js.documents 14.5.0 → 14.6.1

package/dist/layout/reading-order.js

@@ -0,0 +1,200 @@
+import { clusterIntoLines } from "./reconstruct-lines.js";
+import { resolveStandardFont } from "pdf-codec/fonts";
+import { afmWidthOfCode } from "pdf-codec/afm-widths";
+//#region src/layout/reading-order.ts
+const MIN_COLUMN_LINES = 3;
+const MIN_PROSE_MEASURE_CHARS = 20;
+const SPACE_CHAR_CODE = 32;
+const AFM_UNITS_PER_EM = 1e3;
+/**
+* Splits a page's text items into reading-order flows at the column gutters the page's geometry states.
+* @param items - The page's text items in content-stream order, after any table recovery has removed the text it claimed.
+* @returns The flows in reading order, and a locator placing anything else among them. A page with no gutter yields one flow whose lines are `clusterIntoLines(items)`.
+*/
+function readingOrder(items) {
+	const bands = segment(items);
+	const regions = [];
+	collectRegions(bands, [], regions);
+	return {
+		regions,
+		locate: (xPt, yPt) => locate(bands, xPt, yPt)
+	};
+}
+/**
+* Orders two placed blocks for reading: by reading-order path first, then top to bottom by y.
+* @returns Negative when `a` is read first, positive when `b` is, zero when neither precedes the other.
+*/
+function compareReadingPosition(a, b) {
+	const length = Math.min(a.path.length, b.path.length);
+	for (let i = 0; i < length; i++) {
+		const difference = a.path[i] - b.path[i];
+		if (difference !== 0) return difference;
+	}
+	return a.path.length - b.path.length || b.yPt - a.yPt;
+}
+function segment(items) {
+	const lines = clusterIntoLines(items);
+	const split = findColumnSplit(lines);
+	if (split === void 0) return [flowBand(lines)];
+	const before = lines.slice(0, split.startLine);
+	const band = lines.slice(split.startLine, split.endLine);
+	const after = lines.slice(split.endLine);
+	return [
+		...before.length > 0 ? segment(itemsOf(before)) : [],
+		{
+			kind: "columns",
+			gutterXPt: (split.loPt + split.hiPt) / 2,
+			left: segment(split.left),
+			right: segment(split.right),
+			topPt: band[0].baselineY,
+			bottomPt: band.at(-1).baselineY
+		},
+		...after.length > 0 ? segment(itemsOf(after)) : []
+	];
+}
+function flowBand(lines) {
+	return {
+		kind: "flow",
+		lines,
+		topPt: lines[0]?.baselineY ?? Number.POSITIVE_INFINITY,
+		bottomPt: lines.at(-1)?.baselineY ?? Number.NEGATIVE_INFINITY
+	};
+}
+function itemsOf(lines) {
+	return lines.flatMap((line) => line.items);
+}
+function collectRegions(bands, prefix, into) {
+	bands.forEach((band, index) => {
+		if (band.kind === "flow") into.push({
+			lines: band.lines,
+			path: [...prefix, index]
+		});
+		else {
+			collectRegions(band.left, [
+				...prefix,
+				index,
+				0
+			], into);
+			collectRegions(band.right, [
+				...prefix,
+				index,
+				1
+			], into);
+		}
+	});
+}
+function locate(bands, xPt, yPt) {
+	let index = bands.length - 1;
+	for (let i = 0; i + 1 < bands.length; i++) if (yPt >= (bands[i].bottomPt + bands[i + 1].topPt) / 2) {
+		index = i;
+		break;
+	}
+	const band = bands[index];
+	if (band.kind === "flow") return [index];
+	const side = xPt < band.gutterXPt ? 0 : 1;
+	return [
+		index,
+		side,
+		...locate(side === 0 ? band.left : band.right, xPt, yPt)
+	];
+}
+function findColumnSplit(lines) {
+	const items = itemsOf(lines);
+	if (lines.length < MIN_COLUMN_LINES || !items.every(isMeasurable)) return;
+	const runs = gutterRuns(lines, Math.min(...items.map(spaceAdvancePt))).sort((a, b) => b.endLine - b.startLine - (a.endLine - a.startLine) || b.hiPt - b.loPt - (a.hiPt - a.loPt) || a.startLine - b.startLine);
+	for (const run of runs) {
+		const runItems = itemsOf(lines.slice(run.startLine, run.endLine));
+		if (run.hiPt - run.loPt <= Math.max(...runItems.map(spaceAdvancePt))) continue;
