ooxml.js 3.0.1 → 3.1.1

package/dist/typed/docx/read.js

@@ -1,5 +1,5 @@
 import { base64ToBytes } from "../../util/base64.js";
-import { attr, childrenWithTag, elementsWithTag, resolveRelationships, rootElement, textContent } from "../util.js";
+import { attr, childrenWithTag, decodeEntities, elementsWithTag, resolveRelationships, rootElement, textContent } from "../util.js";
 import { DocumentMetadataSchema, readCoreProperties } from "../shared/metadata.js";
 import { sniffImageFormat } from "../../image/sniff.js";
 import { eighthPointsToPt, emuToPt, twipsToPt } from "../shared/units.js";
@@ -7,6 +7,7 @@ import { EMPTY_THEME, readTheme } from "../shared/drawingml.js";
 import { assignSourcePaths } from "../shared/source-path.js";
 import { resolveParagraphProperties, resolveRunProperties } from "./styles.js";
 import { NumberingDefinitionSchema, readNumberingDefinitions } from "./numbering.js";
+import { PROVENANCE_CHANGE_BY_TAG, bookmarkAnchorDescriptor, contentBearingChildren, fieldCharType, indexParagraphContent, insertConstructMarkers, isDeletedChange, readContentControlDescriptor, readProvenanceDescriptor, runInstructionText } from "./constructs.js";
 import { z } from "zod";
 import { COLOR_BLACK, ContentSectionSchema, PAGE_SIZE_LETTER, clampHeadingLevel, rgbHexToColor } from "document-schema.js";
 //#region src/typed/docx/read.ts
@@ -86,13 +87,13 @@ function readRunText(run) {
 	let text = "";
 	for (const child of run.children) {
 		if (child.type !== "element") continue;
-		if (child.tag === "w:t") text += textContent(child);
+		if (child.tag === "w:t" || child.tag === "w:delText") text += textContent(child);
 		else if (child.tag === "w:tab") text += "	";
 		else if (child.tag === "w:br" || child.tag === "w:cr") text += "\n";
 	}
 	return text;
 }
-function readDrawingImage(drawing, rels, pkg) {
+function readDrawingImage(drawing, ctx) {
 	const container = childrenWithTag(drawing, "wp:inline")[0] ?? childrenWithTag(drawing, "wp:anchor")[0];
 	if (container === void 0) return;
 	const extent = childrenWithTag(container, "wp:extent")[0];
@@ -103,8 +104,8 @@ function readDrawingImage(drawing, rels, pkg) {
 	const altText = docPr === void 0 ? void 0 : attr(docPr, "descr") ?? attr(docPr, "title");
 	const blip = elementsWithTag(container.children, "a:blip")[0];
 	const rId = blip === void 0 ? void 0 : attr(blip, "r:embed");
-	const rel = rId === void 0 ? void 0 : rels.get(rId);
-	const mediaPart = rel === void 0 ? void 0 : pkg.parts[rel.target];
+	const rel = rId === void 0 ? void 0 : ctx.rels.get(rId);
+	const mediaPart = rel === void 0 ? void 0 : ctx.pkg.parts[rel.target];
 	if (mediaPart?.kind !== "binary") return;
 	const bytes = base64ToBytes(mediaPart.base64);
 	const format = sniffImageFormat(bytes);
@@ -116,26 +117,26 @@ function readDrawingImage(drawing, rels, pkg) {
 		widthPt: emuToPt(Number(cx)),
 		heightPt: emuToPt(Number(cy))
 	};
-	if (altText !== void 0) image.altText = altText;
+	if (altText !== void 0) image.altText = decodeEntities(altText);
 	return image;
 }
-function collectDrawings(nodes, out) {
+function collectDrawings(nodes, carryDeletions, out) {
 	for (const node of nodes) {
 		if (node.type !== "element") continue;
-		if (node.tag === "w:del") continue;
+		if (!carryDeletions && (node.tag === "w:del" || node.tag === "w:moveFrom")) continue;
