dsh-plugin-lookatstudy 0.22.0 → 0.22.1

package/lib/index.mjs

@@ -4967,22 +4967,105 @@ function decodeHexString(s) {
 	for (const b of bytes) if (b >= 32 && b < 127) out += String.fromCharCode(b);
 	return out;
 }
-/** Pull text out of one decoded content stream. */
+/** Pull text out of one decoded content stream. Linear hand-rolled scan —
+*  the original regex alternation backtracked catastrophically on real
+*  papers' content streams (an unclosed `[` made the lazy TJ arm rescan
+*  megabytes; found 2026-09-14 when the livetest refresh wedged the host for
+*  10+ CPU-minutes on a real arXiv PDF while even a watchdog timer could not
+*  fire — the loop blocked the event loop synchronously). */
+function readLiteral(src, i) {
+	let depth = 1;
+	let j = i + 1;
+	let raw = "";
+	while (j < src.length) {
+		const c = src[j];
+		if (c === "\\") {
+			raw += src.slice(j, j + 2);
+			j += 2;
+			continue;
+		}
+		if (c === "(") {
+			depth++;
+			raw += c;
+			j++;
+			continue;
+		}
+		if (c === ")") {
+			depth--;
+			if (depth === 0) {
+				j++;
+				break;
+			}
+			raw += c;
+			j++;
+			continue;
+		}
+		raw += c;
+		j++;
+	}
+	return {
+		text: decodePdfString(raw),
+		next: j
+	};
+}
+function readHex(src, i) {
+	const close = src.indexOf(">", i);
+	if (close === -1) return {
+		text: decodeHexString(src.slice(i + 1)),
+		next: src.length
+	};
+	return {
+		text: decodeHexString(src.slice(i + 1, close)),
+		next: close + 1
+	};
+}
+function skipWs(src, i) {
+	while (i < src.length && /\s/.test(src[i])) i++;
+	return i;
+}
+function readWord(src, i) {
+	let j = i;
+	while (j < src.length && /[A-Za-z0-9*'"]/.test(src[j])) j++;
+	return src.slice(i, j);
+}
 function textFromContentStream(content) {
 	if (!/BT[\s\S]*?ET/.test(content)) return "";
 	let out = "";
-	const re = /\(((?:\\.|[^\\()])*)\)\s*(Tj|'|")|<([0-9a-fA-F\s]+)>\s*(Tj|'|")|\[((?:\(.*?\)|<[0-9a-fA-F\s]+>|[^\]])*?)\]\s*TJ/g;
-	let m;
-	while ((m = re.exec(content)) !== null) if (m[1] !== void 0) {
-		out += decodePdfString(m[1]);
-		if (m[2] === "'" || m[2] === "\"") out += "\n";
-	} else if (m[3] !== void 0) {
-		out += decodeHexString(m[3]);
-		if (m[4] === "'" || m[4] === "\"") out += "\n";
-	} else if (m[5] !== void 0) {
-		const partRe = /\(((?:\\.|[^\\()])*)\)|<([0-9a-fA-F\s]+)>/g;
-		let pm;
-		while ((pm = partRe.exec(m[5])) !== null) out += pm[1] !== void 0 ? decodePdfString(pm[1]) : decodeHexString(pm[2] ?? "");
+	let i = 0;
+	const n = content.length;
+	while (i < n) {
+		const c = content[i];
+		if (c === "[") {
+			const parts = [];
+			let j = i + 1;
+			while (j < n && content[j] !== "]") {
+				const d = content[j];
+				if (d === "(") {
+					const r = readLiteral(content, j);
+					parts.push(r.text);
+					j = r.next;
+				} else if (d === "<") {
+					const r = readHex(content, j);
+					parts.push(r.text);
+					j = r.next;
+				} else j++;
+			}
+			if (j >= n) break;
+			if (readWord(content, skipWs(content, j + 1)) === "TJ") out += parts.join("");
+			i = j + 1;
+			continue;
+		}
+		if (c === "(" || c === "<") {
+			const r = c === "(" ? readLiteral(content, i) : readHex(content, i);
+			const op = readWord(content, skipWs(content, r.next));
+			if (op === "Tj" || op === "'" || op === "\"") {
+				out += r.text;
+				if (op !== "Tj") out += "\n";
+			}
+			i = r.next;
+			continue;
+		}
+		i++;
 	}
 	return out;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-plugin-lookatstudy",
-  "version": "0.22.0",
+  "version": "0.22.1",
   "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",