@bendyline/squisq-formats 0.1.0

package/dist/ooxml/index.js

@@ -0,0 +1,99 @@
+import {
+  CONTENT_TYPE_CORE_PROPERTIES,
+  CONTENT_TYPE_DOCX_DOCUMENT,
+  CONTENT_TYPE_DOCX_FONT_TABLE,
+  CONTENT_TYPE_DOCX_FOOTNOTES,
+  CONTENT_TYPE_DOCX_NUMBERING,
+  CONTENT_TYPE_DOCX_SETTINGS,
+  CONTENT_TYPE_DOCX_STYLES,
+  CONTENT_TYPE_PPTX_PRESENTATION,
+  CONTENT_TYPE_RELATIONSHIPS,
+  CONTENT_TYPE_XLSX_WORKBOOK,
+  NS_CONTENT_TYPES,
+  NS_CORE_PROPERTIES,
+  NS_DC,
+  NS_DCTERMS,
+  NS_DRAWINGML,
+  NS_MC,
+  NS_PICTURE,
+  NS_PML,
+  NS_R,
+  NS_RELATIONSHIPS,
+  NS_SML,
+  NS_WML,
+  NS_WP_DRAWING,
+  NS_XSI,
+  REL_CORE_PROPERTIES,
+  REL_EXTENDED_PROPERTIES,
+  REL_FONT_TABLE,
+  REL_FOOTNOTES,
+  REL_HYPERLINK,
+  REL_IMAGE,
+  REL_NUMBERING,
+  REL_OFFICE_DOCUMENT,
+  REL_SETTINGS,
+  REL_STYLES,
+  REL_THEME,
+  attrString,
+  createPackage,
+  escapeXml,
+  getCoreProperties,
+  getPartBinary,
+  getPartRelationships,
+  getPartXml,
+  openPackage,
+  selfClosingElement,
+  textElement,
+  xmlDeclaration,
+  xmlElement
+} from "../chunk-KAK4V57E.js";
+export {
+  CONTENT_TYPE_CORE_PROPERTIES,
+  CONTENT_TYPE_DOCX_DOCUMENT,
+  CONTENT_TYPE_DOCX_FONT_TABLE,
+  CONTENT_TYPE_DOCX_FOOTNOTES,
+  CONTENT_TYPE_DOCX_NUMBERING,
+  CONTENT_TYPE_DOCX_SETTINGS,
+  CONTENT_TYPE_DOCX_STYLES,
+  CONTENT_TYPE_PPTX_PRESENTATION,
+  CONTENT_TYPE_RELATIONSHIPS,
+  CONTENT_TYPE_XLSX_WORKBOOK,
+  NS_CONTENT_TYPES,
+  NS_CORE_PROPERTIES,
+  NS_DC,
+  NS_DCTERMS,
+  NS_DRAWINGML,
+  NS_MC,
+  NS_PICTURE,
+  NS_PML,
+  NS_R,
+  NS_RELATIONSHIPS,
+  NS_SML,
+  NS_WML,
+  NS_WP_DRAWING,
+  NS_XSI,
+  REL_CORE_PROPERTIES,
+  REL_EXTENDED_PROPERTIES,
+  REL_FONT_TABLE,
+  REL_FOOTNOTES,
+  REL_HYPERLINK,
+  REL_IMAGE,
+  REL_NUMBERING,
+  REL_OFFICE_DOCUMENT,
+  REL_SETTINGS,
+  REL_STYLES,
+  REL_THEME,
+  attrString,
+  createPackage,
+  escapeXml,
+  getCoreProperties,
+  getPartBinary,
+  getPartRelationships,
+  getPartXml,
+  openPackage,
+  selfClosingElement,
+  textElement,
+  xmlDeclaration,
+  xmlElement
+};
+//# sourceMappingURL=index.js.map

package/dist/ooxml/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/pdf/index.d.ts

