reamkit 0.1.0-alpha.0

package/dist/esm/bidi/char-types.js

@@ -0,0 +1,439 @@
+//#region src/bidi/char-types.ts
+var SINGLE = new Map([
+	[9, "S"],
+	[10, "B"],
+	[11, "S"],
+	[12, "WS"],
+	[13, "B"],
+	[28, "B"],
+	[29, "B"],
+	[30, "B"],
+	[31, "S"],
+	[32, "WS"],
+	[133, "B"],
+	[160, "CS"],
+	[171, "ON"],
+	[187, "ON"],
+	[8234, "LRE"],
+	[8235, "RLE"],
+	[8236, "PDF"],
+	[8237, "LRO"],
+	[8238, "RLO"],
+	[8294, "LRI"],
+	[8295, "RLI"],
+	[8296, "FSI"],
+	[8297, "PDI"],
+	[1564, "AL"],
+	[8232, "WS"],
+	[8233, "B"]
+]);
+var RANGES = [
+	{
+		lo: 0,
+		hi: 8,
+		cls: "BN"
+	},
+	{
+		lo: 14,
+		hi: 27,
+		cls: "BN"
+	},
+	{
+		lo: 33,
+		hi: 34,
+		cls: "ON"
+	},
+	{
+		lo: 35,
+		hi: 37,
+		cls: "ET"
+	},
+	{
+		lo: 38,
+		hi: 42,
+		cls: "ON"
+	},
+	{
+		lo: 43,
+		hi: 43,
+		cls: "ES"
+	},
+	{
+		lo: 44,
+		hi: 44,
+		cls: "CS"
+	},
+	{
+		lo: 45,
+		hi: 45,
+		cls: "ES"
+	},
+	{
+		lo: 46,
+		hi: 47,
+		cls: "CS"
+	},
+	{
+		lo: 48,
+		hi: 57,
+		cls: "EN"
+	},
+	{
+		lo: 58,
+		hi: 58,
+		cls: "CS"
+	},
+	{
+		lo: 59,
+		hi: 64,
+		cls: "ON"
+	},
+	{
+		lo: 65,
+		hi: 90,
+		cls: "L"
+	},
+	{
+		lo: 91,
+		hi: 96,
+		cls: "ON"
+	},
+	{
+		lo: 97,
+		hi: 122,
+		cls: "L"
+	},
+	{
+		lo: 123,
+		hi: 126,
+		cls: "ON"
+	},
+	{
+		lo: 161,
+		hi: 169,
+		cls: "ON"
+	},
+	{
+		lo: 170,
+		hi: 170,
+		cls: "L"
+	},
+	{
+		lo: 172,
+		hi: 175,
+		cls: "ON"
+	},
+	{
+		lo: 176,
+		hi: 177,
+		cls: "ET"
+	},
+	{
+		lo: 178,
+		hi: 179,
+		cls: "EN"
+	},
+	{
+		lo: 180,
+		hi: 180,
+		cls: "ON"
+	},
+	{
+		lo: 181,
+		hi: 181,
+		cls: "L"
+	},
+	{
+		lo: 182,
+		hi: 185,
+		cls: "ON"
+	},
+	{
+		lo: 186,
+		hi: 186,
+		cls: "L"
+	},
+	{
+		lo: 188,
+		hi: 191,
+		cls: "ON"
+	},
+	{
+		lo: 192,
+		hi: 696,
+		cls: "L"
+	},
+	{
+		lo: 768,
+		hi: 879,
+		cls: "NSM"
+	},
+	{
+		lo: 880,
+		hi: 1417,
+		cls: "L"
+	},
+	{
+		lo: 1425,
+		hi: 1469,
+		cls: "NSM"
+	},
+	{
+		lo: 1470,
+		hi: 1470,
+		cls: "R"
+	},
+	{
+		lo: 1471,
+		hi: 1471,
+		cls: "NSM"
+	},
+	{
+		lo: 1472,
+		hi: 1472,
+		cls: "R"
+	},
+	{
+		lo: 1473,
+		hi: 1474,
+		cls: "NSM"
+	},
+	{
+		lo: 1475,
+		hi: 1475,
+		cls: "R"
+	},
+	{
+		lo: 1476,
+		hi: 1477,
+		cls: "NSM"
+	},
+	{
+		lo: 1478,
+		hi: 1478,
+		cls: "R"
+	},
+	{
+		lo: 1479,
+		hi: 1479,
