feat: Add PDF screenshot generation and display (#995)

* Updated pdf2json to 3.1.5

* Extract and store a screenshot from PDF files using pdf2pic

* Installing graphicsmagick and ghostscript

* Generate Missing PDF screenshot with tidyAssets worker for backward support

* Display PDF screenshot instead of the PDF in web if it exists.

* Display PDF screenshot in mobile app if exists.

* Updated pnpm-lock.yaml

* Removed console.log

* Revert the unnecessary changes in package.json

* Revert pnpm-lock changes

* Prevent rendering PDF files if the screenshot is not generated

* refactor: replace useEffect with useMemo for section initialization

* feat: show PDF file download button and handle large PDFs by defaulting to screenshot view

* feat: add file size to openapi spec

* feature: Add Assets preprocessing in fix mode to admin actions

* i18n: add reprocess_assets_fix_mode translation

* i18n: Add missing ar translations

* A bunch of fixes

* Fix openspec schema

---------

Co-authored-by: Mohamed Bassem <me@mbassem.com>

3 files changed, 43 insertions, 0 deletions

diff --git a/packages/shared/assetdb.ts b/packages/shared/assetdb.ts
index 89738fcf..974f7893 100644
--- a/packages/shared/assetdb.ts
+++ b/packages/shared/assetdb.ts
@@ -4,6 +4,7 @@ import { Glob } from "glob";
 import { z } from "zod";
 
 import serverConfig from "./config";
+import logger from "./logger";
 
 const ROOT_PATH = path.join(serverConfig.dataDir, "assets");
 
@@ -241,3 +242,35 @@ export async function* getAllAssets() {
     };
   }
 }
+
+export async function storeScreenshot(
+  screenshot: Buffer | undefined,
+  userId: string,
+  jobId: string,
+) {
+  if (!serverConfig.crawler.storeScreenshot) {
+    logger.info(
+      `[Crawler][${jobId}] Skipping storing the screenshot as per the config.`,
+    );
+    return null;
+  }
+  if (!screenshot) {
+    logger.info(
+      `[Crawler][${jobId}] Skipping storing the screenshot as it's empty.`,
+    );
+    return null;
+  }
+  const assetId = newAssetId();
+  const contentType = "image/png";
+  const fileName = "screenshot.png";
+  await saveAsset({
+    userId,
+    assetId,
+    metadata: { contentType, fileName },
+    asset: screenshot,
+  });
+  logger.info(
+    `[Crawler][${jobId}] Stored the screenshot as assetId: ${assetId}`,
+  );
+  return { assetId, contentType, fileName, size: screenshot.byteLength };
+}
diff --git a/packages/shared/queues.ts b/packages/shared/queues.ts
index cbe58f8d..5484ffb2 100644
--- a/packages/shared/queues.ts
+++ b/packages/shared/queues.ts
@@ -98,6 +98,13 @@ export async function triggerSearchDeletion(bookmarkId: string) {
   });
 }
 
+export async function triggerReprocessingFixMode(bookmarkId: string) {
+  await AssetPreprocessingQueue.enqueue({
+    bookmarkId,
+    fixMode: true,
+  });
+}
+
 export const zvideoRequestSchema = z.object({
   bookmarkId: z.string(),
   url: z.string(),
@@ -143,6 +150,7 @@ export const FeedQueue = new SqliteQueue<ZFeedRequestSchema>(
 // Preprocess Assets
 export const zAssetPreprocessingRequestSchema = z.object({
   bookmarkId: z.string(),
+  fixMode: z.boolean().optional().default(false),
 });
 export type AssetPreprocessingRequest = z.infer<
   typeof zAssetPreprocessingRequestSchema
diff --git a/packages/shared/types/bookmarks.ts b/packages/shared/types/bookmarks.ts
index b6a74474..9644095c 100644
--- a/packages/shared/types/bookmarks.ts
+++ b/packages/shared/types/bookmarks.ts
@@ -17,6 +17,7 @@ export type ZSortOrder = z.infer<typeof zSortOrder>;
 
 export const zAssetTypesSchema = z.enum([
   "screenshot",
+  "assetScreenshot",
   "bannerImage",
   "fullPageArchive",
   "video",
@@ -61,6 +62,7 @@ export const zBookmarkedAssetSchema = z.object({
   assetId: z.string(),
   fileName: z.string().nullish(),
   sourceUrl: z.string().nullish(),
+  size: z.number().nullish(),
 });
 export type ZBookmarkedAsset = z.infer<typeof zBookmarkedAssetSchema>;