@@ -0,0 +1,130 @@
+import { Doc } from '@bendyline/squisq/schemas';
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+
+/**
+ * PDF Export
+ *
+ * Converts a squisq MarkdownDocument (or Doc) into a PDF file
+ * using pdf-lib. Generates paginated, styled output with support for
+ * headings, paragraphs, inline formatting, lists, code blocks,
+ * blockquotes, tables, thematic breaks, and hyperlinks.
+ *
+ * Uses only the 14 standard PDF fonts (no font embedding required),
+ * keeping output size small and rendering fast.
+ *
+ * @example
+ * ```ts
+ * import { parseMarkdown } from '@bendyline/squisq/markdown';
+ * import { markdownDocToPdf } from '@bendyline/squisq-formats/pdf';
+ *
+ * const md = parseMarkdown('# Hello\n\nWorld **bold** text');
+ * const buffer = await markdownDocToPdf(md);
+ * ```
+ */
+
+/** Page size presets. */
+type PdfPageSize = 'letter' | 'a4';
+/**
+ * Options for PDF export.
+ */
+interface PdfExportOptions {
+    /** Document title (PDF metadata). */
+    title?: string;
+    /** Document author (PDF metadata). */
+    author?: string;
+    /** Page size preset. Default: "letter". */
+    pageSize?: PdfPageSize;
+    /** Page margins in points. Default: 72 (1 inch). */
+    margin?: number;
+    /** Default body font size in points. Default: 11. */
+    defaultFontSize?: number;
+}
+/**
+ * Convert a MarkdownDocument to a PDF ArrayBuffer.
+ */
+declare function markdownDocToPdf(doc: MarkdownDocument, options?: PdfExportOptions): Promise<ArrayBuffer>;
+/**
+ * Convert a squisq Doc to a PDF ArrayBuffer.
+ *
+ * Convenience wrapper: Doc → MarkdownDocument → PDF.
+ */
+declare function docToPdf(doc: Doc, options?: PdfExportOptions): Promise<ArrayBuffer>;
+
+/**
+ * PDF Import
+ *
+ * Parses a PDF file and converts its content into a squisq
+ * MarkdownDocument (or Doc) using heuristic detection of headings,
+ * lists, code blocks, tables, blockquotes, and hyperlinks.
+ *
+ * Uses pdfjs-dist (Mozilla pdf.js) for text extraction — a battle-tested,
+ * browser-compatible PDF parser. Since PDFs encode positioned glyphs
+ * rather than semantic structure, all structure detection is inherently
+ * heuristic and works best on simply-formatted documents.
+ *
+ * @example
+ * ```ts
+ * import { pdfToMarkdownDoc } from '@bendyline/squisq-formats/pdf';
+ *
+ * const response = await fetch('document.pdf');
+ * const data = await response.arrayBuffer();
+ * const doc = await pdfToMarkdownDoc(data);
+ * ```
+ */
+
+/**
+ * Options for PDF import.
+ */
+interface PdfImportOptions {
+    /**
+     * Hint for the body font size used in the PDF (in points).
+     * Text items larger than this are considered headings.
+     * If not provided, the importer detects the most common font size.
+     */
+    bodyFontSize?: number;
+    /** Whether to detect tables from column-aligned text. Default: true. */
+    detectTables?: boolean;
+    /** Whether to detect code blocks from monospace fonts. Default: true. */
+    detectCodeBlocks?: boolean;
+    /** Whether to detect blockquotes from indentation. Default: true. */
+    detectBlockquotes?: boolean;
+    /** Whether to detect URLs in text and convert to links. Default: true. */
+    detectLinks?: boolean;
+}
+/**
+ * Convert a PDF file to a MarkdownDocument.
+ *
+ * Structure detection is heuristic — results are best-effort.
+ *
+ * @param data - The raw PDF file as ArrayBuffer, Uint8Array, or Blob
+ * @param options - Import options
+ * @returns A MarkdownDocument representing the detected content
+ */
+declare function pdfToMarkdownDoc(data: ArrayBuffer | Uint8Array | Blob, options?: PdfImportOptions): Promise<MarkdownDocument>;
+/**
+ * Convert a PDF file to a squisq Doc.
+ *
+ * Convenience wrapper: PDF → MarkdownDocument → Doc.
+ */
+declare function pdfToDoc(data: ArrayBuffer | Uint8Array | Blob, options?: PdfImportOptions): Promise<Doc>;
+/**
+ * Configure the pdfjs-dist PDF worker source URL.
+ *
+ * pdfjs-dist requires a worker for PDF parsing. In the **browser**, bundlers
+ * (Vite, webpack) typically handle this automatically, or you can point to a
+ * CDN-hosted worker script. In **Node.js / SSR / test** environments, call
+ * this with a `file://` URL to the worker module **before** any import call.
+ *
+ * @example
+ * ```ts
+ * // Browser — CDN
+ * configurePdfWorker('https://cdn.jsdelivr.net/npm/pdfjs-dist@4/legacy/build/pdf.worker.min.mjs');
+ *
+ * // Node / vitest — file URL
+ * import { pathToFileURL } from 'url';
+ * configurePdfWorker(pathToFileURL(require.resolve('pdfjs-dist/legacy/build/pdf.worker.mjs')).href);
+ * ```
+ */
+declare function configurePdfWorker(workerSrc: string): void;
+
+export { type PdfExportOptions, type PdfImportOptions, configurePdfWorker, docToPdf, markdownDocToPdf, pdfToDoc, pdfToMarkdownDoc };

