@build-qube/papyra-native-wasm32-wasi 0.2.0 → 0.4.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@build-qube/papyra-native-wasm32-wasi",
-  "version": "0.2.0",
+  "version": "0.4.0",
   "main": "papyra.wasi.cjs",
   "files": [
     "papyra.wasm32-wasi.wasm",
@@ -36,7 +36,7 @@
   "browser": "papyra.wasi-browser.js",
   "type": "module",
   "dependencies": {
-    "@napi-rs/wasm-runtime": "~1.2.3",
+    "@napi-rs/wasm-runtime": "~1.2.5",
     "@emnapi/core": "2.0.0-alpha.4",
     "@emnapi/runtime": "2.0.0-alpha.4"
   }

package/papyra.wasi-browser.js

@@ -603,10 +603,15 @@ try {
 export default __napiModule.exports
 export const PageImage = __napiModule.exports.PageImage
 export const PdfDocument = __napiModule.exports.PdfDocument
+export const Sheet = __napiModule.exports.Sheet
+export const TiffImage = __napiModule.exports.TiffImage
+export const Workbook = __napiModule.exports.Workbook
 export const backendName = __napiModule.exports.backendName
 export const configureThreadPool = __napiModule.exports.configureThreadPool
 export const encodeBitmap = __napiModule.exports.encodeBitmap
 export const encodeBitmapSync = __napiModule.exports.encodeBitmapSync
 export const ImageFormat = __napiModule.exports.ImageFormat
+export const loadTiff = __napiModule.exports.loadTiff
+export const loadWorkbook = __napiModule.exports.loadWorkbook
 export const reserveMemory = __napiModule.exports.reserveMemory
 export const runtime = __napiModule.exports.runtime

package/papyra.wasi.cjs

@@ -1,4 +1,4 @@
-// napi-rs-artifact-metadata:{"version":2,"rootEntry":"index.js","exports":["PageImage","PdfDocument","backendName","configureThreadPool","encodeBitmap","encodeBitmapSync","ImageFormat","reserveMemory","runtime"],"managedRootEntries":["browser.js","index.js","papyra.wasm","papyra.debug.wasm"]}
+// napi-rs-artifact-metadata:{"version":2,"rootEntry":"index.js","exports":["PageImage","PdfDocument","Sheet","TiffImage","Workbook","backendName","configureThreadPool","encodeBitmap","encodeBitmapSync","ImageFormat","loadTiff","loadWorkbook","reserveMemory","runtime"],"managedRootEntries":["browser.js","index.js","papyra.wasm","papyra.debug.wasm"]}
 /* eslint-disable */
 /* prettier-ignore */
 
@@ -923,10 +923,15 @@ try {
 module.exports = __napiModule.exports
 module.exports.PageImage = __napiModule.exports.PageImage
 module.exports.PdfDocument = __napiModule.exports.PdfDocument
+module.exports.Sheet = __napiModule.exports.Sheet
+module.exports.TiffImage = __napiModule.exports.TiffImage
+module.exports.Workbook = __napiModule.exports.Workbook
 module.exports.backendName = __napiModule.exports.backendName
 module.exports.configureThreadPool = __napiModule.exports.configureThreadPool
 module.exports.encodeBitmap = __napiModule.exports.encodeBitmap
 module.exports.encodeBitmapSync = __napiModule.exports.encodeBitmapSync
 module.exports.ImageFormat = __napiModule.exports.ImageFormat
+module.exports.loadTiff = __napiModule.exports.loadTiff
+module.exports.loadWorkbook = __napiModule.exports.loadWorkbook
 module.exports.reserveMemory = __napiModule.exports.reserveMemory
 module.exports.runtime = __napiModule.exports.runtime

