reamkit 1.15.3 → 1.15.4

package/dist/esm/core/ir/sheet.d.ts

@@ -80,12 +80,15 @@ export interface SheetFormControl {
 }
 /**
  * An ActiveX control resolved against its activeX part (E-SHEET W10): the control
- * class (`type`, from the `<oleObject progId>`) plus the visible state persisted
- * in the property bag — `caption`, `value` (checked/text/number, as a string)
- * and OptionButton `groupName`. Render-only.
+ * class (`type`, from the `<oleObject progId>` or, for a §18.3.1.19 `<control>`,
+ * its class id) plus the visible state persisted in the property bag —
+ * `caption`, `value` (checked/text/number, as a string) and OptionButton
+ * `groupName`. Render-only.
  */
 export interface SheetActiveXControl {
     readonly type: string;
+    /** §18.3.1.19 `<control name>` — the control's identifier, when it came that way. */
+    readonly name?: string;
     readonly caption?: string;
     readonly value?: string;
     readonly groupName?: string;

package/dist/esm/excel/activex-parser.d.ts

@@ -22,6 +22,15 @@ export declare function parseActiveX(data: Uint8Array): ActiveXProps;
  * unknown progId falls back to a generic `'control'`.
  */
 export declare function activeXType(progId: string | undefined): string;
+/**
+ * The affordance key for an `<ax:ocx ax:classid>`, for controls reached through
+ * §18.3.1.19 `<control>` rather than `<oleObject progId>` — the `<control>`
+ * element carries no progId, so the class id is all there is to type it by.
+ *
+ * @param xmlData The `activeX#.xml` part bytes.
+ * @returns The affordance key, or `'control'` when the class id is unknown.
+ */
+export declare function activeXTypeFromPart(xmlData: Uint8Array): string;
 /**
  * The `<ax:ocx r:id>` of a control whose state is persisted to a binary stream
  * (`persistStreamInit` / `persistStream` / `persistStorage`) rather than to

package/dist/esm/excel/activex-parser.js

@@ -62,6 +62,20 @@ function strAttr(obj, key) {
 	const v = obj[`@_${key}`];
 	return typeof v === "string" ? v : void 0;
 }
+var ACTIVEX_CLASS_TYPES = new Map([["8BD21D40-EC42-11CE-9E0D-00AA006002F3", "checkbox"], ["8BD21D50-EC42-11CE-9E0D-00AA006002F3", "option"]]);
+/**
+* The affordance key for an `<ax:ocx ax:classid>`, for controls reached through
+* §18.3.1.19 `<control>` rather than `<oleObject progId>` — the `<control>`
+* element carries no progId, so the class id is all there is to type it by.
+*
+* @param xmlData The `activeX#.xml` part bytes.
+* @returns The affordance key, or `'control'` when the class id is unknown.
+*/
+function activeXTypeFromPart(xmlData) {
+	const ocx = asObject(parser.parse(decoder.decode(xmlData))["ocx"]);
+	const key = (ocx ? strAttr(ocx, "classid") : void 0)?.replace(/[{}]/g, "").toUpperCase() ?? "";
+	return ACTIVEX_CLASS_TYPES.get(key) ?? "control";
+}
 /**
 * The `<ax:ocx r:id>` of a control whose state is persisted to a binary stream
 * (`persistStreamInit` / `persistStream` / `persistStorage`) rather than to
@@ -209,4 +223,4 @@ function parseActiveXBin(data) {
 	};
 }
 //#endregion
-export { activeXBinRelId, activeXType, parseActiveX, parseActiveXBin };
+export { activeXBinRelId, activeXType, activeXTypeFromPart, parseActiveX, parseActiveXBin };