 		if (node.tag === "w:drawing") {
 			out.push(node);
 			continue;
 		}
-		collectDrawings(node.children, out);
+		collectDrawings(node.children, carryDeletions, out);
 	}
 }
-function readParagraphImages(paragraph, rels, pkg) {
+function readParagraphImages(paragraph, ctx, carryDeletions) {
 	const drawings = [];
-	collectDrawings(paragraph.children, drawings);
+	collectDrawings(paragraph.children, carryDeletions, drawings);
 	const images = [];
 	for (const drawing of drawings) {
-		const image = readDrawingImage(drawing, rels, pkg);
+		const image = readDrawingImage(drawing, ctx);
 		if (image !== void 0) images.push(image);
 	}
 	return images;
@@ -153,23 +154,22 @@ function readRun(run, paragraph, context) {
 		color: props.color
 	};
 }
-function readParagraphRuns(paragraph, context, rels) {
+function readParagraphRuns(paragraph, ctx, carryDeletions) {
 	const runs = [];
 	let fieldState = "none";
 	function walk(nodes, hyperlinkTarget) {
 		for (const node of nodes) {
 			if (node.type !== "element") continue;
 			if (node.tag === "w:r") {
-				const fldChar = childrenWithTag(node, "w:fldChar")[0];
-				if (fldChar !== void 0) {
-					const type = attr(fldChar, "w:fldCharType");
+				const type = fieldCharType(node);
+				if (type !== void 0) {
 					if (type === "begin") fieldState = "code";
 					else if (type === "separate") fieldState = "result";
 					else if (type === "end") fieldState = "none";
 					continue;
 				}
 				if (fieldState === "code") continue;
-				const run = readRun(node, paragraph, context);
+				const run = readRun(node, paragraph, ctx.styles);
 				runs.push(hyperlinkTarget === void 0 ? run : {
 					...run,
 					hyperlink: hyperlinkTarget
@@ -177,21 +177,27 @@ function readParagraphRuns(paragraph, context, rels) {
 			} else if (node.tag === "w:fldSimple") walk(node.children, hyperlinkTarget);
 			else if (node.tag === "w:hyperlink") {
 				const rId = attr(node, "r:id");
-				const target = rId === void 0 ? void 0 : rels.get(rId)?.target;
+				const target = rId === void 0 ? void 0 : ctx.rels.get(rId)?.target;
 				walk(node.children, target ?? hyperlinkTarget);
-			} else if (node.tag === "w:ins") walk(node.children, hyperlinkTarget);
+			} else if (node.tag === "w:ins" || node.tag === "w:moveTo") walk(node.children, hyperlinkTarget);
+			else if (node.tag === "w:del" || node.tag === "w:moveFrom") {
+				if (carryDeletions) walk(node.children, hyperlinkTarget);
+			} else if (node.tag === "w:sdt") {
+				const sdtContent = childrenWithTag(node, "w:sdtContent")[0];
+				if (sdtContent !== void 0) walk(sdtContent.children, hyperlinkTarget);
+			}
 		}
 	}
 	walk(paragraph.children, void 0);
 	return runs;
 }
-function readParagraph(paragraph, context, rels) {
+function readParagraph(paragraph, ctx, carryDeletions) {
 	const pPr = childrenWithTag(paragraph, "w:pPr")[0];
 	const pStyleEl = pPr === void 0 ? void 0 : childrenWithTag(pPr, "w:pStyle")[0];
-	const props = resolveParagraphProperties(paragraph, context);
+	const props = resolveParagraphProperties(paragraph, ctx.styles);
 	return {
 		kind: "paragraph",
-		runs: readParagraphRuns(paragraph, context, rels),
+		runs: readParagraphRuns(paragraph, ctx, carryDeletions),
 		styleId: pStyleEl === void 0 ? void 0 : attr(pStyleEl, "w:val"),
