reamkit 1.15.3 → 1.16.0

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

@@ -1,4 +1,5 @@
 import { ParsedWorksheet } from '../core/spreadsheet-model/index.js';
+import { ThemePalette } from './styles-parser.js';
 /**
  * Parse `worksheet.xml` (§18.3.1.99) into a {@link ParsedWorksheet}: a flat list
  * of cells with absolute row/column addresses and their raw stored values, plus
@@ -8,4 +9,4 @@ import { ParsedWorksheet } from '../core/spreadsheet-model/index.js';
  * and table parts). `r=` is optional on `<row>`/`<c>`; absent indices are inferred
  * from document order. A malformed root or missing `sheetData` yields an empty grid.
  */
-export declare function parseWorksheet(data: Uint8Array): ParsedWorksheet;
+export declare function parseWorksheet(data: Uint8Array, theme?: ThemePalette): ParsedWorksheet;

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

@@ -1,9 +1,14 @@
+import { INDEXED_COLORS } from "../core/indexed-colors.js";
 import { parseCellRef } from "./cell-reference.js";
 import { parseAreaRef } from "./defined-name-ref.js";
+import { resolveInternalEntities } from "../core/opc/xml-entities.js";
+import { decodeXstring } from "./escaped-text.js";
+import { parseDxf } from "./styles-parser.js";
 import { XMLParser } from "fast-xml-parser";
 //#region src/excel/worksheet-parser.ts
 var decoder = new TextDecoder("utf-8");
 var parser = new XMLParser({
+	htmlEntities: true,
 	ignoreAttributes: false,
 	attributeNamePrefix: "@_",
 	textNodeName: "#text",
@@ -21,9 +26,10 @@ var parser = new XMLParser({
 * and table parts). `r=` is optional on `<row>`/`<c>`; absent indices are inferred
 * from document order. A malformed root or missing `sheetData` yields an empty grid.
 */
-function parseWorksheet(data) {
-	const xml = decoder.decode(data);
-	const worksheet = parser.parse(xml)["worksheet"];
+function parseWorksheet(data, theme) {
+	const xml = resolveInternalEntities(decoder.decode(data));
+	const tree = parser.parse(xml);
+	const worksheet = tree["worksheet"] ?? tree["chartsheet"];
 	const emptyExtras = () => ({
 		columns: [],
 		merges: [],
@@ -45,7 +51,12 @@ function parseWorksheet(data) {
 	const pane = parsePane(wsObj);
 	const drawingNode = wsObj["drawing"];
 	const drawingRelId = drawingNode && typeof drawingNode === "object" ? strAttr(drawingNode, "id") : void 0;
-	const conditionalFormats = parseConditionalFormatting(wsObj);
+	const legacyNode = wsObj["legacyDrawing"];
+	const legacyDrawingRelId = legacyNode && typeof legacyNode === "object" ? strAttr(legacyNode, "id") : void 0;
+	const conditionalFormats = [...parseConditionalFormatting(wsObj), ...parseX14ConditionalFormatting(wsObj, {
+		...theme ? { theme } : {},
+		indexed: INDEXED_COLORS
+	})];
 	const dataValidations = parseDataValidations(wsObj);
 	const hyperlinks = parseHyperlinks(wsObj);
 	const headerFooter = parseHeaderFooter(wsObj);
@@ -63,6 +74,7 @@ function parseWorksheet(data) {
 		...colBreaks.length > 0 ? { colBreaks } : {},
 		...pane ? { pane } : {},
 		...drawingRelId !== void 0 ? { drawingRelId } : {},
+		...legacyDrawingRelId !== void 0 ? { legacyDrawingRelId } : {},
 		...conditionalFormats.length > 0 ? { conditionalFormats } : {},
 		...dataValidations.length > 0 ? { dataValidations } : {},
 		...hyperlinks.length > 0 ? { hyperlinks } : {},
@@ -86,6 +98,7 @@ function parseWorksheet(data) {
 	const rows = Array.isArray(rowRaw) ? rowRaw : rowRaw !== void 0 ? [rowRaw] : [];
 	const cells = [];
 	const rowHeights = [];
+	const rowStyles = [];
 	let maxRow = -1;
 	let maxColumn = -1;
