Layout Blocks

1"use client";
2 
3import * as React from "react";
4 
5import { cn } from "@/lib/utils";
6import {
7  FileViewer,
8  FileViewerContent,
9  FileViewerProvider,
10  FileViewerInset,
11  FileViewerSidebar,
12  FileViewerSidebarTrigger,
13  FileViewerViewport,
14} from "@/components/ui/file-viewer";
15import {
16  PdfViewerPages,
17  PdfViewerProvider,
18  type PdfDocumentSource,
19  type PdfViewerHandle,
20} from "@/components/ui/pdf-viewer";
21import {
22  SegmentedDocumentProvider,
23  useSegmentedDocumentViewport,
24} from "@/components/ui/segmented-document-provider";
25import { useSegmentedItemLink } from "@/components/ui/segmented-item-link";
26import {
27  ViewerHeader,
28  ViewerRoot,
29  ViewerSidebarTrigger,
30} from "@/components/ui/viewer";
31 
32import {
33  azureToLayoutDocument,
34  type AzureDocument,
35} from "./layout-blocks-azure";
36import {
37  documentAiPageImages,
38  documentAiToLayoutDocument,
39  type DocumentAiDocument,
40} from "./layout-blocks-document-ai";
41import { createLayoutBlocksViewerModel } from "./layout-blocks-model";
42import { LayoutBlocksPanel } from "./layout-blocks-panel";
43import { layoutDocumentToPdfBlob } from "./layout-blocks-pdf";
44import { createOcrSegmentedDocumentModel } from "./layout-blocks-segmented-document-model";
45import {
46  textractToLayoutDocument,
47  type TextractDocument,
48} from "./layout-blocks-textract";
49import type { LayoutDocument, LayoutLevel } from "./layout-blocks-types";
50import { LayoutOverlayLayer } from "./layout-overlay-layer";
51import { useKeyedMountEffect } from "@/hooks/use-keyed-mount-effect";
52import { joinEffectKey } from "@/lib/effect-key";
53 
54const LOW_CONFIDENCE_THRESHOLD = 0.9;
55const LEVEL_ORDER: LayoutLevel[] = ["block", "paragraph", "line", "word"];
56const EMPTY_PAGE_IMAGES: ReadonlyMap<number, string> = new Map();
57 
58/**
59 * Discriminated source for the OCR viewer. Each provider's raw output is
60 * normalized to a shared {@link LayoutDocument} before rendering, so the viewer
61 * itself is provider-agnostic.
62 */
63export type OcrSource =
64  | {
65      provider: "google-document-ai";
66      output: DocumentAiDocument;
67      pageImages?: ReadonlyMap<number, string>;
68    }
69  | {
70      provider: "aws-textract";
71      output: TextractDocument;
72      pageImages?: ReadonlyMap<number, string>;
73    }
74  | {
75      provider: "azure-document-intelligence";
76      output: AzureDocument;
77      pageImages?: ReadonlyMap<number, string>;
78    };
79 
80export function OcrLayoutBlocks({
81  className,
82  heightClassName = "h-[680px]",
83  source,
84}: {
85  className?: string;
86  heightClassName?: string;
87// … truncated