@yolk-sdk/extractors 0.1.0-canary.98

package/dist/node/worker-admission.mjs

@@ -0,0 +1,107 @@
+import { Effect } from "effect";
+import * as Schema from "effect/Schema";
+//#region src/node/worker-admission.ts
+/**
+* Worker admission. Every Node `FileExtractor` layer in a JavaScript realm (the main thread, or
+* each worker thread or `vm` context that builds the layer) shares one pool of
+* `processWorkerLimit` worker slots, however often the layer is built (a per-request
+* `Effect.provide` builds a new one each time). The pool lives on `globalThis` under a versioned
+* `Symbol.for` key and holds only plain data and plain callbacks, so duplicated copies of this
+* package (and of Effect) in one realm share it. Each layer also has its own pool of
+* `maxConcurrentWorkers` slots (at most `processWorkerLimit`), which can only lower that layer's
+* share. An extraction takes its layer's slot, then a realm slot, and waits for both until its
+* admission deadline; a slot is released only after its worker has terminated.
+*
+* Admission is first come, first served: a freed slot is handed to the longest-waiting extraction
+* whose deadline has not passed, and a new extraction takes a free slot only when nobody waits.
+*/
+/** Worker slots shared by every layer in the realm. */
+const processWorkerLimit = 4;
+const makeSlotPool = (capacity) => ({
+	capacity,
+	active: 0,
+	waiters: /* @__PURE__ */ new Set()
+});
+/** The version names the hand-off protocol above; a change to it needs a new key. */
+const processPoolKey = Symbol.for("@yolk-sdk/extractors/worker-admission/v2");
+/** Waiters are only ever added by `withSlot`, in this module or a copy of it. */
+const WaiterSet = Schema.declare((value) => value instanceof Set);
+/** The pool's shape, checked when another copy of this module may have created it. */
+const isSlotPool = Schema.is(Schema.Struct({
+	capacity: Schema.Number,
+	active: Schema.Number,
+	waiters: WaiterSet
+}));
+/** The realm-wide pool, created by the first copy of this module that asks for it. */
+const processSlotPool = () => {
+	const existing = Object.getOwnPropertyDescriptor(globalThis, processPoolKey)?.value;
+	if (isSlotPool(existing)) return existing;
+	const pool = makeSlotPool(4);
+	Object.defineProperty(globalThis, processPoolKey, {
+		value: pool,
+		configurable: false
+	});
+	return pool;
+};
+/** Running and waiting extractions in the realm-wide pool. */
+const processAdmissionSnapshot = () => {
+	const pool = processSlotPool();
+	return {
+		active: pool.active,
+		waiting: pool.waiters.size
+	};
+};
+/** Hand the slot to the first waiter that can still take it; otherwise it becomes free. */
+const releaseSlot = (pool) => {
+	for (const handOff of pool.waiters) if (handOff()) return;
+	pool.active -= 1;
+};
+/** Whether `deadline` (epoch ms) has passed. */
+const expired = (deadline) => Date.now() > deadline;
+/**
+* Run `self` holding one slot of `pool`, waiting for it until `deadline` (epoch ms) and failing
+* with `busy()` after that. The deadline is checked again whenever a slot would be taken, by a new
+* extraction or by a hand-off, so a late wake-up cannot turn an expired wait into admission.
+*
+* Taking a slot and registering its release happen without an interruption point between them,
+* so a slot is never leaked; only the wait is interruptible. A slot handed to a waiter whose
+* fiber is interrupted before it resumes is released again. The wait uses a real timer, not the
+* Effect `Clock`, so a test clock cannot stall it.
+*/
+const withSlot = (pool, deadline, busy) => (self) => Effect.uninterruptibleMask((restore) => Effect.suspend(() => {
+	let owned = false;
+	return restore(Effect.callback((resume) => {
+		if (expired(deadline)) return resume(Effect.succeed(false));
+		if (pool.active < pool.capacity && pool.waiters.size === 0) {
+			pool.active += 1;
+			owned = true;
+			return resume(Effect.succeed(true));
+		}
+		const handOff = () => {
+			stop();
+			if (expired(deadline)) {
+				queueMicrotask(() => resume(Effect.succeed(false)));
+				return false;
+			}
+			owned = true;
+			queueMicrotask(() => resume(Effect.succeed(true)));
+			return true;
+		};
+		const timer = setTimeout(() => {
+			stop();
+			resume(Effect.succeed(false));
+		}, deadline - Date.now());
+		const stop = () => {
+			clearTimeout(timer);
+			pool.waiters.delete(handOff);
+		};
+		pool.waiters.add(handOff);
+		return Effect.sync(stop);
+	})).pipe(Effect.onInterrupt(() => Effect.sync(() => {
+		if (owned) releaseSlot(pool);
+	})), Effect.flatMap((admitted) => admitted ? restore(self).pipe(Effect.ensuring(Effect.sync(() => releaseSlot(pool)))) : Effect.fail(busy())));
+}));
+//#endregion
+export { makeSlotPool, processAdmissionSnapshot, processSlotPool, processWorkerLimit, withSlot };
+
+//# sourceMappingURL=worker-admission.mjs.map

