@bendyline/squisq-formats 1.3.0 → 1.3.1

package/dist/pdf/index.d.ts

@@ -1,26 +1,152 @@
+import { Doc } from '@bendyline/squisq/schemas';
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+import { ContentContainer } from '@bendyline/squisq/storage';
+
 /**
- * @bendyline/squisq-formats PDF Module
+ * PDF Export
  *
- * Import and export squisq documents (MarkdownDocument / Doc)
- * to/from PDF files.
+ * 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.
  *
- * - **Export** uses pdf-lib (zero dependencies, standard fonts, pure JS).
- * - **Import** uses pdfjs-dist (Mozilla pdf.js) with heuristic structure detection.
+ * Uses only the 14 standard PDF fonts (no font embedding required),
+ * keeping output size small and rendering fast.
  *
- * All operations run in the browser — no server required.
+ * @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;
+    /**
+     * Squisq theme ID to apply (e.g., 'documentary', 'cinematic').
+     * When set, overrides heading, text, and link colors with the theme palette.
+     * Font changes are not supported (pdf-lib uses standard 14 PDF fonts only).
+     */
+    themeId?: string;
+}
+/**
+ * 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>;
+/**
+ * Convert a PDF file to a ContentContainer with markdown + extracted images.
+ *
+ * The container will contain:
+ * - The primary markdown document (index.md)
+ * - Any embedded images under images/ (e.g., images/image1.png)
+ *
+ * Image extraction uses pdfjs-dist's operator list API and requires a browser
+ * environment (canvas is used to encode pixel data to PNG).
+ *
+ * @param data - The raw PDF file as ArrayBuffer, Uint8Array, or Blob
+ * @param options - Import options
+ * @returns A ContentContainer with the document and its media
+ */
+declare function pdfToContainer(data: ArrayBuffer | Uint8Array | Blob, options?: PdfImportOptions): Promise<ContentContainer>;
+/**
+ * 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
- * import {
- *   markdownDocToPdf,
- *   docToPdf,
- *   pdfToMarkdownDoc,
- *   pdfToDoc,
- * } from '@bendyline/squisq-formats/pdf';
+ * // 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);
  * ```
  */
-export { markdownDocToPdf, docToPdf } from './export.js';
-export type { PdfExportOptions } from './export.js';
-export { pdfToMarkdownDoc, pdfToDoc, pdfToContainer, configurePdfWorker } from './import.js';
-export type { PdfImportOptions } from './import.js';
-//# sourceMappingURL=index.d.ts.map
+declare function configurePdfWorker(workerSrc: string): void;
+
+export { type PdfExportOptions, type PdfImportOptions, configurePdfWorker, docToPdf, markdownDocToPdf, pdfToContainer, pdfToDoc, pdfToMarkdownDoc };

package/dist/pdf/index.js

@@ -1,26 +1,17 @@
-/**
- * @bendyline/squisq-formats PDF Module
- *
- * Import and export squisq documents (MarkdownDocument / Doc)
- * to/from PDF files.
- *
- * - **Export** uses pdf-lib (zero dependencies, standard fonts, pure JS).
- * - **Import** uses pdfjs-dist (Mozilla pdf.js) with heuristic structure detection.
- *
- * All operations run in the browser — no server required.
- *
- * @example
- * ```ts
- * import {
- *   markdownDocToPdf,
- *   docToPdf,
- *   pdfToMarkdownDoc,
- *   pdfToDoc,
- * } from '@bendyline/squisq-formats/pdf';
- * ```
- */
-// Export
-export { markdownDocToPdf, docToPdf } from './export.js';
-// Import
-export { pdfToMarkdownDoc, pdfToDoc, pdfToContainer, configurePdfWorker } from './import.js';
+import {
+  configurePdfWorker,
+  docToPdf,
+  markdownDocToPdf,
+  pdfToContainer,
+  pdfToDoc,
+  pdfToMarkdownDoc
+} from "../chunk-S3Y7H2BK.js";
+export {
+  configurePdfWorker,
+  docToPdf,
+  markdownDocToPdf,
+  pdfToContainer,
+  pdfToDoc,
+  pdfToMarkdownDoc
+};
 //# sourceMappingURL=index.js.map

package/dist/pdf/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/pdf/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,SAAS;AACT,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAGzD,SAAS;AACT,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,cAAc,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC"}
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/pptx/index.d.ts

@@ -1,3 +1,80 @@
+import { Doc } from '@bendyline/squisq/schemas';
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+
+/**
+ * PPTX Export
+ *
+ * Converts a squisq MarkdownDocument (or Doc) into a .pptx file
+ * by generating PresentationML + DrawingML XML and assembling the
+ * OOXML package.
