obsidian-annotator
obsidian-annotator copied to clipboard
Ability to edit text copied directly from source when <annotate> button is pressed
It is often the case that the text that is pasted into the top window of the annotation is garbled owing to less than perfect OCR. E.g. displaying "tenn" instead of "term". There are a number of benefits to being able to edit the annotation text: 1. fix bad OCR, 2. add markdown to create links in keywords in the text (workaround has been to add tags, though I'm not sure that these function in the same way and 3. easier to clean up when an annotation spans more than one page, e.g. the last two lines of one page and the first line of the following page. Currently, the headers etc. are included in the annotation text proper.
Assume that the actual pdf reader gets updated from time to time but this is external to Obsidian.
Many thanks. This plug-in is AWESOME as it has freed my workflow from LiquidText.
.
Neglected to mention, that yes, of course, one can edit the annotation by opening as markdown. However, doing so, one can no longer see the correct version of the text in the pdf which makes it sorta tough to remember what the text should be and edit accordingly.
Hmmm. Decided that I needed to edit a badly-garbled annotation in the md view. However, when I went back to view the actual pdf, all the associated annotations and notes no longer displayed Though they still display in the markdown view. Have zero idea as to why this is though it certainly seems to indicate that despite the difficulty of not being able to see the actual pdf while editing an annotation it is all for naught...
When you make edits to the highlighted text in MD and re-open the annotation view, the edited text is sent to the hypothesis annotator.
This has one positive effect and one negative effect.
Positive: Your edits show up in the annotation, so if you fixed garbled text those fixes will be visible
Negative: Hypothesis partially uses the highlighted text to identify where to place the highlight. If you edit it too much it can cause hypothesis to lose track of the highlight.
I might try to improve this in the future.
Thanks, Elias, I think there may have been a bit of "user error" in causing all the annotations to no longer display. I went back to take a look at the md and saw there now seemed to be two instances of the "same" text, the original and the one that I cleaned up. Not sure how that happened as I only added characters to words that were missing in them. For example, "s" seemed not to have been recognized. That said, I decided to delete my "work" and lo and behold my annotations are back in all their garbled glory. So, per your remark below it seems that perhaps an enhancement could 1. allow user to work directly in annotation pane with pdf visible and 2. be limited to certain edits beyond things like changing "rn" to "m" such as adding brackets and other common markdown.... From your experience, does the OCR work better when it has been performed by Hypothesis? I did that once but stopped as the file was c. 10x larger than the Adobe OCR version. Also, is it the case that the Hypothesis pdf reader is also subject to updates by the Hypothesis devs? Hypothetically? Again, thanks. Best, Richard On Tuesday, March 8, 2022, 01:37:30 PM MST, Elias Sundqvist @.***> wrote:
When you edit the highlighted text in md, the edited text is sent to the hypothesis annotator when you open the annotation view.
This has one positive effect and one negative effect.
Positive: Your edits show up in the annotation, so if you fixed garbled text those fixes will be visible
Negative: Hypothesis partially uses the highlighted text to identify where to place the highlight. If you edit it too much it can cause hypothesis to lose track of the highlight.
I might try to improve this in the future.
— Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.Message ID: @.***>