js.documents 4.4.8 → 4.6.0

package/dist/layout/reconstruct.cjs

@@ -1,8 +1,8 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
 const require_model_geometry = require("../model/geometry.cjs");
 const require_model_embedded_drawing = require("../model/embedded-drawing.cjs");
-const require_layout_cell_typing = require("./cell-typing.cjs");
 const require_layout_shared = require("./shared.cjs");
+const require_layout_cell_typing = require("./cell-typing.cjs");
 const require_ports_abort = require("../ports/abort.cjs");
 const require_layout_lattice = require("./lattice.cjs");
 let pdf_codec_fonts = require("pdf-codec/fonts");
@@ -136,19 +136,20 @@ function reconstructWordprocessing(doc, options) {
 	const signal = options?.signal;
 	doc = dropHiddenLayerContent(doc);
 	const headingLevels = headingSizeLevels(doc);
+	const structure = indexStructure(doc);
 	const sections = [];
 	let currentGroup = [];
 	let groupStartPageIndex = 0;
 	for (const page of doc.pages) {
 		require_ports_abort.throwIfAborted(signal);
 		if (currentGroup.length > 0 && !samePageSize(currentGroup[0], page)) {
-			sections.push(buildSection(currentGroup, groupStartPageIndex, doc.images, headingLevels, doc.form));
+			sections.push(buildSection(currentGroup, groupStartPageIndex, doc.images, headingLevels, doc.form, structure));
 			groupStartPageIndex += currentGroup.length;
 			currentGroup = [];
 		}
 		currentGroup.push(page);
 	}
-	if (currentGroup.length > 0) sections.push(buildSection(currentGroup, groupStartPageIndex, doc.images, headingLevels, doc.form));
+	if (currentGroup.length > 0) sections.push(buildSection(currentGroup, groupStartPageIndex, doc.images, headingLevels, doc.form, structure));
 	return {
 		kind: "wordprocessing",
 		metadata: doc.metadata,
@@ -169,6 +170,59 @@ function dropHiddenLayerContent(doc) {
 		}))
 	};
 }
+const HEADING_LEVELS = /* @__PURE__ */ new Map([
+	["H1", 1],
+	["H2", 2],
+	["H3", 3],
+	["H4", 4],
+	["H5", 5],
+	["H6", 6]
+]);
+const DIVISION_TYPES = /* @__PURE__ */ new Set([
+	"Part",
+	"Sect",
+	"Div"
+]);
+const CELL_TYPES = /* @__PURE__ */ new Set(["TD", "TH"]);
+const ROW_TYPES = /* @__PURE__ */ new Set(["TR"]);
+const TABLE_TYPES = /* @__PURE__ */ new Set(["Table"]);
+function indexStructure(doc) {
+	if (doc.structure === void 0 || doc.structure.length === 0) return;
+	const byId = /* @__PURE__ */ new Map();
+	let order = 0;
+	const walk = (element, parent) => {
+		const node = {
+			id: element.id,
+			type: element.type,
+			parent,
+			order: order++
+		};
+		byId.set(element.id, node);
+		for (const child of element.children) walk(child, node);
+	};
+	for (const root of doc.structure) walk(root, void 0);
+	return { nodeOfItem: (item) => item.kind === "link" || item.kind === "internalLink" ? void 0 : byId.get(item.structure ?? "") };
+}
+function nearestOfType(node, types) {
+	for (let current = node; current !== void 0; current = current.parent) if (types.has(current.type)) return current;
+}
+function nearestHeadingLevel(node) {
+	for (let current = node; current !== void 0; current = current.parent) {
+		const level = HEADING_LEVELS.get(current.type);
+		if (level !== void 0) return level;
+	}
+}
+function structureHeadingLevel(items, structure) {
+	if (items.length === 0 || items.some((item) => structure.nodeOfItem(item) === void 0)) return;
+	let level;
+	for (const item of items) {