+		cls: "NSM"
+	},
+	{
+		lo: 1488,
+		hi: 1514,
+		cls: "R"
+	},
+	{
+		lo: 1519,
+		hi: 1524,
+		cls: "R"
+	},
+	{
+		lo: 1536,
+		hi: 1541,
+		cls: "AN"
+	},
+	{
+		lo: 1542,
+		hi: 1544,
+		cls: "AL"
+	},
+	{
+		lo: 1545,
+		hi: 1546,
+		cls: "ET"
+	},
+	{
+		lo: 1547,
+		hi: 1547,
+		cls: "AL"
+	},
+	{
+		lo: 1548,
+		hi: 1548,
+		cls: "CS"
+	},
+	{
+		lo: 1549,
+		hi: 1549,
+		cls: "AL"
+	},
+	{
+		lo: 1550,
+		hi: 1551,
+		cls: "ON"
+	},
+	{
+		lo: 1552,
+		hi: 1562,
+		cls: "NSM"
+	},
+	{
+		lo: 1563,
+		hi: 1563,
+		cls: "AL"
+	},
+	{
+		lo: 1565,
+		hi: 1610,
+		cls: "AL"
+	},
+	{
+		lo: 1611,
+		hi: 1631,
+		cls: "NSM"
+	},
+	{
+		lo: 1632,
+		hi: 1641,
+		cls: "AN"
+	},
+	{
+		lo: 1642,
+		hi: 1642,
+		cls: "ET"
+	},
+	{
+		lo: 1643,
+		hi: 1644,
+		cls: "AN"
+	},
+	{
+		lo: 1645,
+		hi: 1645,
+		cls: "AL"
+	},
+	{
+		lo: 1646,
+		hi: 1647,
+		cls: "AL"
+	},
+	{
+		lo: 1648,
+		hi: 1648,
+		cls: "NSM"
+	},
+	{
+		lo: 1649,
+		hi: 1749,
+		cls: "AL"
+	},
+	{
+		lo: 1750,
+		hi: 1756,
+		cls: "NSM"
+	},
+	{
+		lo: 1757,
+		hi: 1757,
+		cls: "AN"
+	},
+	{
+		lo: 1758,
+		hi: 1764,
+		cls: "NSM"
+	},
+	{
+		lo: 1765,
+		hi: 1766,
+		cls: "AL"
+	},
+	{
+		lo: 1767,
+		hi: 1768,
+		cls: "NSM"
+	},
+	{
+		lo: 1769,
+		hi: 1769,
+		cls: "ON"
+	},
+	{
+		lo: 1770,
+		hi: 1773,
+		cls: "NSM"
+	},
+	{
+		lo: 1774,
+		hi: 1775,
+		cls: "AL"
+	},
+	{
+		lo: 1776,
+		hi: 1785,
+		cls: "EN"
+	},
+	{
+		lo: 1786,
+		hi: 1791,
+		cls: "AL"
+	},
+	{
+		lo: 1792,
+		hi: 1871,
+		cls: "AL"
+	},
+	{
+		lo: 1872,
+		hi: 1919,
+		cls: "AL"
+	},
+	{
+		lo: 1920,
+		hi: 1983,
+		cls: "AL"
+	},
+	{
+		lo: 1984,
+		hi: 2047,
+		cls: "R"
+	},
+	{
+		lo: 2048,
+		hi: 2207,
+		cls: "L"
+	},
+	{
+		lo: 2208,
+		hi: 2303,
+		cls: "AL"
+	},
+	{
+		lo: 64285,
+		hi: 64335,
+		cls: "R"
+	},
+	{
+		lo: 64336,
+		hi: 64975,
+		cls: "AL"
+	},
+	{
+		lo: 65008,
+		hi: 65023,
+		cls: "AL"
+	},
+	{
+		lo: 65136,
+		hi: 65279,
+		cls: "AL"
+	},
+	{
+		lo: 8208,
+		hi: 8231,
+		cls: "ON"
+	},
+	{
+		lo: 8240,
+		hi: 8286,
+		cls: "ON"
+	}
+];
+function bidiClass(cp) {
+	const single = SINGLE.get(cp);
+	if (single !== void 0) return single;
+	for (const r of RANGES) if (cp >= r.lo && cp <= r.hi) return r.cls;
+	return "L";
+}
+//#endregion
+export { bidiClass };

