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, 25 insertions, 0 deletions

diff --git a/packages/trpc/lib/attachments.ts b/packages/trpc/lib/attachments.ts
index f4fda9cd..3ad79a5a 100644
--- a/packages/trpc/lib/attachments.ts
+++ b/packages/trpc/lib/attachments.ts
@@ -6,6 +6,7 @@ import { ZAssetType, zAssetTypesSchema } from "@hoarder/shared/types/bookmarks";
 export function mapDBAssetTypeToUserType(assetType: AssetTypes): ZAssetType {
   const map: Record<AssetTypes, z.infer<typeof zAssetTypesSchema>> = {
     [AssetTypes.LINK_SCREENSHOT]: "screenshot",
+    [AssetTypes.ASSET_SCREENSHOT]: "assetScreenshot",
     [AssetTypes.LINK_FULL_PAGE_ARCHIVE]: "fullPageArchive",
     [AssetTypes.LINK_PRECRAWLED_ARCHIVE]: "precrawledArchive",
     [AssetTypes.LINK_BANNER_IMAGE]: "bannerImage",
@@ -21,6 +22,7 @@ export function mapSchemaAssetTypeToDB(
 ): AssetTypes {
   const map: Record<ZAssetType, AssetTypes> = {
     screenshot: AssetTypes.LINK_SCREENSHOT,
+    assetScreenshot: AssetTypes.ASSET_SCREENSHOT,
     fullPageArchive: AssetTypes.LINK_FULL_PAGE_ARCHIVE,
     precrawledArchive: AssetTypes.LINK_PRECRAWLED_ARCHIVE,
     bannerImage: AssetTypes.LINK_BANNER_IMAGE,
@@ -34,6 +36,7 @@ export function mapSchemaAssetTypeToDB(
 export function humanFriendlyNameForAssertType(type: ZAssetType) {
   const map: Record<ZAssetType, string> = {
     screenshot: "Screenshot",
+    assetScreenshot: "Asset Screenshot",
     fullPageArchive: "Full Page Archive",
     precrawledArchive: "Precrawled Archive",
     bannerImage: "Banner Image",
@@ -47,6 +50,7 @@ export function humanFriendlyNameForAssertType(type: ZAssetType) {
 export function isAllowedToAttachAsset(type: ZAssetType) {
   const map: Record<ZAssetType, boolean> = {
     screenshot: true,
+    assetScreenshot: true,
     fullPageArchive: false,
     precrawledArchive: false,
     bannerImage: true,
@@ -60,6 +64,7 @@ export function isAllowedToAttachAsset(type: ZAssetType) {
 export function isAllowedToDetachAsset(type: ZAssetType) {
   const map: Record<ZAssetType, boolean> = {
     screenshot: true,
+    assetScreenshot: true,
     fullPageArchive: true,
     precrawledArchive: false,
     bannerImage: true,
diff --git a/packages/trpc/routers/admin.ts b/packages/trpc/routers/admin.ts
index c7dd7575..6393c950 100644
--- a/packages/trpc/routers/admin.ts
+++ b/packages/trpc/routers/admin.ts
@@ -9,6 +9,7 @@ import {
   OpenAIQueue,
   SearchIndexingQueue,
   TidyAssetsQueue,
+  triggerReprocessingFixMode,
   triggerSearchReindex,
 } from "@hoarder/shared/queues";
 import {
@@ -154,6 +155,15 @@ export const adminAppRouter = router({
 
     await Promise.all(bookmarkIds.map((b) => triggerSearchReindex(b.id)));
   }),
+  reprocessAssetsFixMode: adminProcedure.mutation(async ({ ctx }) => {
+    const bookmarkIds = await ctx.db.query.bookmarkAssets.findMany({
+      columns: {
+        id: true,
+      },
+    });
+
+    await Promise.all(bookmarkIds.map((b) => triggerReprocessingFixMode(b.id)));
+  }),
   reRunInferenceOnAllBookmarks: adminProcedure
     .input(
       z.object({
diff --git a/packages/trpc/routers/bookmarks.ts b/packages/trpc/routers/bookmarks.ts
index 12ec9ccb..6ab863fb 100644
--- a/packages/trpc/routers/bookmarks.ts
+++ b/packages/trpc/routers/bookmarks.ts
@@ -259,6 +259,7 @@ function toZodSchema(bookmark: BookmarkQueryReturnType): ZBookmark {
         assetId: asset.assetId,
         fileName: asset.fileName,
         sourceUrl: asset.sourceUrl,
+        size: assets.find((a) => a.id == asset.assetId)?.size,
       };
       break;
   }
@@ -441,6 +442,7 @@ export const bookmarksAppRouter = router({
         case BookmarkTypes.ASSET: {
           await AssetPreprocessingQueue.enqueue({
             bookmarkId: bookmark.id,
+            fixMode: false,
           });
           break;
         }
@@ -830,6 +832,7 @@ export const bookmarksAppRouter = router({
                   assetType: bookmarkAssets.assetType,
                   fileName: bookmarkAssets.fileName,
                   sourceUrl: bookmarkAssets.sourceUrl ?? null,
+                  size: null, // This will get filled in the asset loop
                 };
                 break;
               }
@@ -881,6 +884,13 @@ export const bookmarksAppRouter = router({
               }
               acc[bookmarkId].content = content;
             }
+            if (acc[bookmarkId].content.type == BookmarkTypes.ASSET) {
+              const content = acc[bookmarkId].content;
+              if (row.assets.id == content.assetId) {
+                // If this is the bookmark's main aset, caputure its size.
+                content.size = row.assets.size;
+              }
+            }
             acc[bookmarkId].assets.push({
               id: row.assets.id,
               assetType: mapDBAssetTypeToUserType(row.assets.assetType),