package/papyra.wasi.d.cts

@@ -37,18 +37,18 @@ export declare class PdfDocument {
   get pdfVersion(): string | null
   pageSize(index: number): PageDimensions
   /** Render one page on the calling thread. Blocks; prefer `renderPageAsync`. */
-  renderPage(index: number, dpi?: number | undefined | null): RenderedPage
+  renderPage(index: number, dpi?: number | undefined | null, annotations?: boolean | undefined | null): RenderedPage
   /** Render one page off the JS thread. Several of these run concurrently. */
-  renderPageAsync(index: number, dpi?: number | undefined | null): Promise<RenderedPage>
+  renderPageAsync(index: number, dpi?: number | undefined | null, annotations?: boolean | undefined | null): Promise<RenderedPage>
   /** Render one page and keep it in Rust, for encoding. The pixels never reach JS. */
-  renderPageImageAsync(index: number, dpi?: number | undefined | null, signal?: AbortSignal | undefined | null): Promise<PageImage>
+  renderPageImageAsync(index: number, dpi?: number | undefined | null, annotations?: boolean | undefined | null, signal?: AbortSignal | undefined | null): Promise<PageImage>
   /**
    * Convert one page to a standalone SVG document, off the JS thread.
    *
    * There is no DPI: an SVG carries the page's own point dimensions and rasterises at
    * whatever size it is drawn.
    */
-  renderPageSvgAsync(index: number, whiteBackground?: boolean | undefined | null, signal?: AbortSignal | undefined | null): Promise<string>
+  renderPageSvgAsync(index: number, whiteBackground?: boolean | undefined | null, annotations?: boolean | undefined | null, signal?: AbortSignal | undefined | null): Promise<string>
   /** Extract the text of one page, off the JS thread. */
   pageTextAsync(index: number): Promise<PageText>
   /** Extract `[start, end)` in one async task, parallelised internally with rayon. */
@@ -60,6 +60,22 @@ export declare class PdfDocument {
    * common case.
    */
   outline(): Promise<Array<OutlineEntry>>
+  /**
+   * Read the document structure tree (tagged PDF), in pre-order.
+   *
+   * Resolves to an empty array when the document is untagged, which is the common
+   * case. Join to text through `TextLine.mcid` and `StructEntry.content`.
+   */
+  structTree(): Promise<Array<StructEntry>>
+  /**
+   * List the files embedded in the document, in the order it files them.
+   *
+   * Resolves to an empty array when there are none, which is the common case.
+   * Metadata only — fetch the bytes with `attachmentData`.
+   */
+  attachments(): Promise<Array<Attachment>>
+  /** Decompress one embedded file, by its index in `attachments`. */
+  attachmentData(index: number): Promise<Uint8Array>
   /**
    * Read the document information dictionary.
    *
@@ -79,12 +95,147 @@ export declare class PdfDocument {
    */
   pageLabels(): Promise<Array<string>>
   /** Render `[start, end)` in one async task, parallelised internally with rayon. */
-  renderPagesAsync(start: number, end: number, dpi?: number | undefined | null): Promise<Array<RenderedPage>>
+  renderPagesAsync(start: number, end: number, dpi?: number | undefined | null, annotations?: boolean | undefined | null): Promise<Array<RenderedPage>>
+}
+
+export declare class Sheet {
+  get name(): string
+  /** The sheet held more cells than `maxCells`, and this is its first whole rows. */
+  get truncated(): boolean
+  /** One past the last used row, counting from the sheet's first row. */
+  get rows(): number
+  /** One past the last used column, counting from column A. */
+  get cols(): number
+  /**
+   * Merged regions as flat quads — `rowStart, colStart, rowEnd, colEnd`, inclusive —
+   * so a sheet with thousands of them is one allocation, not thousands of objects.
+   */
+  get merges(): Uint32Array
+  /** The styles `CellWindow.styles` indexes. Empty for every format but xlsx. */
+  get styles(): Array<CellStyle>
+  /** Widths and heights as the author set them. Absent for every format but xlsx. */
+  get layout(): SheetLayout | null
+  /**
+   * `[rowStart, rowEnd) x [colStart, colEnd)`, clamped to the sheet.
+   *
+   * Synchronous: the sheet is already parsed, and a window is as large as the
+   * caller asks for — a screenful is a few thousand cells and well under a
+   * millisecond. Ask for a screenful.
+   */
+  window(rowStart: number, rowEnd: number, colStart: number, colEnd: number): CellWindow
+}
+
+export declare class TiffImage {
+  /** Every page in directory order, reduced-resolution copies left out. */
+  get pages(): Array<TiffPageInfo>
+  /** Decode one page to RGBA at no more than the given size, off the JS thread. */
+  renderPage(index: number, options?: TiffRenderOptions | undefined | null): Promise<RenderedPage>
+}
+
+export declare class Workbook {
+  /**
+   * The container actually read: `"xlsx"`, `"xlsb"`, `"xls"`, `"ods"`, `"csv"` or
+   * `"tsv"`. A macro-enabled workbook reads as `"xlsx"`.
+   */
+  get format(): string
+  get sheets(): Array<SheetInfo>
+  /** CSV and TSV only: the encoding the text was decoded from, e.g. `"windows-1252"`. */
+  get encoding(): string | null
+  /** CSV and TSV only: the delimiter, given or sniffed. */
+  get delimiter(): string | null
+  /** Read one sheet's cells, off the JS thread. A sheet is read once and kept. */
+  sheet(index: number): Promise<Sheet>
+}
+
+/**
+ * One file embedded in the document. Metadata only; the bytes come from
+ * `attachmentData`.
+ */
+export interface Attachment {
+  /** The name to save the file under — `/UF`, else `/F`, else its key in the tree. */
+  name: string
+  /** The document's own description of what the file is for. */
+  description?: string
+  /** e.g. `"application/xml"`. Absent more often than not. */
+  mediaType?: string
+  /** Length in bytes, uncompressed, as the document claims it. A hint, not a promise. */
+  size?: number
+  /** ISO 8601. */
+  created?: string
+  /** ISO 8601. */
+  modified?: string
+  /**
+   * `Source`, `Data`, `Alternative`, `Supplement`, `EncryptedPayload` or
+   * `Unspecified` — what the file is to the document. A ZUGFeRD invoice's XML is
+   * `Alternative` or `Data`.
+   */
+  relationship?: string
 }
 
