Skip to content

Add OCR data from PDF and image files as a comment in Joplin, to enable full-text search

License

Notifications You must be signed in to change notification settings

salabur/ocr-joplin-notes

 
 

Repository files navigation

OCR Joplin notes

This script will try to add OCR data, and possibly a preview image, to notes in Joplin. It will only work on notes with a tag, which must be specified on startup.

Use cases

1. As a Joplin user, I want to be able to search for any text, whether it is in the note itself, or in an attachment.

When making the switch from Evernote to Joplin, there was one thing missing for me. Evernote uses OCR on attachments, which makes it possible to do a full text search. This is something that is lacking in Joplin. The excellent rest_uploader has the ability to upload files to Joplin and add OCR data as an HTML comment in the note. For existing Joplin notes, either created manually, by the hotfolder plugin or via the Evernote import, the OCR data will not be present.

2. As a Joplin user, I want to have a preview image in the notes I have imported from Evernote.

Just as notes created by the rest_uploader, I would like to have a nice preview of my PDF documents I've imported from Evernote

What this script won't do

  • It will not run in the background and automatically update your notes. I am considering an option to have a background process check notes with a specific tag to be processed and maybe remove the tag when done.
  • It will not update the OCR data once it has been added. Any changes in the note will be ignored. Workaround is to manually remove the OCR section from the note and invoke the script again.

WARNING This script has the potential to mess up all your Joplin notes. But you are not worried, since you make regular backups of your Joplin notes already. Right??

Installation

For a quick installation, there is a docker version available. See the examples section of this readme.

If you don't like docker, there is the manual installation option. If you already installed rest_uploader you probably almost setup already. All that is left is to install this script and setup the environment variables.

pip install ocr-joplin-notes

Requirements:

  • Joplin's Webclipper must be enabled and your desktop client must be running.
  • Required environment variable JOPLIN_TOKEN. You can find your token in the webclipper settings.
  • Environment variable TESSDATA_PREFIX On Ubuntu, mine is set to /usr/share/tesseract-ocr/4.00/tessdata

Modes

Since this script will update your notes, several modes have been added, so the user of this script can verify if the detection in this script works as should be expected.

Mode: TAG_NOTES

Tags all notes in Joplin based on their possible source and markup type. Apart from adding a tag to every note in Joplin, it does not update any notes. The tags it adds will not be used by this script itself and can be removed. It can however be a fast way to tag all your notes. You'll see the need for tags in the other modes. The format of these tags is: ojn_<markup|html>_. Example:

  • ojn_markup_joplin_desktop
  • ojn_html_evernote.mail.smtp
  • ojn_html_evernote.mobile.android
  • ojn_html_evernote.web.clip

Parameters:

  • [optional] --tag When supplied, it will only process notes which have this tag. Default it will process all notes in Joplin.

Mode: DRY_RUN

Report what will happen if this script would process the notes with a selected tag. It will report back via the output on the screen, as well as add tags to your notes. Every note will be tagged with one of the following tags:

  • ojn_ocr_added
  • ojn_ocr_failed
  • ojn_ocr_skipped

Parameters:

  • [required] --tag="my_tag" Only notes having the specified tag will be scanned.
  • [optional] --exclude_tags="other_tag" Notes having the specified tag will be ignored. This Parameter can be set multiple times to exclude multiple tags.
  • [optional] --add-preview=on|off When on, adds a preview image for every PDF found in an HTML note. Default is on Markdown notes already have a PDF preview in the client.
  • [optional] --autorotation=on|off When on tries to fix any skewed images. Default is on
  • [optional] --language=<3 letter code>. The language to use for the OCR processing. Default in eng

Note OCR is quite a CPU intensive process, and it might take some time for large quantities of files to get processed.

Mode: FULL_RUN

WARNING: This mode will make changes to your notes. Remember those backups I mentioned before.

The FULL_RUN mode will do the same as the DRY_RUN mode, but this time, it will make the changes to your Joplin notes. This is mode you are looking for.

Command line examples

python3 -m ocr_joplin_notes.cli --mode=TAG_NOTES
python3 -m ocr_joplin_notes.cli --mode=DRY_RUN --tag=my_notes_test --language=nld --add-previews=off
python3 -m ocr_joplin_notes.cli --mode=FULL_RUN --tag=my_notes_for_testing --language=get
python3 -m ocr_joplin_notes.cli --mode=FULL_RUN --tag=special_notes --exclude_tags=technical --exclude_tags=art 

Docker example

There is a docker image available, for those who do not want to install all the Python dependencies.

docker run  --env-file ./docker-env  --network="host" plamola/ocr-joplin-notes:0.3.11 python -m ocr_joplin_notes.cli --mode=TAG_NOTES

For this to work, you need to save you Joplin token saved in a file. In the example, the file is called docker-env. The contents of this file will look something like:

JOPLIN_TOKEN=f11db775b76e0f80ab39a932s3f79298d080d

Note: the --network="host" parameter, to allow for access to localhost. This only seems to work on Linux systems.

For Windows and Mac users, the workaround is to add the JOPLIN_SERVER environment variable in the docker-env file:

JOPLIN_SERVER=http://host.docker.internal:41184
# It also works with JOPLIN_SERVER=http://gateway.docker.internal:41184 as the docs indicate

Optional Max image pixels environment variable The OCR process for PDFs will experience errors and stop if your PDFs have large images contained within the file. There is a tunable parameter that allows you to process larger files without errors in exchange for higher memory usage. You can tune this parameter by setting the environment variable MAX_IMAGE_PIXELS in your docker-env file. The default max is 178956970. (this default size can cause errors with PDFs as small as 3MB, so if you experience these errors, you can experiment with increasing the value as seen in the example below)

MAX_IMAGE_PIXELS=400000000

Still to do

  • Daemonize this module Having this script run continuously in the background, monitoring for notes with a predefined tag, which need to receive an OCR treatment
  • Auto tagging Just like the rest-uploader, have the ability to apply tags to the notes based on the OCR data

Closing remarks

  • This is the first Python code I've ever written. It might be crap. It works for me.
  • I don't know if I have the time and energy to keep making updates on this script over time. I might, but it might become abandon-ware someday. That's why they've invented forking.
  • This is an open source project. You can use it, free of charge. Your right to free support and free upgrades does not exist. You might get it, you're not entitled to it.
  • It shouldn't happen, but if this script destroys all your data, you still have your those backups, I told you about. Twice.

About

Add OCR data from PDF and image files as a comment in Joplin, to enable full-text search

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.6%
  • Other 0.4%