package/dist/esm/bidi/index.d.ts

@@ -0,0 +1,14 @@
+import { Direction } from './algorithm.js';
+export type { Direction, BidiResult } from './algorithm.js';
+export type { BidiClass } from './char-types.js';
+export { computeBidi, reorderVisual } from './algorithm.js';
+export { bidiClass } from './char-types.js';
+export interface StringBidi {
+    readonly levels: ReadonlyArray<number>;
+    readonly codePoints: ReadonlyArray<number>;
+    readonly paragraphLevel: number;
+}
+export declare function analyzeString(text: string, dir?: Direction): StringBidi;
+export declare function hasBidiCharacters(text: string): boolean;
+export declare function reverseByCodePoint(text: string): string;
+export declare function reorderTokens(tokenLevels: ReadonlyArray<number>): Array<number>;

package/dist/esm/bidi/index.js

@@ -0,0 +1,25 @@
+import "./char-types.js";
+import { computeBidi } from "./algorithm.js";
+//#region src/bidi/index.ts
+function analyzeString(text, dir = "auto") {
+	const codePoints = [];
+	for (const ch of text) codePoints.push(ch.codePointAt(0));
+	const { levels, paragraphLevel } = computeBidi(codePoints, dir);
+	return {
+		levels,
+		codePoints,
+		paragraphLevel
+	};
+}
+function hasBidiCharacters(text) {
+	for (const ch of text) {
+		const cp = ch.codePointAt(0);
+		if (cp >= 1424 && cp <= 2303 || cp >= 64285 && cp <= 65023 || cp >= 65136 && cp <= 65279) return true;
+	}
+	return false;
+}
+function reverseByCodePoint(text) {
+	return [...text].reverse().join("");
+}
+//#endregion
+export { analyzeString, hasBidiCharacters, reverseByCodePoint };

package/dist/esm/converter/docx-to-pdf.d.ts

@@ -0,0 +1,15 @@
+import { FontBytesByVariant } from '../font/index.js';
+import { FamilyKey, FetchLike } from '../fonts/index.js';
+import { SignatureOptions, StyledRenderOptions } from '../pdf/index.js';
+export interface ConvertDocxOptions extends Omit<StyledRenderOptions, 'registry' | 'styles'> {
+    readonly fontBytes?: Uint8Array;
+    readonly fonts?: FontBytesByVariant;
+    readonly fontFamily?: string;
+    readonly fontFetch?: FetchLike;
+    readonly embedSource?: boolean;
+    readonly signature?: SignatureOptions;
+}
+export declare function convertDocxToPdfSync(docx: Uint8Array, options: ConvertDocxOptions): Uint8Array;
+export declare function detectDocxFamilyKeys(docx: Uint8Array): Set<FamilyKey>;
+export declare function convertDocxToPdf(docx: Uint8Array, options?: ConvertDocxOptions): Promise<Uint8Array>;
+export declare function detectDocxFontFamily(docx: Uint8Array): string | undefined;