+		const itemLevel = nearestHeadingLevel(structure.nodeOfItem(item));
+		if (itemLevel === void 0) return { level: void 0 };
+		if (level === void 0) level = itemLevel;
+		else if (level !== itemLevel) return;
+	}
+	return { level };
+}
 const HEADING_MIN_SIZE_DELTA_PT = 2;
 const MAX_HEADING_LEVEL = 6;
 function headingSizeLevels(doc) {
@@ -193,11 +247,12 @@ function headingLevelOf(paragraph, levels) {
 function samePageSize(a, b) {
 	return a.widthPt === b.widthPt && a.heightPt === b.heightPt;
 }
-function buildSection(pages, startPageIndex, images, headingLevels, form) {
+function buildSection(pages, startPageIndex, images, headingLevels, form, structure) {
 	const blocks = [];
+	const divisionByBlock = /* @__PURE__ */ new WeakMap();
 	pages.forEach((page, i) => {
 		if (i > 0) blocks.push({ kind: "pageBreak" });
-		blocks.push(...reconstructPageBlocks(page, startPageIndex + i, images, headingLevels, form));
+		blocks.push(...reconstructPageBlocks(page, startPageIndex + i, images, headingLevels, form, structure, divisionByBlock));
 	});
 	return {
 		pageSize: {
@@ -205,20 +260,34 @@ function buildSection(pages, startPageIndex, images, headingLevels, form) {
 			heightPt: pages[0].heightPt
 		},
 		margins: ZERO_MARGINS,
-		blocks
+		blocks: wrapDivisionExtents(blocks, divisionByBlock)
 	};
 }
-function reconstructPageBlocks(page, pageIndex, images, headingLevels, form) {
-	const recoveredTable = recoverTable(page, pageIndex);
-	const consumedText = recoveredTable?.consumedText;
-	const textItems = page.items.filter((i) => i.kind === "text" && consumedText?.has(i) !== true);
+function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, structure, divisionByBlock) {
+	const recoveredTables = recoverTables(page, pageIndex, structure);
+	const consumedText = /* @__PURE__ */ new Set();
+	const claimedItems = /* @__PURE__ */ new Set();
+	for (const recovered of recoveredTables) {
+		for (const item of recovered.consumedText) consumedText.add(item);
+		for (const item of recovered.latticeItems) claimedItems.add(item);
+	}
+	const textItems = page.items.filter((i) => i.kind === "text" && !consumedText.has(i));
 	const imageItems = page.items.filter((i) => i.kind === "image");
 	const paragraphs = clusterIntoParagraphs(clusterIntoLines(textItems));
+	const recordDivisions = (block, items) => {
+		if (structure !== void 0) divisionByBlock.set(block, divisionChainOf(items, structure));
+	};
 	const positioned = [];
-	for (const paragraph of paragraphs) positioned.push({
-		yPt: paragraph.lines[0].baselineY,
-		block: paragraphToContentParagraph(paragraph, pageIndex, headingLevelOf(paragraph, headingLevels))
-	});
+	for (const paragraph of paragraphs) {
+		const paragraphItems = paragraph.lines.flatMap((line) => line.items);
+		const structural = structure === void 0 ? void 0 : structureHeadingLevel(paragraphItems, structure);
+		const block = paragraphToContentParagraph(paragraph, pageIndex, structural !== void 0 ? structural.level : headingLevelOf(paragraph, headingLevels));
+		recordDivisions(block, paragraphItems);
+		positioned.push({
+			yPt: paragraph.lines[0].baselineY,
+			block
+		});
+	}
 	for (const img of imageItems) {
 		const asset = images[img.imageId];
 		if (asset !== void 0) {
@@ -235,21 +304,29 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form) {
 				widthPt: img.widthPt,