+ *
+ * Slide segmentation: each H1 or H2 heading starts a new slide.
+ * The heading becomes the slide title; content below becomes the body.
+ *
+ * Supports:
+ * - Theme-based styling (background, text colors, fonts from squisq themes)
+ * - Image embedding (pass images as ArrayBuffers via options)
+ * - Inline formatting (bold, italic, strikethrough, code, links)
+ * - Lists, code blocks, blockquotes, tables, math
+ *
+ * No third-party pptx library — all XML is generated directly using
+ * the shared ooxml/ infrastructure.
+ *
+ * @example
+ * ```ts
+ * import { parseMarkdown } from '@bendyline/squisq/markdown';
+ * import { markdownDocToPptx } from '@bendyline/squisq-formats/pptx';
+ *
+ * const md = parseMarkdown('# Slide 1\n\nHello world\n\n## Slide 2\n\nMore content');
+ * const blob = await markdownDocToPptx(md, { themeId: 'documentary' });
+ * ```
+ */
+
+/**
+ * Options for PPTX export.
+ */
+interface PptxExportOptions {
+    /** Presentation title (appears in core properties) */
+    title?: string;
+    /** Presentation author */
+    author?: string;
+    /** Presentation description */
+    description?: string;
+    /**
+     * Which heading depth triggers a new slide.
+     * - `'h1'` — only H1 headings start slides
+     * - `'h2'` — H1 and H2 headings start slides (default)
+     * - `'heading'` — any heading (H1–H6) starts a slide
+     */
+    slideBreak?: 'h1' | 'h2' | 'heading';
+    /** Default body font family. Default: "Calibri" */
+    defaultFont?: string;
+    /** Default body font size in points. Default: 18 */
+    defaultFontSize?: number;
+    /**
+     * Squisq theme ID to apply (e.g., 'documentary', 'cinematic', 'bold').
+     * Controls slide background, text colors, and fonts.
+     * Falls back to the document's frontmatter `themeId` if not set here.
+     */
+    themeId?: string;
+    /**
+     * Image data keyed by the path/URL used in markdown.
+     * When provided, images are embedded as picture shapes instead of
+     * showing `[Image: alt]` placeholders.
+     */
+    images?: Map<string, ArrayBuffer>;
+}
+/**
+ * Convert a MarkdownDocument to a .pptx ArrayBuffer.
+ */
+declare function markdownDocToPptx(doc: MarkdownDocument, options?: PptxExportOptions): Promise<ArrayBuffer>;
+/**
+ * Convert a squisq Doc to a .pptx ArrayBuffer.
+ *
+ * Convenience wrapper: Doc -> MarkdownDocument -> PPTX.
+ */
+declare function docToPptx(doc: Doc, options?: PptxExportOptions): Promise<ArrayBuffer>;
+
 /**
  * @bendyline/squisq-formats PPTX Module
  *
@@ -14,14 +91,11 @@
  * import { markdownDocToPptx } from '@bendyline/squisq-formats/pptx';
  * ```
  */
-export { markdownDocToPptx, docToPptx } from './export.js';
-export type { PptxExportOptions } from './export.js';
-import type { MarkdownDocument } from '@bendyline/squisq/markdown';
-import type { Doc } from '@bendyline/squisq/schemas';
+
 /**
  * Options for PPTX import (placeholder).
