js.documents 7.20.34 → 7.20.36

package/dist/layout/reconstruct.cjs

@@ -7,6 +7,7 @@ const require_ports_abort = require("../ports/abort.cjs");
 const require_layout_lattice = require("./lattice.cjs");
 let pdf_codec_fonts = require("pdf-codec/fonts");
 let pdf_codec_afm_widths = require("pdf-codec/afm-widths");
+let pdf_codec_text_group = require("pdf-codec/text-group");
 //#region src/layout/reconstruct.ts
 const ZERO_MARGINS = {
 	topPt: 0,
@@ -15,6 +16,9 @@ const ZERO_MARGINS = {
 	leftPt: 0
 };
 const LINE_BASELINE_TOLERANCE_FACTOR = .5;
+function sharesBaseline(anchor, item) {
+	return (0, pdf_codec_text_group.runsShareBaseline)(anchor, item, { baselineToleranceEm: LINE_BASELINE_TOLERANCE_FACTOR });
+}
 const DUPLICATE_PAINT_BUCKET_PT = .1;
 function duplicatePaintKey(item) {
 	const xBucket = Math.round(item.xPt / DUPLICATE_PAINT_BUCKET_PT);
@@ -50,10 +54,9 @@ const REDRAW_PASS_SIMILARITY_THRESHOLD = .6;
 function groupByBaselineInDocumentOrder(items) {
 	const groups = [];
 	for (const item of items) {
-		const tolerance = LINE_BASELINE_TOLERANCE_FACTOR * item.sizePt;
-		const group = groups.find((g) => Math.abs(g.baselineY - item.yPt) <= tolerance);
+		const group = groups.find((g) => sharesBaseline(g.anchor, item));
 		if (group === void 0) groups.push({
-			baselineY: item.yPt,
+			anchor: item,
 			items: [item]
 		});
 		else group.items.push(item);
@@ -126,11 +129,11 @@ function clusterIntoLines(items) {
 	const sorted = [...dropDuplicatePaints(dropFuzzyRedrawnPasses(items))].sort((a, b) => b.yPt - a.yPt || a.xPt - b.xPt);
 	const working = [];
 	for (const item of sorted) {
-		const tolerance = LINE_BASELINE_TOLERANCE_FACTOR * item.sizePt;
-		const line = working.find((l) => Math.abs(l.baselineY - item.yPt) <= tolerance);
+		const line = working.find((l) => sharesBaseline(l.anchor, item));
 		if (line === void 0) working.push({
 			items: [item],
-			baselineY: item.yPt
+			baselineY: item.yPt,
+			anchor: item
 		});
 		else line.items.push(item);
 	}
@@ -160,6 +163,10 @@ function textItemToContentRun(item) {
 	};
 }
 const MIN_WORD_GAP_PT = .5;
+function gapExceeds(previous, next, thresholdPt) {
+	const gap = (0, pdf_codec_text_group.runGapPt)(previous, next);
+	return gap !== void 0 && gap > thresholdPt;
+}
 function textBoxOfItem(item, pageIndex) {
 	const { ascentPt, descentPt } = textItemVerticalExtent(item);
 	return {
@@ -194,9 +201,8 @@ function pushRunsForLine(runs, line, pageIndex) {
 	line.items.forEach((item, itemIndex) => {
 		if (itemIndex > 0) {
 			const prevItem = line.items[itemIndex - 1];
-			const gap = item.xPt - (prevItem.xPt + (prevItem.widthPt ?? 0));
-			if (gap > LARGE_GAP_EM_MULTIPLIER * item.sizePt) runs.push({ text: "	" });
-			else if (gap > MIN_WORD_GAP_PT) runs[runs.length - 1].text += " ";
+			if (gapExceeds(prevItem, item, LARGE_GAP_EM_MULTIPLIER * item.sizePt)) runs.push({ text: "	" });
+			else if (gapExceeds(prevItem, item, MIN_WORD_GAP_PT)) runs[runs.length - 1].text += " ";
 		}
 		const run = textItemToContentRun(item);
 		require_layout_shared.stampFrame(run, pageIndex, textBoxOfItem(item, pageIndex));
@@ -720,7 +726,7 @@ function splitLineByLargeGaps(line) {
 	line.items.forEach((item, i) => {
 		if (i > 0) {
 			const prev = line.items[i - 1];
-			if (item.xPt - (prev.xPt + (prev.widthPt ?? 0)) > LARGE_GAP_EM_MULTIPLIER * item.sizePt) {
+			if (gapExceeds(prev, item, LARGE_GAP_EM_MULTIPLIER * item.sizePt)) {
 				segments.push({
 					items: current,
 					baselineY: line.baselineY
@@ -1340,7 +1346,7 @@ function joinCellText(items) {
 	sorted.forEach((item, i) => {
 		if (i > 0) {
 			const prev = sorted[i - 1];
-			if (item.xPt - (prev.xPt + (prev.widthPt ?? 0)) > MIN_WORD_GAP_PT) text += " ";
+			if (gapExceeds(prev, item, MIN_WORD_GAP_PT)) text += " ";
 		}
 		text += item.text;
 	});

