0.12.1

This is a small hotfix for the 0.12.0 release.

• fix: Fix slice call in the content truncation logic which was resulting in excessive usage of context tokens. Fixes #94 by @MohamedBassem in #4629dac

The bug fixed in this release resulted into mistakingly truncating the first 1500 words in large contents during tag inference (instead of taking only the first 1500 words). This had two side effects for bookmarks with a lot of content (> 1500 words):

1. Due to the truncation of the first 1500 words, you might have got some sub optimal tags for content with >1500 words.
2. For bookmarks with huge content size, you might have got inference failures due to hitting the context limits of the models you're using. And for the ones that were under the context size, they might have taken longer and used more credits than intended.

If this is a problem for you, you can trigger a re-index for the affected bookmarks.

0.12.0

Welcome to the 0.12.0 release of hoarder! This release is special because we have our first full feature contribution by @AhmadMuj. Welcome to the contributors list @AhmadMuj! If you're interested in helping us develop hoarder, you can join the new #development channel on hoarder's discord.

New Features

• Raspberry Pi support is now here!
• You can now upload pdfs and get them indexed and automatically tagged thanks to @AhmadMuj!
• You can now change the layout of the bookmark list. We're starting with three layouts:
  • Masonry which was the layout used before that release.
  • Grid where every card has a fixed height resulting in a strict grid.
  • List which shows all bookmarks under each other in a list (screenshot attached later in the release notes).
• Previously, you were only allowed to upload items from the homepage. Now you can upload directly in a list, tag or even the favourited/archived list.
• Previously, the admin panel only allowed you to recrawl all bookmarks. @AhmadMuj added support for only crawling the failed ones (and showed some stats about them there as well).
• The default OpenAI image model got upgraded to the new gpt-4-turbo which should be more robust and accurate while costing the same.
• You can now add titles to bookmarks (and edit existing ones). Uploaded assets (images and pdfs) now automatically get their filename as a title. A new bookmark action bar got added to the bookmark preview page.

Changes in Behavior

• Previously, archived items were hidden from almost everywhere except the archived list which made it not very different from just deleting the item. Starting from this release, archived items are only hidden from the homepage but will continue appearing in lists, search, etc. This is more aligned with other bookmark managers (e.g. pocket). With this change, you can now use archiving to achieve Inbox Zero on your homepage by archiving the items that no longer require your attention or articles that you've already read.

Fixes

• Crawling robustness:
  • Default crawling timeout increased from 30s to 60s and default navigation timeout increase from 10s to 30s. Both can now be controlled with the env variables CRAWLER_JOB_TIMEOUT_SEC and CRAWLER_NAVIGATE_TIMEOUT_SEC respectively. You might want to increase them further if you're running hoarder on a low-perf machine or have a slow internet connection.
  • Crawling jobs are now retried 5 times before giving up on them.
  • Crawler error messages are now more descriptive to help debug issues.
• You no longer need to refresh the All tags page when tags are changed.
• When a user is deleted, their on disk assets were not correctly deleted. This is now fixed.
• Fixed error when adding notes with a colon in the first word.

Screenshots of the new features

The new action bar in the preview page along with the new support for editing titles.

The new List view and its toggles.

PDF support

Upgrade

• If you're using HOARDER_VERSION=release, run docker compose pull && docker compose up -d.
• If you're pinning it to a specific version, upgrade the version and then run docker compose pull && docker compose up -d.

All Commits

• feature: Add title to bookmarks and allow editing them. Fixes #27 by @MohamedBassem in 81e0b28
• fix(web): Fix some JS console errors in the preview page by @MohamedBassem in 5c9acb1
• feature(web): Allow changing the bookmark grid layout (#98) by @MohamedBassem in 4f17ea6
• fix: Differentiate between pending in db and in redis in admin job stats by @MohamedBassem in cf0df0e
• feature: Recrawl failed links from admin UI (#95) by @AhmadMuj in 95cf8f4
• fix: Increase default navigation timeout to 30s, make it configurable and add retries to crawling jobs by @MohamedBassem in 238c296
• feature: Add PDF support (#88) by @AhmadMuj in #be622e5
• feature(inference): Upgrade the default vision model to the new gpt-4-turbo by @MohamedBassem in 2806701
• fix(crawler): Skip validating URLs in metascrapper as it was already being validated. Fixes #22 by @MohamedBassem in a9242a5
• fix: Delete the API key on logout from phone or extension by @MohamedBassem in f696d33
• fix(web): Fix the toggle theme button allowing clicks from anywhere in the dropdown item by @MohamedBassem in 5ab6c33
• fix: Fix type in the useAddBookmarkToList hook by @MohamedBassem in cae543c
• feature(web): Allow uploading directly into lists/tags. Fixes #69 by @MohamedBassem in fe13408
• refactor(web): Move the tooltip provider to the root of the app by @MohamedBassem in 994691b
• feature: Extract hook logic into separate package and add a new action bar in bookmark preview by @MohamedBassem in 79d61be
• fix(workers): Increase default timeout to 60s, make it configurable and improve logging by @MohamedBassem in 4491831
• fix(docker): Remove BUILDPLATFORM from Dockerfile by @MohamedBassem in e0bb1fc
• fix: Refresh the all tags page automatically when a tag is modified by @MohamedBassem in 044659f
• ui(web): Add a tooltip explaining what archived means by @MohamedBassem in 4cf9908
• fix: Only consider text as URL in editor card if it has an http protocol. Fixes #86 by @MohamedBassem in bc6ee2e
• feature: Change archived meaning to only mean removed from homepage by @MohamedBassem in a81c394
• fix: Delete user's assets when the user is deleted by @MohamedBassem in 91b088d
• build: Increment android app versionCode by @MohamedBassem in 8af95fe

0.11.1

This is a small hotfix release for v0.11.0.

