@bendyline/squisq-formats 0.1.0

package/src/pdf/index.ts

@@ -0,0 +1,29 @@
+/**
+ * @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';
+export type { PdfExportOptions } from './export.js';
+
+// Import
+export { pdfToMarkdownDoc, pdfToDoc, configurePdfWorker } from './import.js';
+export type { PdfImportOptions } from './import.js';

package/src/pdf/styles.ts

@@ -0,0 +1,180 @@
+/**
+ * PDF Style / Layout Constants
+ *
+ * Shared constants used by both PDF export and import. Defines page
+ * dimensions, margins, font sizes, colours, and spacing for rendering
+ * MarkdownDocument content as a paginated PDF.
+ */
+
+// ============================================
+// Page Layout
+// ============================================
+
+/** Standard US Letter page width in points (8.5 in × 72 pt/in). */
+export const PAGE_WIDTH_LETTER = 612;
+
+/** Standard US Letter page height in points (11 in × 72 pt/in). */
+export const PAGE_HEIGHT_LETTER = 792;
+
+/** A4 page width in points (210 mm ≈ 595 pt). */
+export const PAGE_WIDTH_A4 = 595.28;
+
+/** A4 page height in points (297 mm ≈ 842 pt). */
+export const PAGE_HEIGHT_A4 = 841.89;
+
+/** Default page margins in points (1 inch = 72 pt). */
+export const DEFAULT_MARGIN = 72;
+
+// ============================================
+// Fonts — pdf-lib standard font names
+// ============================================
+
+export const FONT_REGULAR = 'Helvetica';
+export const FONT_BOLD = 'Helvetica-Bold';
+export const FONT_ITALIC = 'Helvetica-Oblique';
+export const FONT_BOLD_ITALIC = 'Helvetica-BoldOblique';
+export const FONT_MONO = 'Courier';
+export const FONT_MONO_BOLD = 'Courier-Bold';
+
+// ============================================
+// Font Sizes
+// ============================================
+
+/** Default body font size in points. */
+export const DEFAULT_FONT_SIZE = 11;
+
+/** Heading font sizes indexed by depth (1-based). */
+export const HEADING_SIZES: Record<number, number> = {
+  1: 24,
+  2: 20,
+  3: 16,
+  4: 14,
+  5: 13,
+  6: 12,
+};
+
+/** Font size used for code blocks and inline code. */
+export const CODE_FONT_SIZE = 10;
+
+// ============================================
+// Spacing (in points)
+// ============================================
+
+/** Line height multiplier (leading). */
+export const LINE_HEIGHT_FACTOR = 1.4;
+
+/** Space above a heading. */
+export const HEADING_SPACE_BEFORE = 18;
+
+/** Space below a heading. */
+export const HEADING_SPACE_AFTER = 6;
+
+/** Space between paragraphs. */
+export const PARAGRAPH_SPACING = 8;
+
+/** Extra indent per list nesting level. */
+export const LIST_INDENT = 24;
+
+/** Bullet character for unordered lists. */
+export const BULLET_CHAR = '\u2022'; // •
+
+/** Indent for blockquotes. */
+export const BLOCKQUOTE_INDENT = 24;
+
+/** Width of the blockquote left bar. */
+export const BLOCKQUOTE_BAR_WIDTH = 3;
+
+// ============================================
+// Colours (RGB 0-1)
+// ============================================
+
+export const COLOR_TEXT = { r: 0.12, g: 0.12, b: 0.14 };
+export const COLOR_HEADING = { r: 0.08, g: 0.08, b: 0.1 };
+export const COLOR_LINK = { r: 0.05, g: 0.27, b: 0.73 };
+export const COLOR_CODE_BG = { r: 0.95, g: 0.95, b: 0.97 };
+export const COLOR_CODE_TEXT = { r: 0.2, g: 0.2, b: 0.22 };
+export const COLOR_BLOCKQUOTE_BAR = { r: 0.75, g: 0.75, b: 0.78 };
+export const COLOR_BLOCKQUOTE_TEXT = { r: 0.35, g: 0.35, b: 0.38 };
+export const COLOR_THEMATIC_BREAK = { r: 0.78, g: 0.78, b: 0.8 };
+export const COLOR_TABLE_BORDER = { r: 0.75, g: 0.75, b: 0.78 };
+export const COLOR_TABLE_HEADER_BG = { r: 0.93, g: 0.93, b: 0.95 };
+
+// ============================================
+// Table Constants
+// ============================================
+
+/** Table cell horizontal padding in points. */
+export const TABLE_CELL_PAD_X = 6;
+
+/** Table cell vertical padding in points. */
+export const TABLE_CELL_PAD_Y = 4;
+
+/** Table border line width. */
+export const TABLE_BORDER_WIDTH = 0.5;
+
+// ============================================
+// Import Heuristic Thresholds
+// ============================================
+
+/**
+ * Minimum font size (in PDF points) to consider as a heading.
+ * Anything ≥ this and larger than the detected body size is a heading.
+ */
+export const IMPORT_HEADING_MIN_SIZE = 13;
+
+/**
+ * Font size ranges mapped to heading depth for import heuristics.
+ * Applied when a text item's font size exceeds the body size.
+ */
+export const IMPORT_HEADING_SIZE_RANGES: Array<{ min: number; depth: number }> = [
+  { min: 22, depth: 1 },
+  { min: 18, depth: 2 },
+  { min: 15, depth: 3 },
+  { min: 13.5, depth: 4 },
+  { min: 12.5, depth: 5 },
+  { min: 12, depth: 6 },
+];
+
+/**
+ * Y-distance (in points) between lines that implies a paragraph break
+ * rather than a continuation of the same paragraph.
+ */
+export const IMPORT_PARAGRAPH_GAP = 4;
+
+/**
+ * Characters that indicate an unordered list bullet at the start of a line.
+ */
+export const IMPORT_BULLET_CHARS = new Set([
+  '\u2022',
+  '\u2023',
+  '\u25E6',
+  '\u2043',
+  '-',
+  '*',
+  '\u2013',
+  '\u2014',
+  '\u25AA',
+  '\u25AB',
+]);
+
+/**
+ * Regex to detect an ordered list prefix (e.g. "1.", "2)", "iv.").
+ */
+export const IMPORT_ORDERED_PREFIX = /^(\d+|[a-z]+|[ivxlcdm]+)[.)]\s/i;
+
+/**
+ * Column alignment tolerance — text items whose x-positions differ by
+ * less than this are considered to be in the same column (for table detection).
+ */
+export const IMPORT_COLUMN_TOLERANCE = 8;
+
+/**
+ * Minimum number of consecutive rows with the same column count to
+ * consider them a table.
+ */
+export const IMPORT_TABLE_MIN_ROWS = 2;
+
+/**
+ * URL pattern for detecting links in extracted text.
+ */
+export const IMPORT_URL_PATTERN = /https?:\/\/[^\s)>\]]+/g;