 		headingLevel: props.outlineLvl === void 0 ? void 0 : clampHeadingLevel(props.outlineLvl + 1),
 		alignment: props.alignment,
@@ -252,7 +258,7 @@ function readCellBorders(tcPr) {
 	if (bottom !== void 0) borders.bottom = bottom;
 	return Object.keys(borders).length === 0 ? void 0 : borders;
 }
-function readRawCell(tc, context, rels, pkg) {
+function readRawCell(tc, ctx, carryDeletions) {
 	const tcPr = childrenWithTag(tc, "w:tcPr")[0];
 	const gridSpanEl = tcPr === void 0 ? void 0 : childrenWithTag(tcPr, "w:gridSpan")[0];
 	const gridSpanVal = gridSpanEl === void 0 ? void 0 : attr(gridSpanEl, "w:val");
@@ -263,7 +269,7 @@ function readRawCell(tc, context, rels, pkg) {
 		isVMergeContinuation: vMergeVal === "continue",
 		background: readCellShading(tcPr),
 		borders: readCellBorders(tcPr),
-		blocks: readBodyBlocks(tc.children, context, rels, pkg)
+		blocks: readBlockScope(tc.children, ctx, carryDeletions)
 	};
 }
 function readRowHeightPt(tr) {
@@ -273,11 +279,11 @@ function readRowHeightPt(tr) {
 	const val = trHeight === void 0 ? void 0 : attr(trHeight, "w:val");
 	return val === void 0 ? void 0 : twipsToPt(Number(val));
 }
-function readTable(tbl, context, rels, pkg) {
+function readTable(tbl, ctx, carryDeletions) {
 	const tblGrid = childrenWithTag(tbl, "w:tblGrid")[0];
 	const columnWidthsPt = tblGrid === void 0 ? [] : childrenWithTag(tblGrid, "w:gridCol").map((col) => twipsToPt(Number(attr(col, "w:w") ?? "0")));
 	const trs = childrenWithTag(tbl, "w:tr");
-	const rawRows = trs.map((tr) => childrenWithTag(tr, "w:tc").map((tc) => readRawCell(tc, context, rels, pkg)));
+	const rawRows = trs.map((tr) => childrenWithTag(tr, "w:tc").map((tc) => readRawCell(tc, ctx, carryDeletions)));
 	const rowColumnIndices = rawRows.map((row) => {
 		const indices = [];
 		let col = 0;
@@ -312,63 +318,254 @@ function readTable(tbl, context, rels, pkg) {
 		}))
 	};
 }
-function readBodyBlocks(nodes, context, rels, pkg) {
-	const blocks = [];
+function newFlowState() {
+	return {
+		blocks: [],
+		extents: [],
+		sectionBreaks: [],
+		bookmarkEvents: [],
+		openFields: [],
+		order: 0
+	};
+}
+function resolveBookmarkExtents(events) {
+	const byId = /* @__PURE__ */ new Map();
+	for (const event of events) {
+		const existing = byId.get(event.id);
+		if (existing === void 0) byId.set(event.id, [event]);
+		else existing.push(event);
+	}
+	const extents = [];
+	for (const halves of byId.values()) {
+		const start = halves.filter((half) => half.kind === "start");
+		const end = halves.filter((half) => half.kind === "end");
+		const open = start[0];
+		const close = end[0];
+		if (start.length !== 1 || end.length !== 1 || open === void 0 || close === void 0) continue;
+		if (open.name === void 0 || !open.qualified || !close.qualified || close.index < open.index) continue;
+		extents.push({
+			startIndex: open.index,
+			endIndex: close.index,
+			order: open.order,
+			descriptor: bookmarkAnchorDescriptor(open.name)
+		});
+	}
+	return extents;
+}
+function wholeParagraphTrackedChange(index) {
+	const content = contentBearingChildren(index);
+	const first = content[0];
+	if (first === void 0) return;
+	const change = PROVENANCE_CHANGE_BY_TAG.get(first.tag);
+	if (change === void 0 || !content.every((child) => child.tag === first.tag)) return;
