@laminapdf/wasm 0.2.0-alpha.3 → 0.2.0-alpha.5

package/dist/lamina_pdf_extract.d.ts

@@ -32,11 +32,110 @@ export interface Diagnostic {
 }
 
 
+
+/** One link on a page, in page space, and what following it does (by `kind`). */
+export type PageLinkJs = {
+    /** Its rectangle: `[x0, y0, x1, y1]`, top-left and bottom-right. */
+    rect: [number, number, number, number];
+    /** The areas that follow it, eight numbers per area; absent when the rectangle is the area. */
+    quads?: number[];
+    /** Its description (`/Contents`, document text). */
+    contents?: string;
+} & LinkToJs;
+
+
 /**
  * A conformance profile.
  */
 export type ProfileInit = "PDF/UA-1" | "PDF/A-2b" | "PDF/A-2u" | "PDF/A-3b";
 
+/**
+ * A document's properties, for a properties view.
+ */
+export interface PropertiesJs {
+    /**
+     * `/Title`.
+     */
+    title?: string;
+    /**
+     * `/Author`.
+     */
+    author?: string;
+    /**
+     * `/Subject`.
+     */
+    subject?: string;
+    /**
+     * `/Keywords`.
+     */
+    keywords?: string;
+    /**
+     * `/Creator`: the application the document was made with.
+     */
+    creator?: string;
+    /**
+     * `/Producer`: the application that wrote the PDF.
+     */
+    producer?: string;
+    /**
+     * `/CreationDate` as ISO 8601.
+     */
+    created?: string;
+    /**
+     * `/ModDate` as ISO 8601.
+     */
+    modified?: string;
+    /**
+     * The XMP packet as text.
+     */
+    xmp?: string;
+    /**
+     * The PDF version.
+     */
+    version?: string;
+    /**
+     * The number of pages.
+     */
+    pageCount: number;
+    /**
+     * The file's size in bytes.
+     */
+    fileSize: number;
+    /**
+     * The file is linearized ("fast web view").
+     */
+    linearized: boolean;
+    /**
+     * The catalog says the document is tagged.
+     */
+    tagged: boolean;
+    /**
+     * How the file is encrypted; absent when it isn't.
+     */
+    encryption?: EncryptionJs;
+    /**
+     * The fonts the pages use.
+     */
+    fonts: FontJs[];
+    /**
+     * The signatures present, not verified.
+     */
+    signatures: number;
+    /**
+     * The signature fields' full names.
+     */
+    signatureFields: string[];
+    /**
+     * What the document holds besides its pages.
+     */
+    contains: ContainsJs;
+}
+
+/**
+ * A font's encoding: a named one (`WinAnsiEncoding`, `Identity-H`), or the font's own.
+ */
+export type EncodingJs = { kind: "named"; name: string } | { kind: "custom" };
+
 /**
  * A job's progress (`laminapdf::Step::Progress`).
  */
@@ -50,11 +149,35 @@ export interface ProgressJs {
      */
     total?: number;
     /**
-     * The phase (`parse`, `layout`, `extract`, `render`, `write`, `sign`).
+     * The phase (`parse`, `layout`, `extract`, `render`, `write`, `sign`, `model`).
      */
     phase: string;
 }
 
+/**
+ * A kid of a structure element: an element, or marked content on a page.
+ */
+export type StructureKidJs = ({ type: "node" } & StructureNodeJs) | { type: "content"; page: number; mcid: number };
+
+/**
+ * A layout region, its box `[x, y, width, height]` in top-left points (ADR 0027), as the
+ * output's boxes are.
+ */
+export interface RegionInit {
+    /**
+     * What it is.
+     */
+    label: RegionLabelName;
+    /**
+     * `[x, y, width, height]`.
+     */
+    bbox: [number, number, number, number];
+    /**
+     * The model's confidence, 0 to 1.
+     */
+    score: number;
+}
+
 /**
  * A list of ops with its schema version ([ADR 0006] §6): `{ "v": 1, "ops": [ … ] }`, used
  * for saved op logs, batch processing and sync messages. Unknown fields are rejected, and so
@@ -76,11 +199,114 @@ export interface OpLog {
     ops: Op[];
 }
 