 	let currentRow = -1;
@@ -96,6 +109,11 @@ function parseWorksheet(data) {
 		currentRow = explicitRow !== void 0 ? explicitRow : currentRow + 1;
 		const height = parseRowHeight(obj, currentRow);
 		if (height) rowHeights.push(height);
+		const rowStyle = Number(strAttr(obj, "s"));
+		if (boolAttr(obj, "customFormat") && Number.isFinite(rowStyle) && rowStyle > 0) rowStyles.push({
+			row: currentRow,
+			styleIndex: rowStyle
+		});
 		const cellRaw = obj["c"];
 		const rowCells = Array.isArray(cellRaw) ? cellRaw : cellRaw !== void 0 ? [cellRaw] : [];
 		let prevCol = -1;
@@ -108,6 +126,7 @@ function parseWorksheet(data) {
 			if (parsed.column > maxColumn) maxColumn = parsed.column;
 		}
 	}
+	const colStyles = parseColumnStyles(wsObj);
 	return {
 		cells,
 		maxRow,
@@ -115,6 +134,8 @@ function parseWorksheet(data) {
 		columns: parseColumns(wsObj),
 		merges: parseMerges(wsObj),
 		rowHeights,
+		...colStyles.length > 0 ? { columnStyles: colStyles } : {},
+		...rowStyles.length > 0 ? { rowStyles } : {},
 		...printModel
 	};
 }
@@ -126,15 +147,17 @@ function parseRowIndex(obj) {
 }
 function parseRowHeight(obj, rowIndex) {
 	if (rowIndex === void 0) return void 0;
+	const hidden = boolAttr(obj, "hidden");
 	const htRaw = strAttr(obj, "ht");
-	if (htRaw === void 0) return void 0;
-	const heightPt = Number(htRaw);
-	if (!Number.isFinite(heightPt)) return void 0;
+	const heightPt = htRaw !== void 0 ? Number(htRaw) : NaN;
+	if (!Number.isFinite(heightPt) && !hidden) return void 0;
 	const customRaw = strAttr(obj, "customHeight");
+	const customHeight = customRaw === "1" || customRaw === "true";
 	return {
 		row: rowIndex,
-		heightPt,
-		customHeight: customRaw === "1" || customRaw === "true"
+		heightPt: Number.isFinite(heightPt) ? heightPt : 0,
+		customHeight,
+		...hidden ? { hidden } : {}
 	};
 }
 function parsePageMargins(ws) {
@@ -167,13 +190,18 @@ function parsePageSetup(ws) {
 	const scale = parseNumericAttr(obj, "scale");
 	const fitToWidth = parseNumericAttr(obj, "fitToWidth");
 	const fitToHeight = parseNumericAttr(obj, "fitToHeight");
-	if (paperSize === void 0 && orientation === void 0 && scale === void 0 && fitToWidth === void 0 && fitToHeight === void 0) return;
+	const printerSettingsRelId = strAttr(obj, "id");
+	const commentsRaw = strAttr(obj, "cellComments");
+	const cellComments = commentsRaw === "none" || commentsRaw === "asDisplayed" || commentsRaw === "atEnd" ? commentsRaw : void 0;
+	if (paperSize === void 0 && orientation === void 0 && scale === void 0 && fitToWidth === void 0 && fitToHeight === void 0 && printerSettingsRelId === void 0 && cellComments === void 0) return;
 	return {
 		...paperSize !== void 0 ? { paperSize: Math.round(paperSize) } : {},
 		...orientation !== void 0 ? { orientation } : {},
 		...scale !== void 0 ? { scale } : {},
 		...fitToWidth !== void 0 ? { fitToWidth: Math.round(fitToWidth) } : {},
-		...fitToHeight !== void 0 ? { fitToHeight: Math.round(fitToHeight) } : {}
+		...fitToHeight !== void 0 ? { fitToHeight: Math.round(fitToHeight) } : {},
+		...printerSettingsRelId !== void 0 ? { printerSettingsRelId } : {},
+		...cellComments !== void 0 ? { cellComments } : {}
 	};
 }
 function parseFitToPage(ws) {
@@ -194,11 +222,13 @@ function parsePrintOptions(ws) {
 		return raw === "1" || raw === "true";
 	};
 	const gridLines = flag("gridLines");
+	const headings = flag("headings");
 	const horizontalCentered = flag("horizontalCentered");
 	const verticalCentered = flag("verticalCentered");
-	if (gridLines === void 0 && horizontalCentered === void 0 && verticalCentered === void 0) return;