+	return {
+		element: first,
+		change
+	};
+}
+function recordParagraphBookmarks(index, paragraphIndex, endIndex, state) {
+	index.elements.forEach((element, position) => {
+		if (element.tag !== "w:bookmarkStart" && element.tag !== "w:bookmarkEnd") return;
+		const id = attr(element, "w:id");
+		if (id === void 0) return;
+		const leading = index.firstContentIndex === -1 || position < index.firstContentIndex;
+		const trailing = index.lastContentIndex === -1 || position > index.lastContentIndex;
+		if (element.tag === "w:bookmarkStart") {
+			const name = attr(element, "w:name");
+			state.bookmarkEvents.push({
+				id,
+				name: name === void 0 ? void 0 : decodeEntities(name),
+				kind: "start",
+				index: leading ? paragraphIndex : endIndex,
+				qualified: leading || trailing,
+				order: state.order++
+			});
+			return;
+		}
+		state.bookmarkEvents.push({
+			id,
+			name: void 0,
+			kind: "end",
+			index: trailing ? endIndex : paragraphIndex,
+			qualified: leading || trailing,
+			order: state.order++
+		});
+	});
+}
+function scanParagraphFields(index, paragraphIndex, endIndex, state) {
+	const content = contentBearingChildren(index);
+	content.forEach((child, position) => {
+		if (child.tag === "w:fldSimple") {
+			if (content.length === 1) state.extents.push({
+				startIndex: paragraphIndex,
+				endIndex,
+				order: state.order++,
+				descriptor: {
+					kind: "field",
+					instruction: decodeEntities(attr(child, "w:instr") ?? "")
+				}
+			});
+			return;
+		}
+		if (child.tag !== "w:r") return;
+		const type = fieldCharType(child);
+		if (type === "begin") {
+			state.openFields.push({
+				instruction: "",
+				inCode: true,
+				startIndex: paragraphIndex,
+				qualifiedStart: position === 0,
+				order: state.order++
+			});
+			return;
+		}
+		if (type === "separate") {
+			const open = state.openFields[state.openFields.length - 1];
+			if (open !== void 0) open.inCode = false;
+			return;
+		}
+		if (type === "end") {
+			const open = state.openFields.pop();
+			if (open !== void 0 && open.qualifiedStart && position === content.length - 1) state.extents.push({
+				startIndex: open.startIndex,
+				endIndex,
+				order: open.order,
+				descriptor: {
+					kind: "field",
+					instruction: open.instruction
+				}
+			});
+			return;
+		}
+		const open = state.openFields[state.openFields.length - 1];
+		if (open?.inCode === true) open.instruction += runInstructionText(child);
+	});
+}
+function collectParagraph(paragraph, ctx, state, carryDeletions) {
+	const index = indexParagraphContent(paragraph);
+	const tracked = wholeParagraphTrackedChange(index);
+	const paragraphDeleted = carryDeletions || tracked !== void 0 && isDeletedChange(tracked.change);
+	if (hasPageBreakBefore(paragraph)) state.blocks.push({ kind: "pageBreak" });
+	const paragraphIndex = state.blocks.length;
+	state.blocks.push(readParagraph(paragraph, ctx, paragraphDeleted));
+	state.blocks.push(...readParagraphImages(paragraph, ctx, paragraphDeleted));
+	const endIndex = state.blocks.length;
+	if (tracked !== void 0) state.extents.push({
+		startIndex: paragraphIndex,
+		endIndex,
+		order: state.order++,
+		descriptor: readProvenanceDescriptor(tracked.element, tracked.change)
+	});
+	recordParagraphBookmarks(index, paragraphIndex, endIndex, state);