+		const left = runItems.filter((item) => rightEdge(item) <= run.loPt);
+		const right = runItems.filter((item) => item.xPt >= run.hiPt);
+		if (isRunningProse(left, run.hiPt - run.loPt) && isRunningProse(right, run.hiPt - run.loPt)) return {
+			...run,
+			left,
+			right
+		};
+	}
+}
+function isMeasurable(item) {
+	return item.widthPt !== void 0 && item.writingMode === void 0 && (item.rotationDeg ?? 0) === 0;
+}
+function rightEdge(item) {
+	return item.xPt + (item.widthPt ?? 0);
+}
+function spaceAdvancePt(item) {
+	const { standardName } = resolveStandardFont(item.font.family, item.font.weight === "bold", item.font.style === "italic");
+	const advance = afmWidthOfCode(standardName, SPACE_CHAR_CODE);
+	if (advance === void 0) throw new Error(`${standardName} has no AFM width for the space glyph`);
+	return advance / AFM_UNITS_PER_EM * item.sizePt;
+}
+function freeSpans(line) {
+	const sorted = [...line.items].sort((a, b) => a.xPt - b.xPt);
+	const spans = [];
+	let reachedPt = Number.NEGATIVE_INFINITY;
+	for (const item of sorted) {
+		if (item.xPt > reachedPt) spans.push({
+			loPt: reachedPt,
+			hiPt: item.xPt
+		});
+		reachedPt = Math.max(reachedPt, rightEdge(item));
+	}
+	spans.push({
+		loPt: reachedPt,
+		hiPt: Number.POSITIVE_INFINITY
+	});
+	return spans;
+}
+function gutterRuns(lines, minWidthPt) {
+	const runs = [];
+	let open = /* @__PURE__ */ new Map();
+	const record = (span, endLine) => {
+		if (Number.isFinite(span.loPt) && Number.isFinite(span.hiPt) && endLine - span.startLine >= MIN_COLUMN_LINES) runs.push({
+			...span,
+			endLine
+		});
+	};
+	lines.forEach((line, lineIndex) => {
+		const next = /* @__PURE__ */ new Map();
+		const keep = (loPt, hiPt, startLine) => {
+			if (hiPt - loPt <= minWidthPt) return;
+			const key = `${String(loPt)}|${String(hiPt)}`;
+			const existing = next.get(key);
+			if (existing === void 0 || startLine < existing.startLine) next.set(key, {
+				loPt,
+				hiPt,
+				startLine
+			});
+		};
+		const spans = freeSpans(line);
+		for (const strip of open.values()) for (const span of spans) keep(Math.max(strip.loPt, span.loPt), Math.min(strip.hiPt, span.hiPt), strip.startLine);
+		for (const span of spans) keep(span.loPt, span.hiPt, lineIndex);
+		for (const [key, strip] of open) if (next.get(key)?.startLine !== strip.startLine) record(strip, lineIndex);
+		open = next;
+	});
+	for (const strip of open.values()) record(strip, lines.length);
+	return runs;
+}
+function isRunningProse(items, gutterWidthPt) {
+	const lines = clusterIntoLines(items);
+	if (lines.length < MIN_COLUMN_LINES) return false;
+	const characters = items.reduce((sum, item) => sum + item.text.length, 0);
+	const advancePt = items.reduce((sum, item) => sum + (item.widthPt ?? 0), 0);
+	if (characters === 0 || advancePt === 0) return false;
+	const leftPt = Math.min(...items.map((item) => item.xPt));
+	const rightPt = Math.max(...items.map(rightEdge));
+	if ((rightPt - leftPt) / (advancePt / characters) < MIN_PROSE_MEASURE_CHARS) return false;
+	for (const line of lines) for (const span of freeSpans(line)) if (Number.isFinite(span.loPt) && Number.isFinite(span.hiPt) && span.hiPt - span.loPt >= gutterWidthPt) return false;
+	let wrapped = 0;
+	for (let i = 0; i + 1 < lines.length; i++) {
+		const lineRightPt = Math.max(...lines[i].items.map(rightEdge));
+		const nextFirst = lines[i + 1].items[0];
+		if (rightPt - lineRightPt < firstWordWidthPt(nextFirst) + spaceAdvancePt(nextFirst)) wrapped++;
+	}
+	return wrapped * 2 > lines.length - 1;
+}
+function firstWordWidthPt(item) {
+	const text = item.text.trimStart();
+	const firstWord = text.split(/\s/u)[0] ?? text;
+	return text.length === 0 ? 0 : (item.widthPt ?? 0) * firstWord.length / item.text.length;
+}
+//#endregion
+export { compareReadingPosition, readingOrder };

