@build-qube/papyra-native-wasm32-wasi 0.1.0 → 0.3.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@build-qube/papyra-native-wasm32-wasi",
-  "version": "0.1.0",
+  "version": "0.3.0",
   "main": "papyra.wasi.cjs",
   "files": [
     "papyra.wasm32-wasi.wasm",

package/papyra.wasi.d.cts

@@ -33,13 +33,22 @@ export declare class PdfDocument {
   /** Load an encrypted PDF. */
   static loadWithPassword(data: Uint8Array, password: string): PdfDocument
   get pageCount(): number
+  /** The PDF specification version the file declares, e.g. `"1.7"`. */
+  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, 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. */
@@ -51,8 +60,67 @@ 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.
+   *
+   * Synchronous: hayro parses it while loading, so this only decodes eight strings.
+   */
+  info(): DocumentInfo
+  /**
+   * Read the links on one page.
+   *
+   * Resolves to an empty array for a page with no links, which is most pages.
+   */
+  pageLinks(index: number): Promise<Array<PageLink>>
+  /**
+   * Read the label printed on each page, one entry per page.
+   *
+   * Resolves to an empty array when the document defines no `/PageLabels`.
+   */
+  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>>
+}
+
+/**
+ * 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"`. */
@@ -67,6 +135,21 @@ export declare function backendName(): string
  */
 export declare function configureThreadPool(threads: number, stackSizeMb?: number | undefined | null): number
 
+/** The document information dictionary. Every field is independently optional. */
+export interface DocumentInfo {
+  title?: string
+  author?: string
+  subject?: string
+  keywords?: string
+  /** The application that authored the original document, e.g. `AutoCAD`. */
+  creator?: string
+  /** The application that wrote the PDF itself, e.g. `Ghostscript`. */
+  producer?: string
+  /** ISO 8601, converted from the PDF's own date form. */
+  created?: string
+  modified?: string
+}
+
 /** Encode a raw RGBA8 buffer off the JS thread. */
 export declare function encodeBitmap(data: Uint8Array, width: number, height: number, stride: number, format: ImageFormat, quality?: number | undefined | null, signal?: AbortSignal | undefined | null): Promise<Uint8Array>
 
@@ -86,6 +169,18 @@ export declare const enum ImageFormat {
   Jpeg = 'jpeg'
 }
 
+/**
+ * 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".
@@ -123,6 +218,26 @@ export interface PageDimensions {
   height: number
 }
 
+/**
+ * One link annotation.
+ *
+ * The rectangle is in the page-as-rendered space — pixels from the top-left at
+ * 72 DPI, rotation and crop box applied — so it is the same space as `PageText` and
+ * scales to any render with one multiply.
+ */
+export interface PageLink {
+  x0: number
+  y0: number
+  x1: number
+  y1: number
+  /** Set when the link points into this document; `uri` is then absent. */
+  dest?: OutlineDestination
+  /** Set when the link points at a URI; `dest` is then absent. */
+  uri?: string
+  /** The annotation's `/Contents` — its tooltip. Usually absent. */
+  alt?: string
+}
+
 /**
  * The text of one page.
  *
@@ -163,6 +278,26 @@ export declare function reserveMemory(megabytes: number): number
 /** `"native"` or `"wasm"`. The wrapper uses this to pick a concurrency strategy. */
 export declare function runtime(): string
 
+/** 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.
  *
@@ -188,4 +323,9 @@ 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
 }