package/dist/node/worker-admission.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"worker-admission.mjs","names":[],"sources":["../../src/node/worker-admission.ts"],"sourcesContent":["import { Effect } from 'effect'\nimport * as Schema from 'effect/Schema'\n\n/**\n * Worker admission. Every Node `FileExtractor` layer in a JavaScript realm (the main thread, or\n * each worker thread or `vm` context that builds the layer) shares one pool of\n * `processWorkerLimit` worker slots, however often the layer is built (a per-request\n * `Effect.provide` builds a new one each time). The pool lives on `globalThis` under a versioned\n * `Symbol.for` key and holds only plain data and plain callbacks, so duplicated copies of this\n * package (and of Effect) in one realm share it. Each layer also has its own pool of\n * `maxConcurrentWorkers` slots (at most `processWorkerLimit`), which can only lower that layer's\n * share. An extraction takes its layer's slot, then a realm slot, and waits for both until its\n * admission deadline; a slot is released only after its worker has terminated.\n *\n * Admission is first come, first served: a freed slot is handed to the longest-waiting extraction\n * whose deadline has not passed, and a new extraction takes a free slot only when nobody waits.\n */\n\n/** Worker slots shared by every layer in the realm. */\nexport const processWorkerLimit = 4\n\n/**\n * Takes a freed slot for its waiter and returns `true`, or returns `false` when the waiter's\n * deadline has passed (it then fails with `busy`). Either way it leaves the queue. It never runs\n * the waiter's fiber: the resume is deferred to a microtask, so a waiter that finishes at once\n * cannot release (and hand off) again inside this call, and a long queue drains in constant stack.\n */\nexport type SlotHandOff = () => boolean\n\n/**\n * A counting pool of worker slots with a FIFO queue (a `Set` iterates in insertion order). While\n * anyone waits, every slot is taken: a released slot passes to a waiter without being counted\n * free.\n */\nexport type SlotPool = {\n  readonly capacity: number\n  active: number\n  readonly waiters: Set<SlotHandOff>\n}\n\nexport const makeSlotPool = (capacity: number): SlotPool => ({\n  capacity,\n  active: 0,\n  waiters: new Set()\n})\n\n/** The version names the hand-off protocol above; a change to it needs a new key. */\nconst processPoolKey = Symbol.for('@yolk-sdk/extractors/worker-admission/v2')\n\n/** Waiters are only ever added by `withSlot`, in this module or a copy of it. */\nconst WaiterSet = Schema.declare((value): value is Set<SlotHandOff> => value instanceof Set)\n\n/** The pool's shape, checked when another copy of this module may have created it. */\nconst isSlotPool = Schema.is(\n  Schema.Struct({ capacity: Schema.Number, active: Schema.Number, waiters: WaiterSet })\n)\n\n/** The realm-wide pool, created by the first copy of this module that asks for it. */\nexport const processSlotPool = (): SlotPool => {\n  const existing: unknown = Object.getOwnPropertyDescriptor(globalThis, processPoolKey)?.value\n\n  if (isSlotPool(existing)) return existing\n\n  const pool = makeSlotPool(processWorkerLimit)\n\n  Object.defineProperty(globalThis, processPoolKey, { value: pool, configurable: false })\n\n  return pool\n}\n\n/** Running and waiting extractions in the realm-wide pool. */\nexport const processAdmissionSnapshot = () => {\n  const pool = processSlotPool()\n\n  return { active: pool.active, waiting: pool.waiters.size }\n}\n\n/** Hand the slot to the first waiter that can still take it; otherwise it becomes free. */\nconst releaseSlot = (pool: SlotPool) => {\n  for (const handOff of pool.waiters) if (handOff()) return\n\n  pool.active -= 1\n}\n\n/** Whether `deadline` (epoch ms) has passed. */\nconst expired = (deadline: number) => Date.now() > deadline\n\n/**\n * Run `self` holding one slot of `pool`, waiting for it until `deadline` (epoch ms) and failing\n * with `busy()` after that. The deadline is checked again whenever a slot would be taken, by a new\n * extraction or by a hand-off, so a late wake-up cannot turn an expired wait into admission.\n *\n * Taking a slot and registering its release happen without an interruption point between them,\n * so a slot is never leaked; only the wait is interruptible. A slot handed to a waiter whose\n * fiber is interrupted before it resumes is released again. The wait uses a real timer, not the\n * Effect `Clock`, so a test clock cannot stall it.\n */\nexport const withSlot =\n  <E2>(pool: SlotPool, deadline: number, busy: () => E2) =>\n  <A, E, R>(self: Effect.Effect<A, E, R>): Effect.Effect<A, E | E2, R> =>\n    Effect.uninterruptibleMask(restore =>\n      Effect.suspend(() => {\n        // Set once this fiber owns a slot, before its release is registered.\n        let owned = false\n\n        const admission = Effect.callback<boolean>(resume => {\n          if (expired(deadline)) return resume(Effect.succeed(false))\n\n          if (pool.active < pool.capacity && pool.waiters.size === 0) {\n            pool.active += 1\n            owned = true\n\n            return resume(Effect.succeed(true))\n          }\n\n          const handOff: SlotHandOff = () => {\n            stop()\n\n            if (expired(deadline)) {\n              queueMicrotask(() => resume(Effect.succeed(false)))\n\n              return false\n            }\n\n            // The slot is this fiber's from here on. If it is interrupted before the deferred\n            // resume runs, Effect ignores that resume and `onInterrupt` passes the slot on.\n            owned = true\n            queueMicrotask(() => resume(Effect.succeed(true)))\n\n            return true\n          }\n\n          const timer = setTimeout(() => {\n            stop()\n            resume(Effect.succeed(false))\n          }, deadline - Date.now())\n\n          const stop = () => {\n            clearTimeout(timer)\n            pool.waiters.delete(handOff)\n          }\n\n          pool.waiters.add(handOff)\n\n          return Effect.sync(stop)\n        })\n\n        return restore(admission).pipe(\n          // Interrupted after a hand-off but before resuming: pass the slot on.\n          Effect.onInterrupt(() =>\n            Effect.sync(() => {\n              if (owned) releaseSlot(pool)\n            })\n          ),\n          Effect.flatMap((admitted): Effect.Effect<A, E | E2, R> =>\n            admitted\n              ? restore(self).pipe(Effect.ensuring(Effect.sync(() => releaseSlot(pool))))\n              : Effect.fail(busy())\n          )\n        )\n      })\n    )\n"],"mappings":";;;;;;;;;;;;;;;;;;AAmBA,MAAa,qBAAqB;AAqBlC,MAAa,gBAAgB,cAAgC;CAC3D;CACA,QAAQ;CACR,yBAAS,IAAI,IAAI;AACnB;;AAGA,MAAM,iBAAiB,OAAO,IAAI,0CAA0C;;AAG5E,MAAM,YAAY,OAAO,SAAS,UAAqC,iBAAiB,GAAG;;AAG3F,MAAM,aAAa,OAAO,GACxB,OAAO,OAAO;CAAE,UAAU,OAAO;CAAQ,QAAQ,OAAO;CAAQ,SAAS;AAAU,CAAC,CACtF;;AAGA,MAAa,wBAAkC;CAC7C,MAAM,WAAoB,OAAO,yBAAyB,YAAY,cAAc,GAAG;CAEvF,IAAI,WAAW,QAAQ,GAAG,OAAO;CAEjC,MAAM,OAAO,aAAA,CAA+B;CAE5C,OAAO,eAAe,YAAY,gBAAgB;EAAE,OAAO;EAAM,cAAc;CAAM,CAAC;CAEtF,OAAO;AACT;;AAGA,MAAa,iCAAiC;CAC5C,MAAM,OAAO,gBAAgB;CAE7B,OAAO;EAAE,QAAQ,KAAK;EAAQ,SAAS,KAAK,QAAQ;CAAK;AAC3D;;AAGA,MAAM,eAAe,SAAmB;CACtC,KAAK,MAAM,WAAW,KAAK,SAAS,IAAI,QAAQ,GAAG;CAEnD,KAAK,UAAU;AACjB;;AAGA,MAAM,WAAW,aAAqB,KAAK,IAAI,IAAI;;;;;;;;;;;AAYnD,MAAa,YACN,MAAgB,UAAkB,UAC7B,SACR,OAAO,qBAAoB,YACzB,OAAO,cAAc;CAEnB,IAAI,QAAQ;CA4CZ,OAAO,QA1CW,OAAO,UAAkB,WAAU;EACnD,IAAI,QAAQ,QAAQ,GAAG,OAAO,OAAO,OAAO,QAAQ,KAAK,CAAC;EAE1D,IAAI,KAAK,SAAS,KAAK,YAAY,KAAK,QAAQ,SAAS,GAAG;GAC1D,KAAK,UAAU;GACf,QAAQ;GAER,OAAO,OAAO,OAAO,QAAQ,IAAI,CAAC;EACpC;EAEA,MAAM,gBAA6B;GACjC,KAAK;GAEL,IAAI,QAAQ,QAAQ,GAAG;IACrB,qBAAqB,OAAO,OAAO,QAAQ,KAAK,CAAC,CAAC;IAElD,OAAO;GACT;GAIA,QAAQ;GACR,qBAAqB,OAAO,OAAO,QAAQ,IAAI,CAAC,CAAC;GAEjD,OAAO;EACT;EAEA,MAAM,QAAQ,iBAAiB;GAC7B,KAAK;GACL,OAAO,OAAO,QAAQ,KAAK,CAAC;EAC9B,GAAG,WAAW,KAAK,IAAI,CAAC;EAExB,MAAM,aAAa;GACjB,aAAa,KAAK;GAClB,KAAK,QAAQ,OAAO,OAAO;EAC7B;EAEA,KAAK,QAAQ,IAAI,OAAO;EAExB,OAAO,OAAO,KAAK,IAAI;CACzB,CAEuB,CAAC,EAAE,KAExB,OAAO,kBACL,OAAO,WAAW;EAChB,IAAI,OAAO,YAAY,IAAI;CAC7B,CAAC,CACH,GACA,OAAO,SAAS,aACd,WACI,QAAQ,IAAI,EAAE,KAAK,OAAO,SAAS,OAAO,WAAW,YAAY,IAAI,CAAC,CAAC,CAAC,IACxE,OAAO,KAAK,KAAK,CAAC,CACxB,CACF;AACF,CAAC,CACH"}

package/dist/node/xlsx-hyperlinks.d.mts

@@ -0,0 +1,34 @@
+import { StrippedHyperlinkTags } from "./office-archive.mjs";
+import { CellRange } from "./xlsx-range.mjs";
+
+//#region src/node/xlsx-hyperlinks.d.ts
+type XlsxHyperlink = {
+  readonly range: CellRange; /** Normalized `http:`, `https:`, or `mailto:` URL, at most 2,048 characters. */
+  readonly target: string; /** Label for cells without text, at most 1,024 characters (longer labels end in `…`). */
+  readonly display?: string;
+};
+/** Hyperlinks per worksheet name, in document order (a later link wins on overlap). */
+type XlsxHyperlinks = ReadonlyMap<string, ReadonlyArray<XlsxHyperlink>>;
+/** A sheet of the workbook and the validated worksheet part SheetJS reads for it, if any. */
+type HyperlinkSheet = {
+  readonly name: string;
+  readonly part: string | undefined;
+};
+/**
+ * Map the hyperlink tags removed from a workbook to its sheets (the same sheet list and worksheet
+ * parts SheetJS reads, from `buildSheetJsInput`) through each worksheet's relationships.
+ */
+declare const resolveXlsxHyperlinks: (parts: Readonly<Record<string, Uint8Array>>, hyperlinkTags: StrippedHyperlinkTags, sheets: ReadonlyArray<HyperlinkSheet>) => XlsxHyperlinks;
+type HyperlinkLookup = {
+  /** The winning link at a cell. Rows must be queried in ascending order. */readonly at: (row: number, column: number) => XlsxHyperlink | undefined;
+};
+/**
+ * Index links clipped to the visited range for row-major lookup without scanning every link per
+ * cell: a segment tree over columns whose nodes hold max-heaps (by document order) of the links
+ * covering them. Links enter when the sweep reaches their first row and are discarded lazily once
+ * it passes their last row. Work is O((links + cells) · log(columns) · log(links)).
+ */
+declare const makeHyperlinkLookup: (links: ReadonlyArray<XlsxHyperlink>, visited: CellRange) => HyperlinkLookup;
+//#endregion
+export { HyperlinkLookup, HyperlinkSheet, XlsxHyperlink, XlsxHyperlinks, makeHyperlinkLookup, resolveXlsxHyperlinks };
+//# sourceMappingURL=xlsx-hyperlinks.d.mts.map