 				heightPt: img.heightPt
 			});
+			recordDivisions(block, [img]);
 			positioned.push({
 				yPt: img.yPt,
 				block
 			});
 		}
 	}
-	if (recoveredTable !== void 0) positioned.push({
-		yPt: recoveredTable.topYPt,
-		block: recoveredTable.table
-	});
-	const recoveredVectors = recoverPageVectors(page, pageIndex, recoveredTable?.latticeItems ?? NO_ITEMS);
-	if (recoveredVectors !== void 0) positioned.push({
-		yPt: recoveredVectors.topYPt,
-		block: recoveredVectors.block
-	});
+	for (const recovered of recoveredTables) {
+		recordDivisions(recovered.table, [...recovered.consumedText]);
+		positioned.push({
+			yPt: recovered.topYPt,
+			block: recovered.table
+		});
+	}
+	const recoveredVectors = recoverPageVectors(page, pageIndex, claimedItems);
+	if (recoveredVectors !== void 0) {
+		const vectorItems = page.items.filter((item) => !claimedItems.has(item) && layoutItemToVector(item, page.heightPt) !== void 0);
+		recordDivisions(recoveredVectors.block, vectorItems);
+		positioned.push({
+			yPt: recoveredVectors.topYPt,
+			block: recoveredVectors.block
+		});
+	}
 	positioned.sort((a, b) => b.yPt - a.yPt);
 	const blocks = positioned.map((p) => p.block);
 	reconcileLinks(blocks, page, pageIndex);
@@ -257,6 +334,49 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form) {
 	appendFormFieldConstructs(blocks, form, pageIndex);
 	return blocks;
 }
+function divisionChainOf(items, structure) {
+	let prefix;
+	for (const item of items) {
+		const chain = [];
+		for (let current = structure.nodeOfItem(item); current !== void 0; current = current.parent) if (DIVISION_TYPES.has(current.type)) chain.push(current.id);
+		chain.reverse();
+		if (prefix === void 0) {
+			prefix = chain;
+			continue;
+		}
+		let shared = 0;
+		while (shared < prefix.length && shared < chain.length && prefix[shared] === chain[shared]) shared += 1;
+		prefix = prefix.slice(0, shared);
+	}
+	return prefix ?? [];
+}
+function wrapDivisionExtents(blocks, divisionByBlock) {
+	const out = [];
+	const open = [];
+	for (const block of blocks) {
+		const chain = divisionByBlock.get(block);
+		if (chain !== void 0) {
+			while (open.length > 0 && !chain.includes(open[open.length - 1])) {
+				out.push({ kind: "constructEnd" });
+				open.pop();
+			}
+			for (const id of chain) {
+				if (open.includes(id)) continue;
+				out.push({
+					kind: "constructStart",
+					descriptor: { kind: "division" }
+				});
+				open.push(id);
+			}
+		}
+		out.push(block);
+	}
+	while (open.length > 0) {
+		out.push({ kind: "constructEnd" });
+		open.pop();
+	}
+	return out;
+}
 function framesIntersect(a, b) {
 	return a.xPt < b.xPt + b.widthPt && b.xPt < a.xPt + a.widthPt && a.yPt < b.yPt + b.heightPt && b.yPt < a.yPt + a.heightPt;
 }
@@ -420,9 +540,10 @@ function paragraphToContentParagraph(paragraph, pageIndex, headingLevel) {
 }
 function reconstructPresentation(doc, options) {
 	const signal = options?.signal;
+	const structure = indexStructure(doc);
 	const slides = doc.pages.map((page, pageIndex) => {
 		require_ports_abort.throwIfAborted(signal);
-		return reconstructSlide(page, pageIndex, doc.images);
+		return reconstructSlide(page, pageIndex, doc.images, structure);
 	});
 	return {
 		kind: "presentation",
@@ -430,10 +551,15 @@ function reconstructPresentation(doc, options) {
 		slides
 	};
 }
-function reconstructSlide(page, pageIndex, images) {