package/dist/pdf/index.js

@@ -0,0 +1,15 @@
+import {
+  configurePdfWorker,
+  docToPdf,
+  markdownDocToPdf,
+  pdfToDoc,
+  pdfToMarkdownDoc
+} from "../chunk-TBPD5PCU.js";
+export {
+  configurePdfWorker,
+  docToPdf,
+  markdownDocToPdf,
+  pdfToDoc,
+  pdfToMarkdownDoc
+};
+//# sourceMappingURL=index.js.map

package/dist/pdf/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/pptx/index.d.ts

@@ -0,0 +1,58 @@
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+import { Doc } from '@bendyline/squisq/schemas';
+
+/**
+ * @bendyline/squisq-formats PPTX Module (Stub)
+ *
+ * Placeholder for PowerPoint .pptx import/export support.
+ * Will use PresentationML (`<p:presentation>`, `<p:sld>`) via the
+ * shared ooxml/ infrastructure.
+ *
+ * @example
+ * ```ts
+ * import { markdownDocToPptx } from '@bendyline/squisq-formats/pptx';
+ * ```
+ */
+
+/**
+ * Options for PPTX export (placeholder).
+ */
+interface PptxExportOptions {
+    /** Presentation title */
+    title?: string;
+    /** Presentation author */
+    author?: string;
+}
+/**
+ * Options for PPTX import (placeholder).
+ */
+interface PptxImportOptions {
+    /** Whether to extract embedded images as data URIs */
+    extractImages?: boolean;
+}
+/**
+ * Convert a MarkdownDocument to a .pptx Blob.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+declare function markdownDocToPptx(_doc: MarkdownDocument, _options?: PptxExportOptions): Promise<Blob>;
+/**
+ * Convert a squisq Doc to a .pptx Blob.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+declare function docToPptx(_doc: Doc, _options?: PptxExportOptions): Promise<Blob>;
+/**
+ * Convert a .pptx file to a MarkdownDocument.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+declare function pptxToMarkdownDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<MarkdownDocument>;
+/**
+ * Convert a .pptx file to a squisq Doc.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+declare function pptxToDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<Doc>;
+
+export { type PptxExportOptions, type PptxImportOptions, docToPptx, markdownDocToPptx, pptxToDoc, pptxToMarkdownDoc };

package/dist/pptx/index.js

@@ -0,0 +1,13 @@
+import {
+  docToPptx,
+  markdownDocToPptx,
+  pptxToDoc,
+  pptxToMarkdownDoc
+} from "../chunk-532L4D5D.js";
+export {
+  docToPptx,
+  markdownDocToPptx,
+  pptxToDoc,
+  pptxToMarkdownDoc
+};
+//# sourceMappingURL=index.js.map

package/dist/pptx/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/xlsx/index.d.ts