package/dist/node/xlsx-hyperlinks.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-hyperlinks.d.mts","names":[],"sources":["../../src/node/xlsx-hyperlinks.ts"],"mappings":";;;;KAQY,aAAA;EAAA,SACD,KAAA,EAAO,SAAS,EADf;EAAA,SAGD,MAAA;WAEA,OAAA;AAAA;;KAIC,cAAA,GAAiB,WAAA,SAAoB,aAAA,CAAc,aAAA;;KAkEnD,cAAA;EAAA,SACD,IAAA;EAAA,SACA,IAAI;AAAA;;;;;cAOF,qBAAA,GACX,KAAA,EAAO,QAAA,CAAS,MAAA,SAAe,UAAA,IAC/B,aAAA,EAAe,qBAAA,EACf,MAAA,EAAQ,aAAA,CAAc,cAAA,MACrB,cAAA;AAAA,KAoCS,eAAA;EAnHiB,oFAqHlB,EAAA,GAAK,GAAA,UAAa,MAAA,aAAmB,aAAa;AAAA;;AArHe;AAkE5E;;;;cA4Da,mBAAA,GACX,KAAA,EAAO,aAAA,CAAc,aAAA,GACrB,OAAA,EAAS,SAAA,KACR,eAAA"}

package/dist/node/xlsx-hyperlinks.mjs

@@ -0,0 +1,159 @@
+import { prefixedAttribute, xmlAttributes } from "./xml-text.mjs";
+import { indexParts, relationships, relationshipsPathFor, utf8Text } from "./xlsx-parts.mjs";
+import { parseCellRange } from "./xlsx-range.mjs";
+import { Buffer } from "node:buffer";
+//#region src/node/xlsx-hyperlinks.ts
+/** Longer targets are dropped: a truncated URL would point somewhere else. */
+const maxHyperlinkTargetCharacters = 2048;
+/** Longer display labels are cut to this many characters, ending with an ellipsis. */
+const maxHyperlinkDisplayCharacters = 1024;
+const capDisplay = (display) => {
+	if (display.length <= maxHyperlinkDisplayCharacters) return display;
+	let kept = display.slice(0, maxHyperlinkDisplayCharacters - 1);
+	const last = kept.charCodeAt(kept.length - 1);
+	if (last >= 55296 && last <= 56319) kept = kept.slice(0, -1);
+	return `${kept}\u2026`;
+};
+const shownProtocols = new Set([
+	"http:",
+	"https:",
+	"mailto:"
+]);
+const shownTarget = (target) => {
+	if (target.length > maxHyperlinkTargetCharacters || !URL.canParse(target)) return void 0;
+	const url = new URL(target);
+	if (!shownProtocols.has(url.protocol) || url.href.length > maxHyperlinkTargetCharacters) return void 0;
+	return url.href;
+};
+const hyperlinkFrom = (tag, sheetRelationships) => {
+	const attributes = xmlAttributes(Buffer.from(tag, "latin1").toString("utf8"));
+	const ref = attributes.get("ref");
+	const range = ref === void 0 ? void 0 : parseCellRange(ref);
+	const id = prefixedAttribute(attributes, "id");
+	const relationship = id === void 0 ? void 0 : sheetRelationships.get(id);
+	if (range === void 0 || relationship === void 0 || !relationship.external) return void 0;
+	const location = attributes.get("location");
+	const target = shownTarget(location === void 0 || location.length === 0 ? relationship.target : `${relationship.target}#${location}`);
+	if (target === void 0) return void 0;
+	const display = attributes.get("display")?.trim();
+	return display === void 0 || display.length === 0 ? {
+		range,
+		target
+	} : {
+		range,
+		target,
+		display: capDisplay(display)
+	};
+};
+/**
+* Map the hyperlink tags removed from a workbook to its sheets (the same sheet list and worksheet
+* parts SheetJS reads, from `buildSheetJsInput`) through each worksheet's relationships.
+*/
+const resolveXlsxHyperlinks = (parts, hyperlinkTags, sheets) => {
+	const bySheet = /* @__PURE__ */ new Map();
+	if (hyperlinkTags.size === 0) return bySheet;
+	const index = indexParts(parts);
+	const text = (path) => utf8Text(index.find(path)?.bytes);
+	for (const { name, part } of sheets) {
+		const tags = part === void 0 ? void 0 : hyperlinkTags.get(part);
+		if (part === void 0 || tags === void 0 || bySheet.has(name)) continue;
+		const sheetRelationships = relationships(text(relationshipsPathFor(part)));
+		const links = tags.flatMap((linkTag) => {
+			const link = hyperlinkFrom(linkTag, sheetRelationships);
+			return link === void 0 ? [] : [link];
+		});
+		if (links.length > 0) bySheet.set(name, links);
+	}
+	return bySheet;
+};
+/**
+* Index links clipped to the visited range for row-major lookup without scanning every link per
+* cell: a segment tree over columns whose nodes hold max-heaps (by document order) of the links
+* covering them. Links enter when the sweep reaches their first row and are discarded lazily once
+* it passes their last row. Work is O((links + cells) · log(columns) · log(links)).
+*/
+const makeHyperlinkLookup = (links, visited) => {
+	const clipped = links.flatMap((link, order) => {
+		const firstRow = Math.max(link.range.start.r, visited.start.r);
+		const lastRow = Math.min(link.range.end.r, visited.end.r);
+		const firstColumn = Math.max(link.range.start.c, visited.start.c) - visited.start.c;
+		const lastColumn = Math.min(link.range.end.c, visited.end.c) - visited.start.c;
+		return firstRow > lastRow || firstColumn > lastColumn ? [] : [{
+			order,
+			firstRow,
+			lastRow,
+			firstColumn,
+			lastColumn,
+			link
+		}];
+	}).sort((left, right) => left.firstRow - right.firstRow || left.order - right.order);
+	if (clipped.length === 0) return { at: () => void 0 };
+	const width = visited.end.c - visited.start.c + 1;
+	let leaves = 1;
+	while (leaves < width) leaves *= 2;
+	const heaps = [];
+	const push = (node, entry) => {
+		const heap = heaps[node] ?? [];
+		heaps[node] = heap;
+		heap.push(entry);
+		let index = heap.length - 1;
+		while (index > 0) {
+			const parent = index - 1 >> 1;
+			const parentEntry = heap[parent];
+			if (parentEntry === void 0 || parentEntry.order >= entry.order) break;
+			heap[index] = parentEntry;
+			heap[parent] = entry;
+			index = parent;
+		}
+	};
+	const pop = (heap) => {
+		const last = heap.pop();
+		if (last === void 0 || heap.length === 0) return;
+		heap[0] = last;
+		let index = 0;
+		for (;;) {
+			const left = index * 2 + 1;
+			const right = left + 1;
+			let largest = index;
+			const largestOrder = () => heap[largest]?.order ?? -1;
+			if ((heap[left]?.order ?? -1) > largestOrder()) largest = left;
+			if ((heap[right]?.order ?? -1) > largestOrder()) largest = right;
+			const swapped = heap[largest];
+			if (largest === index || swapped === void 0) return;
+			heap[largest] = last;
+			heap[index] = swapped;
+			index = largest;
+		}
+	};
+	const insert = (entry) => {
+		let low = entry.firstColumn + leaves;
+		let high = entry.lastColumn + leaves + 1;
+		while (low < high) {
+			if ((low & 1) === 1) push(low++, entry);
+			if ((high & 1) === 1) push(--high, entry);
+			low >>= 1;
+			high >>= 1;
+		}
+	};
+	let nextEntry = 0;
+	return { at: (row, column) => {
+		for (let entry = clipped[nextEntry]; entry !== void 0 && entry.firstRow <= row;) {
+			insert(entry);
+			nextEntry += 1;
+			entry = clipped[nextEntry];
+		}
+		let best;
+		for (let node = column - visited.start.c + leaves; node >= 1; node >>= 1) {
+			const heap = heaps[node];
+			if (heap === void 0) continue;
+			while (heap[0] !== void 0 && heap[0].lastRow < row) pop(heap);
+			const top = heap[0];
+			if (top !== void 0 && (best === void 0 || top.order > best.order)) best = top;
+		}
+		return best?.link;
+	} };
+};
+//#endregion
+export { makeHyperlinkLookup, resolveXlsxHyperlinks };
+
+//# sourceMappingURL=xlsx-hyperlinks.mjs.map