package/dist/esm/converter/docx-to-pdf.js

@@ -0,0 +1,227 @@
+import { FontRegistry } from "../font/font-registry.js";
+import { fetchFontSet, resolveFamilyKey } from "../fonts/remote-fonts.js";
+import { parseChart } from "../ooxml/drawingml/chart-parser.js";
+import { DEFAULT_THEME_PALETTE, makeColorResolver } from "../ooxml/drawingml/colors.js";
+import { parseTheme } from "../ooxml/drawingml/theme-parser.js";
+import { OpcPackage } from "../opc/package.js";
+import { parseCoreProperties } from "../opc/core-properties.js";
+import "../opc/index.js";
+import { EMPTY_SECTION, parseDocument, parseHeaderFooter, parseSections } from "../ooxml/wordproc/document-parser.js";
+import { EMPTY_STYLE_SHEET, parseStyles } from "../ooxml/wordproc/styles-parser.js";
+import { loadEmbeddedFonts } from "../ooxml/wordproc/font-table.js";
+import { EMPTY_NUMBERING, parseNumbering } from "../ooxml/wordproc/numbering-parser.js";
+import { EMPTY_SETTINGS, parseSettings } from "../ooxml/wordproc/settings-parser.js";
+import "../ooxml/wordproc/index.js";
+import { signPdf } from "../pdf/signature.js";
+import { renderStyledPdf } from "../pdf/styled-page-renderer.js";
+import "../pdf/index.js";
+//#region src/converter/docx-to-pdf.ts
+var STYLES_PART = "word/styles.xml";
+var NUMBERING_PART = "word/numbering.xml";
+var SETTINGS_PART = "word/settings.xml";
+var CORE_PROPS_PART = "docProps/core.xml";
+var MAIN_DOCUMENT_PART = "word/document.xml";
+var REL_HEADER = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/header";
+var REL_FOOTER = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/footer";
+var REL_IMAGE = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/image";
+var REL_THEME = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/theme";
+var THEME_PART = "word/theme/theme1.xml";
+var REL_CHART = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/chart";
+function convertDocxToPdfSync(docx, options) {
+	const fonts = options.fonts ?? (options.fontBytes ? { regular: options.fontBytes } : void 0);
+	if (!fonts) throw new Error("convertDocxToPdfSync requires options.fonts or options.fontBytes");
+	const registry = FontRegistry.fromBytes(fonts);
+	const pkg = OpcPackage.open(docx);
+	const main = pkg.getMainDocument();
+	const resolveColor = buildColorResolver(pkg);
+	const body = parseDocument(main.data, resolveColor);
+	const rawSections = parseSections(main.data);
+	const stylesData = pkg.getPart(STYLES_PART);
+	const styles = stylesData ? parseStyles(stylesData) : EMPTY_STYLE_SHEET;
+	const numberingData = pkg.getPart(NUMBERING_PART);
+	const numbering = numberingData ? parseNumbering(numberingData) : EMPTY_NUMBERING;
+	const settingsData = pkg.getPart(SETTINGS_PART);
+	const settings = settingsData ? parseSettings(settingsData) : EMPTY_SETTINGS;
+	const sections = rawSections.length > 0 ? rawSections.map((s) => ({
+		...s,
+		properties: settings.evenAndOddHeaders ? {
+			...s.properties,
+			evenAndOddHeaders: true
+		} : s.properties
+	})) : [{
+		properties: settings.evenAndOddHeaders ? {
+			...EMPTY_SECTION,
+			evenAndOddHeaders: true
+		} : EMPTY_SECTION,
+		endIndex: body.length
+	}];
