yom-format 0.1.0

package/dist/pipeline/convert.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Mirrors the Swift implementation's `YOMConverter.Stage`
+ * (`swift/Sources/YOMConverter/YOMConverter.swift`) so Web and native
+ * progress reporting read the same way. `rendering`, `encoding`, and
+ * `extractingText` all advance together per-page here since this pipeline
+ * does all three in one pass per page (see `convertPdfToYom`).
+ */
+export type ConversionStage = "rendering" | "encoding" | "extractingText" | "buildingSearchIndex" | "writingPackfile" | "validating";
+export type ProgressCallback = (stage: ConversionStage, fraction: number) => void;
+export declare class ConversionCancelledError extends Error {
+    constructor();
+}
+export interface ConvertOptions {
+    title: string;
+    onProgress?: ProgressCallback;
+    isCancelled?: () => boolean;
+}
+/**
+ * Converts a source PDF's bytes into a validated `.yom` package, entirely in
+ * the browser. Ports the shape of the Swift implementation's
+ * `YOMConverter.convert` + `YOMConversionPipeline.buildWriterInput`
+ * (`swift/Sources/YOMConverter/`), with two known scope differences:
+ * image-only pages get no OCR text yet (a vendored OCR engine for that is
+ * not yet implemented), and PDF export metadata is generated but not
+ * consumed by an export feature yet.
+ */
+export declare function convertPdfToYom(pdfBytes: Uint8Array, options: ConvertOptions): Promise<Uint8Array>;
+//# sourceMappingURL=convert.d.ts.map

package/dist/pipeline/convert.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"convert.d.ts","sourceRoot":"","sources":["../../src/pipeline/convert.ts"],"names":[],"mappings":"AAaA;;;;;;GAMG;AACH,MAAM,MAAM,eAAe,GACvB,WAAW,GACX,UAAU,GACV,gBAAgB,GAChB,qBAAqB,GACrB,iBAAiB,GACjB,YAAY,CAAC;AAEjB,MAAM,MAAM,gBAAgB,GAAG,CAAC,KAAK,EAAE,eAAe,EAAE,QAAQ,EAAE,MAAM,KAAK,IAAI,CAAC;AAElF,qBAAa,wBAAyB,SAAQ,KAAK;;CAIlD;AAED,MAAM,WAAW,cAAc;IAC7B,KAAK,EAAE,MAAM,CAAC;IACd,UAAU,CAAC,EAAE,gBAAgB,CAAC;IAC9B,WAAW,CAAC,EAAE,MAAM,OAAO,CAAC;CAC7B;AAED;;;;;;;;GAQG;AACH,wBAAsB,eAAe,CAAC,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,cAAc,GAAG,OAAO,CAAC,UAAU,CAAC,CAgGxG"}

package/dist/pipeline/convert.js

@@ -0,0 +1,111 @@
+import { BlobKind } from "../yom/generated/yomu/yom/blob-kind.js";
+import { Compression } from "../yom/generated/yomu/yom/compression.js";
+import { ReadingDirection } from "../yom/generated/yomu/yom/reading-direction.js";
+import { writeYOM } from "../yom/writer.js";
+import { validateYOM } from "../yom/validator.js";
+import { buildNormalizedTextBlob } from "../yom/textNormalizer.js";
+import { buildSearchIndex } from "../yom/searchIndex.js";
+import { loadPdfDocument, extractOutline, extractPageLabels } from "./pdfDocument.js";
+import { renderPage } from "./renderPage.js";
+import { extractPageText } from "./textExtract.js";
+import { encodePageAvif, encodeThumbnailAvif } from "./avif.js";
+import { buildOcrLayoutBlob } from "./ocrLayoutBuilder.js";
+export class ConversionCancelledError extends Error {
+    constructor() {
+        super("conversion cancelled");
+    }
+}
+/**
+ * Converts a source PDF's bytes into a validated `.yom` package, entirely in
+ * the browser. Ports the shape of the Swift implementation's
+ * `YOMConverter.convert` + `YOMConversionPipeline.buildWriterInput`
+ * (`swift/Sources/YOMConverter/`), with two known scope differences:
+ * image-only pages get no OCR text yet (a vendored OCR engine for that is
+ * not yet implemented), and PDF export metadata is generated but not
+ * consumed by an export feature yet.
+ */
+export async function convertPdfToYom(pdfBytes, options) {
+    const { title, onProgress = () => { }, isCancelled = () => false } = options;
+    const doc = await loadPdfDocument(pdfBytes);
+    const pageCount = doc.numPages;
+    if (pageCount === 0)
+        throw new Error("PDF has no pages");
+    const [outline, pageLabels] = await Promise.all([extractOutline(doc), extractPageLabels(doc)]);
+    const pageTexts = [];
+    const pageInputs = [];
+    for (let i = 0; i < pageCount; i++) {
+        if (isCancelled())
+            throw new ConversionCancelledError();
+        const page = await doc.getPage(i + 1);
+        const rendered = await renderPage(page);
+        onProgress("rendering", (i + 1) / pageCount);
+        const [pageAvif, thumbAvif] = await Promise.all([
+            encodePageAvif(rendered.image),
+            encodeThumbnailAvif(rendered.thumbnail),
+        ]);
+        onProgress("encoding", (i + 1) / pageCount);
+        const pageText = await extractPageText(page, rendered.viewport);
+        pageTexts.push(pageText);
+        onProgress("extractingText", (i + 1) / pageCount);
+        // Geometry uses PDF-space `page.view` (pdf.js does not expose distinct
+        // mediaBox/cropBox), matching the native converter's own simplification
+        // of collapsing contentBox into cropBox (`YOMPDFExportMetadata.swift`).
+        const pdfBox = page.view.slice();
+        pageInputs.push({
+            pageLabel: pageLabels[i] ?? `${i + 1}`,
+            pixelWidth: rendered.image.width,
+            pixelHeight: rendered.image.height,
+            dpi: 150,
+            colorSpace: "sRGB",
+            colorProfileRef: "",
+            mediaBox: pdfBox,
+            cropBox: pdfBox,
+            contentBox: pdfBox,
+            rotation: page.rotate,
+            readingDirection: ReadingDirection.LeftToRight,
+            pageImage: { kind: BlobKind.PageImage, data: pageAvif, compression: Compression.None },
+            thumbnail: { kind: BlobKind.ThumbnailImage, data: thumbAvif, compression: Compression.None },
+        });
+        page.cleanup();
+    }
+    if (isCancelled())
+        throw new ConversionCancelledError();
+    onProgress("buildingSearchIndex", 0);
+    const ocrLayoutBytes = buildOcrLayoutBlob(pageTexts);
+    const normalized = buildNormalizedTextBlob(pageTexts.map((p) => p.text));
+    const searchIndex = buildSearchIndex(normalized.normalizedPageTexts);
+    const searchTerms = searchIndex.searchTerms.map((t) => ({
+        term: t.term,
+        dictionaryOffset: BigInt(t.dictionaryOffset),
+        dictionaryLength: BigInt(t.dictionaryLength),
+        postingsOffset: BigInt(t.postingsOffset),
+        postingsLength: BigInt(t.postingsLength),
+    }));
+    onProgress("buildingSearchIndex", 1);
+    const writerInput = {
+        book: {
+            title,
+            readingDirection: ReadingDirection.LeftToRight,
+            createdAtUnix: BigInt(Math.floor(Date.now() / 1000)),
+        },
+        pages: pageInputs,
+        ocrLayout: { kind: BlobKind.OCRLayout, data: ocrLayoutBytes, compression: Compression.Zstd },
+        normalizedText: { kind: BlobKind.NormalizedText, data: normalized.data, compression: Compression.Zstd },
+        searchDictionary: { kind: BlobKind.SearchDictionary, data: searchIndex.dictionaryBlob, compression: Compression.Zstd },
+        searchPostings: { kind: BlobKind.SearchPostings, data: searchIndex.postingsBlob, compression: Compression.None },
+        searchTerms,
+        pdfExport: { pageLabels, outline },
+    };
+    if (isCancelled())
+        throw new ConversionCancelledError();
+    onProgress("writingPackfile", 0);
+    const packfileBytes = await writeYOM(writerInput);
+    onProgress("writingPackfile", 1);
+    if (isCancelled())
+        throw new ConversionCancelledError();
+    onProgress("validating", 0);
+    await validateYOM(packfileBytes);
+    onProgress("validating", 1);
+    return packfileBytes;
+}
+//# sourceMappingURL=convert.js.map