+/**
+ * A model result the job waits for.
+ */
+export interface ModelRequestJs {
+    /**
+     * `layout` (regions) or `ocr` (words).
+     */
+    kind: string;
+    /**
+     * The 0-based page to render.
+     */
+    page: number;
+    /**
+     * The resolution to render it at.
+     */
+    dpi: number;
+    /**
+     * OCR's languages (BCP 47); empty for the engine's default.
+     */
+    langs: string[];
+}
+
+/**
+ * A model the host runs, as the output's `generator.models` names it.
+ */
+export interface ModelInfoInit {
+    /**
+     * Its name.
+     */
+    name: string;
+    /**
+     * Its version or revision.
+     */
+    version: string;
+    /**
+     * Its weights' SHA-256, lowercase hex.
+     */
+    sha256: string;
+    /**
+     * Its weights' SPDX licence.
+     */
+    licence: string;
+}
+
+/**
+ * A page and how to show it.
+ */
+export interface TargetJs {
+    /**
+     * The page's 0-based index.
+     */
+    page: number;
+    /**
+     * How to show it, in page space.
+     */
+    view: TargetViewJs;
+}
+
 /**
  * A page-label numbering style.
  */
 export type LabelStyleInit = "decimal" | "upperRoman" | "lowerRoman" | "upperLetters" | "lowerLetters";
 
+/**
+ * A region's kind.
+ */
+export type RegionLabelName = "text" | "title" | "section-header" | "list-item" | "table" | "picture" | "caption" | "footnote" | "formula" | "page-header" | "page-footer" | "other";
+
+/**
+ * A step's progress, and the model result the job waits for, if any.
+ */
+export interface ExtractProgressJs {
+    /**
+     * Pages done.
+     */
+    done: number;
+    /**
+     * Pages asked for.
+     */
+    total?: number;
+    /**
+     * `extract`, or `model` while a request is open.
+     */
+    phase: string;
+    /**
+     * The result the job waits for: hand it to `provide`.
+     */
+    request?: ModelRequestJs;
+}
+
+/**
+ * An OCR word, its box as a region's.
+ */
+export interface WordInit {
+    /**
+     * Its text.
+     */
+    text: string;
+    /**
+     * `[x, y, width, height]`.
+     */
+    bbox: [number, number, number, number];
+    /**
+     * The engine's confidence, 0 to 1.
+     */
+    confidence: number;
+}
+
 /**
  * An attached file's options.
  */
@@ -181,6 +407,37 @@ export interface SetInfo {
  */
 export type FitInit = { type: "page" } | { type: "width"; top?: number } | { type: "height"; left?: number } | { type: "xyz"; left?: number; top?: number; zoom?: number } | { type: "rect"; x: number; y: number; width: number; height: number } | { type: "bounds" };
 
+/**
+ * How a file is encrypted, and what its author permits.
+ */
+export interface EncryptionJs {
+    /**
+     * `/V`.
+     */
+    version: number;
+    /**
+     * `/R`.
+     */
+    revision: number;
+    /**
+     * The key length in bits.
+     */
+    keyBits: number;
+    /**
+     * The cipher streams use.
+     */
+    cipher: CipherJs;
+    /**
+     * What `/P` permits; absent when the handler gives no permissions.
+     */
+    permissions?: PermissionsJs;
+}
+
+/**
+ * How a target shows its page, in page space; an unset coordinate is left out.
+ */
+export type TargetViewJs = { kind: "at"; x?: number; y?: number; zoom?: number } | { kind: "fit" } | { kind: "fit-width"; y?: number } | { kind: "fit-height"; x?: number } | { kind: "fit-rect"; x0: number; y0: number; x1: number; y1: number };
+
 /**
  * How a viewer lays pages out.
  */
@@ -237,6 +494,58 @@ export interface ViewerPreferencesInit {
  */
 export type RenumberJs = "keep" | "compact";
 
+/**
+ * One element of the structure tree (`laminapdf::StructureNode`; its strings are document text).
+ */
+export interface StructureNodeJs {
+    /**
+     * Its standard structure type (`H1`, `P`, `LI`, `Figure` …), through the `/RoleMap`.
+     */
+    kind: string;
+    /**
+     * `/Alt`: alternate text.
+     */
+    alt?: string;
+    /**
+     * `/Lang`: the language of its content.
+     */
+    lang?: string;
+    /**
+     * `/ActualText`: the text its content stands for.
+     */
+    actualText?: string;
+    /**
+     * Its kids, in reading order.
+     */
+    kids: StructureKidJs[];
+}
+
+/**
+ * One font the document uses.
+ */
+export interface FontJs {
+    /**
+     * Its name without a subset tag.
+     */
+    name?: string;
+    /**
+     * `Type0`, `Type1`, `MMType1`, `Type3` or `TrueType`.
+     */
+    kind?: string;
+    /**
+     * The file holds the font.
+     */
+    embedded: boolean;
+    /**
+     * The file holds only the glyphs used.
+     */
+    subset: boolean;
+    /**
+     * Its encoding; absent when the font gives none.
+     */
+    encoding?: EncodingJs;
+}
+
 /**
  * One outline item and its children.
  */
@@ -271,6 +580,28 @@ export interface OutlineEntryInit {
     italic?: boolean;
 }
 
