reamkit 1.5.0 → 1.6.0

package/dist/esm/pdf-reader/tagged.js

@@ -0,0 +1,88 @@
+import { pt } from "../core/ir/units.js";
+import { buildFlowDoc, paragraphBlock } from "./flow-build.js";
+import { extractPageText } from "./text.js";
+import { readStructTree } from "./struct-tree.js";
+//#region src/pdf-reader/tagged.ts
+var ASSUMED_CONTENT_WIDTH_PT = 468;
+function reconstructTaggedPdf(file) {
+	const root = readStructTree(file);
+	if (!root) return void 0;
+	const pageText = file.pages().map((page) => {
+		const byMcid = /* @__PURE__ */ new Map();
+		for (const run of extractPageText(file, page)) if (run.mcid !== void 0) byMcid.set(run.mcid, (byMcid.get(run.mcid) ?? "") + run.text);
+		return byMcid;
+	});
+	const textOf = (node) => squash(node.mcids.map(({ page, mcid }) => pageText[page]?.get(mcid) ?? "").join(" "));
+	const collectText = (node) => squash([textOf(node), ...node.children.map(collectText)].join(" "));
+	function emit(node, out) {
+		if (node.type === "Table") {
+			const table = buildTable(node);
+			if (table) out.push(table);
+			return;
+		}
+		if (node.type === "LI") {
+			const text = collectText(node);
+			if (text.length > 0) out.push(paragraphBlock(text, void 0));
+			return;
+		}
+		if (node.children.length === 0) {
+			const text = textOf(node);
+			if (text.length > 0) out.push(paragraphBlock(text, headingLevel(node.type)));
+			return;
+		}
+		for (const child of node.children) emit(child, out);
+	}
+	function buildTable(tableNode) {
+		const rows = [];
+		const collectRows = (n) => {
+			for (const child of n.children) if (child.type === "TR") rows.push(buildRow(child));
+			else if (child.type === "THead" || child.type === "TBody" || child.type === "TFoot") collectRows(child);
+		};
+		collectRows(tableNode);
+		if (rows.length === 0) return void 0;
+		const numCols = Math.max(1, ...rows.map((r) => r.cells.reduce((s, c) => s + (c.properties.colSpan ?? 1), 0)));
+		const colWidth = pt(Math.max(1, ASSUMED_CONTENT_WIDTH_PT / numCols));
+		return {
+			kind: "table",
+			table: {
+				properties: {},
+				grid: Array.from({ length: numCols }, () => colWidth),
+				rows
+			}
+		};
+	}
+	function buildRow(trNode) {
+		const cells = [];
+		let allHeader = false;
+		for (const cell of trNode.children) {
+			if (cell.type !== "TH" && cell.type !== "TD") continue;
+			if (cells.length === 0) allHeader = true;
+			if (cell.type !== "TH") allHeader = false;
+			const content = [];
+			for (const child of cell.children) emit(child, content);
+			if (content.length === 0) content.push(paragraphBlock(textOf(cell), void 0));
+			const colSpan = cell.colSpan ?? 1;
+			cells.push({
+				properties: colSpan > 1 ? { colSpan } : {},
+				content
+			});
+		}
+		return {
+			properties: allHeader && cells.length > 0 ? { isHeader: true } : {},
+			cells
+		};
+	}
+	const body = [];
+	emit(root, body);
+	if (body.length === 0) return void 0;
+	return buildFlowDoc(body);
+}
+function squash(text) {
+	return text.replace(/\s+/g, " ").trim();
+}
+function headingLevel(type) {
+	const m = /^H([1-6])$/.exec(type);
+	return m ? Number(m[1]) - 1 : void 0;
+}
+//#endregion
+export { reconstructTaggedPdf };

package/dist/esm/pdf-reader/text.d.ts

@@ -0,0 +1,3 @@
+import { TextRun } from './content.js';
+import { PdfFile, PdfPage } from './document.js';
+export declare function extractPageText(file: PdfFile, page: PdfPage): Array<TextRun>;

package/dist/esm/pdf-reader/text.js

@@ -0,0 +1,18 @@
+import { interpretContent } from "./content.js";
+import { buildContentFont } from "./font.js";
+//#region src/pdf-reader/text.ts
+function extractPageText(file, page) {
+	const fonts = /* @__PURE__ */ new Map();
+	if (page.resources) {
+		const fontContainer = file.get(page.resources, "Font");
+		if (fontContainer instanceof Map) for (const [fontName, fontRef] of fontContainer) {
+			const fontDict = file.resolve(fontRef);
+			if (fontDict instanceof Map) try {
+				fonts.set(fontName, buildContentFont(file, fontDict));
+			} catch {}
+		}
+	}
+	return interpretContent(file.pageContent(page), fonts);
+}
+//#endregion
+export { extractPageText };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "reamkit",
-  "version": "1.5.0",
-  "description": "Ream — DOCX/XLSX to PDF, SVG, HTML and back to DOCX/XLSX, built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
+  "version": "1.6.0",
+  "description": "Ream — convert DOCX, XLSX and PDF to PDF, SVG, HTML, DOCX and XLSX, built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
   "license": "MIT",
   "author": "Alex Krassavin <info@reamkit.dev>",
   "repository": {