package/dist/pipeline/convert.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"convert.js","sourceRoot":"","sources":["../../src/pipeline/convert.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,wCAAwC,CAAC;AAClE,OAAO,EAAE,WAAW,EAAE,MAAM,0CAA0C,CAAC;AACvE,OAAO,EAAE,gBAAgB,EAAE,MAAM,gDAAgD,CAAC;AAClF,OAAO,EAAE,QAAQ,EAA0C,MAAM,kBAAkB,CAAC;AACpF,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EAAE,uBAAuB,EAAE,MAAM,0BAA0B,CAAC;AACnE,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACzD,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAC;AACtF,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,EAAE,eAAe,EAAiB,MAAM,kBAAkB,CAAC;AAClE,OAAO,EAAE,cAAc,EAAE,mBAAmB,EAAE,MAAM,WAAW,CAAC;AAChE,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAC;AAmB3D,MAAM,OAAO,wBAAyB,SAAQ,KAAK;IACjD;QACE,KAAK,CAAC,sBAAsB,CAAC,CAAC;IAChC,CAAC;CACF;AAQD;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,QAAoB,EAAE,OAAuB;IACjF,MAAM,EAAE,KAAK,EAAE,UAAU,GAAG,GAAG,EAAE,GAAE,CAAC,EAAE,WAAW,GAAG,GAAG,EAAE,CAAC,KAAK,EAAE,GAAG,OAAO,CAAC;IAE5E,MAAM,GAAG,GAAG,MAAM,eAAe,CAAC,QAAQ,CAAC,CAAC;IAC5C,MAAM,SAAS,GAAG,GAAG,CAAC,QAAQ,CAAC;IAC/B,IAAI,SAAS,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,kBAAkB,CAAC,CAAC;IAEzD,MAAM,CAAC,OAAO,EAAE,UAAU,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,GAAG,CAAC,EAAE,iBAAiB,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;IAE/F,MAAM,SAAS,GAAe,EAAE,CAAC;IACjC,MAAM,UAAU,GAAyB,EAAE,CAAC;IAE5C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,EAAE,CAAC,EAAE,EAAE,CAAC;QACnC,IAAI,WAAW,EAAE;YAAE,MAAM,IAAI,wBAAwB,EAAE,CAAC;QAExD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACtC,MAAM,QAAQ,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,CAAC;QACxC,UAAU,CAAC,WAAW,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;QAE7C,MAAM,CAAC,QAAQ,EAAE,SAAS,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YAC9C,cAAc,CAAC,QAAQ,CAAC,KAAK,CAAC;YAC9B,mBAAmB,CAAC,QAAQ,CAAC,SAAS,CAAC;SACxC,CAAC,CAAC;QACH,UAAU,CAAC,UAAU,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;QAE5C,MAAM,QAAQ,GAAG,MAAM,eAAe,CAAC,IAAI,EAAE,QAAQ,CAAC,QAAQ,CAAC,CAAC;QAChE,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACzB,UAAU,CAAC,gBAAgB,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;QAElD,uEAAuE;QACvE,wEAAwE;QACxE,wEAAwE;QACxE,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC;QAEjC,UAAU,CAAC,IAAI,CAAC;YACd,SAAS,EAAE,UAAU,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,GAAG,CAAC,EAAE;YACtC,UAAU,EAAE,QAAQ,CAAC,KAAK,CAAC,KAAK;YAChC,WAAW,EAAE,QAAQ,CAAC,KAAK,CAAC,MAAM;YAClC,GAAG,EAAE,GAAG;YACR,UAAU,EAAE,MAAM;YAClB,eAAe,EAAE,EAAE;YACnB,QAAQ,EAAE,MAAM;YAChB,OAAO,EAAE,MAAM;YACf,UAAU,EAAE,MAAM;YAClB,QAAQ,EAAE,IAAI,CAAC,MAAM;YACrB,gBAAgB,EAAE,gBAAgB,CAAC,WAAW;YAC9C,SAAS,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,SAAS,EAAE,IAAI,EAAE,QAAQ,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;YACtF,SAAS,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,cAAc,EAAE,IAAI,EAAE,SAAS,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;SAC7F,CAAC,CAAC;QAEH,IAAI,CAAC,OAAO,EAAE,CAAC;IACjB,CAAC;IAED,IAAI,WAAW,EAAE;QAAE,MAAM,IAAI,wBAAwB,EAAE,CAAC;IAExD,UAAU,CAAC,qBAAqB,EAAE,CAAC,CAAC,CAAC;IACrC,MAAM,cAAc,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;IACrD,MAAM,UAAU,GAAG,uBAAuB,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IACzE,MAAM,WAAW,GAAG,gBAAgB,CAAC,UAAU,CAAC,mBAAmB,CAAC,CAAC;IACrE,MAAM,WAAW,GAAsB,WAAW,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACzE,IAAI,EAAE,CAAC,CAAC,IAAI;QACZ,gBAAgB,EAAE,MAAM,CAAC,CAAC,CAAC,gBAAgB,CAAC;QAC5C,gBAAgB,EAAE,MAAM,CAAC,CAAC,CAAC,gBAAgB,CAAC;QAC5C,cAAc,EAAE,MAAM,CAAC,CAAC,CAAC,cAAc,CAAC;QACxC,cAAc,EAAE,MAAM,CAAC,CAAC,CAAC,cAAc,CAAC;KACzC,CAAC,CAAC,CAAC;IACJ,UAAU,CAAC,qBAAqB,EAAE,CAAC,CAAC,CAAC;IAErC,MAAM,WAAW,GAAgB;QAC/B,IAAI,EAAE;YACJ,KAAK;YACL,gBAAgB,EAAE,gBAAgB,CAAC,WAAW;YAC9C,aAAa,EAAE,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC,CAAC;SACrD;QACD,KAAK,EAAE,UAAU;QACjB,SAAS,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,SAAS,EAAE,IAAI,EAAE,cAAc,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;QAC5F,cAAc,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,cAAc,EAAE,IAAI,EAAE,UAAU,CAAC,IAAI,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;QACvG,gBAAgB,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,gBAAgB,EAAE,IAAI,EAAE,WAAW,CAAC,cAAc,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;QACtH,cAAc,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,cAAc,EAAE,IAAI,EAAE,WAAW,CAAC,YAAY,EAAE,WAAW,EAAE,WAAW,CAAC,IAAI,EAAE;QAChH,WAAW;QACX,SAAS,EAAE,EAAE,UAAU,EAAE,OAAO,EAAE;KACnC,CAAC;IAEF,IAAI,WAAW,EAAE;QAAE,MAAM,IAAI,wBAAwB,EAAE,CAAC;IAExD,UAAU,CAAC,iBAAiB,EAAE,CAAC,CAAC,CAAC;IACjC,MAAM,aAAa,GAAG,MAAM,QAAQ,CAAC,WAAW,CAAC,CAAC;IAClD,UAAU,CAAC,iBAAiB,EAAE,CAAC,CAAC,CAAC;IAEjC,IAAI,WAAW,EAAE;QAAE,MAAM,IAAI,wBAAwB,EAAE,CAAC;IAExD,UAAU,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC;IAC5B,MAAM,WAAW,CAAC,aAAa,CAAC,CAAC;IACjC,UAAU,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC;IAE5B,OAAO,aAAa,CAAC;AACvB,CAAC"}