+/**
+ * One outline item.
+ */
+export interface OutlineNodeJs {
+    /**
+     * Its title (document text).
+     */
+    title: string;
+    /**
+     * Whether it opens with its children shown.
+     */
+    open: boolean;
+    /**
+     * Where it goes, when its destination names a page of this document.
+     */
+    target?: TargetJs;
+    /**
+     * Its children, in order.
+     */
+    children: OutlineNodeJs[];
+}
+
 /**
  * One page's facts (`laminapdf::PageInfo`).
  */
@@ -402,6 +733,10 @@ export interface ExtractInit {
      * How chunks are cut (`chunks`, and `json` to include them).
      */
     chunk?: ChunkInit;
+    /**
+     * Models the host runs (`extractJob` only): the job asks for their results.
+     */
+    models?: ModelsInit;
 }
 
 /**
@@ -558,6 +893,11 @@ export interface DocumentInfoJs {
  */
 export type PdfVersionInit = "1.4" | "1.5" | "1.6" | "1.7" | "2.0";
 
+/**
+ * The cipher a file's streams use (ISO 32000-2 §7.6): the one list, which the `.d.ts` gets as a union.
+ */
+export type CipherJs = "rc4" | "aes-128" | "aes-256" | "identity" | "other";
+
 /**
  * The cross-reference form.
  */
@@ -609,6 +949,28 @@ export interface MetadataInit {
  */
 export type ModificationClass = "metadata" | "formFill" | "annotation" | "signature" | "pageStructure" | "content" | "document";
 
+/**
+ * The models the host runs for an extraction job.
+ */
+export interface ModelsInit {
+    /**
+     * A layout model: every page's regions are asked for.
+     */
+    layout?: ModelInfoInit;
+    /**
+     * An OCR engine: pages with no visible text are asked for.
+     */
+    ocr?: ModelInfoInit;
+    /**
+     * The resolution pages are asked for at; 150 by default.
+     */
+    dpi?: number;
+    /**
+     * OCR's languages (BCP 47).
+     */
+    langs?: string[];
+}
+
 /**
  * The pages an edit touched.
  */
@@ -634,6 +996,67 @@ export type SaveModeJs = "auto" | "incremental" | "fullRewrite";
  */
 export type ExtractFormat = "markdown" | "json" | "text" | "chunks";
 
+/**
+ * What `provide` takes.
+ */
+export type ModelResultInit = { kind: "layout"; regions: RegionInit[] } | { kind: "ocr"; words: WordInit[] };
+
+/**
+ * What a document holds besides its pages.
+ */
+export interface ContainsJs {
+    /**
+     * Some action runs JavaScript.
+     */
+    javascript: boolean;
+    /**
+     * The number of embedded files.
+     */
+    attachments: number;
+    /**
+     * The document has a form with at least one field.
+     */
+    forms: boolean;
+}
+
+/**
+ * What a file's author permits: a request to readers, not a lock.
+ */
+export interface PermissionsJs {
+    /**
+     * Printing.
+     */
+    print: boolean;
+    /**
+     * Printing at full quality.
+     */
+    printHighQuality: boolean;
+    /**
+     * Changing the contents.
+     */
+    modify: boolean;
+    /**
+     * Copying text and graphics.
+     */
+    copy: boolean;
+    /**
+     * Adding or changing annotations.
+     */
+    annotate: boolean;
+    /**
+     * Filling form fields.
+     */
+    fillForms: boolean;
+    /**
+     * Extracting text for accessibility.
+     */
+    accessibility: boolean;
+    /**
+     * Inserting, rotating or deleting pages.
+     */
+    assemble: boolean;
+}
+
 /**
  * What a save wrote (`laminapdf::SaveReport`).
  */
@@ -669,6 +1092,11 @@ export interface SaveReportJs {
  */
 export type OutputJs = "whole" | "appendedOnly";
 
