dsh-plugin-lookatstudy 0.25.1 → 0.25.3

package/lib/index.mjs

@@ -1130,6 +1130,21 @@ function normalizeSpeechText(md) {
 	s = s.replace(/`[^`\n]*`/g, "");
 	s = s.replace(/!\[[^\]]*\]\([^)\n]*\)/g, "");
 	s = s.replace(/\[([^\]]+)\]\([^)\n]*\)/g, "$1");
+	s = s.replace(/\[\]\([^)\n]*\)/g, "");
+	s = s.replace(/<br\s*\/?>/gi, "\n");
+	s = s.replace(/<https?:\/\/[^>\s]+>/g, "");
+	s = s.replace(/https?:\/\/[^\s)\]]+/g, "");
+	s = s.replace(/<(script|style)\b[^>]*>[\s\S]*?<\/\1>/gi, "\n");
+	s = s.replace(/<[^>\n]+>/g, "");
+	s = s.replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, "\"").replace(/&#0?39;/g, "'").replace(/&nbsp;/g, " ").replace(/&#(\d+);/g, (_m, d) => {
+		try {
+			return String.fromCodePoint(Number(d));
+		} catch {
+			return "";
+		}
+	}).replace(/&[a-z]+;/gi, "");
+	s = s.replace(/\[([^\]]+)\]\[[^\]]*\]/g, "$1");
+	s = s.replace(/\[\^[^\]]*\]/g, "");
 	s = s.replace(/^[ \t]{0,3}#{1,6}[ \t]+/gm, "");
 	s = s.replace(/^[ \t]*(?:[-*+]|\d+\.)[ \t]+/gm, "");
 	s = s.replace(/^[ \t]*>[ \t]?/gm, "");
@@ -3122,6 +3137,7 @@ function workbenchState(state, now) {
 			speechText: normalizeSpeechText(normalizeMathNotation(ref.lesson.body)),
 			...ref.lesson.translation !== void 0 ? {
 				translationHtml: renderMarkdown(normalizeMathNotation(ref.lesson.translation)),
+				translationSpeechText: normalizeSpeechText(normalizeMathNotation(ref.lesson.translation)),
 				translationLang: ref.lesson.translationLang ?? ""
 			} : {}
 		};
@@ -6582,12 +6598,19 @@ function validateDesign(design, validFiles) {
 	};
 }
 /**
-* Heading similarity (word-level overlap) — the translation misalignment
-* guard. Ported verbatim from upstream import-pipeline.ts (their audit F16,
-* 2026-09-13): ordinal alignment assumes both heading lists share structure;
-* a machine-translated file with a dropped header shifts every ordinal, and
-* A's translation would silently land on B's node — overlap too low skips the
-* segment (宁缺毋错). Empty titles cannot be checked and pass conservatively.
+* Heading similarity — the translation misalignment guard. Ported from
+* upstream import-pipeline.ts (their audit F16, 2026-09-13): ordinal
+* alignment assumes both heading lists share structure; a machine-translated
+* file with a dropped header shifts every ordinal, and A's translation would
+* silently land on B's node — overlap too low skips the segment (宁缺毋错).
+* Empty titles cannot be checked and pass conservatively.
+*
+* 0.25.2 (owner real-import catch): upstream's word overlap normalizes on
+* whitespace, and a CJK title is ONE token — paraphrased Chinese headings
+* (课程总览与环境搭建指南 vs 课程总览) could never overlap and the guard
+* skipped every non-verbatim segment (38/110 paired on the owner's import;
+* the 38 were exact matches). CJK strings now compare by character
+* containment first, then character-bigram overlap ≥ 0.3.