+	if (gridLines === void 0 && headings === void 0 && horizontalCentered === void 0 && verticalCentered === void 0) return;
 	return {
 		...gridLines !== void 0 ? { gridLines } : {},
+		...headings !== void 0 ? { headings } : {},
 		...horizontalCentered !== void 0 ? { horizontalCentered } : {},
 		...verticalCentered !== void 0 ? { verticalCentered } : {}
 	};
@@ -267,7 +297,7 @@ function parseSheetFormatPr(ws) {
 		...base !== void 0 && base > 0 ? { baseColWidthChars: base } : {}
 	};
 }
-function parseColumns(ws) {
+function parseColumnStyles(ws) {
 	const colsNode = ws["cols"];
 	if (!colsNode || typeof colsNode !== "object") return [];
 	const colRaw = colsNode["col"];
@@ -278,15 +308,41 @@ function parseColumns(ws) {
 		const obj = c;
 		const min = Number(strAttr(obj, "min"));
 		const max = Number(strAttr(obj, "max"));
-		const width = Number(strAttr(obj, "width"));
-		if (Number.isFinite(min) && Number.isFinite(max) && Number.isFinite(width)) out.push({
+		const styleIndex = Number(strAttr(obj, "style"));
+		if (!Number.isFinite(min) || !Number.isFinite(max)) continue;
+		if (!Number.isFinite(styleIndex) || styleIndex <= 0) continue;
+		out.push({
 			min,
 			max,
-			widthChars: width
+			styleIndex
 		});
 	}
 	return out;
 }
+function parseColumns(ws) {
+	const colsNode = ws["cols"];
+	if (!colsNode || typeof colsNode !== "object") return [];
+	const colRaw = colsNode["col"];
+	const items = Array.isArray(colRaw) ? colRaw : colRaw !== void 0 ? [colRaw] : [];
+	const out = [];
+	for (const c of items) {
+		if (!c || typeof c !== "object") continue;
+		const obj = c;
+		const min = Number(strAttr(obj, "min"));
+		const max = Number(strAttr(obj, "max"));
+		const width = Number(strAttr(obj, "width"));
+		if (Number.isFinite(min) && Number.isFinite(max) && Number.isFinite(width)) {
+			const hidden = boolAttr(obj, "hidden");
+			out.push({
+				min,
+				max,
+				widthChars: width,
+				...hidden ? { hidden } : {}
+			});
+		}
+	}
+	return out;
+}
 function parseMerges(ws) {
 	const mergeNode = ws["mergeCells"];
 	if (!mergeNode || typeof mergeNode !== "object") return [];
@@ -339,6 +395,62 @@ function parseConditionalFormatting(ws) {
 	}
 	return out;
 }
+/** The rule types the extension states as a formula rather than a literal. */
+var TEXT_RULES = new Set([
+	"containsText",
+	"notContainsText",
+	"beginsWith",
+	"endsWith"
+]);
+function parseX14ConditionalFormatting(ws, colors) {
+	const out = [];
+	for (const ext of toArray(asObjectNode(ws["extLst"])?.["ext"])) {
+		const group = asObjectNode(asObjectNode(ext)?.["conditionalFormattings"]);
+		if (!group) continue;
+		for (const cf of toArray(group["conditionalFormatting"])) {
+			const obj = asObjectNode(cf);
+			if (!obj) continue;
+			const sqref = typeof obj["sqref"] === "string" ? obj["sqref"] : void 0;
+			if (!sqref) continue;
+			const ranges = parseSqref(sqref);
+			if (ranges.length === 0) continue;
+			const rules = [];
+			for (const rn of toArray(obj["cfRule"])) {
+				const node = asObjectNode(rn);
+				if (!node) continue;
+				const rule = parseCfRule(asBaseCfRule(node));
+				if (!rule) continue;
+				const dxf = parseDxf(node["dxf"], colors);
+				rules.push("dxfId" in rule && Object.keys(dxf).length > 0 ? {
+					...rule,
+					dxf
+				} : rule);
+			}
+			if (rules.length > 0) out.push({
+				ranges,
+				rules
+			});
+		}
+	}
+	return out;
+}
+/**
+* Rewrite a `<x14:cfRule>` into the shape {@link parseCfRule} reads, so one
+* reader serves both spellings of the same rule.