+	const headersFooters = loadHeadersFootersForSections(pkg, sections, resolveColor);
+	const images = loadImages(pkg);
+	const charts = loadCharts(pkg, resolveColor);
+	const embeddedFonts = loadEmbeddedFonts(pkg);
+	const coreData = pkg.getPart(CORE_PROPS_PART);
+	const coreProps = coreData ? parseCoreProperties(coreData) : void 0;
+	const { fontBytes: _ignoreA, fonts: _ignoreB, info: callerInfo, embedSource, attachments: callerAttachments, signature: _ignoreSig, ...renderOptions } = options;
+	const info = mergeInfo(coreProps, callerInfo);
+	const language = detectDocxLanguage(stylesData, main.data);
+	const attachments = [...callerAttachments ?? []];
+	if (embedSource && options.pdfA?.startsWith("PDF/A-3")) attachments.push({
+		name: "source.docx",
+		bytes: docx,
+		mimeType: "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+		relationship: "Source",
+		description: "Source Word document"
+	});
+	return renderStyledPdf(body, {
+		registry,
+		styles,
+		numbering,
+		sections,
+		headersFooters,
+		images,
+		charts,
+		...embeddedFonts.size > 0 ? { embeddedFonts } : {},
+		...info ? { info } : {},
+		...language ? { language } : {},
+		...attachments.length > 0 ? { attachments } : {},
+		...renderOptions
+	});
+}
+function detectDocxLanguage(stylesData, documentData) {
+	const re = /<w:lang\b[^>]*\bw:val="([^"]+)"/;
+	const decoder = new TextDecoder();
+	for (const data of [stylesData, documentData]) {
+		if (!data) continue;
+		const m = re.exec(decoder.decode(data));
+		if (m?.[1]) return m[1];
+	}
+}
+function detectDocxFamilyKeys(docx) {
+	const keys = new Set(["roboto"]);
+	let pkg;
+	try {
+		pkg = OpcPackage.open(docx);
+	} catch {
+		return keys;
+	}
+	const decoder = new TextDecoder("utf-8");
+	const re = /<w:rFonts[^>]*\bw:ascii="([^"]+)"/g;
+	for (const part of [STYLES_PART, MAIN_DOCUMENT_PART]) {
+		const data = pkg.getPart(part);
+		if (!data) continue;
+		const xml = decoder.decode(data);
+		let m;
+		while ((m = re.exec(xml)) !== null) keys.add(resolveFamilyKey(m[1]));
+	}
+	return keys;
+}
+async function convertDocxToPdf(docx, options = {}) {
+	const { signature } = options;
+	const pdf = await buildUnsignedDocxPdf(docx, signature ? {
+		...options,
+		signaturePlaceholder: signature
+	} : options);
+	return signature ? signPdf(pdf, signature) : pdf;
+}
+async function buildUnsignedDocxPdf(docx, options) {
+	if (options.fonts ?? options.fontBytes) return convertDocxToPdfSync(docx, options);
+	const fetchOpt = options.fontFetch ? { fetch: options.fontFetch } : {};
+	const keys = options.fontFamily ? new Set([resolveFamilyKey(options.fontFamily)]) : detectDocxFamilyKeys(docx);
+	if (keys.size <= 1) {
+		const family = keys.values().next().value;
+		const fonts = await fetchFontSet({
+			...family ? { family } : {},
+			...fetchOpt
+		});
+		return convertDocxToPdfSync(docx, {
+			...options,
+			fonts
+		});
+	}
+	const registriesByFamily = /* @__PURE__ */ new Map();
+	let baseBytes;
+	for (const key of keys) {
+		const bytes = await fetchFontSet({
+			family: key,
+			...fetchOpt
+		});
+		registriesByFamily.set(key, FontRegistry.fromBytes(bytes));
+		if (key === "roboto" || !baseBytes) baseBytes = bytes;
+	}
+	return convertDocxToPdfSync(docx, {