package/dist/layout/reconstruct.js

@@ -6,6 +6,7 @@ import { throwIfAborted } from "../ports/abort.js";
 import { detectGridLattice, findColumnIndex, findRowIndex } from "./lattice.js";
 import { resolveStandardFont } from "pdf-codec/fonts";
 import { STANDARD_METRICS } from "pdf-codec/afm-widths";
+import { runGapPt, runsShareBaseline } from "pdf-codec/text-group";
 //#region src/layout/reconstruct.ts
 const ZERO_MARGINS = {
 	topPt: 0,
@@ -14,6 +15,9 @@ const ZERO_MARGINS = {
 	leftPt: 0
 };
 const LINE_BASELINE_TOLERANCE_FACTOR = .5;
+function sharesBaseline(anchor, item) {
+	return runsShareBaseline(anchor, item, { baselineToleranceEm: LINE_BASELINE_TOLERANCE_FACTOR });
+}
 const DUPLICATE_PAINT_BUCKET_PT = .1;
 function duplicatePaintKey(item) {
 	const xBucket = Math.round(item.xPt / DUPLICATE_PAINT_BUCKET_PT);
@@ -49,10 +53,9 @@ const REDRAW_PASS_SIMILARITY_THRESHOLD = .6;
 function groupByBaselineInDocumentOrder(items) {
 	const groups = [];
 	for (const item of items) {
-		const tolerance = LINE_BASELINE_TOLERANCE_FACTOR * item.sizePt;
-		const group = groups.find((g) => Math.abs(g.baselineY - item.yPt) <= tolerance);
+		const group = groups.find((g) => sharesBaseline(g.anchor, item));
 		if (group === void 0) groups.push({
-			baselineY: item.yPt,
+			anchor: item,
 			items: [item]
 		});
 		else group.items.push(item);
@@ -125,11 +128,11 @@ function clusterIntoLines(items) {
 	const sorted = [...dropDuplicatePaints(dropFuzzyRedrawnPasses(items))].sort((a, b) => b.yPt - a.yPt || a.xPt - b.xPt);
 	const working = [];
 	for (const item of sorted) {
-		const tolerance = LINE_BASELINE_TOLERANCE_FACTOR * item.sizePt;
-		const line = working.find((l) => Math.abs(l.baselineY - item.yPt) <= tolerance);
+		const line = working.find((l) => sharesBaseline(l.anchor, item));
 		if (line === void 0) working.push({
 			items: [item],
-			baselineY: item.yPt
+			baselineY: item.yPt,
+			anchor: item
 		});
 		else line.items.push(item);
 	}
@@ -159,6 +162,10 @@ function textItemToContentRun(item) {
 	};
 }
 const MIN_WORD_GAP_PT = .5;
+function gapExceeds(previous, next, thresholdPt) {
+	const gap = runGapPt(previous, next);
+	return gap !== void 0 && gap > thresholdPt;
+}
 function textBoxOfItem(item, pageIndex) {
 	const { ascentPt, descentPt } = textItemVerticalExtent(item);
 	return {
@@ -193,9 +200,8 @@ function pushRunsForLine(runs, line, pageIndex) {
 	line.items.forEach((item, itemIndex) => {
 		if (itemIndex > 0) {
 			const prevItem = line.items[itemIndex - 1];
-			const gap = item.xPt - (prevItem.xPt + (prevItem.widthPt ?? 0));
-			if (gap > LARGE_GAP_EM_MULTIPLIER * item.sizePt) runs.push({ text: "	" });
-			else if (gap > MIN_WORD_GAP_PT) runs[runs.length - 1].text += " ";
+			if (gapExceeds(prevItem, item, LARGE_GAP_EM_MULTIPLIER * item.sizePt)) runs.push({ text: "	" });
+			else if (gapExceeds(prevItem, item, MIN_WORD_GAP_PT)) runs[runs.length - 1].text += " ";
 		}
 		const run = textItemToContentRun(item);
 		stampFrame(run, pageIndex, textBoxOfItem(item, pageIndex));
@@ -719,7 +725,7 @@ function splitLineByLargeGaps(line) {
 	line.items.forEach((item, i) => {
 		if (i > 0) {
 			const prev = line.items[i - 1];
-			if (item.xPt - (prev.xPt + (prev.widthPt ?? 0)) > LARGE_GAP_EM_MULTIPLIER * item.sizePt) {
+			if (gapExceeds(prev, item, LARGE_GAP_EM_MULTIPLIER * item.sizePt)) {
 				segments.push({
 					items: current,
 					baselineY: line.baselineY
@@ -1339,7 +1345,7 @@ function joinCellText(items) {
 	sorted.forEach((item, i) => {
 		if (i > 0) {
 			const prev = sorted[i - 1];
-			if (item.xPt - (prev.xPt + (prev.widthPt ?? 0)) > MIN_WORD_GAP_PT) text += " ";
+			if (gapExceeds(prev, item, MIN_WORD_GAP_PT)) text += " ";
 		}
 		text += item.text;
 	});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "js.documents",
-  "version": "7.20.34",
+  "version": "7.20.36",
   "description": "Bidirectional docx/pptx <-> PDF conversion and a read+write editable OOXML document model, built on ooxml.js and Zod 4 codecs.",
   "type": "module",
   "repository": {
@@ -86,7 +86,7 @@
     "markdown-codec": "7.0.1",
     "odf.js": "8.0.1",
     "ooxml.js": "9.1.2",
-    "pdf-codec": "5.0.1",
+    "pdf-codec": "5.2.0",
     "ppt-codec": "2.0.1",
     "rtf-codec": "5.0.1",
     "temml": "0.13.4",