package/dist/pipeline/index.d.ts

@@ -0,0 +1,17 @@
+/**
+ * Browser-only PDF-to-YOM conversion pipeline: renders a PDF with pdf.js,
+ * encodes pages/thumbnails as AVIF, extracts embedded text and outline, and
+ * writes a validated `.yom` package. Requires a DOM (Canvas/OffscreenCanvas)
+ * environment; it will not run under plain Node.
+ *
+ * Kept as a separate subpath (`yom-format/pipeline`) so the core read/write
+ * API in `yom-format` stays free of the pdf.js and image-codec dependencies
+ * this pulls in.
+ */
+export { convertPdfToYom, ConversionCancelledError, type ConversionStage, type ProgressCallback, type ConvertOptions, } from "./convert.js";
+export { loadPdfDocument, extractOutline, extractPageLabels, type OutlineEntryData } from "./pdfDocument.js";
+export { renderPage, THUMBNAIL_MAX_DIMENSION, type RenderedPage } from "./renderPage.js";
+export { extractPageText, type PageText, type PageTextToken } from "./textExtract.js";
+export { encodePageAvif, encodeThumbnailAvif } from "./avif.js";
+export { buildOcrLayoutBlob } from "./ocrLayoutBuilder.js";
+//# sourceMappingURL=index.d.ts.map

package/dist/pipeline/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/pipeline/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EACL,eAAe,EACf,wBAAwB,EACxB,KAAK,eAAe,EACpB,KAAK,gBAAgB,EACrB,KAAK,cAAc,GACpB,MAAM,cAAc,CAAC;AAEtB,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,iBAAiB,EAAE,KAAK,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AAE7G,OAAO,EAAE,UAAU,EAAE,uBAAuB,EAAE,KAAK,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAEzF,OAAO,EAAE,eAAe,EAAE,KAAK,QAAQ,EAAE,KAAK,aAAa,EAAE,MAAM,kBAAkB,CAAC;AAEtF,OAAO,EAAE,cAAc,EAAE,mBAAmB,EAAE,MAAM,WAAW,CAAC;AAEhE,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAC"}

package/dist/pipeline/index.js

@@ -0,0 +1,17 @@
+/**
+ * Browser-only PDF-to-YOM conversion pipeline: renders a PDF with pdf.js,
+ * encodes pages/thumbnails as AVIF, extracts embedded text and outline, and
+ * writes a validated `.yom` package. Requires a DOM (Canvas/OffscreenCanvas)
+ * environment; it will not run under plain Node.
+ *
+ * Kept as a separate subpath (`yom-format/pipeline`) so the core read/write
+ * API in `yom-format` stays free of the pdf.js and image-codec dependencies
+ * this pulls in.
+ */
+export { convertPdfToYom, ConversionCancelledError, } from "./convert.js";
+export { loadPdfDocument, extractOutline, extractPageLabels } from "./pdfDocument.js";
+export { renderPage, THUMBNAIL_MAX_DIMENSION } from "./renderPage.js";
+export { extractPageText } from "./textExtract.js";
+export { encodePageAvif, encodeThumbnailAvif } from "./avif.js";
+export { buildOcrLayoutBlob } from "./ocrLayoutBuilder.js";
+//# sourceMappingURL=index.js.map