-	const recoveredTable = recoverTable(page, pageIndex);
-	const consumedText = recoveredTable?.consumedText;
-	const textItems = page.items.filter((i) => i.kind === "text" && consumedText?.has(i) !== true);
+function reconstructSlide(page, pageIndex, images, structure) {
+	const recoveredTables = recoverTables(page, pageIndex, structure);
+	const consumedText = /* @__PURE__ */ new Set();
+	const claimedItems = /* @__PURE__ */ new Set();
+	for (const recovered of recoveredTables) {
+		for (const item of recovered.consumedText) consumedText.add(item);
+		for (const item of recovered.latticeItems) claimedItems.add(item);
+	}
+	const textItems = page.items.filter((i) => i.kind === "text" && !consumedText.has(i));
 	const imageItems = page.items.filter((i) => i.kind === "image");
 	const textShapes = clusterIntoBlocks(clusterIntoLines(textItems)).map((block) => blockToShape(block, page.heightPt, pageIndex));
 	const imageShapes = [];
@@ -441,14 +567,14 @@ function reconstructSlide(page, pageIndex, images) {
 		const shape = imageToShape(img, page.heightPt, pageIndex, images);
 		if (shape !== void 0) imageShapes.push(shape);
 	}
-	const recoveredVectors = recoverPageVectors(page, pageIndex, recoveredTable?.latticeItems ?? NO_ITEMS);
+	const recoveredVectors = recoverPageVectors(page, pageIndex, claimedItems);
 	const vectorShapes = recoveredVectors === void 0 ? [] : [wrapBlockInShape(recoveredVectors.block, {
 		xPt: 0,
 		yPt: 0,
 		widthPt: page.widthPt,
 		heightPt: page.heightPt
 	}, page.heightPt, pageIndex)];
-	const tableShapes = recoveredTable === void 0 ? [] : [wrapBlockInShape(recoveredTable.table, recoveredTable.frame, page.heightPt, pageIndex)];
+	const tableShapes = recoveredTables.map((recovered) => wrapBlockInShape(recovered.table, recovered.frame, page.heightPt, pageIndex));
 	return {
 		size: {
 			widthPt: page.widthPt,
@@ -894,6 +1020,112 @@ function recoverPageVectors(page, pageIndex, excluded) {
 		topYPt: page.heightPt - topYDownPt
 	};
 }
+function recoverTables(page, pageIndex, structure) {
+	if (structure !== void 0) {
+		const tagged = recoverTaggedTables(page, pageIndex, structure);
+		if (tagged.length > 0) return tagged;
+	}
+	const lattice = recoverTable(page, pageIndex);
+	return lattice === void 0 ? [] : [lattice];
+}
+function recoverTaggedTables(page, pageIndex, structure) {
+	const cellsByTable = /* @__PURE__ */ new Map();
+	for (const item of page.items) {
+		if (item.kind !== "text") continue;
+		const cell = nearestOfType(structure.nodeOfItem(item), CELL_TYPES);
+		const row = nearestOfType(cell?.parent, ROW_TYPES);
+		const table = nearestOfType(row?.parent, TABLE_TYPES);
+		if (cell === void 0 || row === void 0 || table === void 0) continue;
+		const entry = cellsByTable.get(table.id) ?? {
+			table,
+			cells: []
+		};
+		entry.cells.push({
+			table,
+			row,
+			cell,
+			items: [item]
+		});
+		cellsByTable.set(table.id, entry);
+	}
+	if (cellsByTable.size === 0) return [];
+	const latticeItems = require_layout_lattice.detectGridLattice(page.items)?.sourceItems ?? NO_ITEMS;
+	const recovered = [];
+	for (const { cells } of cellsByTable.values()) {
+		const rowsById = /* @__PURE__ */ new Map();
+		for (const cellEntry of cells) {
+			const rowEntry = rowsById.get(cellEntry.row.id) ?? {
+				row: cellEntry.row,
+				cells: []
+			};
+			rowEntry.cells.push(cellEntry);