+/**
+ * What following a link does, by `kind`.
+ */
+export type LinkToJs = { kind: "dest"; target?: TargetJs } | { kind: "named"; name: string; target?: TargetJs } | { kind: "uri"; uri: string } | { kind: "remote"; file?: string } | { kind: "named-action"; name: string } | { kind: "unsupported"; action: string };
+
 /**
  * What happens to hidden text and page furniture.
  */
@@ -873,6 +1301,16 @@ export class WasmDocument {
      * error (nothing is applied).
      */
     apply(ops: Op[]): EditedJs;
+    /**
+     * An extraction stepped a page at a time, whose models the host runs: each step returns
+     * its progress, with the result it waits for under `request` (render that page at `dpi`,
+     * run the model, hand the result to `provide`), and once every page is done, the output in
+     * `options.format` (the extract and text modules).
+     *
+     * # Errors
+     * As `extract`, for the options and the pages asked for.
+     */
+    extractJob(options: ExtractInit): WasmExtractJob;
     /**
      * The document's text for retrieval and LLMs, as one string: Markdown, plain text, the
      * `laminapdf.extract/1` JSON, or the chunks as a JSON array (the extract and text modules).
@@ -893,6 +1331,21 @@ export class WasmDocument {
      * Reading errors.
      */
     info(): DocumentInfoJs;
+    /**
+     * Whether the document says it is tagged (`/MarkInfo /Marked true`; the extract and text modules).
+     *
+     * # Errors
+     * Reading errors.
+     */
+    isTagged(): boolean;
+    /**
+     * The named destination `name`, or `undefined` when the document has none by that name or
+     * it names no page here (the extract and text modules).
+     *
+     * # Errors
+     * `data-pending` and the reading errors.
+     */
+    namedDestination(name: string): TargetJs | undefined;
     /**
      * Opens a document from its bytes.
      *
@@ -910,6 +1363,14 @@ export class WasmDocument {
      * As `openBytes`, and `data-pending`.
      */
     static openSource(source: WasmSparseSource, options: OpenOptionsInit | undefined): WasmDocument;
+    /**
+     * The outline's top-level items, each with its children; empty without an outline (the
+     * extract and text modules).
+     *
+     * # Errors
+     * `data-pending` and the reading errors.
+     */
+    outline(): OutlineNodeJs[];
     /**
      * A handle on the page with `id` (`p12.0`).
      *
@@ -938,6 +1399,13 @@ export class WasmDocument {
      * Reading errors.
      */
     pageInfos(): PageInfoJs[];
+    /**
+     * The visible links on page `index`, in annotation order (the extract and text modules).
+     *
+     * # Errors
+     * `page-index-out-of-range`, `data-pending` and the reading errors.
+     */
+    pageLinks(index: number): PageLinkJs[];
     /**
      * A handle on page `index` that follows the page by its ID (F2).
      *
@@ -945,6 +1413,13 @@ export class WasmDocument {
      * `page-index-out-of-range`; reading errors.
      */
     page(index: number): WasmPage;
+    /**
+     * The document's properties (the extract and text modules). Reads every object once.
+     *
+     * # Errors
+     * `data-pending` and the reading errors.
+     */
+    properties(): PropertiesJs;
     /**
      * Rebases onto `bytes`, the output of a save whose rebase failed (`not-rebased`).
      *
@@ -959,6 +1434,15 @@ export class WasmDocument {
      * As [`WasmDocument::undo`].
      */
     redo(): UndoOutcomeJs | undefined;
+    /**
+     * The document for reader mode, as JSON: whether it can be shown and why not, its
+     * language, its blocks in the structure tree's order (figures, formulas and tables as crops
+     * with their alternate text) and the checks' measures (the extract module only).
+     *
+     * # Errors
+     * `data-pending` and the reading errors.
+     */
+    reflow(): string;
     /**
      * A save as a job, stepped with a budget of work units (ADR 0028 §3, §4).
      *
@@ -973,6 +1457,14 @@ export class WasmDocument {
      * The save's errors (`would-invalidate-signatures`, `requires-rewrite`, …).
      */
     save(options: SaveOptionsInit | undefined): { bytes: Uint8Array<ArrayBuffer>; report: SaveReportJs };
+    /**
+     * The structure tree's top-level elements in reading order, or `undefined` without one (the
+     * extract and text modules, beside `WasmPage.text`, whose runs it links by MCID).
+     *
+     * # Errors
+     * Reading errors.
+     */
+    structureTree(): StructureNodeJs[] | undefined;
     /**
      * Undoes the newest transaction; `undefined` when there is nothing to undo.
      *
@@ -986,6 +1478,38 @@ export class WasmDocument {
     warnings(): WasmWarningCursor;
 }
 