package/dist/esm/excel/print-model.js

@@ -160,11 +160,14 @@ function worksheetToBody(worksheet, sharedStrings, styles, date1904, print) {
 	if (worksheet.maxRow < 0 || worksheet.maxColumn < 0) return [];
 	let usedRow = -1;
 	let usedCol = -1;
+	const contentAt = /* @__PURE__ */ new Set();
 	for (const c of worksheet.cells) if (c.rawValue !== "" || c.inlineText !== void 0) {
+		contentAt.add(key(c.row, c.column));
 		if (c.row > usedRow) usedRow = c.row;
 		if (c.column > usedCol) usedCol = c.column;
 	}
 	for (const m of worksheet.merges) {
+		if (!contentAt.has(key(m.startRow, m.startColumn))) continue;
 		if (m.endRow > usedRow) usedRow = m.endRow;
 		if (m.endColumn > usedCol) usedCol = m.endColumn;
 	}
@@ -186,7 +189,7 @@ function worksheetToBody(worksheet, sharedStrings, styles, date1904, print) {
 		colEnd = Math.min(usedCol, print.printArea.endColumn);
 	}
 	if (rowEnd < rowStart || colEnd < colStart) return [];
-	const MAX_GRID_COLS = 1024;
+	const MAX_GRID_COLS = 16384;
 	const MAX_GRID_ROWS = 5e4;
 	const MAX_GRID_CELLS = 1e6;
 	const wantRows = rowEnd - rowStart + 1;

package/dist/esm/excel/worksheet-parser.js

@@ -792,7 +792,7 @@ function parseCell(c, fallbackRow, fallbackCol) {
 	if (ref) try {
 		address = parseCellRef(ref);
 	} catch {
-		address = implied;
+		address = numericRef(ref, fallbackRow) ?? implied;
 	}
 	else address = implied;
 	const type = validateCellType(strAttr(obj, "t") ?? "n");
@@ -822,6 +822,38 @@ function parseCell(c, fallbackRow, fallbackCol) {
 		...Number.isFinite(styleIndex) ? { styleIndex } : {}
 	};
 }
+var NUMERIC_REF = /^(\d+)_(\d+)$/;
+/**
+* Decode a numeric `"4_2"`-style reference, but only when the file corroborates
+* the reading.
+*
+* This is not a spec spelling and there is no obligation to understand it. What
+* makes it safe to act on is that the file states its own convention and can be
+* checked against itself: every ref inside `<row r="2">` ends in `_2`. So the
+* row half must agree with the row the cell actually sits in — and when it does
+* not, the reference has told us nothing and document order (§18.3.1.4) remains
+* the answer. A guess that cannot be checked is worse than the fallback; this
+* one can be.
+*
+* The alternative is not harmless: these cells are sparse (columns 1, 4, 5, 7,
+* 11, 13, 22 …), so packing them consecutively files every value under the
+* wrong heading.
+*
+* @param ref      The unparseable `r` attribute.
+* @param rowIndex The 0-based row the cell sits in.
+* @returns The address, or undefined when the reading is not corroborated.
+*/
+function numericRef(ref, rowIndex) {
+	const m = NUMERIC_REF.exec(ref);
+	if (!m) return void 0;
+	const column = Number(m[1]);
+	const row = Number(m[2]);
+	if (column < 1 || row !== rowIndex + 1) return void 0;
+	return {
+		column: column - 1,
+		row: rowIndex
+	};
+}
 function validateCellType(t) {
 	if (t === "n" || t === "s" || t === "str" || t === "b" || t === "d" || t === "e" || t === "inlineStr") return t;
 	return "n";