package/dist/layout/reconstruct-presentations-prod.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CA4no_aw.cjs");
+const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CsvGlvP2.cjs");
 exports.layoutItemToVector = require_reconstruct_presentations_prod.layoutItemToVector;
 exports.lineToParagraph = require_reconstruct_presentations_prod.lineToParagraph;
 exports.reconstructDrawing = require_reconstruct_presentations_prod.reconstructDrawing;

package/dist/layout/reconstruct-presentations-prod.js

@@ -1,2 +1,2 @@
-import { a as reconstructSheet, i as reconstructPresentation, n as lineToParagraph, o as recoverPageVectors, r as reconstructDrawing, s as splitLineByLargeGaps, t as layoutItemToVector } from "../reconstruct-presentations-prod-CC8ct-5O.js";
+import { a as reconstructSheet, i as reconstructPresentation, n as lineToParagraph, o as recoverPageVectors, r as reconstructDrawing, s as splitLineByLargeGaps, t as layoutItemToVector } from "../reconstruct-presentations-prod-D1KbGPxa.js";
 export { layoutItemToVector, lineToParagraph, reconstructDrawing, reconstructPresentation, reconstructSheet, recoverPageVectors, splitLineByLargeGaps };

package/dist/layout/reconstruct-tables.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CA4no_aw.cjs");
+const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CsvGlvP2.cjs");
 exports.buildGridFromLattice = require_reconstruct_presentations_prod.buildGridFromLattice;
 exports.buildGridFromTextClustering = require_reconstruct_presentations_prod.buildGridFromTextClustering;
 exports.recoverTables = require_reconstruct_presentations_prod.recoverTables;

package/dist/layout/reconstruct-tables.js

@@ -1,2 +1,2 @@
-import { b as recoverTables, v as buildGridFromLattice, y as buildGridFromTextClustering } from "../reconstruct-presentations-prod-CC8ct-5O.js";
+import { b as recoverTables, v as buildGridFromLattice, y as buildGridFromTextClustering } from "../reconstruct-presentations-prod-D1KbGPxa.js";
 export { buildGridFromLattice, buildGridFromTextClustering, recoverTables };

package/dist/layout/reconstruct.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CA4no_aw.cjs");
+const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CsvGlvP2.cjs");
 exports.CELL_TYPES = require_reconstruct_presentations_prod.CELL_TYPES;
 exports.DEFAULT_COLUMN_WIDTH_FALLBACK_PT = require_reconstruct_presentations_prod.DEFAULT_COLUMN_WIDTH_FALLBACK_PT;
 exports.NO_ITEMS = require_reconstruct_presentations_prod.NO_ITEMS;

package/dist/layout/reconstruct.js

@@ -1,2 +1,2 @@
-import { _ as reconstructWordprocessing, c as CELL_TYPES, d as ROW_TYPES, f as TABLE_TYPES, g as reconstructSpreadsheet, h as nearestOfType, l as DEFAULT_COLUMN_WIDTH_FALLBACK_PT, m as indexStructure, p as ZERO_MARGINS, u as NO_ITEMS } from "../reconstruct-presentations-prod-CC8ct-5O.js";
+import { _ as reconstructWordprocessing, c as CELL_TYPES, d as ROW_TYPES, f as TABLE_TYPES, g as reconstructSpreadsheet, h as nearestOfType, l as DEFAULT_COLUMN_WIDTH_FALLBACK_PT, m as indexStructure, p as ZERO_MARGINS, u as NO_ITEMS } from "../reconstruct-presentations-prod-D1KbGPxa.js";
 export { CELL_TYPES, DEFAULT_COLUMN_WIDTH_FALLBACK_PT, NO_ITEMS, ROW_TYPES, TABLE_TYPES, ZERO_MARGINS, indexStructure, nearestOfType, reconstructSpreadsheet, reconstructWordprocessing };

package/dist/{reconstruct-presentations-prod-CA4no_aw.cjs → reconstruct-presentations-prod-CsvGlvP2.cjs}