@@ -0,0 +1,58 @@
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+import { Doc } from '@bendyline/squisq/schemas';
+
+/**
+ * @bendyline/squisq-formats XLSX Module (Stub)
+ *
+ * Placeholder for Excel .xlsx import/export support.
+ * Will use SpreadsheetML (`<spreadsheet>`, `<worksheet>`, `<sheetData>`)
+ * via the shared ooxml/ infrastructure.
+ *
+ * @example
+ * ```ts
+ * import { markdownDocToXlsx } from '@bendyline/squisq-formats/xlsx';
+ * ```
+ */
+
+/**
+ * Options for XLSX export (placeholder).
+ */
+interface XlsxExportOptions {
+    /** Workbook title */
+    title?: string;
+    /** Workbook author */
+    author?: string;
+}
+/**
+ * Options for XLSX import (placeholder).
+ */
+interface XlsxImportOptions {
+    /** Which sheet to import (0-based index or name). Default: 0 */
+    sheet?: number | string;
+}
+/**
+ * Convert a MarkdownDocument to a .xlsx Blob.
+ *
+ * @throws Error — XLSX support is not yet implemented
+ */
+declare function markdownDocToXlsx(_doc: MarkdownDocument, _options?: XlsxExportOptions): Promise<Blob>;
+/**
+ * Convert a squisq Doc to a .xlsx Blob.
+ *
+ * @throws Error — XLSX support is not yet implemented
+ */
+declare function docToXlsx(_doc: Doc, _options?: XlsxExportOptions): Promise<Blob>;
+/**
+ * Convert a .xlsx file to a MarkdownDocument.
+ *
+ * @throws Error — XLSX support is not yet implemented
+ */
+declare function xlsxToMarkdownDoc(_data: ArrayBuffer | Blob, _options?: XlsxImportOptions): Promise<MarkdownDocument>;
+/**
+ * Convert a .xlsx file to a squisq Doc.
+ *
+ * @throws Error — XLSX support is not yet implemented
+ */
+declare function xlsxToDoc(_data: ArrayBuffer | Blob, _options?: XlsxImportOptions): Promise<Doc>;
+
+export { type XlsxExportOptions, type XlsxImportOptions, docToXlsx, markdownDocToXlsx, xlsxToDoc, xlsxToMarkdownDoc };

package/dist/xlsx/index.js

@@ -0,0 +1,13 @@
+import {
+  docToXlsx,
+  markdownDocToXlsx,
+  xlsxToDoc,
+  xlsxToMarkdownDoc
+} from "../chunk-67KIJHV2.js";
+export {
+  docToXlsx,
+  markdownDocToXlsx,
+  xlsxToDoc,
+  xlsxToMarkdownDoc
+};
+//# sourceMappingURL=index.js.map

package/dist/xlsx/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/package.json

@@ -0,0 +1,85 @@
+{
+  "name": "@bendyline/squisq-formats",
+  "version": "0.1.0",
+  "description": "Document format converters — DOCX, PDF, OOXML import/export",
+  "license": "MIT",
+  "author": "Bendyline",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/nicknisi/squisq.git",
+    "directory": "packages/formats"
+  },
+  "homepage": "https://github.com/nicknisi/squisq",
+  "keywords": [
+    "docx",
+    "pdf",
+    "ooxml",
+    "pptx",
+    "xlsx",
+    "document",
+    "converter"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "type": "module",
+  "main": "./dist/index.js",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "files": [
+    "dist",
+    "src"
+  ],
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "default": "./dist/index.js"
+    },
+    "./docx": {
+      "types": "./dist/docx/index.d.ts",
+      "import": "./dist/docx/index.js",
+      "default": "./dist/docx/index.js"
+    },
+    "./pptx": {
+      "types": "./dist/pptx/index.d.ts",
+      "import": "./dist/pptx/index.js",
+      "default": "./dist/pptx/index.js"
+    },
+    "./xlsx": {
+      "types": "./dist/xlsx/index.d.ts",
+      "import": "./dist/xlsx/index.js",
+      "default": "./dist/xlsx/index.js"
+    },
+    "./ooxml": {
+      "types": "./dist/ooxml/index.d.ts",
+      "import": "./dist/ooxml/index.js",
+      "default": "./dist/ooxml/index.js"
+    },
+    "./pdf": {
+      "types": "./dist/pdf/index.d.ts",
+      "import": "./dist/pdf/index.js",
+      "default": "./dist/pdf/index.js"
+    },
+    "./html": {
+      "types": "./dist/html/index.d.ts",
+      "import": "./dist/html/index.js",
+      "default": "./dist/html/index.js"
+    }
+  },
+  "scripts": {
+    "build": "tsup",
+    "typecheck": "tsc --noEmit",
+    "prepublishOnly": "npm run build"
+  },
+  "dependencies": {
+    "@bendyline/squisq": "0.1.1",
+    "jszip": "^3.10.1",
+    "pdf-lib": "^1.17.1",
+    "pdfjs-dist": "^4.9.155"
+  },
+  "devDependencies": {
+    "tsup": "^8.0.0",
+    "typescript": "^5.3.0"
+  }
+}