+		...options,
+		fonts: baseBytes,
+		registriesByFamily
+	});
+}
+function mergeInfo(core, caller) {
+	if (!core && !caller) return void 0;
+	return {
+		...core?.title ? { title: core.title } : {},
+		...core?.creator ? { author: core.creator } : {},
+		...core?.subject ? { subject: core.subject } : {},
+		...core?.keywords ? { keywords: core.keywords } : {},
+		...core?.created ? { creationDate: core.created } : {},
+		...core?.modified ? { modificationDate: core.modified } : {},
+		...caller
+	};
+}
+function loadImages(pkg) {
+	const rels = pkg.getPartRelationships(MAIN_DOCUMENT_PART);
+	if (rels.length === 0) return /* @__PURE__ */ new Map();
+	const out = /* @__PURE__ */ new Map();
+	for (const rel of rels) {
+		if (rel.type !== REL_IMAGE) continue;
+		const resolved = pkg.resolveRelatedPart(MAIN_DOCUMENT_PART, rel);
+		if (!resolved) continue;
+		out.set(rel.id, resolved.data);
+	}
+	return out;
+}
+function loadCharts(pkg, resolveColor) {
+	const out = /* @__PURE__ */ new Map();
+	for (const rel of pkg.getPartRelationships(MAIN_DOCUMENT_PART)) {
+		if (rel.type !== REL_CHART) continue;
+		const resolved = pkg.resolveRelatedPart(MAIN_DOCUMENT_PART, rel);
+		if (!resolved) continue;
+		const chart = parseChart(resolved.data, resolveColor);
+		if (chart) out.set(rel.id, chart);
+	}
+	return out;
+}
+function buildColorResolver(pkg) {
+	const themeData = loadTheme(pkg);
+	if (!themeData) return makeColorResolver(DEFAULT_THEME_PALETTE);
+	const palette = new Map(DEFAULT_THEME_PALETTE);
+	for (const [slot, hex] of parseTheme(themeData)) palette.set(slot, hex);
+	return makeColorResolver(palette);
+}
+function loadTheme(pkg) {
+	for (const rel of pkg.getPartRelationships(MAIN_DOCUMENT_PART)) {
+		if (rel.type !== REL_THEME) continue;
+		const resolved = pkg.resolveRelatedPart(MAIN_DOCUMENT_PART, rel);
+		if (resolved) return resolved.data;
+	}
+	return pkg.getPart(THEME_PART);
+}
+function loadHeadersFootersForSections(pkg, sections, resolveColor) {
+	const wanted = /* @__PURE__ */ new Set();
+	for (const s of sections) {
+		for (const h of s.properties.headers) wanted.add(h.relationshipId);
+		for (const f of s.properties.footers) wanted.add(f.relationshipId);
+	}
+	if (wanted.size === 0) return /* @__PURE__ */ new Map();
+	const rels = pkg.getPartRelationships(MAIN_DOCUMENT_PART);
+	if (rels.length === 0) return /* @__PURE__ */ new Map();
+	const out = /* @__PURE__ */ new Map();
+	for (const rel of rels) {
+		if (!wanted.has(rel.id)) continue;
+		if (rel.type !== REL_HEADER && rel.type !== REL_FOOTER) continue;
+		const resolved = pkg.resolveRelatedPart(MAIN_DOCUMENT_PART, rel);
+		if (!resolved) continue;
+		out.set(rel.id, parseHeaderFooter(resolved.data, resolveColor));
+	}
+	return out;
+}
+//#endregion
+export { convertDocxToPdf, convertDocxToPdfSync };

package/dist/esm/converter/index.d.ts

@@ -0,0 +1,4 @@
+export { convertDocxToPdf, convertDocxToPdfSync, detectDocxFontFamily, } from './docx-to-pdf.js';
+export type { ConvertDocxOptions } from './docx-to-pdf.js';
+export { convertXlsxToPdf, convertXlsxToPdfSync } from './xlsx-to-pdf.js';
+export type { ConvertXlsxOptions } from './xlsx-to-pdf.js';