 */
 function headingsSimilar(a, b) {
 	const norm = (s) => s.toLowerCase().replace(/[^\p{L}\p{N}]+/gu, " ").trim().split(/\s+/).filter(Boolean);
@@ -6595,6 +6618,24 @@ function headingsSimilar(a, b) {
 	const B = norm(b);
 	if (A.length === 0 || B.length === 0) return true;
 	if (A.join(" ") === B.join(" ")) return true;
+	const han = (s) => s.toLowerCase().replace(/[^\p{Script=Han}\p{N}]/gu, "");
+	const ca = han(a);
+	const cb = han(b);
+	if (ca.length >= 2 || cb.length >= 2) {
+		if (ca === "" || cb === "") return true;
+		if (ca.includes(cb) || cb.includes(ca)) return true;
+		const bigramsOf = (s) => {
+			const out = /* @__PURE__ */ new Set();
+			const chars = Array.from(s);
+			for (let i = 0; i + 1 < chars.length; i++) out.add(chars[i] + chars[i + 1]);
+			return out;
+		};
+		const ba = bigramsOf(ca);
+		const bb = bigramsOf(cb);
+		let hit = 0;
+		for (const g of ba) if (bb.has(g)) hit++;
+		return hit / Math.max(1, ba.size) >= .3;
+	}
 	const setB = new Set(B);
 	return A.filter((w) => setB.has(w)).length / A.length >= .3;
 }
@@ -6675,7 +6716,8 @@ function buildCourseFromDesign(courseTitle, validated, contents, translations) {
 				let tIndex = titleIndex;
 				let keep = true;
 				if (lesson.anchor === null || titleIndex === -1) tIndex = -1;
-				else if (titleIndex >= tHeadings.length || !headingsSimilar(lesson.title, tHeadings[titleIndex].title)) keep = false;
+				else if (titleIndex >= tHeadings.length) keep = false;
+				else if (!(tHeadings.length === headings.length && tHeadings.every((h, i) => h.level === headings[i].level)) && !headingsSimilar(lesson.title, tHeadings[titleIndex].title)) keep = false;
 				if (keep) translation = replaceImagesByPosition(sliceLessonBody(pair.content, tHeadings, tIndex, isFirst), originalImagesOf(body));
 			}
 			return {
@@ -7739,7 +7781,7 @@ ${p.text}`));
 					translations: {
 						type: "integer",
 						required: true,
-						description: "Lessons that landed with a paired translation (the interface-language mirror existed and its anchor matched) — their teach tab offers 原文/对照/译文."
+						description: "Lessons that landed with a paired translation (the interface-language mirror existed and its anchor matched) — their teach tab offers 原文/译文."
 					},
 					created: {
 						type: "boolean",
@@ -7754,7 +7796,7 @@ ${p.text}`));
 			},
 			render: (_args, value) => [{
 				type: "text",
-				text: value.created === false ? `“${value.title}” (id ${value.courseId}) was ALREADY imported — the design did not create a new course. Continue with the existing course, or delete it first (study_delete_course) to rebuild from this design.` : `Imported designed course “${value.title}” (${value.sections} sections, ${value.lessons} lessons${value.droppedLessons > 0 ? `, ${value.droppedLessons} hallucinated lesson(s) dropped` : ""})${typeof value.translations === "number" && value.translations > 0 ? ` — ${value.translations} lesson(s) carry the learner's-language translation (原文/对照/译文 in the teach tab)` : ""}${typeof value.languageTarget === "string" ? ` — language course (${value.languageTarget}): keep reading passages, example sentences, and quiz language material in the target language` : ""}. First lesson: “${value.firstLessonTitle}” (id ${value.firstLessonId}). Present the course map to the learner.`
+				text: value.created === false ? `“${value.title}” (id ${value.courseId}) was ALREADY imported — the design did not create a new course. Continue with the existing course, or delete it first (study_delete_course) to rebuild from this design.` : `Imported designed course “${value.title}” (${value.sections} sections, ${value.lessons} lessons${value.droppedLessons > 0 ? `, ${value.droppedLessons} hallucinated lesson(s) dropped` : ""})${typeof value.translations === "number" && value.translations > 0 ? ` — ${value.translations} lesson(s) carry the learner's-language translation (原文/译文 in the teach tab)` : ""}${typeof value.languageTarget === "string" ? ` — language course (${value.languageTarget}): keep reading passages, example sentences, and quiz language material in the target language` : ""}. First lesson: “${value.firstLessonTitle}” (id ${value.firstLessonId}). Present the course map to the learner.`
 			}]
 		},
 		async execute(args, exec) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-plugin-lookatstudy",
-  "version": "0.25.1",
+  "version": "0.25.3",
   "packageManager": "pnpm@11.7.0",
   "description": "Turn any markdown, local folder, or GitHub learning repo into a guided course inside DeepSeek Harness: gated skill-tree progression, BKT mastery tracking, SM-2 spaced repetition.",
   "type": "module",