@@ -5,6 +5,7 @@ const require_layout_cell_typing = require("./layout/cell-typing.cjs");
 const require_ports_abort = require("./ports/abort.cjs");
 const require_layout_reconstruct_lines = require("./layout/reconstruct-lines.cjs");
 const require_layout_lattice = require("./layout/lattice.cjs");
+const require_layout_reading_order = require("./layout/reading-order.cjs");
 let document_schema_js = require("document-schema.js");
 let pdf_codec_layout = require("pdf-codec/layout");
 //#region src/layout/reconstruct-tables.ts
@@ -519,19 +520,19 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 	}
 	const textItems = page.items.filter((i) => i.kind === "text" && !consumedText.has(i));
 	const imageItems = page.items.filter((i) => i.kind === "image");
-	const lines = require_layout_reconstruct_lines.clusterIntoLines(textItems);
-	const paragraphs = require_layout_reconstruct_lines.clusterIntoParagraphs(lines);
+	const order = require_layout_reading_order.readingOrder(textItems);
 	const recordDivisions = (block, items) => {
 		if (structure !== void 0) divisionByBlock.set(block, divisionChainOf(items, structure));
 	};
 	const positioned = [];
-	for (const paragraph of paragraphs) {
+	for (const { lines, path } of order.regions) for (const paragraph of require_layout_reconstruct_lines.clusterIntoParagraphs(lines)) {
 		const paragraphItems = paragraph.lines.flatMap((line) => line.items);
 		const structural = structure === void 0 ? void 0 : structureHeadingLevel(paragraphItems, structure);
 		const headingLevel = structural !== void 0 ? structural.level : headingLevelOf(paragraph, headingLevels);
 		const block = require_layout_reconstruct_lines.paragraphToContentParagraph(paragraph, pageIndex, headingLevel);
 		recordDivisions(block, paragraphItems);
 		positioned.push({
+			path,
 			yPt: paragraph.lines[0].baselineY,
 			block
 		});
@@ -554,6 +555,7 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 			});
 			recordDivisions(block, [img]);
 			positioned.push({
+				path: order.locate(img.xPt + img.widthPt / 2, img.yPt),
 				yPt: img.yPt,
 				block
 			});
@@ -562,6 +564,7 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 	for (const recovered of recoveredTables) {
 		recordDivisions(recovered.table, [...recovered.consumedText]);
 		positioned.push({
+			path: order.locate(recovered.frame.xPt + recovered.frame.widthPt / 2, recovered.topYPt),
 			yPt: recovered.topYPt,
 			block: recovered.table
 		});
@@ -571,11 +574,12 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 		const vectorItems = page.items.filter((item) => !claimedItems.has(item) && layoutItemToVector(item, page.heightPt) !== void 0);
 		recordDivisions(recoveredVectors.block, vectorItems);
 		positioned.push({
+			path: order.locate(Number.NEGATIVE_INFINITY, recoveredVectors.topYPt),
 			yPt: recoveredVectors.topYPt,
 			block: recoveredVectors.block
 		});
 	}
-	positioned.sort((a, b) => b.yPt - a.yPt);
+	positioned.sort(require_layout_reading_order.compareReadingPosition);
 	const blocks = positioned.map((p) => p.block);
 	reconcileLinks(blocks, page, pageIndex);
 	appendAnnotationConstructs(blocks, page, pageIndex);
@@ -772,7 +776,7 @@ function reconstructSlide(page, pageIndex, images, structure) {
 	}
 	const textItems = page.items.filter((i) => i.kind === "text" && !consumedText.has(i));
 	const imageItems = page.items.filter((i) => i.kind === "image");
-	const textShapes = clusterIntoBlocks(require_layout_reconstruct_lines.clusterIntoLines(textItems)).map((block) => blockToShape(block, page.heightPt, pageIndex));
+	const textShapes = require_layout_reading_order.readingOrder(textItems).regions.flatMap((region) => clusterIntoBlocks(region.lines)).map((block) => blockToShape(block, page.heightPt, pageIndex));
 	const imageShapes = [];
 	for (const img of imageItems) {
 		const shape = imageToShape(img, page.heightPt, pageIndex, images);

package/dist/{reconstruct-presentations-prod-CC8ct-5O.js → reconstruct-presentations-prod-D1KbGPxa.js}