package/dist/pipeline/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/pipeline/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EACL,eAAe,EACf,wBAAwB,GAIzB,MAAM,cAAc,CAAC;AAEtB,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,iBAAiB,EAAyB,MAAM,kBAAkB,CAAC;AAE7G,OAAO,EAAE,UAAU,EAAE,uBAAuB,EAAqB,MAAM,iBAAiB,CAAC;AAEzF,OAAO,EAAE,eAAe,EAAqC,MAAM,kBAAkB,CAAC;AAEtF,OAAO,EAAE,cAAc,EAAE,mBAAmB,EAAE,MAAM,WAAW,CAAC;AAEhE,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAC"}

package/dist/pipeline/ocrLayoutBuilder.d.ts

@@ -0,0 +1,10 @@
+import type { PageText } from "./textExtract.js";
+/**
+ * Builds the pre-compression OCR FlatBuffers blob (`writer.ts` zstd-
+ * compresses it per the fixed policy) from per-page token layout. Each
+ * page's `text` field is the concatenation of its own tokens' text, in
+ * order, with no separators -- `textStart`/`textLength` index directly into
+ * that buffer, not into the original PDF text run structure.
+ */
+export declare function buildOcrLayoutBlob(pages: PageText[]): Uint8Array;
+//# sourceMappingURL=ocrLayoutBuilder.d.ts.map

package/dist/pipeline/ocrLayoutBuilder.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ocrLayoutBuilder.d.ts","sourceRoot":"","sources":["../../src/pipeline/ocrLayoutBuilder.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,QAAQ,EAAiB,MAAM,kBAAkB,CAAC;AAIhE;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,QAAQ,EAAE,GAAG,UAAU,CAsChE"}

package/dist/pipeline/ocrLayoutBuilder.js

@@ -0,0 +1,62 @@
+import { Builder } from "flatbuffers";
+import { OCRLayoutT } from "../yom/generated/yomu/yom/ocrlayout.js";
+import { OCRPageT } from "../yom/generated/yomu/yom/ocrpage.js";
+import { OCRBlockT } from "../yom/generated/yomu/yom/ocrblock.js";
+import { OCRLineT } from "../yom/generated/yomu/yom/ocrline.js";
+import { OCRTokenT } from "../yom/generated/yomu/yom/ocrtoken.js";
+import { concatBytes } from "../yom/bytes.js";
+/**
+ * Builds the pre-compression OCR FlatBuffers blob (`writer.ts` zstd-
+ * compresses it per the fixed policy) from per-page token layout. Each
+ * page's `text` field is the concatenation of its own tokens' text, in
+ * order, with no separators -- `textStart`/`textLength` index directly into
+ * that buffer, not into the original PDF text run structure.
+ */
+export function buildOcrLayoutBlob(pages) {
+    const encoder = new TextEncoder();
+    const builder = new Builder(1024);
+    builder.forceDefaults(true);
+    const ocrPages = pages.map((page, pageIndex) => {
+        const tokenBytesList = page.tokens.map((t) => encoder.encode(t.text));
+        let cursor = 0;
+        const tokens = page.tokens.map((token, i) => {
+            const bytes = tokenBytesList[i];
+            const textStart = cursor;
+            cursor += bytes.length;
+            return new OCRTokenT(textStart, bytes.length, token.bbox, 1.0, token.lineId, token.blockId);
+        });
+        const pageText = new TextDecoder().decode(concatBytes(tokenBytesList));
+        const lineBboxes = new Map();
+        const blockBboxes = new Map();
+        for (const token of page.tokens) {
+            unionBbox(lineBboxes, token.lineId, token.bbox);
+            unionBbox(blockBboxes, token.blockId, token.bbox);
+        }
+        const lines = Array.from(lineBboxes.keys())
+            .sort((a, b) => a - b)
+            .map((id) => new OCRLineT(id, blockIdForLine(page.tokens, id), lineBboxes.get(id)));
+        const blocks = Array.from(blockBboxes.keys())
+            .sort((a, b) => a - b)
+            .map((id) => new OCRBlockT(id, blockBboxes.get(id)));
+        return new OCRPageT(pageIndex, pageText, blocks, lines, tokens);
+    });
+    const layout = new OCRLayoutT(ocrPages);
+    const offset = layout.pack(builder);
+    builder.finish(offset);
+    return builder.asUint8Array();
+}
+function blockIdForLine(tokens, lineId) {
+    return tokens.find((t) => t.lineId === lineId)?.blockId ?? 0;
+}
+function unionBbox(map, id, bbox) {
+    const existing = map.get(id);
+    if (!existing) {
+        map.set(id, [...bbox]);
+        return;
+    }
+    existing[0] = Math.min(existing[0], bbox[0]);
+    existing[1] = Math.min(existing[1], bbox[1]);
+    existing[2] = Math.max(existing[2], bbox[2]);
+    existing[3] = Math.max(existing[3], bbox[3]);
+}
+//# sourceMappingURL=ocrLayoutBuilder.js.map