package/dist/node/xlsx-hyperlinks.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-hyperlinks.mjs","names":[],"sources":["../../src/node/xlsx-hyperlinks.ts"],"sourcesContent":["import { Buffer } from 'node:buffer'\nimport type { StrippedHyperlinkTags } from './office-archive.ts'\nimport { indexParts, relationships, relationshipsPathFor, utf8Text } from './xlsx-parts.ts'\nimport type { Relationship } from './xlsx-parts.ts'\nimport { parseCellRange } from './xlsx-range.ts'\nimport type { CellRange } from './xlsx-range.ts'\nimport { prefixedAttribute, xmlAttributes } from './xml-text.ts'\n\nexport type XlsxHyperlink = {\n  readonly range: CellRange\n  /** Normalized `http:`, `https:`, or `mailto:` URL, at most 2,048 characters. */\n  readonly target: string\n  /** Label for cells without text, at most 1,024 characters (longer labels end in `…`). */\n  readonly display?: string\n}\n\n/** Hyperlinks per worksheet name, in document order (a later link wins on overlap). */\nexport type XlsxHyperlinks = ReadonlyMap<string, ReadonlyArray<XlsxHyperlink>>\n\n/** Longer targets are dropped: a truncated URL would point somewhere else. */\nconst maxHyperlinkTargetCharacters = 2048\n\n/** Longer display labels are cut to this many characters, ending with an ellipsis. */\nconst maxHyperlinkDisplayCharacters = 1024\n\nconst capDisplay = (display: string) => {\n  if (display.length <= maxHyperlinkDisplayCharacters) return display\n\n  let kept = display.slice(0, maxHyperlinkDisplayCharacters - 1)\n  const last = kept.charCodeAt(kept.length - 1)\n\n  // Never leave half of a surrogate pair before the ellipsis.\n  if (last >= 0xd800 && last <= 0xdbff) kept = kept.slice(0, -1)\n\n  return `${kept}\\u2026`\n}\n\nconst shownProtocols = new Set(['http:', 'https:', 'mailto:'])\n\nconst shownTarget = (target: string) => {\n  if (target.length > maxHyperlinkTargetCharacters || !URL.canParse(target)) return undefined\n\n  const url = new URL(target)\n\n  if (!shownProtocols.has(url.protocol) || url.href.length > maxHyperlinkTargetCharacters)\n    return undefined\n\n  return url.href\n}\n\nconst hyperlinkFrom = (\n  tag: string,\n  sheetRelationships: ReadonlyMap<string, Relationship>\n): XlsxHyperlink | undefined => {\n  // Tags were captured as Latin-1 text of the original bytes; attribute values are UTF-8.\n  const attributes = xmlAttributes(Buffer.from(tag, 'latin1').toString('utf8'))\n  const ref = attributes.get('ref')\n  const range = ref === undefined ? undefined : parseCellRange(ref)\n  const id = prefixedAttribute(attributes, 'id')\n  const relationship = id === undefined ? undefined : sheetRelationships.get(id)\n\n  // Internal `location`-only links (`#Sheet2!A1`) are omitted; external links keep their\n  // location as a fragment, as SheetJS does.\n  if (range === undefined || relationship === undefined || !relationship.external) return undefined\n\n  const location = attributes.get('location')\n\n  const target = shownTarget(\n    location === undefined || location.length === 0\n      ? relationship.target\n      : `${relationship.target}#${location}`\n  )\n\n  if (target === undefined) return undefined\n\n  const display = attributes.get('display')?.trim()\n\n  return display === undefined || display.length === 0\n    ? { range, target }\n    : { range, target, display: capDisplay(display) }\n}\n\n/** A sheet of the workbook and the validated worksheet part SheetJS reads for it, if any. */\nexport type HyperlinkSheet = {\n  readonly name: string\n  readonly part: string | undefined\n}\n\n/**\n * Map the hyperlink tags removed from a workbook to its sheets (the same sheet list and worksheet\n * parts SheetJS reads, from `buildSheetJsInput`) through each worksheet's relationships.\n */\nexport const resolveXlsxHyperlinks = (\n  parts: Readonly<Record<string, Uint8Array>>,\n  hyperlinkTags: StrippedHyperlinkTags,\n  sheets: ReadonlyArray<HyperlinkSheet>\n): XlsxHyperlinks => {\n  const bySheet = new Map<string, ReadonlyArray<XlsxHyperlink>>()\n\n  if (hyperlinkTags.size === 0) return bySheet\n\n  const index = indexParts(parts)\n  const text = (path: string) => utf8Text(index.find(path)?.bytes)\n\n  for (const { name, part } of sheets) {\n    const tags = part === undefined ? undefined : hyperlinkTags.get(part)\n\n    if (part === undefined || tags === undefined || bySheet.has(name)) continue\n\n    const sheetRelationships = relationships(text(relationshipsPathFor(part)))\n\n    const links = tags.flatMap(linkTag => {\n      const link = hyperlinkFrom(linkTag, sheetRelationships)\n\n      return link === undefined ? [] : [link]\n    })\n\n    if (links.length > 0) bySheet.set(name, links)\n  }\n\n  return bySheet\n}\n\ntype ClippedLink = {\n  readonly order: number\n  readonly firstRow: number\n  readonly lastRow: number\n  readonly firstColumn: number\n  readonly lastColumn: number\n  readonly link: XlsxHyperlink\n}\n\nexport type HyperlinkLookup = {\n  /** The winning link at a cell. Rows must be queried in ascending order. */\n  readonly at: (row: number, column: number) => XlsxHyperlink | undefined\n}\n\n/**\n * Index links clipped to the visited range for row-major lookup without scanning every link per\n * cell: a segment tree over columns whose nodes hold max-heaps (by document order) of the links\n * covering them. Links enter when the sweep reaches their first row and are discarded lazily once\n * it passes their last row. Work is O((links + cells) · log(columns) · log(links)).