+/**
+ * An extraction in progress, from `WasmDocument.extractJob`; it holds its document weakly.
+ */
+export class WasmExtractJob {
+    private constructor();
+    free(): void;
+    [Symbol.dispose](): void;
+    /**
+     * Stops the job: its next step fails with `cancelled`.
+     */
+    cancel(): void;
+    /**
+     * Hands the job the result of its open request: `{ kind: "layout", regions }` or
+     * `{ kind: "ocr", words }`, boxes as `[x, y, width, height]` in top-left points.
+     *
+     * # Errors
+     * `model-result-unexpected` when nothing was asked for or `result` is the other kind;
+     * `invalid-argument` / `unknown-option` for a malformed result.
+     */
+    provide(result: ModelResultInit): void;
+    /**
+     * Extracts up to `budget` pages: `{ state: "progress", done, total, phase, request? }` or
+     * `{ state: "done", output }`, the output in the job's format. While `request` is set, a
+     * step does nothing until `provide` answers it.
+     *
+     * # Errors
+     * `cancelled`, `document-changed`, `job-finished`, `disposed` when the document is freed,
+     * and the reading errors.
+     */
+    step(budget: number): ({ state: "progress" } & ExtractProgressJs) | { state: "done"; output: string };
+}
+
 /**
  * An image to draw (decoded or passed through once, shared by every use).
  */
@@ -1071,7 +1595,8 @@ export class WasmPageBuilder {
     [Symbol.dispose](): void;
     /**
      * Opens an artifact until the matching `endTag`: `kind` is `layout`, `page`, `pagination`
-     * or `background`.
+     * or `background`, or a pagination artifact that says what it is: `header` (attached to the
+     * top edge), `footer` (the bottom) or `watermark`.
      *
      * # Errors
      * `invalid-argument` for another kind.
@@ -1340,6 +1865,7 @@ export interface InitOutput {
     readonly memory: WebAssembly.Memory;
     readonly __wbg_wasmbuilder_free: (a: number, b: number) => void;
     readonly __wbg_wasmdocument_free: (a: number, b: number) => void;
+    readonly __wbg_wasmextractjob_free: (a: number, b: number) => void;
     readonly __wbg_wasmimage_free: (a: number, b: number) => void;
     readonly __wbg_wasmpage_free: (a: number, b: number) => void;
     readonly __wbg_wasmpagebuilder_free: (a: number, b: number) => void;
@@ -1362,21 +1888,32 @@ export interface InitOutput {
     readonly wasmdocument_apply: (a: number, b: any) => [number, number, number];
     readonly wasmdocument_applyLog: (a: number, b: number, c: number) => [number, number, number];
     readonly wasmdocument_extract: (a: number, b: any) => [number, number, number, number];
+    readonly wasmdocument_extractJob: (a: number, b: any) => [number, number, number];
     readonly wasmdocument_hasUnsavedChanges: (a: number) => number;
     readonly wasmdocument_info: (a: number) => [number, number, number];
+    readonly wasmdocument_isTagged: (a: number) => [number, number, number];
+    readonly wasmdocument_namedDestination: (a: number, b: number, c: number) => [number, number, number];
     readonly wasmdocument_openBytes: (a: number, b: number, c: any) => [number, number, number];
     readonly wasmdocument_openSource: (a: number, b: any) => [number, number, number];
+    readonly wasmdocument_outline: (a: number) => [number, number, number];
     readonly wasmdocument_page: (a: number, b: number) => [number, number, number];
     readonly wasmdocument_pageById: (a: number, b: number, c: number) => [number, number, number];
     readonly wasmdocument_pageCount: (a: number) => [number, number, number];
     readonly wasmdocument_pageInfo: (a: number, b: number) => [number, number, number];
     readonly wasmdocument_pageInfos: (a: number) => [number, number, number];
+    readonly wasmdocument_pageLinks: (a: number, b: number) => [number, number, number];
+    readonly wasmdocument_properties: (a: number) => [number, number, number];
     readonly wasmdocument_rebase: (a: number, b: number, c: number) => [number, number];
     readonly wasmdocument_redo: (a: number) => [number, number, number];
+    readonly wasmdocument_reflow: (a: number) => [number, number, number, number];
     readonly wasmdocument_save: (a: number, b: any) => [number, number, number];
     readonly wasmdocument_saveJob: (a: number, b: any) => [number, number, number];
+    readonly wasmdocument_structureTree: (a: number) => [number, number, number];
     readonly wasmdocument_undo: (a: number) => [number, number, number];
     readonly wasmdocument_warnings: (a: number) => number;
+    readonly wasmextractjob_cancel: (a: number) => void;
+    readonly wasmextractjob_provide: (a: number, b: any) => [number, number];
+    readonly wasmextractjob_step: (a: number, b: number) => [number, number, number];
     readonly wasmimage_height: (a: number) => number;
     readonly wasmimage_jpeg: (a: number, b: number) => [number, number, number];
     readonly wasmimage_png: (a: number, b: number) => [number, number, number];