5ef2efe fix: Incorrect counts for tags in the AllTags page. Fixes #78

0.11.0

Welcome to the v0.11.0 release of Hoarder. This packs a bunch of new stuff and fixes.

New Features

• There's now a CLI tool to enable mass manipulations for your bookmarks. You can use it, for example, to mass import bookmarks from your previous tool. Check its docs here, and you can check the initial version of importing bookmarks docs here.
• You can now attach notes to your bookmarks.
• You can now delete tags in the tag view page.
• The All Tags page now differentiates between manually attached tags and AI attached tags.
• You can specify the language of the auto generated tags with INFERENCE_LANG=<lang> env variable. It currently defaults to english. Thanks @COOHU-Kr for the suggestion!
• The web UI now allows uploading multiple images at the same time. Webp support also got added.
• Clicking ⌘ + E or (ctrl + E) will now focus on the new bookmark editor. Thanks @chanunnaki for the idea!
• Current server version is now shown in the admin panel. It'll also check github API to know if there are new version. You can disable the new version check with DISABLE_NEW_RELEASE_CHECK=true.
• In case you missed it, the android app is almost there (under google's mercy). If you can't wait, you can download the APK from this issue

Fixes

• Search results are now sorted by relevance.
• In some rare instances, the crawler was getting stuck parsing some pages preventing any other links from getting crawled. There's now a 30s timeout to solve this problem.
• Fix the sharp is missing error in the web container logs.
• When logging out, people were getting redirected to localhost:3000 by mistake. Add NEXTAUTH_URL=<server_url> to your env file to fix that. The installation docs has been updated.

Upgrade

• [Optional but hightly recommended] The initial docker compose was using an old chrome version (v100). Bump the chrome container version to v123 by changing the chrome container image:
  • from: gcr.io/zenika-hub/alpine-chrome:100
  • to: gcr.io/zenika-hub/alpine-chrome:123
• If you're using HOARDER_VERSION=release, run docker compose pull && docker compose up -d.
• If you're pinning it to a specific version, upgrade the version and then run docker compose pull && docker compose up -d.

Some screenshots from the release

The new All Tags page

Attaching notes to bookmarks

Server version in admin panel

The new CLI

0.10.1

Mostly a lot of minor fixes:

• Fix: Attempt to increase the reliability of the ollama inference by working around some of the errors in the ollama package.
• Fix: There was a limit of 2k chars on notes. This is no longer there.
• Fix: When a bookmark is deleted in the search page, it wasn't getting removed from the grid. This is now fixed.
• feature: In the editor card, you can now save the bookmark by pressing Ctrl + Enter as well as the Cmd + Enter. This should be more intuitive to windows users.
• feature: You can now upload webp images.
• feature: Maximum asset size was 4MB. Now it's configurable with the MAX_ASSET_SIZE_MB env variable. This variable still defaults to 4MB.
• ui: Red destructive color looks slightly better now in dark mode.

0.10.0

Main features:

• Local LLM support via ollama. Check the configuration docs for how to set it up.

Other minor fixes:

• Drop CORS restrictions on the API because it is actually intended to be used cross origins).
• Fix the "Favourites" typo in the sidebar.
• Allow specifying a specific redis database index to use instead of the default 0 one.
• Document the DISABLE_SIGNUPS flag.

0.9.0

It's announcement day 🎉🎉

• The browser now runs in its own container which resulted into a smaller overall image sizes.
• We have a new logo. This release updates the logo everywhere.
• Allow showing demo mode creds in sign in page.
• Minor UI fixes

0.8.0

• web: Add night mode support
• docker: Run chrome in its own container

0.7.0

• A new mobile app got introduced
• Now it's possible to hoard images and get them auto tagged
• The bookmark preview page got revamped
• An inline editor card to replace the nav bar buttons
• Fix flicking in search which makes feel much smoother now. Also reduced the debounce duration for faster reaction time.
• Added missing indicies to the db.
• The bookmark grid is now paginated.
• A new revamped tags editor
• A new landing page at https://hoarder.app
• Add support for removing items from lists.
• We're now publishing both macos and linux docker images

0.6.0

feature: Preview page for the stored content of links
fix: Use puppeteer adblocker to block cookies notices
fix: Teach openai to ignore cookie consent popups
feature: Initial attempt at making the app a PWA
feature: Add an 'All Lists' page
feature: Show a loading indicator when tags are still being fetched
feature: Make the first user an admin automatically