package/dist/pipeline/ocrLayoutBuilder.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ocrLayoutBuilder.js","sourceRoot":"","sources":["../../src/pipeline/ocrLayoutBuilder.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,aAAa,CAAC;AACtC,OAAO,EAAE,UAAU,EAAE,MAAM,wCAAwC,CAAC;AACpE,OAAO,EAAE,QAAQ,EAAE,MAAM,sCAAsC,CAAC;AAChE,OAAO,EAAE,SAAS,EAAE,MAAM,uCAAuC,CAAC;AAClE,OAAO,EAAE,QAAQ,EAAE,MAAM,sCAAsC,CAAC;AAChE,OAAO,EAAE,SAAS,EAAE,MAAM,uCAAuC,CAAC;AAClE,OAAO,EAAE,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAK9C;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,KAAiB;IAClD,MAAM,OAAO,GAAG,IAAI,WAAW,EAAE,CAAC;IAClC,MAAM,OAAO,GAAG,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;IAClC,OAAO,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;IAE5B,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,EAAE;QAC7C,MAAM,cAAc,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACtE,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,CAAC,EAAE,EAAE;YAC1C,MAAM,KAAK,GAAG,cAAc,CAAC,CAAC,CAAC,CAAC;YAChC,MAAM,SAAS,GAAG,MAAM,CAAC;YACzB,MAAM,IAAI,KAAK,CAAC,MAAM,CAAC;YACvB,OAAO,IAAI,SAAS,CAAC,SAAS,EAAE,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,IAAI,EAAE,GAAG,EAAE,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,OAAO,CAAC,CAAC;QAC9F,CAAC,CAAC,CAAC;QAEH,MAAM,QAAQ,GAAG,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,WAAW,CAAC,cAAc,CAAC,CAAC,CAAC;QAEvE,MAAM,UAAU,GAAG,IAAI,GAAG,EAAgB,CAAC;QAC3C,MAAM,WAAW,GAAG,IAAI,GAAG,EAAgB,CAAC;QAC5C,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAChC,SAAS,CAAC,UAAU,EAAE,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;YAChD,SAAS,CAAC,WAAW,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QACpD,CAAC;QAED,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;aACxC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC;aACrB,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,IAAI,QAAQ,CAAC,EAAE,EAAE,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,GAAG,CAAC,EAAE,CAAE,CAAC,CAAC,CAAC;QACvF,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;aAC1C,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC;aACrB,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,IAAI,SAAS,CAAC,EAAE,EAAE,WAAW,CAAC,GAAG,CAAC,EAAE,CAAE,CAAC,CAAC,CAAC;QAExD,OAAO,IAAI,QAAQ,CAAC,SAAS,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC;IAClE,CAAC,CAAC,CAAC;IAEH,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,QAAQ,CAAC,CAAC;IACxC,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACpC,OAAO,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;IACvB,OAAO,OAAO,CAAC,YAAY,EAAE,CAAC;AAChC,CAAC;AAED,SAAS,cAAc,CAAC,MAAuB,EAAE,MAAc;IAC7D,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,MAAM,CAAC,EAAE,OAAO,IAAI,CAAC,CAAC;AAC/D,CAAC;AAED,SAAS,SAAS,CAAC,GAAsB,EAAE,EAAU,EAAE,IAAU;IAC/D,MAAM,QAAQ,GAAG,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;IAC7B,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC;QACvB,OAAO;IACT,CAAC;IACD,QAAQ,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,QAAQ,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,QAAQ,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,QAAQ,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;AAC/C,CAAC"}

package/dist/pipeline/pdfDocument.d.ts

@@ -0,0 +1,12 @@
+import type { PDFDocumentProxy } from "pdfjs-dist";
+export declare function loadPdfDocument(data: Uint8Array): Promise<PDFDocumentProxy>;
+export interface OutlineEntryData {
+    title: string;
+    pageId: number;
+    children: OutlineEntryData[];
+}
+/** Extracts the PDF outline/TOC, resolving each destination to a page index. */
+export declare function extractOutline(doc: PDFDocumentProxy): Promise<OutlineEntryData[]>;
+/** Page labels, falling back to 1-based numbering when the PDF has none. */
+export declare function extractPageLabels(doc: PDFDocumentProxy): Promise<string[]>;
+//# sourceMappingURL=pdfDocument.d.ts.map

package/dist/pipeline/pdfDocument.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdfDocument.d.ts","sourceRoot":"","sources":["../../src/pipeline/pdfDocument.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AA4BnD,wBAAsB,eAAe,CAAC,IAAI,EAAE,UAAU,GAAG,OAAO,CAAC,gBAAgB,CAAC,CAKjF;AAED,MAAM,WAAW,gBAAgB;IAC/B,KAAK,EAAE,MAAM,CAAC;IACd,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,gBAAgB,EAAE,CAAC;CAC9B;AAED,gFAAgF;AAChF,wBAAsB,cAAc,CAAC,GAAG,EAAE,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAIvF;AAiCD,4EAA4E;AAC5E,wBAAsB,iBAAiB,CAAC,GAAG,EAAE,gBAAgB,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAIhF"}

package/dist/pipeline/pdfDocument.js

@@ -0,0 +1,76 @@
+import { GlobalWorkerOptions, getDocument } from "pdfjs-dist";
+// Runs inside `convert.worker.ts`; pdf.js spawns (or fakes, if the browser
+// disallows a nested worker) its own core worker from this URL.
+GlobalWorkerOptions.workerSrc = new URL("pdfjs-dist/build/pdf.worker.mjs", import.meta.url).href;
+/**
+ * Minimal `CanvasFactory` so pdf.js's *internal* auxiliary canvases (soft
+ * masks, tiling patterns) work in a Worker, where `document` does not exist.
+ * Our own top-level per-page render target is a separate `OffscreenCanvas`
+ * created directly in `renderPage.ts`.
+ */
+class OffscreenCanvasFactory {
+    create(width, height) {
+        const canvas = new OffscreenCanvas(width, height);
+        const context = canvas.getContext("2d");
+        return { canvas, context };
+    }
+    reset(canvasAndContext, width, height) {
+        canvasAndContext.canvas.width = width;
+        canvasAndContext.canvas.height = height;
+    }
+    destroy(canvasAndContext) {
+        canvasAndContext.canvas = null;
+        canvasAndContext.context = null;
+    }
+}
+export async function loadPdfDocument(data) {
+    // pdf.js instantiates this itself (`new CanvasFactory(...)`), so the
+    // option must be the class, not an instance of it.
+    const task = getDocument({ data, CanvasFactory: OffscreenCanvasFactory });
+    return task.promise;
+}
+/** Extracts the PDF outline/TOC, resolving each destination to a page index. */
+export async function extractOutline(doc) {
+    const raw = await doc.getOutline();
+    if (!raw)
+        return [];
+    return resolveOutlineNodes(doc, raw);
+}
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+async function resolveOutlineNodes(doc, nodes) {
+    const result = [];
+    for (const node of nodes) {
+        const pageId = await resolvePageId(doc, node.dest);
+        const children = node.items?.length ? await resolveOutlineNodes(doc, node.items) : [];
+        result.push({ title: typeof node.title === "string" ? node.title : "", pageId, children });
+    }
+    return result;
+}
+async function resolvePageId(doc, 
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+dest) {
+    try {
+        let resolved = dest;
+        if (typeof resolved === "string") {
+            resolved = await doc.getDestination(resolved);
+        }
+        if (!Array.isArray(resolved) || resolved.length === 0)
+            return 0;
+        const ref = resolved[0];
+        if (typeof ref === "number")
+            return ref;
+        const index = await doc.getPageIndex(ref);
+        return index >= 0 ? index : 0;
+    }
+    catch {
+        return 0;
+    }
+}
+/** Page labels, falling back to 1-based numbering when the PDF has none. */
+export async function extractPageLabels(doc) {
+    const labels = await doc.getPageLabels();
+    if (labels)
+        return labels;
+    return Array.from({ length: doc.numPages }, (_, i) => `${i + 1}`);
+}
+//# sourceMappingURL=pdfDocument.js.map