@@ -5,6 +5,7 @@ import { inferCellValue } from "./layout/cell-typing.js";
 import { throwIfAborted } from "./ports/abort.js";
 import { HALF_POINT_BUCKET_PT, bucketCounts, clusterIntoLines, clusterIntoParagraphs, estimateModalLineSpacing, fontSizesClose, gapExceeds, lineBox, modeOf, paragraphToContentParagraph, pushRunsForLine, textBoxOfItem, textItemToContentRun, textItemVerticalExtent } from "./layout/reconstruct-lines.js";
 import { detectGridLattice, findColumnIndex, findRowIndex } from "./layout/lattice.js";
+import { compareReadingPosition, readingOrder } from "./layout/reading-order.js";
 import { denseTableRows } from "document-schema.js";
 import { sourcePageIndex } from "pdf-codec/layout";
 //#region src/layout/reconstruct-tables.ts
@@ -519,19 +520,19 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 	}
 	const textItems = page.items.filter((i) => i.kind === "text" && !consumedText.has(i));
 	const imageItems = page.items.filter((i) => i.kind === "image");
-	const lines = clusterIntoLines(textItems);
-	const paragraphs = clusterIntoParagraphs(lines);
+	const order = readingOrder(textItems);
 	const recordDivisions = (block, items) => {
 		if (structure !== void 0) divisionByBlock.set(block, divisionChainOf(items, structure));
 	};
 	const positioned = [];
-	for (const paragraph of paragraphs) {
+	for (const { lines, path } of order.regions) for (const paragraph of clusterIntoParagraphs(lines)) {
 		const paragraphItems = paragraph.lines.flatMap((line) => line.items);
 		const structural = structure === void 0 ? void 0 : structureHeadingLevel(paragraphItems, structure);
 		const headingLevel = structural !== void 0 ? structural.level : headingLevelOf(paragraph, headingLevels);
 		const block = paragraphToContentParagraph(paragraph, pageIndex, headingLevel);
 		recordDivisions(block, paragraphItems);
 		positioned.push({
+			path,
 			yPt: paragraph.lines[0].baselineY,
 			block
 		});
@@ -554,6 +555,7 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 			});
 			recordDivisions(block, [img]);
 			positioned.push({
+				path: order.locate(img.xPt + img.widthPt / 2, img.yPt),
 				yPt: img.yPt,
 				block
 			});
@@ -562,6 +564,7 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 	for (const recovered of recoveredTables) {
 		recordDivisions(recovered.table, [...recovered.consumedText]);
 		positioned.push({
+			path: order.locate(recovered.frame.xPt + recovered.frame.widthPt / 2, recovered.topYPt),
 			yPt: recovered.topYPt,
 			block: recovered.table
 		});
@@ -571,11 +574,12 @@ function reconstructPageBlocks(page, pageIndex, images, headingLevels, form, str
 		const vectorItems = page.items.filter((item) => !claimedItems.has(item) && layoutItemToVector(item, page.heightPt) !== void 0);
 		recordDivisions(recoveredVectors.block, vectorItems);
 		positioned.push({
+			path: order.locate(Number.NEGATIVE_INFINITY, recoveredVectors.topYPt),
 			yPt: recoveredVectors.topYPt,
 			block: recoveredVectors.block
 		});
 	}
-	positioned.sort((a, b) => b.yPt - a.yPt);
+	positioned.sort(compareReadingPosition);
 	const blocks = positioned.map((p) => p.block);
 	reconcileLinks(blocks, page, pageIndex);
 	appendAnnotationConstructs(blocks, page, pageIndex);
@@ -772,7 +776,7 @@ function reconstructSlide(page, pageIndex, images, structure) {
 	}
 	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 textShapes = readingOrder(textItems).regions.flatMap((region) => clusterIntoBlocks(region.lines)).map((block) => blockToShape(block, page.heightPt, pageIndex));
 	const imageShapes = [];
 	for (const img of imageItems) {
 		const shape = imageToShape(img, page.heightPt, pageIndex, images);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "js.documents",
-  "version": "14.5.0",
+  "version": "14.6.1",
   "description": "Converts between docx, pptx, odt, odp, ods, odg, xlsx, csv, svg, markdown, rtf, doc, xls, ppt and PDF through one shared, Zod-typed content schema (wpd read-only), plus read+write editors for OOXML and OpenDocument. Worker-isomorphic.",
   "type": "module",
   "repository": {
@@ -86,7 +86,7 @@
     "markdown-codec": "11.1.16",
     "odf.js": "11.2.2",
     "ooxml.js": "12.2.32",
-    "pdf-codec": "5.4.0",
+    "pdf-codec": "5.5.0",
     "ppt-codec": "2.2.23",
     "rtf-codec": "8.1.24",
     "temml": "0.13.4",