\n */\nexport const makeHyperlinkLookup = (\n  links: ReadonlyArray<XlsxHyperlink>,\n  visited: CellRange\n): HyperlinkLookup => {\n  const clipped = links\n    .flatMap((link, order): ReadonlyArray<ClippedLink> => {\n      const firstRow = Math.max(link.range.start.r, visited.start.r)\n      const lastRow = Math.min(link.range.end.r, visited.end.r)\n      const firstColumn = Math.max(link.range.start.c, visited.start.c) - visited.start.c\n      const lastColumn = Math.min(link.range.end.c, visited.end.c) - visited.start.c\n\n      return firstRow > lastRow || firstColumn > lastColumn\n        ? []\n        : [{ order, firstRow, lastRow, firstColumn, lastColumn, link }]\n    })\n    .sort((left, right) => left.firstRow - right.firstRow || left.order - right.order)\n\n  if (clipped.length === 0) return { at: () => undefined }\n\n  const width = visited.end.c - visited.start.c + 1\n  let leaves = 1\n\n  while (leaves < width) leaves *= 2\n\n  const heaps: Array<Array<ClippedLink> | undefined> = []\n\n  const push = (node: number, entry: ClippedLink) => {\n    const heap = heaps[node] ?? []\n    heaps[node] = heap\n    heap.push(entry)\n\n    let index = heap.length - 1\n\n    while (index > 0) {\n      const parent = (index - 1) >> 1\n      const parentEntry = heap[parent]\n\n      if (parentEntry === undefined || parentEntry.order >= entry.order) break\n\n      heap[index] = parentEntry\n      heap[parent] = entry\n      index = parent\n    }\n  }\n\n  const pop = (heap: Array<ClippedLink>) => {\n    const last = heap.pop()\n\n    if (last === undefined || heap.length === 0) return\n\n    heap[0] = last\n\n    let index = 0\n\n    for (;;) {\n      const left = index * 2 + 1\n      const right = left + 1\n      let largest = index\n\n      const largestOrder = () => heap[largest]?.order ?? -1\n\n      if ((heap[left]?.order ?? -1) > largestOrder()) largest = left\n\n      if ((heap[right]?.order ?? -1) > largestOrder()) largest = right\n\n      const swapped = heap[largest]\n\n      if (largest === index || swapped === undefined) return\n\n      heap[largest] = last\n      heap[index] = swapped\n      index = largest\n    }\n  }\n\n  const insert = (entry: ClippedLink) => {\n    let low = entry.firstColumn + leaves\n    let high = entry.lastColumn + leaves + 1\n\n    while (low < high) {\n      if ((low & 1) === 1) push(low++, entry)\n\n      if ((high & 1) === 1) push(--high, entry)\n\n      low >>= 1\n      high >>= 1\n    }\n  }\n\n  let nextEntry = 0\n\n  return {\n    at: (row, column) => {\n      for (let entry = clipped[nextEntry]; entry !== undefined && entry.firstRow <= row;) {\n        insert(entry)\n        nextEntry += 1\n        entry = clipped[nextEntry]\n      }\n\n      let best: ClippedLink | undefined\n\n      for (let node = column - visited.start.c + leaves; node >= 1; node >>= 1) {\n        const heap = heaps[node]\n\n        if (heap === undefined) continue\n\n        while (heap[0] !== undefined && heap[0].lastRow < row) pop(heap)\n\n        const top = heap[0]\n\n        if (top !== undefined && (best === undefined || top.order > best.order)) best = top\n      }\n\n      return best?.link\n    }\n  }\n}\n"],"mappings":";;;;;;AAoBA,MAAM,+BAA+B;;AAGrC,MAAM,gCAAgC;AAEtC,MAAM,cAAc,YAAoB;CACtC,IAAI,QAAQ,UAAU,+BAA+B,OAAO;CAE5D,IAAI,OAAO,QAAQ,MAAM,GAAG,gCAAgC,CAAC;CAC7D,MAAM,OAAO,KAAK,WAAW,KAAK,SAAS,CAAC;CAG5C,IAAI,QAAQ,SAAU,QAAQ,OAAQ,OAAO,KAAK,MAAM,GAAG,EAAE;CAE7D,OAAO,GAAG,KAAK;AACjB;AAEA,MAAM,iBAAiB,IAAI,IAAI;CAAC;CAAS;CAAU;AAAS,CAAC;AAE7D,MAAM,eAAe,WAAmB;CACtC,IAAI,OAAO,SAAS,gCAAgC,CAAC,IAAI,SAAS,MAAM,GAAG,OAAO,KAAA;CAElF,MAAM,MAAM,IAAI,IAAI,MAAM;CAE1B,IAAI,CAAC,eAAe,IAAI,IAAI,QAAQ,KAAK,IAAI,KAAK,SAAS,8BACzD,OAAO,KAAA;CAET,OAAO,IAAI;AACb;AAEA,MAAM,iBACJ,KACA,uBAC8B;CAE9B,MAAM,aAAa,cAAc,OAAO,KAAK,KAAK,QAAQ,EAAE,SAAS,MAAM,CAAC;CAC5E,MAAM,MAAM,WAAW,IAAI,KAAK;CAChC,MAAM,QAAQ,QAAQ,KAAA,IAAY,KAAA,IAAY,eAAe,GAAG;CAChE,MAAM,KAAK,kBAAkB,YAAY,IAAI;CAC7C,MAAM,eAAe,OAAO,KAAA,IAAY,KAAA,IAAY,mBAAmB,IAAI,EAAE;CAI7E,IAAI,UAAU,KAAA,KAAa,iBAAiB,KAAA,KAAa,CAAC,aAAa,UAAU,OAAO,KAAA;CAExF,MAAM,WAAW,WAAW,IAAI,UAAU;CAE1C,MAAM,SAAS,YACb,aAAa,KAAA,KAAa,SAAS,WAAW,IAC1C,aAAa,SACb,GAAG,aAAa,OAAO,GAAG,UAChC;CAEA,IAAI,WAAW,KAAA,GAAW,OAAO,KAAA;CAEjC,MAAM,UAAU,WAAW,IAAI,SAAS,GAAG,KAAK;CAEhD,OAAO,YAAY,KAAA,KAAa,QAAQ,WAAW,IAC/C;EAAE;EAAO;CAAO,IAChB;EAAE;EAAO;EAAQ,SAAS,WAAW,OAAO;CAAE;AACpD;;;;;AAYA,MAAa,yBACX,OACA,eACA,WACmB;CACnB,MAAM,0BAAU,IAAI,IAA0C;CAE9D,IAAI,cAAc,SAAS,GAAG,OAAO;CAErC,MAAM,QAAQ,WAAW,KAAK;CAC9B,MAAM,QAAQ,SAAiB,SAAS,MAAM,KAAK,IAAI,GAAG,KAAK;CAE/D,KAAK,MAAM,EAAE,MAAM,UAAU,QAAQ;EACnC,MAAM,OAAO,SAAS,KAAA,IAAY,KAAA,IAAY,cAAc,IAAI,IAAI;EAEpE,IAAI,SAAS,KAAA,KAAa,SAAS,KAAA,KAAa,QAAQ,IAAI,IAAI,GAAG;EAEnE,MAAM,qBAAqB,cAAc,KAAK,qBAAqB,IAAI,CAAC,CAAC;EAEzE,MAAM,QAAQ,KAAK,SAAQ,YAAW;GACpC,MAAM,OAAO,cAAc,SAAS,kBAAkB;GAEtD,OAAO,SAAS,KAAA,IAAY,CAAC,IAAI,CAAC,IAAI;EACxC,CAAC;EAED,IAAI,MAAM,SAAS,GAAG,QAAQ,IAAI,MAAM,KAAK;CAC/C;CAEA,OAAO;AACT;;;;;;;AAsBA,MAAa,uBACX,OACA,YACoB;CACpB,MAAM,UAAU,MACb,SAAS,MAAM,UAAsC;EACpD,MAAM,WAAW,KAAK,IAAI,KAAK,MAAM,MAAM,GAAG,QAAQ,MAAM,CAAC;EAC7D,MAAM,UAAU,KAAK,IAAI,KAAK,MAAM,IAAI,GAAG,QAAQ,IAAI,CAAC;EACxD,MAAM,cAAc,KAAK,IAAI,KAAK,MAAM,MAAM,GAAG,QAAQ,MAAM,CAAC,IAAI,QAAQ,MAAM;EAClF,MAAM,aAAa,KAAK,IAAI,KAAK,MAAM,IAAI,GAAG,QAAQ,IAAI,CAAC,IAAI,QAAQ,MAAM;EAE7E,OAAO,WAAW,WAAW,cAAc,aACvC,CAAC,IACD,CAAC;GAAE;GAAO;GAAU;GAAS;GAAa;GAAY;EAAK,CAAC;CAClE,CAAC,EACA,MAAM,MAAM,UAAU,KAAK,WAAW,MAAM,YAAY,KAAK,QAAQ,MAAM,KAAK;CAEnF,IAAI,QAAQ,WAAW,GAAG,OAAO,EAAE,UAAU,KAAA,EAAU;CAEvD,MAAM,QAAQ,QAAQ,IAAI,IAAI,QAAQ,MAAM,IAAI;CAChD,IAAI,SAAS;CAEb,OAAO,SAAS,OAAO,UAAU;CAEjC,MAAM,QAA+C,CAAC;CAEtD,MAAM,QAAQ,MAAc,UAAuB;EACjD,MAAM,OAAO,MAAM,SAAS,CAAC;EAC7B,MAAM,QAAQ;EACd,KAAK,KAAK,KAAK;EAEf,IAAI,QAAQ,KAAK,SAAS;EAE1B,OAAO,QAAQ,GAAG;GAChB,MAAM,SAAU,QAAQ,KAAM;GAC9B,MAAM,cAAc,KAAK;GAEzB,IAAI,gBAAgB,KAAA,KAAa,YAAY,SAAS,MAAM,OAAO;GAEnE,KAAK,SAAS;GACd,KAAK,UAAU;GACf,QAAQ;EACV;CACF;CAEA,MAAM,OAAO,SAA6B;EACxC,MAAM,OAAO,KAAK,IAAI;EAEtB,IAAI,SAAS,KAAA,KAAa,KAAK,WAAW,GAAG;EAE7C,KAAK,KAAK;EAEV,IAAI,QAAQ;EAEZ,SAAS;GACP,MAAM,OAAO,QAAQ,IAAI;GACzB,MAAM,QAAQ,OAAO;GACrB,IAAI,UAAU;GAEd,MAAM,qBAAqB,KAAK,UAAU,SAAS;GAEnD,KAAK,KAAK,OAAO,SAAS,MAAM,aAAa,GAAG,UAAU;GAE1D,KAAK,KAAK,QAAQ,SAAS,MAAM,aAAa,GAAG,UAAU;GAE3D,MAAM,UAAU,KAAK;GAErB,IAAI,YAAY,SAAS,YAAY,KAAA,GAAW;GAEhD,KAAK,WAAW;GAChB,KAAK,SAAS;GACd,QAAQ;EACV;CACF;CAEA,MAAM,UAAU,UAAuB;EACrC,IAAI,MAAM,MAAM,cAAc;EAC9B,IAAI,OAAO,MAAM,aAAa,SAAS;EAEvC,OAAO,MAAM,MAAM;GACjB,KAAK,MAAM,OAAO,GAAG,KAAK,OAAO,KAAK;GAEtC,KAAK,OAAO,OAAO,GAAG,KAAK,EAAE,MAAM,KAAK;GAExC,QAAQ;GACR,SAAS;EACX;CACF;CAEA,IAAI,YAAY;CAEhB,OAAO,EACL,KAAK,KAAK,WAAW;EACnB,KAAK,IAAI,QAAQ,QAAQ,YAAY,UAAU,KAAA,KAAa,MAAM,YAAY,MAAM;GAClF,OAAO,KAAK;GACZ,aAAa;GACb,QAAQ,QAAQ;EAClB;EAEA,IAAI;EAEJ,KAAK,IAAI,OAAO,SAAS,QAAQ,MAAM,IAAI,QAAQ,QAAQ,GAAG,SAAS,GAAG;GACxE,MAAM,OAAO,MAAM;GAEnB,IAAI,SAAS,KAAA,GAAW;GAExB,OAAO,KAAK,OAAO,KAAA,KAAa,KAAK,GAAG,UAAU,KAAK,IAAI,IAAI;GAE/D,MAAM,MAAM,KAAK;GAEjB,IAAI,QAAQ,KAAA,MAAc,SAAS,KAAA,KAAa,IAAI,QAAQ,KAAK,QAAQ,OAAO;EAClF;EAEA,OAAO,MAAM;CACf,EACF;AACF"}