package/dist/esm/excel/xlsx-reader.js

@@ -21,7 +21,7 @@ import { parsePivotTablePart } from "./pivot-table-parser.js";
 import { parseSlicerCachePart, parseSlicerPart } from "./slicer-parser.js";
 import { parseLegacyComments, parsePersons, parseThreadedComments } from "./comments-parser.js";
 import { parseFormControlProps } from "./form-control-parser.js";
-import { activeXBinRelId, activeXType, parseActiveX, parseActiveXBin } from "./activex-parser.js";
+import { activeXBinRelId, activeXType, activeXTypeFromPart, parseActiveX, parseActiveXBin } from "./activex-parser.js";
 import { parseSheetShapes } from "./sheet-shape-parser.js";
 import { projectSheetDoc } from "./sheet-to-flow.js";
 //#region src/excel/xlsx-reader.ts
@@ -34,6 +34,7 @@ var SLICER_CACHE_REL_TAIL = "/slicerCache";
 var THREADED_COMMENTS_REL_TAIL = "/threadedComment";
 var PERSON_REL_TAIL = "/person";
 var MAX_SLICER_ITEMS = 256;
+var ACTIVEX_PART = /^xl\/activeX\//i;
 /**
 * Read a `.xlsx` and project it to a {@link FlowDoc} in one step:
 * {@link readXlsxToSheetDoc} then {@link projectSheetDoc}.
@@ -219,13 +220,33 @@ function readXlsxToSheetDoc(xlsx) {
 			}
 			if (resolvedComments.length > 0) comments = resolvedComments;
 		}
+		const wsRels = pkg.getPartRelationships(resolved.path);
+		const resolvedAx = [];
+		const activeXState = (part) => {
+			const props = parseActiveX(part.data);
+			const binRelId = activeXBinRelId(part.data);
+			const binRel = binRelId ? pkg.getPartRelationships(part.path).find((r) => r.id === binRelId) : void 0;
+			const binPart = binRel ? pkg.resolveRelatedPart(part.path, binRel) : void 0;
+			return {
+				type: "control",
+				...binPart ? parseActiveXBin(binPart.data) : {},
+				...props
+			};
+		};
 		let formControls;
 		if (worksheet.formControls && worksheet.formControls.length > 0) {
-			const wsRels = pkg.getPartRelationships(resolved.path);
 			const resolvedControls = [];
 			for (const fc of worksheet.formControls) {
 				const rel = wsRels.find((r) => r.id === fc.relId);
 				const part = rel ? pkg.resolveRelatedPart(resolved.path, rel) : void 0;
+				if (part && ACTIVEX_PART.test(part.path)) {
+					resolvedAx.push({
+						...activeXState(part),
+						type: activeXTypeFromPart(part.data),
+						...fc.name ? { name: fc.name } : {}
+					});
+					continue;
+				}
 				const props = part ? parseFormControlProps(part.data) : {};
 				resolvedControls.push({
 					...fc.name ? { name: fc.name } : {},
@@ -234,29 +255,15 @@ function readXlsxToSheetDoc(xlsx) {
 			}
 			if (resolvedControls.length > 0) formControls = resolvedControls;
 		}
-		let activeXControls;
-		if (worksheet.oleObjects && worksheet.oleObjects.length > 0) {
-			const wsRels = pkg.getPartRelationships(resolved.path);
-			const resolvedAx = [];
-			for (const ole of worksheet.oleObjects) {
-				const rel = wsRels.find((r) => r.id === ole.relId);
-				const part = rel ? pkg.resolveRelatedPart(resolved.path, rel) : void 0;
-				const props = part ? parseActiveX(part.data) : {};
-				let binProps = {};
-				if (part) {
-					const binRelId = activeXBinRelId(part.data);
-					const binRel = binRelId ? pkg.getPartRelationships(part.path).find((r) => r.id === binRelId) : void 0;
-					const binPart = binRel ? pkg.resolveRelatedPart(part.path, binRel) : void 0;
-					if (binPart) binProps = parseActiveXBin(binPart.data);
-				}
-				resolvedAx.push({
-					type: activeXType(ole.progId),
-					...binProps,
-					...props
-				});
-			}
-			if (resolvedAx.length > 0) activeXControls = resolvedAx;
+		if (worksheet.oleObjects && worksheet.oleObjects.length > 0) for (const ole of worksheet.oleObjects) {
+			const rel = wsRels.find((r) => r.id === ole.relId);
+			const part = rel ? pkg.resolveRelatedPart(resolved.path, rel) : void 0;
+			resolvedAx.push({
+				...part ? activeXState(part) : {},
+				type: activeXType(ole.progId)
+			});
 		}
+		const activeXControls = resolvedAx.length > 0 ? resolvedAx : void 0;
 		const grid = tables || pivotTables ? {
 			...worksheet,
 			...tables ? { tables } : {},

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.15.3",
+  "version": "1.15.4",
   "description": "Ream — convert DOCX, XLSX, PPTX and PDF to PDF, SVG, HTML, DOCX and XLSX, built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
   "license": "MIT",
   "author": "Alex Krassavin <info@reamkit.dev>",