+*/
+function asBaseCfRule(node) {
+	const formulas = toArray(node["f"]).filter((f) => typeof f === "string" || typeof f === "number");
+	const out = {
+		...node,
+		formula: formulas.length === 1 ? formulas[0] : formulas
+	};
+	if (out["@_dxfId"] === void 0 && node["dxf"] !== void 0) out["@_dxfId"] = "0";
+	if (TEXT_RULES.has(String(out["@_type"])) && formulas.length > 0) {
+		out["@_type"] = "expression";
+		out["formula"] = formulas[0];
+	}
+	return out;
+}
 var DV_TYPES = new Set([
 	"none",
 	"whole",
@@ -588,13 +700,16 @@ function parseDataBarRule(obj, priority) {
 	if (!colorHex) return void 0;
 	const minLength = parseNumericAttr(dbObj, "minLength");
 	const maxLength = parseNumericAttr(dbObj, "maxLength");
+	const showValueRaw = strAttr(dbObj, "showValue");
+	const showValue = showValueRaw === "0" || showValueRaw === "false" ? false : void 0;
 	return {
 		type: "dataBar",
 		priority,
 		cfvos,
 		colorHex,
 		...minLength !== void 0 ? { minLength } : {},
-		...maxLength !== void 0 ? { maxLength } : {}
+		...maxLength !== void 0 ? { maxLength } : {},
+		...showValue === false ? { showValue } : {}
 	};
 }
 function parseCellIsRule(obj, priority) {
@@ -736,9 +851,15 @@ function collectControls(node, out, seen) {
 		if (!relId || seen.has(relId)) continue;
 		seen.add(relId);
 		const name = strAttr(obj, "name");
+		const shapeId = strAttr(obj, "shapeId");
+		const controlPr = asObjectNode(obj["controlPr"]);
+		const printAttr = controlPr ? strAttr(controlPr, "print") : void 0;
+		const print = printAttr === "0" || printAttr === "false" ? false : void 0;
 		out.push({
 			relId,
-			...name ? { name } : {}
+			...name ? { name } : {},
+			...shapeId ? { shapeId } : {},
+			...print === false ? { print } : {}
 		});
 	}
 }
@@ -752,7 +873,7 @@ function parseOleObjects(ws) {
 function collectOleObjects(node, out, seen) {
 	if (!node) return;
 	const direct = toArray(node["oleObject"]);
-	const fromChoice = toArray(asObjectNode(node["AlternateContent"])?.["Choice"]).flatMap((c) => toArray(asObjectNode(c)?.["oleObject"]));
+	const fromChoice = toArray(node["AlternateContent"]).flatMap((ac) => toArray(asObjectNode(ac)?.["Choice"])).flatMap((c) => toArray(asObjectNode(c)?.["oleObject"]));
 	for (const o of [...direct, ...fromChoice]) {
 		const obj = asObjectNode(o);
 		if (!obj) continue;
@@ -800,10 +921,13 @@ function parseCell(c, fallbackRow, fallbackCol) {
 	const styleIndex = styleStr !== void 0 ? Number(styleStr) : void 0;
 	const v = obj["v"];
 	const rawValue = textOf(v);
+	const vmStr = type === "e" ? strAttr(obj, "vm") : void 0;
+	const vm = vmStr !== void 0 ? Number(vmStr) : void 0;
 	const base = {
 		column: address.column,
 		row: address.row,
-		type
+		type,
+		...vm !== void 0 && Number.isInteger(vm) && vm > 0 ? { valueMetadataIndex: vm } : {}
 	};
 	if (type === "inlineStr") {
 		const is = obj["is"];
@@ -831,11 +955,11 @@ function strAttr(obj, key) {
 	return typeof v === "string" ? v : void 0;
 }
 function textOf(node) {
-	if (typeof node === "string") return node;
+	if (typeof node === "string") return decodeXstring(node);
 	if (typeof node === "number") return String(node);
 	if (!node || typeof node !== "object") return "";
 	const inner = node["#text"];
-	if (typeof inner === "string") return inner;
+	if (typeof inner === "string") return decodeXstring(inner);
 	if (typeof inner === "number") return String(inner);
 	return "";
 }
@@ -846,12 +970,9 @@ function inlineStringText(is) {
 	const t = obj["t"];
 	const direct = textOf(t);
 	if (direct) return direct.length > MAX_CELL_CHARS ? direct.slice(0, MAX_CELL_CHARS) : direct;
-	const r = obj["r"];
-	if (Array.isArray(r)) {
-		const joined = r.map((rr) => textOf(rr?.["t"])).join("");
-		return joined.length > MAX_CELL_CHARS ? joined.slice(0, MAX_CELL_CHARS) : joined;
-	}
-	return "";
+	const raw = obj["r"];
+	const joined = (Array.isArray(raw) ? raw : raw !== void 0 ? [raw] : []).map((rr) => textOf(rr?.["t"])).join("");
+	return joined.length > MAX_CELL_CHARS ? joined.slice(0, MAX_CELL_CHARS) : joined;
 }
 //#endregion
 export { parseWorksheet };