package/dist/node/xlsx-parts.d.mts

@@ -0,0 +1,29 @@
+//#region src/node/xlsx-parts.d.ts
+/** The main workbook part every validated XLSX archive contains. */
+declare const workbookPartPath = "xl/workbook.xml";
+declare const utf8Text: (bytes: Uint8Array | undefined) => string;
+/** Resolve a relationship target against the directory of its source part (`xl/`, …). */
+declare const resolvePartPath: (baseDirectory: string, target: string) => string;
+declare const directoryOf: (path: string) => string;
+declare const relationshipsPathFor: (partPath: string) => string;
+/**
+ * Validated parts by case-insensitive name, as SheetJS and OPC look them up. Archive validation
+ * rejects names that differ only in case, so each lower-cased name has one part.
+ */
+type PartIndex = {
+  /** The validated entry name and bytes of a part, matched ignoring case. */readonly find: (path: string) => {
+    readonly name: string;
+    readonly bytes: Uint8Array;
+  } | undefined;
+};
+declare const indexParts: (parts: Readonly<Record<string, Uint8Array>>) => PartIndex;
+type Relationship = {
+  readonly target: string; /** The `Type` attribute as written (relationship types are case-sensitive URIs). */
+  readonly type: string | undefined;
+  readonly external: boolean;
+};
+/** Relationships by `Id`; the first relationship with an `Id` wins. */
+declare const relationships: (xml: string) => ReadonlyMap<string, Relationship>;
+//#endregion
+export { PartIndex, Relationship, directoryOf, indexParts, relationships, relationshipsPathFor, resolvePartPath, utf8Text, workbookPartPath };
+//# sourceMappingURL=xlsx-parts.d.mts.map

package/dist/node/xlsx-parts.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-parts.d.mts","names":[],"sources":["../../src/node/xlsx-parts.ts"],"mappings":";;cAIa,gBAAA;AAAA,cAEA,QAAA,GAAY,KAA6B,EAAtB,UAAU;;cAI7B,eAAA,GAAmB,aAAA,UAAuB,MAAc;AAAA,cAYxD,WAAA,GAAe,IAAY;AAAA,cAE3B,oBAAA,GAAwB,QAAgB;;;;AAlBC;KAyB1C,SAAA;EAXX,oFAaU,IAAA,GAAO,IAAA;IAAA,SAA4B,IAAA;IAAA,SAAuB,KAAA,EAAO,UAAU;EAAA;AAAA;AAAA,cAGzE,UAAA,GAAc,KAAA,EAAO,QAAA,CAAS,MAAA,SAAe,UAAA,OAAe,SAAA;AAAA,KAqB7D,YAAA;EAAA,SACD,MAAA,UAlCE;EAAA,SAoCF,IAAA;EAAA,SACA,QAAA;AAAA;AArC0C;AAAA,cAyCxC,aAAA,GAAiB,GAAA,aAAc,WAAW,SAAS,YAAA"}

package/dist/node/xlsx-parts.mjs

@@ -0,0 +1,49 @@
+import { xmlAttributes } from "./xml-text.mjs";
+import { Buffer } from "node:buffer";
+//#region src/node/xlsx-parts.ts
+/** The main workbook part every validated XLSX archive contains. */
+const workbookPartPath = "xl/workbook.xml";
+const utf8Text = (bytes) => bytes === void 0 ? "" : Buffer.from(bytes).toString("utf8");
+/** Resolve a relationship target against the directory of its source part (`xl/`, …). */
+const resolvePartPath = (baseDirectory, target) => {
+	const segments = [];
+	const path = target.startsWith("/") ? target.slice(1) : `${baseDirectory}${target}`;
+	for (const segment of path.split("/")) if (segment === "..") segments.pop();
+	else if (segment !== "." && segment !== "") segments.push(segment);
+	return segments.join("/");
+};
+const directoryOf = (path) => path.slice(0, path.lastIndexOf("/") + 1);
+const relationshipsPathFor = (partPath) => `${directoryOf(partPath)}_rels/${partPath.slice(partPath.lastIndexOf("/") + 1)}.rels`;
+const indexParts = (parts) => {
+	const names = /* @__PURE__ */ new Map();
+	for (const name of Object.keys(parts)) names.set(name.toLowerCase(), name);
+	return { find: (path) => {
+		const name = names.get(path.toLowerCase());
+		const bytes = name === void 0 ? void 0 : parts[name];
+		return name === void 0 || bytes === void 0 ? void 0 : {
+			name,
+			bytes
+		};
+	} };
+};
+const startTagPattern = (localName) => new RegExp(`<(?:[\\w.-]+:)?${localName}(?=[\\s/>])[^<>]*>`, "g");
+const relationshipTag = startTagPattern("Relationship");
+/** Relationships by `Id`; the first relationship with an `Id` wins. */
+const relationships = (xml) => {
+	const byId = /* @__PURE__ */ new Map();
+	for (const [tag] of xml.matchAll(relationshipTag)) {
+		const attributes = xmlAttributes(tag);
+		const id = attributes.get("Id");
+		const target = attributes.get("Target");
+		if (id !== void 0 && target !== void 0 && !byId.has(id)) byId.set(id, {
+			target,
+			type: attributes.get("Type"),
+			external: attributes.get("TargetMode") === "External"
+		});
+	}
+	return byId;
+};
+//#endregion
+export { directoryOf, indexParts, relationships, relationshipsPathFor, resolvePartPath, utf8Text, workbookPartPath };
+
+//# sourceMappingURL=xlsx-parts.mjs.map

