feat: add pdf image renderer

This commit is contained in:
David Nguyen
2026-01-27 14:39:16 +11:00
parent 7d38e18f93
commit 4fb3c2cb0f
73 changed files with 2045 additions and 1105 deletions
@@ -1,13 +1,19 @@
import { PDF } from '@libpdf/core';
import { DocumentDataType } from '@prisma/client';
import { base64 } from '@scure/base';
import pMap from 'p-map';
import { match } from 'ts-pattern';
import type { TDocumentDataMeta } from '@documenso/lib/types/document-data';
import { ZDocumentDataMetaSchema } from '@documenso/lib/types/document-data';
import { env } from '@documenso/lib/utils/env';
import { prisma } from '@documenso/prisma';
import { AppError } from '../../errors/app-error';
import { pdfToImages } from '../../server-only/ai/pdf-to-images';
import { createDocumentData } from '../../server-only/document-data/create-document-data';
import { normalizePdf } from '../../server-only/pdf/normalize-pdf';
import { getEnvelopeItemPageImageS3Key } from '../../utils/envelope-images';
import { uploadS3File } from './server-actions';
type File = {
@@ -41,7 +47,67 @@ export const putPdfFileServerSide = async (file: File) => {
const { type, data } = await putFileServerSide(file);
return await createDocumentData({ type, data });
const newDocumentData = await createDocumentData({ type, data });
void extractAndStorePdfImages(arrayBuffer, newDocumentData.id);
return newDocumentData;
};
/**
* Extract and stores page images and metadata to S3.
*/
export const extractAndStorePdfImages = async (
arrayBuffer: ArrayBuffer,
documentDataId: string,
) => {
const images = await pdfToImages(new Uint8Array(arrayBuffer));
const pageMetadata = images.map((image) => ({
originalWidth: image.originalWidth,
originalHeight: image.originalHeight,
scale: image.scale,
scaledWidth: image.scaledWidth,
scaledHeight: image.scaledHeight,
}));
const documentDataMetadata = ZDocumentDataMetaSchema.parse({
pages: pageMetadata,
} satisfies TDocumentDataMeta);
const updatedDocumentData = await prisma.documentData.update({
where: { id: documentDataId },
data: {
metadata: documentDataMetadata,
},
});
if (
env('NEXT_PUBLIC_UPLOAD_TRANSPORT') === 's3' &&
updatedDocumentData.type === DocumentDataType.S3_PATH
) {
await pMap(
images,
async (image) => {
const imageBlob = new Blob([new Uint8Array(image.image)], { type: 'image/jpeg' });
const pageIndex = image.pageIndex;
const s3Key = getEnvelopeItemPageImageS3Key(updatedDocumentData.data, pageIndex);
const imageFile = new File([imageBlob], `${pageIndex}.jpeg`, {
type: 'image/jpeg',
});
const { key } = await uploadS3File(imageFile, s3Key);
return key;
},
{ concurrency: 100 },
);
}
return pageMetadata;
};
/**
@@ -63,10 +129,14 @@ export const putNormalizedPdfFileServerSide = async (
arrayBuffer: async () => Promise.resolve(normalized),
});
return await createDocumentData({
const newDocumentData = await createDocumentData({
type: documentData.type,
data: documentData.data,
});
void extractAndStorePdfImages(normalized, newDocumentData.id);
return newDocumentData;
};
/**
@@ -91,13 +91,13 @@ export const getPresignGetUrl = async (key: string) => {
/**
* Uploads a file to S3.
*/
export const uploadS3File = async (file: File) => {
export const uploadS3File = async (file: File, keyOverride?: string) => {
const client = getS3Client();
// Get the basename and extension for the file
const { name, ext } = path.parse(file.name);
const key = `${alphaid(12)}/${slugify(name)}${ext}`;
const key = keyOverride ?? `${alphaid(12)}/${slugify(name)}${ext}`;
const fileBuffer = await file.arrayBuffer();
@@ -124,6 +124,28 @@ export const deleteS3File = async (key: string) => {
);
};
/**
* Be careful about using this function as we don't allow the
* frontend to ever pull a file from S3 directly.
*/
export const UNSAFE_getS3File = async (key: string) => {
// Additional safeguard to prevent path traversal.
if (key.includes('..') || key.startsWith('/')) {
throw new Error('Invalid S3 key');
}
const client = getS3Client();
const response = await client.send(
new GetObjectCommand({
Bucket: env('NEXT_PRIVATE_UPLOAD_BUCKET'),
Key: key,
}),
);
return response.Body;
};
const getS3Client = () => {
const NEXT_PUBLIC_UPLOAD_TRANSPORT = env('NEXT_PUBLIC_UPLOAD_TRANSPORT');