@build-qube/papyra-native-wasm32-wasi 0.2.0 → 0.3.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@build-qube/papyra-native-wasm32-wasi",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "main": "papyra.wasi.cjs",
   "files": [
     "papyra.wasm32-wasi.wasm",

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.
    *
@@ -79,7 +95,32 @@ 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>>
+}
+
+/**
+ * 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"`. */
@@ -128,6 +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".
@@ -225,6 +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.
  *
@@ -250,4 +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
 }