package/dist/node/xlsx-parts.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-parts.mjs","names":[],"sources":["../../src/node/xlsx-parts.ts"],"sourcesContent":["import { Buffer } from 'node:buffer'\nimport { xmlAttributes } from './xml-text.ts'\n\n/** The main workbook part every validated XLSX archive contains. */\nexport const workbookPartPath = 'xl/workbook.xml'\n\nexport const utf8Text = (bytes: Uint8Array | undefined) =>\n  bytes === undefined ? '' : Buffer.from(bytes).toString('utf8')\n\n/** Resolve a relationship target against the directory of its source part (`xl/`, …). */\nexport const resolvePartPath = (baseDirectory: string, target: string) => {\n  const segments: Array<string> = []\n  const path = target.startsWith('/') ? target.slice(1) : `${baseDirectory}${target}`\n\n  for (const segment of path.split('/')) {\n    if (segment === '..') segments.pop()\n    else if (segment !== '.' && segment !== '') segments.push(segment)\n  }\n\n  return segments.join('/')\n}\n\nexport const directoryOf = (path: string) => path.slice(0, path.lastIndexOf('/') + 1)\n\nexport const relationshipsPathFor = (partPath: string) =>\n  `${directoryOf(partPath)}_rels/${partPath.slice(partPath.lastIndexOf('/') + 1)}.rels`\n\n/**\n * Validated parts by case-insensitive name, as SheetJS and OPC look them up. Archive validation\n * rejects names that differ only in case, so each lower-cased name has one part.\n */\nexport type PartIndex = {\n  /** The validated entry name and bytes of a part, matched ignoring case. */\n  readonly find: (path: string) => { readonly name: string; readonly bytes: Uint8Array } | undefined\n}\n\nexport const indexParts = (parts: Readonly<Record<string, Uint8Array>>): PartIndex => {\n  const names = new Map<string, string>()\n\n  for (const name of Object.keys(parts)) names.set(name.toLowerCase(), name)\n\n  return {\n    find: path => {\n      const name = names.get(path.toLowerCase())\n      const bytes = name === undefined ? undefined : parts[name]\n\n      return name === undefined || bytes === undefined ? undefined : { name, bytes }\n    }\n  }\n}\n\n// `[^<>]` keeps each candidate inside one tag, so scans stay linear in the part size.\nconst startTagPattern = (localName: string) =>\n  new RegExp(`<(?:[\\\\w.-]+:)?${localName}(?=[\\\\s/>])[^<>]*>`, 'g')\n\nconst relationshipTag = startTagPattern('Relationship')\n\nexport type Relationship = {\n  readonly target: string\n  /** The `Type` attribute as written (relationship types are case-sensitive URIs). */\n  readonly type: string | undefined\n  readonly external: boolean\n}\n\n/** Relationships by `Id`; the first relationship with an `Id` wins. */\nexport const relationships = (xml: string): ReadonlyMap<string, Relationship> => {\n  const byId = new Map<string, Relationship>()\n\n  for (const [tag] of xml.matchAll(relationshipTag)) {\n    const attributes = xmlAttributes(tag)\n    const id = attributes.get('Id')\n    const target = attributes.get('Target')\n\n    if (id !== undefined && target !== undefined && !byId.has(id))\n      byId.set(id, {\n        target,\n        type: attributes.get('Type'),\n        external: attributes.get('TargetMode') === 'External'\n      })\n  }\n\n  return byId\n}\n"],"mappings":";;;;AAIA,MAAa,mBAAmB;AAEhC,MAAa,YAAY,UACvB,UAAU,KAAA,IAAY,KAAK,OAAO,KAAK,KAAK,EAAE,SAAS,MAAM;;AAG/D,MAAa,mBAAmB,eAAuB,WAAmB;CACxE,MAAM,WAA0B,CAAC;CACjC,MAAM,OAAO,OAAO,WAAW,GAAG,IAAI,OAAO,MAAM,CAAC,IAAI,GAAG,gBAAgB;CAE3E,KAAK,MAAM,WAAW,KAAK,MAAM,GAAG,GAClC,IAAI,YAAY,MAAM,SAAS,IAAI;MAC9B,IAAI,YAAY,OAAO,YAAY,IAAI,SAAS,KAAK,OAAO;CAGnE,OAAO,SAAS,KAAK,GAAG;AAC1B;AAEA,MAAa,eAAe,SAAiB,KAAK,MAAM,GAAG,KAAK,YAAY,GAAG,IAAI,CAAC;AAEpF,MAAa,wBAAwB,aACnC,GAAG,YAAY,QAAQ,EAAE,QAAQ,SAAS,MAAM,SAAS,YAAY,GAAG,IAAI,CAAC,EAAE;AAWjF,MAAa,cAAc,UAA2D;CACpF,MAAM,wBAAQ,IAAI,IAAoB;CAEtC,KAAK,MAAM,QAAQ,OAAO,KAAK,KAAK,GAAG,MAAM,IAAI,KAAK,YAAY,GAAG,IAAI;CAEzE,OAAO,EACL,OAAM,SAAQ;EACZ,MAAM,OAAO,MAAM,IAAI,KAAK,YAAY,CAAC;EACzC,MAAM,QAAQ,SAAS,KAAA,IAAY,KAAA,IAAY,MAAM;EAErD,OAAO,SAAS,KAAA,KAAa,UAAU,KAAA,IAAY,KAAA,IAAY;GAAE;GAAM;EAAM;CAC/E,EACF;AACF;AAGA,MAAM,mBAAmB,cACvB,IAAI,OAAO,kBAAkB,UAAU,qBAAqB,GAAG;AAEjE,MAAM,kBAAkB,gBAAgB,cAAc;;AAUtD,MAAa,iBAAiB,QAAmD;CAC/E,MAAM,uBAAO,IAAI,IAA0B;CAE3C,KAAK,MAAM,CAAC,QAAQ,IAAI,SAAS,eAAe,GAAG;EACjD,MAAM,aAAa,cAAc,GAAG;EACpC,MAAM,KAAK,WAAW,IAAI,IAAI;EAC9B,MAAM,SAAS,WAAW,IAAI,QAAQ;EAEtC,IAAI,OAAO,KAAA,KAAa,WAAW,KAAA,KAAa,CAAC,KAAK,IAAI,EAAE,GAC1D,KAAK,IAAI,IAAI;GACX;GACA,MAAM,WAAW,IAAI,MAAM;GAC3B,UAAU,WAAW,IAAI,YAAY,MAAM;EAC7C,CAAC;CACL;CAEA,OAAO;AACT"}

package/dist/node/xlsx-range.d.mts

@@ -0,0 +1,21 @@
+//#region src/node/xlsx-range.d.ts
+/** Zero-based cell position. */
+type CellPosition = {
+  readonly r: number;
+  readonly c: number;
+};
+type CellRange = {
+  readonly start: CellPosition;
+  readonly end: CellPosition;
+};
+/**
+ * Strictly parse `A1` or `A1:B2` within Excel's grid. Permissive SheetJS range decoders wrap or
+ * normalize invalid and overflowed references, so they are never used on untrusted refs.
+ */
+declare const parseCellRange: (ref: string) => CellRange | undefined;
+declare const cellCount: (range: CellRange) => number;
+/** `A1`-style address of a zero-based position. */
+declare const encodeCellAddress: (position: CellPosition) => string;
+//#endregion
+export { CellPosition, CellRange, cellCount, encodeCellAddress, parseCellRange };
+//# sourceMappingURL=xlsx-range.d.mts.map

package/dist/node/xlsx-range.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-range.d.mts","names":[],"sources":["../../src/node/xlsx-range.ts"],"mappings":";;KACY,YAAA;EAAA,SACD,CAAA;EAAA,SACA,CAAC;AAAA;AAAA,KAGA,SAAA;EAAA,SACD,KAAA,EAAO,YAAA;EAAA,SACP,GAAA,EAAK,YAAY;AAAA;;;;;cA6Bf,cAAA,GAAkB,GAAA,aAAc,SAAS;AAAA,cAiBzC,SAAA,GAAa,KAAgB,EAAT,SAAS;AA9Cd;AAAA,cAkDf,iBAAA,GAAqB,QAAsB,EAAZ,YAAY"}

package/dist/node/xlsx-range.mjs

@@ -0,0 +1,49 @@
+//#region src/node/xlsx-range.ts
+const maxExcelRows = 1048576;
+const maxExcelColumns = 16384;
+const cellAddressPattern = /^([A-Z]{1,3})([1-9][0-9]{0,6})$/;
+const cellAddress = (value) => {
+	const match = cellAddressPattern.exec(value);
+	if (match?.[1] === void 0 || match[2] === void 0) return void 0;
+	let column = 0;
+	for (const letter of match[1]) column = column * 26 + letter.charCodeAt(0) - 64;
+	const row = Number(match[2]);
+	if (!Number.isSafeInteger(row) || row > maxExcelRows || column > maxExcelColumns) return void 0;
+	return {
+		r: row - 1,
+		c: column - 1
+	};
+};
+/**
+* Strictly parse `A1` or `A1:B2` within Excel's grid. Permissive SheetJS range decoders wrap or
+* normalize invalid and overflowed references, so they are never used on untrusted refs.
+*/
+const parseCellRange = (ref) => {
+	if (ref.length > 21) return void 0;
+	const parts = ref.split(":");
+	const first = parts[0];
+	if (first === void 0 || parts.length > 2) return void 0;
+	const start = cellAddress(first);
+	const end = cellAddress(parts[1] ?? first);
+	if (start === void 0 || end === void 0 || end.r < start.r || end.c < start.c) return void 0;
+	return {
+		start,
+		end
+	};
+};
+const cellCount = (range) => (range.end.r - range.start.r + 1) * (range.end.c - range.start.c + 1);
+/** `A1`-style address of a zero-based position. */
+const encodeCellAddress = (position) => {
+	let column = "";
+	let remaining = position.c + 1;
+	while (remaining > 0) {
+		const letter = (remaining - 1) % 26;
+		column = String.fromCharCode(65 + letter) + column;
+		remaining = Math.floor((remaining - 1) / 26);
+	}
+	return `${column}${position.r + 1}`;
+};
+//#endregion
+export { cellCount, encodeCellAddress, parseCellRange };
+
+//# sourceMappingURL=xlsx-range.mjs.map