+	scanParagraphFields(index, paragraphIndex, endIndex, state);
+	const pPr = childrenWithTag(paragraph, "w:pPr")[0];
+	const sectPr = pPr === void 0 ? void 0 : childrenWithTag(pPr, "w:sectPr")[0];
+	if (sectPr !== void 0) state.sectionBreaks.push({
+		index: state.blocks.length,
+		sectPr
+	});
+}
+function collectFlowNodes(nodes, ctx, state, carryDeletions) {
 	for (const node of nodes) {
 		if (node.type !== "element") continue;
 		if (node.tag === "w:p") {
-			if (hasPageBreakBefore(node)) blocks.push({ kind: "pageBreak" });
-			blocks.push(readParagraph(node, context, rels));
-			blocks.push(...readParagraphImages(node, rels, pkg));
-		} else if (node.tag === "w:tbl") blocks.push(readTable(node, context, rels, pkg));
-		else if (node.tag === "w:sdt") {
+			collectParagraph(node, ctx, state, carryDeletions);
+			continue;
+		}
+		if (node.tag === "w:tbl") {
+			state.blocks.push(readTable(node, ctx, carryDeletions));
+			continue;
+		}
+		if (node.tag === "w:sdt") {
+			const order = state.order++;
+			const startIndex = state.blocks.length;
 			const sdtContent = childrenWithTag(node, "w:sdtContent")[0];
-			if (sdtContent !== void 0) blocks.push(...readBodyBlocks(sdtContent.children, context, rels, pkg));
-		} else if (node.tag === "w:ins") blocks.push(...readBodyBlocks(node.children, context, rels, pkg));
-		else if (node.tag === "mc:AlternateContent") {
+			if (sdtContent !== void 0) collectFlowNodes(sdtContent.children, ctx, state, carryDeletions);
+			state.extents.push({
+				startIndex,
+				endIndex: state.blocks.length,
+				order,
+				descriptor: readContentControlDescriptor(node)
+			});
+			continue;
+		}
+		const change = PROVENANCE_CHANGE_BY_TAG.get(node.tag);
+		if (change !== void 0) {
+			const order = state.order++;
+			const startIndex = state.blocks.length;
+			collectFlowNodes(node.children, ctx, state, carryDeletions || isDeletedChange(change));
+			state.extents.push({
+				startIndex,
+				endIndex: state.blocks.length,
+				order,
+				descriptor: readProvenanceDescriptor(node, change)
+			});
+			continue;
+		}
+		if (node.tag === "mc:AlternateContent") {
 			const target = childrenWithTag(node, "mc:Fallback")[0] ?? childrenWithTag(node, "mc:Choice")[0];
-			if (target !== void 0) blocks.push(...readBodyBlocks(target.children, context, rels, pkg));
+			if (target !== void 0) collectFlowNodes(target.children, ctx, state, carryDeletions);
+			continue;
 		}
-	}
-	return blocks;
-}
-function readSections(body, context, rels, pkg) {
-	const sections = [];
-	let currentBlocks = [];
-	for (const node of body.children) {
-		if (node.type !== "element") continue;
-		if (node.tag === "w:sectPr") {
-			sections.push({
-				pageSize: readPageSize(node),
-				margins: readMargins(node),
-				blocks: currentBlocks
+		if (node.tag === "w:bookmarkStart") {
+			const id = attr(node, "w:id");
+			const name = attr(node, "w:name");
+			if (id !== void 0) state.bookmarkEvents.push({
+				id,
+				name: name === void 0 ? void 0 : decodeEntities(name),
+				kind: "start",
+				index: state.blocks.length,
+				qualified: true,
+				order: state.order++
 			});
-			currentBlocks = [];
 			continue;
 		}
-		if (node.tag === "w:p") {
-			const pPr = childrenWithTag(node, "w:pPr")[0];