package/src/pptx/index.ts

@@ -0,0 +1,78 @@
+/**
+ * @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';
+ * ```
+ */
+
+import type { MarkdownDocument } from '@bendyline/squisq/markdown';
+import type { Doc } from '@bendyline/squisq/schemas';
+
+/**
+ * Options for PPTX export (placeholder).
+ */
+export interface PptxExportOptions {
+  /** Presentation title */
+  title?: string;
+  /** Presentation author */
+  author?: string;
+}
+
+/**
+ * Options for PPTX import (placeholder).
+ */
+export 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
+ */
+export async function markdownDocToPptx(
+  _doc: MarkdownDocument,
+  _options?: PptxExportOptions,
+): Promise<Blob> {
+  throw new Error('PPTX export is not yet implemented');
+}
+
+/**
+ * Convert a squisq Doc to a .pptx Blob.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+export async function docToPptx(_doc: Doc, _options?: PptxExportOptions): Promise<Blob> {
+  throw new Error('PPTX export is not yet implemented');
+}
+
+/**
+ * Convert a .pptx file to a MarkdownDocument.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+export async function pptxToMarkdownDoc(
+  _data: ArrayBuffer | Blob,
+  _options?: PptxImportOptions,
+): Promise<MarkdownDocument> {
+  throw new Error('PPTX import is not yet implemented');
+}
+
+/**
+ * Convert a .pptx file to a squisq Doc.
+ *
+ * @throws Error — PPTX support is not yet implemented
+ */
+export async function pptxToDoc(
+  _data: ArrayBuffer | Blob,
+  _options?: PptxImportOptions,
+): Promise<Doc> {
+  throw new Error('PPTX import is not yet implemented');
+}

package/src/xlsx/index.ts

@@ -0,0 +1,78 @@
+/**
+ * @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';
+ * ```
+ */
+
+import type { MarkdownDocument } from '@bendyline/squisq/markdown';
+import type { Doc } from '@bendyline/squisq/schemas';
+
+/**
+ * Options for XLSX export (placeholder).
+ */
+export interface XlsxExportOptions {
+  /** Workbook title */
+  title?: string;
+  /** Workbook author */
+  author?: string;
+}
+
+/**
+ * Options for XLSX import (placeholder).
+ */
+export 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
+ */
+export async function markdownDocToXlsx(
+  _doc: MarkdownDocument,
+  _options?: XlsxExportOptions,
+): Promise<Blob> {
+  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: Doc, _options?: XlsxExportOptions): Promise<Blob> {
+  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: ArrayBuffer | Blob,
+  _options?: XlsxImportOptions,
+): Promise<MarkdownDocument> {
+  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: ArrayBuffer | Blob,
+  _options?: XlsxImportOptions,
+): Promise<Doc> {
+  throw new Error('XLSX import is not yet implemented');
+}