 /** Name of the rendering backend, currently always `"hayro"`. */
 export declare function backendName(): string
 
+/** One edge of a cell border. */
+export interface BorderEdge {
+  /** `"thin"`, `"medium"`, `"thick"`, `"dashed"`, `"dotted"`, `"double"` or `"hair"`. */
+  style: string
+  /** `0xRRGGBB`; absent for automatic. */
+  color?: number
+}
+
+/** How a cell looks. Shared by index: cells name a style, they do not carry one. */
+export interface CellStyle {
+  bold: boolean
+  italic: boolean
+  underline: boolean
+  strike: boolean
+  /** Font size relative to the workbook's default font. */
+  fontScale: number
+  /** `0xRRGGBB`; absent for automatic. */
+  color?: number
+  /** `0xRRGGBB` background; absent for none. */
+  fill?: number
+  borderTop?: BorderEdge
+  borderRight?: BorderEdge
+  borderBottom?: BorderEdge
+  borderLeft?: BorderEdge
+  /**
+   * `"left"`, `"center"`, `"right"`, `"justify"`, `"centerContinuous"`, `"fill"` or
+   * `"distributed"`. Absent means text left, numbers right.
+   */
+  horizontal?: string
+  /** `"top"`, `"center"`, `"bottom"`, `"justify"` or `"distributed"`. Absent means bottom. */
+  vertical?: string
+  wrap: boolean
+  indent: number
+}
+
+/** A rectangle of cells, packed. See `Window` in `papyra-tables`. */
+export interface CellWindow {
+  rowStart: number
+  colStart: number
+  rows: number
+  cols: number
+  /**
+   * One kind tag per cell, row-major: 0 empty, 1 number, 2 text, 3 bool, 4 date,
+   * 5 duration, 6 error.
+   */
+  kinds: Uint8Array
+  /**
+   * One number per cell: the value, a bool as 0/1, a date's serial, a duration in
+   * days. NaN when there is none.
+   */
+  numbers: Float64Array
+  /** Every cell's display text, concatenated. */
+  text: string
+  /** UTF-16 offsets into `text`, one more than there are cells. */
+  offsets: Uint32Array
+  /** One index into `Sheet.styles` per cell. 0 is the default style. */
+  styles: Uint16Array
+  /** The colour a number format chose per cell, as `0x01RRGGBB`; 0 for none. */
+  colors: Uint32Array
+}
+
 /**
  * Size the rayon pool used by `renderPagesAsync`.
  *
@@ -128,6 +279,24 @@ export declare const enum ImageFormat {
   Jpeg = 'jpeg'
 }
 