-			const sectPr = pPr === void 0 ? void 0 : childrenWithTag(pPr, "w:sectPr")[0];
-			if (hasPageBreakBefore(node)) currentBlocks.push({ kind: "pageBreak" });
-			currentBlocks.push(readParagraph(node, context, rels));
-			currentBlocks.push(...readParagraphImages(node, rels, pkg));
-			if (sectPr !== void 0) {
-				sections.push({
-					pageSize: readPageSize(sectPr),
-					margins: readMargins(sectPr),
-					blocks: currentBlocks
-				});
-				currentBlocks = [];
-			}
+		if (node.tag === "w:bookmarkEnd") {
+			const id = attr(node, "w:id");
+			if (id !== void 0) state.bookmarkEvents.push({
+				id,
+				name: void 0,
+				kind: "end",
+				index: state.blocks.length,
+				qualified: true,
+				order: state.order++
+			});
 			continue;
 		}
-		currentBlocks.push(...readBodyBlocks([node], context, rels, pkg));
+		if (node.tag === "w:sectPr") state.sectionBreaks.push({
+			index: state.blocks.length,
+			sectPr: node
+		});
 	}
-	if (currentBlocks.length > 0 || sections.length === 0) sections.push({
-		pageSize: PAGE_SIZE_LETTER,
-		margins: DEFAULT_MARGINS,
-		blocks: currentBlocks
-	});
+}
+function readBlockScope(nodes, ctx, carryDeletions) {
+	const state = newFlowState();
+	collectFlowNodes(nodes, ctx, state, carryDeletions);
+	return insertConstructMarkers(state.blocks, [...state.extents, ...resolveBookmarkExtents(state.bookmarkEvents)]);
+}
+function readSections(body, ctx) {
+	const state = newFlowState();
+	collectFlowNodes(body.children, ctx, state, false);
+	const extents = [...state.extents, ...resolveBookmarkExtents(state.bookmarkEvents)];
+	function sliceSection(pageSize, margins, from, to) {
+		const contained = extents.filter((extent) => extent.startIndex >= from && extent.endIndex <= to).map((extent) => ({
+			...extent,
+			startIndex: extent.startIndex - from,
+			endIndex: extent.endIndex - from
+		}));
+		return {
+			pageSize,
+			margins,
+			blocks: insertConstructMarkers(state.blocks.slice(from, to), contained)
+		};
+	}
+	const sections = [];
+	let from = 0;
+	for (const sectionBreak of state.sectionBreaks) {
+		sections.push(sliceSection(readPageSize(sectionBreak.sectPr), readMargins(sectionBreak.sectPr), from, sectionBreak.index));
+		from = sectionBreak.index;
+	}
+	if (from < state.blocks.length || sections.length === 0) sections.push(sliceSection(PAGE_SIZE_LETTER, DEFAULT_MARGINS, from, state.blocks.length));
 	sections.forEach((section, sectionIndex) => assignSourcePaths(section.blocks, `sections[${sectionIndex}]`));
 	return sections;
 }
@@ -423,13 +620,17 @@ function readDocx(pkg) {
 	const body = childrenWithTag(documentRoot, "w:body")[0];
 	if (body === void 0) throw new Error(`readDocx: ${DOCUMENT_PART_PATH} has no w:body element`);
 	const docRels = resolveRelationships(pkg, DOCUMENT_PART_PATH);
-	const context = {
-		stylesRoot: rootElement(pkg.parts[STYLES_PART_PATH]),
-		theme: readDocumentTheme(pkg, docRels)
+	const ctx = {
+		styles: {
+			stylesRoot: rootElement(pkg.parts[STYLES_PART_PATH]),
+			theme: readDocumentTheme(pkg, docRels)
+		},
+		rels: docRels,
+		pkg
 	};
 	return {
 		metadata: readCoreProperties(pkg),
-		sections: readSections(body, context, docRels, pkg),
+		sections: readSections(body, ctx),
 		comments: readComments(pkg),
 		footnotes: readFootnotes(pkg),
 		headers: readHeaderFooterText(pkg, "word/header"),