  */
-export interface PptxImportOptions {
+interface PptxImportOptions {
     /** Whether to extract embedded images as data URIs */
     extractImages?: boolean;
 }
@@ -30,11 +104,12 @@ export interface PptxImportOptions {
  *
  * @throws Error — PPTX import is not yet implemented
  */
-export declare function pptxToMarkdownDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<MarkdownDocument>;
+declare function pptxToMarkdownDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<MarkdownDocument>;
 /**
  * Convert a .pptx file to a squisq Doc.
  *
  * @throws Error — PPTX import is not yet implemented
  */
-export declare function pptxToDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<Doc>;
-//# sourceMappingURL=index.d.ts.map
+declare function pptxToDoc(_data: ArrayBuffer | Blob, _options?: PptxImportOptions): Promise<Doc>;
+
+export { type PptxExportOptions, type PptxImportOptions, docToPptx, markdownDocToPptx, pptxToDoc, pptxToMarkdownDoc };

package/dist/pptx/index.js

@@ -1,35 +1,15 @@
-/**
- * @bendyline/squisq-formats PPTX Module
- *
- * PowerPoint .pptx export support using PresentationML (`<p:presentation>`,
- * `<p:sld>`) via the shared ooxml/ infrastructure.
- *
- * Slide segmentation: each H1/H2 heading starts a new slide by default.
- * Inline formatting (bold, italic, code, links) is preserved as DrawingML runs.
- *
- * Import is not yet implemented.
- *
- * @example
- * ```ts
- * import { markdownDocToPptx } from '@bendyline/squisq-formats/pptx';
- * ```
- */
-// Export
-export { markdownDocToPptx, docToPptx } from './export.js';
-/**
- * Convert a .pptx file to a MarkdownDocument.
- *
- * @throws Error — PPTX import is not yet implemented
- */
-export async function pptxToMarkdownDoc(_data, _options) {
-    throw new Error('PPTX import is not yet implemented');
-}
-/**
- * Convert a .pptx file to a squisq Doc.
- *
- * @throws Error — PPTX import is not yet implemented
- */
-export async function pptxToDoc(_data, _options) {
-    throw new Error('PPTX import is not yet implemented');
-}
+import {
+  docToPptx,
+  markdownDocToPptx,
+  pptxToDoc,
+  pptxToMarkdownDoc
+} from "../chunk-VN2KEOYB.js";
+import "../chunk-7DEUGIOO.js";
+import "../chunk-U4MRIFKL.js";
+export {
+  docToPptx,
+  markdownDocToPptx,
+  pptxToDoc,
+  pptxToMarkdownDoc
+};
 //# sourceMappingURL=index.js.map

package/dist/pptx/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/pptx/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,SAAS;AACT,OAAO,EAAE,iBAAiB,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AAe3D;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,KAAyB,EACzB,QAA4B;IAE5B,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAC7B,KAAyB,EACzB,QAA4B;IAE5B,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC"}
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/xlsx/index.d.ts

@@ -1,3 +1,6 @@
+import { MarkdownDocument } from '@bendyline/squisq/markdown';
+import { Doc } from '@bendyline/squisq/schemas';
+
 /**
  * @bendyline/squisq-formats XLSX Module (Stub)
  *
@@ -10,12 +13,11 @@
  * import { markdownDocToXlsx } from '@bendyline/squisq-formats/xlsx';
  * ```
  */
-import type { MarkdownDocument } from '@bendyline/squisq/markdown';
-import type { Doc } from '@bendyline/squisq/schemas';
+
 /**
  * Options for XLSX export (placeholder).
  */
-export interface XlsxExportOptions {
+interface XlsxExportOptions {
     /** Workbook title */
     title?: string;
     /** Workbook author */
@@ -24,7 +26,7 @@ export interface XlsxExportOptions {
 /**
  * Options for XLSX import (placeholder).
  */
-export interface XlsxImportOptions {
+interface XlsxImportOptions {
     /** Which sheet to import (0-based index or name). Default: 0 */
     sheet?: number | string;
 }
@@ -33,23 +35,24 @@ export interface XlsxImportOptions {
  *
  * @throws Error — XLSX support is not yet implemented
  */
-export declare function markdownDocToXlsx(_doc: MarkdownDocument, _options?: XlsxExportOptions): Promise<Blob>;
+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
  */
-export declare function docToXlsx(_doc: Doc, _options?: XlsxExportOptions): Promise<Blob>;
+declare function docToXlsx(_doc: Doc, _options?: XlsxExportOptions): Promise<Blob>;
 /**
  * Convert a .xlsx file to a MarkdownDocument.