package/dist/pipeline/pdfDocument.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdfDocument.js","sourceRoot":"","sources":["../../src/pipeline/pdfDocument.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAG9D,2EAA2E;AAC3E,gEAAgE;AAChE,mBAAmB,CAAC,SAAS,GAAG,IAAI,GAAG,CAAC,iCAAiC,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC;AAEjG;;;;;GAKG;AACH,MAAM,sBAAsB;IAC1B,MAAM,CAAC,KAAa,EAAE,MAAc;QAClC,MAAM,MAAM,GAAG,IAAI,eAAe,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;QAClD,MAAM,OAAO,GAAG,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;QACxC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IAC7B,CAAC;IACD,KAAK,CAAC,gBAA6C,EAAE,KAAa,EAAE,MAAc;QAChF,gBAAgB,CAAC,MAAM,CAAC,KAAK,GAAG,KAAK,CAAC;QACtC,gBAAgB,CAAC,MAAM,CAAC,MAAM,GAAG,MAAM,CAAC;IAC1C,CAAC;IACD,OAAO,CAAC,gBAAsE;QAC5E,gBAAgB,CAAC,MAAM,GAAG,IAAI,CAAC;QAC/B,gBAAgB,CAAC,OAAO,GAAG,IAAI,CAAC;IAClC,CAAC;CACF;AAED,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,IAAgB;IACpD,qEAAqE;IACrE,mDAAmD;IACnD,MAAM,IAAI,GAAG,WAAW,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,sBAA0C,EAAE,CAAC,CAAC;IAC9F,OAAO,IAAI,CAAC,OAAO,CAAC;AACtB,CAAC;AAQD,gFAAgF;AAChF,MAAM,CAAC,KAAK,UAAU,cAAc,CAAC,GAAqB;IACxD,MAAM,GAAG,GAAG,MAAM,GAAG,CAAC,UAAU,EAAE,CAAC;IACnC,IAAI,CAAC,GAAG;QAAE,OAAO,EAAE,CAAC;IACpB,OAAO,mBAAmB,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;AACvC,CAAC;AAED,8DAA8D;AAC9D,KAAK,UAAU,mBAAmB,CAAC,GAAqB,EAAE,KAAY;IACpE,MAAM,MAAM,GAAuB,EAAE,CAAC;IACtC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,MAAM,GAAG,MAAM,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC;QACnD,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,mBAAmB,CAAC,GAAG,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QACtF,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,OAAO,IAAI,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC;IAC7F,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,KAAK,UAAU,aAAa,CAC1B,GAAqB;AACrB,8DAA8D;AAC9D,IAAuC;IAEvC,IAAI,CAAC;QACH,IAAI,QAAQ,GAAG,IAAI,CAAC;QACpB,IAAI,OAAO,QAAQ,KAAK,QAAQ,EAAE,CAAC;YACjC,QAAQ,GAAG,MAAM,GAAG,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC;QAChD,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,CAAC,CAAC;QAChE,MAAM,GAAG,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QACxB,IAAI,OAAO,GAAG,KAAK,QAAQ;YAAE,OAAO,GAAG,CAAC;QACxC,MAAM,KAAK,GAAG,MAAM,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC;QAC1C,OAAO,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAChC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,CAAC,CAAC;IACX,CAAC;AACH,CAAC;AAED,4EAA4E;AAC5E,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,GAAqB;IAC3D,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC,aAAa,EAAE,CAAC;IACzC,IAAI,MAAM;QAAE,OAAO,MAAM,CAAC;IAC1B,OAAO,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;AACpE,CAAC"}

package/dist/pipeline/renderPage.d.ts

@@ -0,0 +1,10 @@
+import type { PDFPageProxy } from "pdfjs-dist";
+import type { PageViewport } from "pdfjs-dist";
+export declare const THUMBNAIL_MAX_DIMENSION = 320;
+export interface RenderedPage {
+    image: ImageData;
+    thumbnail: ImageData;
+    viewport: PageViewport;
+}
+export declare function renderPage(page: PDFPageProxy): Promise<RenderedPage>;
+//# sourceMappingURL=renderPage.d.ts.map

package/dist/pipeline/renderPage.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"renderPage.d.ts","sourceRoot":"","sources":["../../src/pipeline/renderPage.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAC/C,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAS/C,eAAO,MAAM,uBAAuB,MAAM,CAAC;AAE3C,MAAM,WAAW,YAAY;IAC3B,KAAK,EAAE,SAAS,CAAC;IACjB,SAAS,EAAE,SAAS,CAAC;IACrB,QAAQ,EAAE,YAAY,CAAC;CACxB;AAED,wBAAsB,UAAU,CAAC,IAAI,EAAE,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC,CA6B1E"}

package/dist/pipeline/renderPage.js

@@ -0,0 +1,36 @@
+/**
+ * Render resolution and thumbnail sizing. `RENDER_DPI` favors legibility for
+ * the primary reading surface (mirrors the intent behind
+ * `AVIFCodec.pageQuality` on the native side, expressed here as resolution
+ * rather than a platform-specific encoder quality knob).
+ */
+const RENDER_DPI = 150;
+export const THUMBNAIL_MAX_DIMENSION = 320;
+export async function renderPage(page) {
+    const scale = RENDER_DPI / 72;
+    const viewport = page.getViewport({ scale });
+    const width = Math.max(1, Math.round(viewport.width));
+    const height = Math.max(1, Math.round(viewport.height));
+    const canvas = new OffscreenCanvas(width, height);
+    const ctx = canvas.getContext("2d");
+    if (!ctx)
+        throw new Error("failed to acquire 2D rendering context");
+    await page.render({
+        canvas: null,
+        canvasContext: ctx,
+        viewport,
+    }).promise;
+    const image = ctx.getImageData(0, 0, width, height);
+    const longestSide = Math.max(width, height);
+    const thumbScale = longestSide > THUMBNAIL_MAX_DIMENSION ? THUMBNAIL_MAX_DIMENSION / longestSide : 1;
+    const thumbWidth = Math.max(1, Math.round(width * thumbScale));
+    const thumbHeight = Math.max(1, Math.round(height * thumbScale));
+    const thumbCanvas = new OffscreenCanvas(thumbWidth, thumbHeight);
+    const thumbCtx = thumbCanvas.getContext("2d");
+    if (!thumbCtx)
+        throw new Error("failed to acquire 2D rendering context");
+    thumbCtx.drawImage(canvas, 0, 0, thumbWidth, thumbHeight);
+    const thumbnail = thumbCtx.getImageData(0, 0, thumbWidth, thumbHeight);
+    return { image, thumbnail, viewport };
+}
+//# sourceMappingURL=renderPage.js.map