+			rowsById.set(cellEntry.row.id, rowEntry);
+		}
+		const rows = [...rowsById.values()].sort((a, b) => a.row.order - b.row.order);
+		for (const row of rows) row.cells.sort((a, b) => a.cell.order - b.cell.order);
+		const columnCount = Math.max(...rows.map((row) => row.cells.length));
+		const columnWidthsPt = [];
+		for (let j = 0; j < columnCount; j++) {
+			let minX = Number.POSITIVE_INFINITY;
+			let maxX = Number.NEGATIVE_INFINITY;
+			for (const row of rows) {
+				const cellEntry = row.cells[j];
+				if (cellEntry === void 0) continue;
+				for (const item of cellEntry.items) {
+					minX = Math.min(minX, item.xPt);
+					maxX = Math.max(maxX, item.xPt + (item.widthPt ?? 0));
+				}
+			}
+			const width = maxX - minX;
+			columnWidthsPt.push(width > 0 ? width : DEFAULT_COLUMN_WIDTH_FALLBACK_PT);
+		}
+		const allItems = cells.flatMap((cellEntry) => cellEntry.items);
+		const contentRows = rows.map((row) => {
+			const contentCells = [];
+			for (let j = 0; j < columnCount; j++) {
+				const cellEntry = row.cells[j];
+				const cell = { blocks: cellBlocksFromItems(cellEntry?.items ?? [], pageIndex) };
+				if (cellEntry !== void 0) require_layout_shared.stampFrame(cell, pageIndex, textItemsPdfBox(cellEntry.items));
+				contentCells.push(cell);
+			}
+			return { cells: contentCells };
+		});
+		const box = textItemsPdfBox(allItems);
+		const table = {
+			kind: "table",
+			rows: contentRows,
+			columnWidthsPt
+		};
+		require_layout_shared.stampFrame(table, pageIndex, box);
+		recovered.push({
+			table,
+			frame: require_model_geometry.flipY(box, page.heightPt),
+			topYPt: box.yPt + box.heightPt,
+			consumedText: new Set(allItems),
+			latticeItems
+		});
+	}
+	return recovered;
+}
+function textItemsPdfBox(items) {
+	let minX = Number.POSITIVE_INFINITY;
+	let maxX = Number.NEGATIVE_INFINITY;
+	let minY = Number.POSITIVE_INFINITY;
+	let maxY = Number.NEGATIVE_INFINITY;
+	for (const item of items) {
+		const { ascentPt, descentPt } = textItemVerticalExtent(item);
+		minX = Math.min(minX, item.xPt);
+		maxX = Math.max(maxX, item.xPt + (item.widthPt ?? 0));
+		minY = Math.min(minY, item.yPt - descentPt);
+		maxY = Math.max(maxY, item.yPt + ascentPt);
+	}
+	return {
+		xPt: minX,
+		yPt: minY,
+		widthPt: maxX - minX,
+		heightPt: maxY - minY
+	};
+}
 function cellBlocksFromItems(items, pageIndex) {
 	return clusterIntoLines(items).map((line) => lineToParagraph(line, pageIndex));
 }
@@ -998,23 +1230,7 @@ function buildCellsFromGroups(groups, pageIndex, context) {
 			value: inferredCellValue(displayText, row, column, context),
 			displayText
 		};
-		let minX = Number.POSITIVE_INFINITY;
-		let maxX = Number.NEGATIVE_INFINITY;
-		let minY = Number.POSITIVE_INFINITY;
-		let maxY = Number.NEGATIVE_INFINITY;
-		for (const item of items) {
-			const { ascentPt, descentPt } = textItemVerticalExtent(item);
-			minX = Math.min(minX, item.xPt);
-			maxX = Math.max(maxX, item.xPt + (item.widthPt ?? 0));
-			minY = Math.min(minY, item.yPt - descentPt);
-			maxY = Math.max(maxY, item.yPt + ascentPt);
-		}
-		require_layout_shared.stampFrame(cell, pageIndex, {
-			xPt: minX,
-			yPt: minY,
-			widthPt: maxX - minX,
-			heightPt: maxY - minY
-		});
+		require_layout_shared.stampFrame(cell, pageIndex, textItemsPdfBox(items));
 		cells.push(cell);
 	}
 	cells.sort((a, b) => a.row - b.row || a.column - b.column);