  *
  * @throws Error — XLSX support is not yet implemented
  */
-export declare function xlsxToMarkdownDoc(_data: ArrayBuffer | Blob, _options?: XlsxImportOptions): Promise<MarkdownDocument>;
+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
  */
-export declare function xlsxToDoc(_data: ArrayBuffer | Blob, _options?: XlsxImportOptions): Promise<Doc>;
-//# sourceMappingURL=index.d.ts.map
+declare function xlsxToDoc(_data: ArrayBuffer | Blob, _options?: XlsxImportOptions): Promise<Doc>;
+
+export { type XlsxExportOptions, type XlsxImportOptions, docToXlsx, markdownDocToXlsx, xlsxToDoc, xlsxToMarkdownDoc };

package/dist/xlsx/index.js

@@ -1,45 +1,13 @@
-/**
- * @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';
- * ```
- */
-/**
- * Convert a MarkdownDocument to a .xlsx Blob.
- *
- * @throws Error — XLSX support is not yet implemented
- */
-export async function markdownDocToXlsx(_doc, _options) {
-    throw new Error('XLSX export is not yet implemented');
-}
-/**
- * Convert a squisq Doc to a .xlsx Blob.
- *
- * @throws Error — XLSX support is not yet implemented
- */
-export async function docToXlsx(_doc, _options) {
-    throw new Error('XLSX export is not yet implemented');
-}
-/**
- * Convert a .xlsx file to a MarkdownDocument.
- *
- * @throws Error — XLSX support is not yet implemented
- */
-export async function xlsxToMarkdownDoc(_data, _options) {
-    throw new Error('XLSX import is not yet implemented');
-}
-/**
- * Convert a .xlsx file to a squisq Doc.
- *
- * @throws Error — XLSX support is not yet implemented
- */
-export async function xlsxToDoc(_data, _options) {
-    throw new Error('XLSX import is not yet implemented');
-}
+import {
+  docToXlsx,
+  markdownDocToXlsx,
+  xlsxToDoc,
+  xlsxToMarkdownDoc
+} from "../chunk-67KIJHV2.js";
+export {
+  docToXlsx,
+  markdownDocToXlsx,
+  xlsxToDoc,
+  xlsxToMarkdownDoc
+};
 //# sourceMappingURL=index.js.map

package/dist/xlsx/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/xlsx/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAuBH;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,IAAsB,EACtB,QAA4B;IAE5B,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,IAAS,EAAE,QAA4B;IACrE,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,KAAyB,EACzB,QAA4B;IAE5B,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAC7B,KAAyB,EACzB,QAA4B;IAE5B,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;AACxD,CAAC"}
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bendyline/squisq-formats",
-  "version": "1.3.0",
+  "version": "1.3.1",
   "description": "Document format converters — DOCX, PDF, OOXML import/export",
   "license": "MIT",
   "author": "Bendyline",
@@ -84,7 +84,7 @@
     "typecheck": "tsc --noEmit"
   },
   "dependencies": {
-    "@bendyline/squisq": "1.4.0",
+    "@bendyline/squisq": "1.4.1",
     "jszip": "3.10.1",
     "pdf-lib": "1.17.1",
     "pdfjs-dist": "4.10.38"

package/dist/__tests__/container.test.d.ts

@@ -1,8 +0,0 @@
-/**
- * Tests for container ZIP serialization: containerToZip, zipToContainer.
- *
- * Verifies round-trip fidelity: write files to a container, serialize to ZIP,
- * deserialize back, and confirm all files are intact.
- */
-export {};
-//# sourceMappingURL=container.test.d.ts.map

package/dist/__tests__/container.test.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"container.test.d.ts","sourceRoot":"","sources":["../../src/__tests__/container.test.ts"],"names":[],"mappings":"AAAA;;;;;GAKG"}