package/dist/pipeline/renderPage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"renderPage.js","sourceRoot":"","sources":["../../src/pipeline/renderPage.ts"],"names":[],"mappings":"AAGA;;;;;GAKG;AACH,MAAM,UAAU,GAAG,GAAG,CAAC;AACvB,MAAM,CAAC,MAAM,uBAAuB,GAAG,GAAG,CAAC;AAQ3C,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,IAAkB;IACjD,MAAM,KAAK,GAAG,UAAU,GAAG,EAAE,CAAC;IAC9B,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;IACtD,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC;IAExD,MAAM,MAAM,GAAG,IAAI,eAAe,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IAClD,MAAM,GAAG,GAAG,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;IACpC,IAAI,CAAC,GAAG;QAAE,MAAM,IAAI,KAAK,CAAC,wCAAwC,CAAC,CAAC;IAEpE,MAAM,IAAI,CAAC,MAAM,CAAC;QAChB,MAAM,EAAE,IAAI;QACZ,aAAa,EAAE,GAA0C;QACzD,QAAQ;KACT,CAAC,CAAC,OAAO,CAAC;IACX,MAAM,KAAK,GAAG,GAAG,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC;IAEpD,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IAC5C,MAAM,UAAU,GAAG,WAAW,GAAG,uBAAuB,CAAC,CAAC,CAAC,uBAAuB,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC;IACrG,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,GAAG,UAAU,CAAC,CAAC,CAAC;IAC/D,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,UAAU,CAAC,CAAC,CAAC;IAEjE,MAAM,WAAW,GAAG,IAAI,eAAe,CAAC,UAAU,EAAE,WAAW,CAAC,CAAC;IACjE,MAAM,QAAQ,GAAG,WAAW,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;IAC9C,IAAI,CAAC,QAAQ;QAAE,MAAM,IAAI,KAAK,CAAC,wCAAwC,CAAC,CAAC;IACzE,QAAQ,CAAC,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC,EAAE,UAAU,EAAE,WAAW,CAAC,CAAC;IAC1D,MAAM,SAAS,GAAG,QAAQ,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,UAAU,EAAE,WAAW,CAAC,CAAC;IAEvE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,CAAC;AACxC,CAAC"}

package/dist/pipeline/textExtract.d.ts

@@ -0,0 +1,29 @@
+import type { PDFPageProxy, PageViewport } from "pdfjs-dist";
+export interface PageTextToken {
+    text: string;
+    /** Normalized `[x0, y0, x1, y1]`, 0..1 relative to the page. */
+    bbox: [number, number, number, number];
+    lineId: number;
+    blockId: number;
+}
+export interface PageText {
+    text: string;
+    tokens: PageTextToken[];
+}
+/**
+ * Extracts the PDF's embedded text layer as page text plus a token layout
+ * suitable for the OCR FlatBuffers schema. This is not OCR -- it reuses the
+ * PDF's own text layer positions, which is exactly the "prefer existing text
+ * layer" half of the format's OCR/text rule (see docs/format-spec.md §6,
+ * "Search/OCR quality variance"). Image-only pages (no embedded text) get an
+ * empty token list here; filling that gap with a vendored OCR engine is not
+ * yet implemented.
+ *
+ * Block/line/bbox derivation is intentionally simple (one block per page,
+ * lines split on pdf.js's `hasEOL`, an axis-aligned box approximated from
+ * each item's transform and width) since the schema only requires a valid
+ * byte range, not layout-analysis-grade precision -- true OCR is
+ * responsible for real layout quality.
+ */
+export declare function extractPageText(page: PDFPageProxy, viewport: PageViewport): Promise<PageText>;
+//# sourceMappingURL=textExtract.d.ts.map

package/dist/pipeline/textExtract.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"textExtract.d.ts","sourceRoot":"","sources":["../../src/pipeline/textExtract.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAE7D,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,MAAM,CAAC;IACb,gEAAgE;IAChE,IAAI,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;IACvC,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,QAAQ;IACvB,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,aAAa,EAAE,CAAC;CACzB;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAsB,eAAe,CAAC,IAAI,EAAE,YAAY,EAAE,QAAQ,EAAE,YAAY,GAAG,OAAO,CAAC,QAAQ,CAAC,CAiCnG"}

package/dist/pipeline/textExtract.js

@@ -0,0 +1,48 @@
+/**
+ * Extracts the PDF's embedded text layer as page text plus a token layout
+ * suitable for the OCR FlatBuffers schema. This is not OCR -- it reuses the
+ * PDF's own text layer positions, which is exactly the "prefer existing text
+ * layer" half of the format's OCR/text rule (see docs/format-spec.md §6,
+ * "Search/OCR quality variance"). Image-only pages (no embedded text) get an
+ * empty token list here; filling that gap with a vendored OCR engine is not
+ * yet implemented.
+ *
+ * Block/line/bbox derivation is intentionally simple (one block per page,
+ * lines split on pdf.js's `hasEOL`, an axis-aligned box approximated from
+ * each item's transform and width) since the schema only requires a valid
+ * byte range, not layout-analysis-grade precision -- true OCR is
+ * responsible for real layout quality.
+ */
+export async function extractPageText(page, viewport) {
+    const content = await page.getTextContent();
+    const tokens = [];
+    const textParts = [];
+    let lineId = 0;
+    const blockId = 0;
+    for (const item of content.items) {
+        if (!("str" in item))
+            continue; // skip TextMarkedContent entries
+        textParts.push(item.str);
+        if (item.str.length > 0) {
+            const [a, b, c, d, e, f] = item.transform;
+            const fontHeight = Math.hypot(c, d) || Math.hypot(a, b) || 1;
+            const x0 = e;
+            const y0 = f;
+            const x1 = e + a * item.width;
+            const y1 = f + fontHeight;
+            const [vx0, vy0] = viewport.convertToViewportPoint(x0, y0);
+            const [vx1, vy1] = viewport.convertToViewportPoint(x1, y1);
+            const bbox = [
+                Math.min(vx0, vx1) / viewport.width,
+                Math.min(vy0, vy1) / viewport.height,
+                Math.max(vx0, vx1) / viewport.width,
+                Math.max(vy0, vy1) / viewport.height,
+            ];
+            tokens.push({ text: item.str, bbox, lineId, blockId });
+        }
+        if (item.hasEOL)
+            lineId += 1;
+    }
+    return { text: textParts.join(""), tokens };
+}
+//# sourceMappingURL=textExtract.js.map