+/** Open a TIFF and list its pages, off the JS thread. */
+export declare function loadTiff(data: Uint8Array): Promise<TiffImage>
+
+/** Open a spreadsheet or delimited text file. */
+export declare function loadWorkbook(data: Uint8Array, options?: WorkbookOptions | undefined | null): Promise<Workbook>
+
+/**
+ * The outline walk, off the JS thread.
+ *
+ * One run of page content owned by a structure element.
+ */
+export interface MarkedContent {
+  /** 0-based page index. */
+  page: number
+  /** Matches `TextLine.mcid` on that page. */
+  mcid: number
+}
+
 /**
  * Where an outline entry points. Coordinates are PDF points from the page's
  * bottom-left; `null` means "leave this axis unchanged".
@@ -225,6 +394,46 @@ export declare function reserveMemory(megabytes: number): number
 /** `"native"` or `"wasm"`. The wrapper uses this to pick a concurrency strategy. */
 export declare function runtime(): string
 
+/** A sheet's entry in the workbook. */
+export interface SheetInfo {
+  name: string
+  /** `"worksheet"`, `"chartsheet"`, `"dialog"`, `"macro"` or `"vba"`. */
+  kind: string
+  /** `"visible"`, `"hidden"` or `"veryHidden"`. */
+  visibility: string
+}
+
+/** Column widths and row heights in pixels, as Excel lays the sheet out. */
+export interface SheetLayout {
+  defaultColWidth: number
+  defaultRowHeight: number
+  /** Inclusive column spans as flat triples: `first, last, width`. Width 0 is hidden. */
+  cols: Float64Array
+  /** Rows with their own height as flat pairs: `row, height`. Height 0 is hidden. */
+  rows: Float64Array
+  showGridLines: boolean
+}
+
+/** One structure-tree element. Flat and in pre-order; `level` carries the tree. */
+export interface StructEntry {
+  /** Standard structure type, after `/RoleMap`: `"H1"`, `"P"`, `"Table"`, `"TD"`, … */
+  role: string
+  /** The document's own tag, before `/RoleMap`. Equal to `role` unless remapped. */
+  rawRole: string
+  /** Nesting depth. Top-level elements are 0. */
+  level: number
+  /** Content this element owns directly. Empty for a pure container. */
+  content: Array<MarkedContent>
+  /** `/Alt` — replacement text for a figure. */
+  alt?: string
+  /** `/ActualText` — text replacing the content rather than describing it. */
+  actualText?: string
+  /** `/Lang` — a BCP 47 tag for this subtree. */
+  lang?: string
+  /** `/T` — a human-readable title, e.g. a table's caption. */
+  title?: string
+}
+
 /**
  * A run of glyphs sharing one baseline.
  *
@@ -250,4 +459,50 @@ export interface TextLine {
   ascent: number
   /** Extent below the baseline. */
   descent: number
+  /**
+   * Marked-content id of the line's first glyph, joining to `StructNode.content` on
+   * a tagged page. `null` on an untagged page, and for content outside any sequence.
+   */
+  mcid?: number
+}
+
+/** One page of a TIFF, as shown: `Orientation` already applied. */
+export interface TiffPageInfo {
+  width: number
+  height: number
+  /** Horizontal resolution in dots per inch, when the file states one. */
+  xDpi?: number
+  /** Vertical resolution in dots per inch, when the file states one. */
+  yDpi?: number
+  /**
+   * `"none"`, `"ccitt-rle"`, `"ccitt-g3"`, `"ccitt-g4"`, `"lzw"`, `"old-jpeg"`,
+   * `"jpeg"`, `"deflate"`, `"packbits"` or `"other"`.
+   */
+  compression: string
+  bitsPerSample: number
+  samplesPerPixel: number
+  /** TIFF `Orientation`, 1–8. */
+  orientation: number
+}
+
+/** Output limits for a page decode. Both optional; the page is never enlarged. */
+export interface TiffRenderOptions {
+  /** Defaults to 4096 x 4096. A float on the JS side, so read as one. */
+  maxPixels?: number
+  maxWidth?: number
+}
+
+/** How to read the bytes. Every field is optional and sniffed when absent. */
+export interface WorkbookOptions {
+  /**
+   * `"xlsx"`, `"xlsm"`, `"xlsb"`, `"xls"`, `"ods"`, `"csv"` or `"tsv"`. Only needed
+   * to call a file TSV, or to skip sniffing.
+   */
+  format?: string
+  /** CSV and TSV only: a single ASCII character. */
+  delimiter?: string
+  /** CSV and TSV only: a WHATWG encoding label such as `"windows-1252"`. */
+  encoding?: string
+  /** Cells held per sheet before it is truncated to its first whole rows. */
+  maxCells?: number
 }