && md.includes("\n") ? md.slice(md.indexOf("\n") + 1).trim() : md;
+		const firstHeading = md.startsWith("# ") ? md.split("\n")[0].slice(2).trim() : "";
+		const fileTitle = tocLabels.get(opfDirKey(item.href)) || firstHeading || UNTITLED_MARK;
+		if (/^(contents|table of contents)$/i.test(fileTitle) && body0.replace(/\s/g, "").length < 4e3) continue;
+		if (body0.replace(/\s/g, "").length < 1500 && /project gutenberg|copyright/i.test(body0)) continue;
+		if (/^(版权信息|版权|版权页|著作权|版權)/.test(fileTitle)) continue;
+		if ((body0.match(/^(书名|作者|译者|编者|丛书|出版社|出版时间|出版年|ISBN|版次|定价|装帧|字数|品牌|出品方)[：:]/gm) ?? []).length >= 3 && body0.replace(/\s/g, "").length < 2500) continue;
+		if (/^目\s*录$/.test(fileTitle)) {
+			const lines = body0.split("\n").map((l) => l.trim()).filter(Boolean);
+			if (lines.length >= 3 && lines.filter((l) => l.includes("](")).length / lines.length >= .6) continue;
+		}
+		const body = sanitizeEpubBody(body0);
+		if (body.replace(/\s/g, "").length < 8) continue;
+		const titleIsLicense = /FULL PROJECT GUTENBERG|LICENSE/i.test(fileTitle);
+		const split = splitChaptersInBody(body, titleIsLicense ? { minHighConf: 1 } : void 0);
+		if (split && split.length >= (titleIsLicense ? 1 : 2)) {
+			for (const seg of split) {
+				chapters.push({
+					path: `chapters/${String(n).padStart(2, "0")}-${sanitizeFileName(seg.title)}.md`,
+					title: seg.title,
+					markdown: `# ${seg.title}\n\n${seg.content}`
+				});
+				n++;
+			}
+			n--;
+			continue;
+		}
+		if (titleIsLicense) {
+			if (body.replace(/\s/g, "").length > 2e3) chapters.push({
+				path: `chapters/${String(n).padStart(2, "0")}-appendix.md`,
+				title: "附录",
+				markdown: `# 附录\n\n${body}`
+			});
+			continue;
+		}
+		chapters.push({
+			path: `chapters/${String(n).padStart(2, "0")}-${sanitizeFileName(fileTitle)}.md`,
+			title: fileTitle,
+			markdown: `# ${fileTitle}\n\n${body}`
+		});
+	}
+	const absorbed = /* @__PURE__ */ new Set();
+	for (let i = 0; i + 1 < chapters.length; i++) {
+		const stub = chapters[i];
+		const next = chapters[i + 1];
+		if (absorbed.has(stub.path)) continue;
+		const stubBody = stub.markdown.includes("\n") ? stub.markdown.slice(stub.markdown.indexOf("\n") + 1).trim() : "";
+		if (stub.title === UNTITLED_MARK || stub.title === "附录") continue;
+		if (stubBody.replace(/\s/g, "").length > 300) continue;
+		if (next.title !== UNTITLED_MARK) continue;
+		const nextBody = next.markdown.includes("\n") ? next.markdown.slice(next.markdown.indexOf("\n") + 1).trim() : next.markdown;
+		chapters[i] = {
+			path: stub.path,
+			title: stub.title,
+			markdown: `# ${stub.title}\n\n${stubBody}\n\n${nextBody}`
+		};
+		absorbed.add(stub.path);
+		chapters.splice(i + 1, 1);
+		i--;
+	}
+	if (chapters.length === 0) throw new Error("epub 里没有可识别的章节文本");
+	return {
+		title: bookTitle,
+		chapters
+	};
+}
+/** 文件夹导入路径用:整本书压平成一个 markdown(全部 H1 降为 H2,给结构设计当 anchor 拆章)。 */
+function parseEpubFlat(buf) {
+	return parseEpub(buf).chapters.map((c) => c.markdown.replace(/^# /gm, "## ")).join("\n\n");
+}
+//#endregion
+export { parseEpubFlat };