package/dist/pipeline/textExtract.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"textExtract.js","sourceRoot":"","sources":["../../src/pipeline/textExtract.ts"],"names":[],"mappings":"AAeA;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,IAAkB,EAAE,QAAsB;IAC9E,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;IAC5C,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,MAAM,OAAO,GAAG,CAAC,CAAC;IAElB,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,EAAE,CAAC;QACjC,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC;YAAE,SAAS,CAAC,iCAAiC;QACjE,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAEzB,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC;YAC1C,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC;YAC7D,MAAM,EAAE,GAAG,CAAC,CAAC;YACb,MAAM,EAAE,GAAG,CAAC,CAAC;YACb,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC;YAC9B,MAAM,EAAE,GAAG,CAAC,GAAG,UAAU,CAAC;YAC1B,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,sBAAsB,CAAC,EAAE,EAAE,EAAE,CAAqB,CAAC;YAC/E,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,sBAAsB,CAAC,EAAE,EAAE,EAAE,CAAqB,CAAC;YAC/E,MAAM,IAAI,GAAqC;gBAC7C,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,KAAK;gBACnC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM;gBACpC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,KAAK;gBACnC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM;aACrC,CAAC;YACF,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC,CAAC;QACzD,CAAC;QAED,IAAI,IAAI,CAAC,MAAM;YAAE,MAAM,IAAI,CAAC,CAAC;IAC/B,CAAC;IAED,OAAO,EAAE,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC;AAC9C,CAAC"}

package/dist/yom/blobLimits.d.ts

@@ -0,0 +1,15 @@
+import type { BlobKind } from "./generated/yomu/yom/blob-kind.js";
+/**
+ * Gap 2 reader-side size caps. Enforces, from a blob's *declared* sizes
+ * (before any bytes are read or decompressed):
+ *
+ *  - `uncompressed_length <= MAX_BLOB_UNCOMPRESSED[kind]`
+ *  - inflation ratio `uncompressed / max(compressed, 1) <= MAX_INFLATION_RATIO`
+ *
+ * so a hostile file can't drive an unbounded allocation. Offset/length
+ * against-the-source bounds are checked by the caller (reader/validator),
+ * which know the exact source extent. These are reader limits, not format
+ * requirements -- see `constants.ts`.
+ */
+export declare function enforceBlobSizePolicy(kind: BlobKind, compressedLength: number, uncompressedLength: number): void;
+//# sourceMappingURL=blobLimits.d.ts.map

package/dist/yom/blobLimits.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"blobLimits.d.ts","sourceRoot":"","sources":["../../src/yom/blobLimits.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,mCAAmC,CAAC;AAIlE;;;;;;;;;;;GAWG;AACH,wBAAgB,qBAAqB,CAAC,IAAI,EAAE,QAAQ,EAAE,gBAAgB,EAAE,MAAM,EAAE,kBAAkB,EAAE,MAAM,GAAG,IAAI,CAahH"}

package/dist/yom/blobLimits.js

@@ -0,0 +1,25 @@
+import { MAX_INFLATION_RATIO, maxBlobUncompressed } from "./constants.js";
+import { BlobTooLargeError } from "./errors.js";
+/**
+ * Gap 2 reader-side size caps. Enforces, from a blob's *declared* sizes
+ * (before any bytes are read or decompressed):
+ *
+ *  - `uncompressed_length <= MAX_BLOB_UNCOMPRESSED[kind]`
+ *  - inflation ratio `uncompressed / max(compressed, 1) <= MAX_INFLATION_RATIO`
+ *
+ * so a hostile file can't drive an unbounded allocation. Offset/length
+ * against-the-source bounds are checked by the caller (reader/validator),
+ * which know the exact source extent. These are reader limits, not format
+ * requirements -- see `constants.ts`.
+ */
+export function enforceBlobSizePolicy(kind, compressedLength, uncompressedLength) {
+    const cap = maxBlobUncompressed(kind);
+    if (uncompressedLength > cap) {
+        throw new BlobTooLargeError(`blob kind ${kind} declares uncompressed length ${uncompressedLength} > cap ${cap}`);
+    }
+    const ratio = uncompressedLength / Math.max(compressedLength, 1);
+    if (ratio > MAX_INFLATION_RATIO) {
+        throw new BlobTooLargeError(`blob kind ${kind} inflation ratio ${ratio.toFixed(1)} > cap ${MAX_INFLATION_RATIO}`);
+    }
+}
+//# sourceMappingURL=blobLimits.js.map

package/dist/yom/blobLimits.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"blobLimits.js","sourceRoot":"","sources":["../../src/yom/blobLimits.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,MAAM,gBAAgB,CAAC;AAC1E,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAEhD;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,qBAAqB,CAAC,IAAc,EAAE,gBAAwB,EAAE,kBAA0B;IACxG,MAAM,GAAG,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;IACtC,IAAI,kBAAkB,GAAG,GAAG,EAAE,CAAC;QAC7B,MAAM,IAAI,iBAAiB,CACzB,aAAa,IAAI,iCAAiC,kBAAkB,UAAU,GAAG,EAAE,CACpF,CAAC;IACJ,CAAC;IACD,MAAM,KAAK,GAAG,kBAAkB,GAAG,IAAI,CAAC,GAAG,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC;IACjE,IAAI,KAAK,GAAG,mBAAmB,EAAE,CAAC;QAChC,MAAM,IAAI,iBAAiB,CACzB,aAAa,IAAI,oBAAoB,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,UAAU,mBAAmB,EAAE,CACrF,CAAC;IACJ,CAAC;AACH,CAAC"}