package/dist/node/xlsx-range.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-range.mjs","names":[],"sources":["../../src/node/xlsx-range.ts"],"sourcesContent":["/** Zero-based cell position. */\nexport type CellPosition = {\n  readonly r: number\n  readonly c: number\n}\n\nexport type CellRange = {\n  readonly start: CellPosition\n  readonly end: CellPosition\n}\n\nconst maxExcelRows = 1_048_576\n\nconst maxExcelColumns = 16_384\n\nconst cellAddressPattern = /^([A-Z]{1,3})([1-9][0-9]{0,6})$/\n\nconst cellAddress = (value: string): CellPosition | undefined => {\n  const match = cellAddressPattern.exec(value)\n\n  if (match?.[1] === undefined || match[2] === undefined) return undefined\n\n  let column = 0\n\n  for (const letter of match[1]) column = column * 26 + letter.charCodeAt(0) - 64\n\n  const row = Number(match[2])\n\n  if (!Number.isSafeInteger(row) || row > maxExcelRows || column > maxExcelColumns) return undefined\n\n  return { r: row - 1, c: column - 1 }\n}\n\n/**\n * Strictly parse `A1` or `A1:B2` within Excel's grid. Permissive SheetJS range decoders wrap or\n * normalize invalid and overflowed references, so they are never used on untrusted refs.\n */\nexport const parseCellRange = (ref: string): CellRange | undefined => {\n  if (ref.length > 21) return undefined\n\n  const parts = ref.split(':')\n  const first = parts[0]\n\n  if (first === undefined || parts.length > 2) return undefined\n\n  const start = cellAddress(first)\n  const end = cellAddress(parts[1] ?? first)\n\n  if (start === undefined || end === undefined || end.r < start.r || end.c < start.c)\n    return undefined\n\n  return { start, end }\n}\n\nexport const cellCount = (range: CellRange) =>\n  (range.end.r - range.start.r + 1) * (range.end.c - range.start.c + 1)\n\n/** `A1`-style address of a zero-based position. */\nexport const encodeCellAddress = (position: CellPosition) => {\n  let column = ''\n  let remaining = position.c + 1\n\n  while (remaining > 0) {\n    const letter = (remaining - 1) % 26\n    column = String.fromCharCode(65 + letter) + column\n    remaining = Math.floor((remaining - 1) / 26)\n  }\n\n  return `${column}${position.r + 1}`\n}\n"],"mappings":";AAWA,MAAM,eAAe;AAErB,MAAM,kBAAkB;AAExB,MAAM,qBAAqB;AAE3B,MAAM,eAAe,UAA4C;CAC/D,MAAM,QAAQ,mBAAmB,KAAK,KAAK;CAE3C,IAAI,QAAQ,OAAO,KAAA,KAAa,MAAM,OAAO,KAAA,GAAW,OAAO,KAAA;CAE/D,IAAI,SAAS;CAEb,KAAK,MAAM,UAAU,MAAM,IAAI,SAAS,SAAS,KAAK,OAAO,WAAW,CAAC,IAAI;CAE7E,MAAM,MAAM,OAAO,MAAM,EAAE;CAE3B,IAAI,CAAC,OAAO,cAAc,GAAG,KAAK,MAAM,gBAAgB,SAAS,iBAAiB,OAAO,KAAA;CAEzF,OAAO;EAAE,GAAG,MAAM;EAAG,GAAG,SAAS;CAAE;AACrC;;;;;AAMA,MAAa,kBAAkB,QAAuC;CACpE,IAAI,IAAI,SAAS,IAAI,OAAO,KAAA;CAE5B,MAAM,QAAQ,IAAI,MAAM,GAAG;CAC3B,MAAM,QAAQ,MAAM;CAEpB,IAAI,UAAU,KAAA,KAAa,MAAM,SAAS,GAAG,OAAO,KAAA;CAEpD,MAAM,QAAQ,YAAY,KAAK;CAC/B,MAAM,MAAM,YAAY,MAAM,MAAM,KAAK;CAEzC,IAAI,UAAU,KAAA,KAAa,QAAQ,KAAA,KAAa,IAAI,IAAI,MAAM,KAAK,IAAI,IAAI,MAAM,GAC/E,OAAO,KAAA;CAET,OAAO;EAAE;EAAO;CAAI;AACtB;AAEA,MAAa,aAAa,WACvB,MAAM,IAAI,IAAI,MAAM,MAAM,IAAI,MAAM,MAAM,IAAI,IAAI,MAAM,MAAM,IAAI;;AAGrE,MAAa,qBAAqB,aAA2B;CAC3D,IAAI,SAAS;CACb,IAAI,YAAY,SAAS,IAAI;CAE7B,OAAO,YAAY,GAAG;EACpB,MAAM,UAAU,YAAY,KAAK;EACjC,SAAS,OAAO,aAAa,KAAK,MAAM,IAAI;EAC5C,YAAY,KAAK,OAAO,YAAY,KAAK,EAAE;CAC7C;CAEA,OAAO,GAAG,SAAS,SAAS,IAAI;AAClC"}

package/dist/node/xlsx-routing.d.mts

@@ -0,0 +1,36 @@
+//#region src/node/xlsx-routing.d.ts
+/**
+ * Early, exact rejection of XLSX input that SheetJS 0.20.3 would hand to its ODS, Numbers, or
+ * binary (XLSB) parsers, so users get a clear error instead of "Could not read XLSX". These checks
+ * are not the security guarantee: SheetJS only ever receives the allowlisted archive built by
+ * `buildSheetJsInput` (`xlsx-sheetjs-input.ts`), which contains no marker entries and no `.bin`
+ * parts whatever these checks decide.
+ */
+/**
+ * An entry name as SheetJS looks it up: its ZIP reader (`cfb_add`) keeps a name that already
+ * starts with `Root Entry/` and otherwise stores `"Root Entry/" + name` with the first `//`
+ * collapsed; `safegetzipfile` strips `Root Entry/`, treats `\` and `/` alike, and compares
+ * lower-cased names.
+ */
+declare const sheetJsEntryPath: (name: string) => string;
+/**
+ * Whether SheetJS would treat this entry as an ODS, UOC, or Numbers marker. `CFB.find` matches
+ * `Index.zip` by base name, and any `Root Entry/` name (any case) is rejected outright.
+ */
+declare const isAlternateFormatEntry: (name: string) => boolean;
+/**
+ * Whether a relationships part has a `<Relationship>` whose `Target` ends in `.bin` and whose
+ * `Type` (read exactly as SheetJS reads it: case-sensitive, missing counts as a sheet) is not on
+ * the allowlist.
+ */
+declare const relationshipsRouteToBinary: (content: Uint8Array) => boolean;
+/**
+ * Whether `[Content_Types].xml` has an `<Override>` (any prefix, as SheetJS reads it) with an
+ * XLSB content type, or a `PartName` ending in `.bin` whose content type is not on the allowlist.
+ * `<Default>` entries are ignored: SheetJS does not route by them, and its own XLSX writer emits
+ * `<Default Extension="bin">` with the XLSB workbook type.
+ */
+declare const contentTypesRouteToBinary: (content: Uint8Array) => boolean;
+//#endregion
+export { contentTypesRouteToBinary, isAlternateFormatEntry, relationshipsRouteToBinary, sheetJsEntryPath };
+//# sourceMappingURL=xlsx-routing.d.mts.map

package/dist/node/xlsx-routing.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-routing.d.mts","names":[],"sources":["../../src/node/xlsx-routing.ts"],"mappings":";;AAiBA;;;;AAA6C;AAiB7C;;;;AAAmD;AA0FnD;;cA3Ga,gBAAA,GAAoB,IAAY;;AA2GiB;AAgC9D;;cA1Ha,sBAAA,GAA0B,IAAY;;AA0HU;;;;cAhChD,0BAAA,GAA8B,OAAmB,EAAV,UAAU;;;;;;;cAgCjD,yBAAA,GAA6B,OAAmB,EAAV,UAAU"}