ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML #437

DiegoPino · 2024-04-04T21:18:45Z

New ML supporting Code for Annotations + JS Triggers
Newest PDF.js via module import/async (might need to pass the actual paths from PHP, but this also works)
Theme level/template level accessibility and margins. Anyone can override the template
New Option (simple/not so simple). Once I have the code the not so simple will enable native PDF.js search by using the pdfViewer class and the bus/dispatch/search controller. UPDATE: Or maybe I use the PDF Search Controller directly without the viewer. That would give me more control.
Mirador 3.3.0 (default) and Mirador 4.0 alpha 2. (The former has CSS issues by the way, z-index)
IIIF Content Search API with VTT and Plain Text (time/space mutation) processing + new settings
JS improvements + small CSS
New Twig template based Ajax/JS processors (data properties) for Views (on load/on click) and for Annotations (miniOCR) saved in Flavors with new Solr based Endpoint in our Annotations REST API

Not a lot since we have so far next/prev. But it does work. Tab selection. Passes the tests.

DiegoPino · 2024-04-04T21:19:45Z

js/pdfs_strawberry.js

-
-              var scale = 1.5;
-              var viewport = page.getViewport({scale: scale});
+      import("https://cdn.jsdelivr.net/npm/[email protected]/build/pdf.min.mjs").then((pdfjsLib) => {


These two here could be settings passed by PHP .... have to think about it

@alliomeria

… pure text to the Content Search API @alliomeria (will explain tomorrow)

@alliomeria

… too Next step is to make 3 different queries based on what processors we have (so far this just does OCR so X/Y and Time) And add JMESPATH expressions to extract VTTs and Plain Text from annotations associated to a canvas. @alliomeria more soon! (we are close :)

+ Plus a whole veggie garden of code to support VTT

@alliomeria

@alliomeria 4.0.0 alpha-2 at least allows an user to pick a subtitle, does not enable one by default though.

…st yet So we can still dispatch an event, but not from our Window Dom, but from the parent/initialized Archipelago DIV (will it work? who knows ... but probably was not working before anyways hahaha)

…ernal items level. So basically a new JMESPATH selector (wish i had better things to do on a weekend)

…rocessors

…ion OR a Canvas as target. None of these work with Mirador sadly (at least so far)

@alliomeria

.. but i have not yet implemented the extra "hits" structure for V1 or "Annotations" for V2 that allow the before and after to be used. @alliomeria will explain during our call

A single file_uuid per hit.

see https://www.w3.org/TR/media-frags/#mf-advanced

@alliomeria

@alliomeria you will have a great presentation. I'm not sure about OR (me) though!

…s on single images Not ready. Lot's to do still. This is 🌰

But i will have to remove tons of code afterwards (means this commit basically does not work because of a Drupal bug that leads to all Drupal.views stored globally being deleted when two ajax views are present in the same page and one uses the Pager of one... damn)

…extual one Also, for contextual ones, use the "exception" value instead of NULL So we can avoid passing "NULL" and end being not validated because of that. $executable->exposed_data = $filters; was the trick ..

…king

So far so good. But the Actual Search needs adapting for non OCR so we return a proper annotation

@alliomeria

@alliomeria need to share the IIIF Manifest that drives this. Very similar to VTT

When multiple canvas, videos,audios, text are all present in the Content Search Results

@alliomeria

…ht take that time @alliomeria i think i got it this time. But i can't say for sure. At least nothing breaks and the canvas swapping does work

DiegoPino · 2024-06-19T15:08:49Z

@alliomeria Merging. Will keep working on the new issue/features in other branches, this one just became too huge.
TODO:

Document the new data-property (twig/html) driven Ajax/JS interactions. (for us too bc I will forget how this all works). I will make a pull request tomorrow AM + OpenAPI one too

First pass on mjs and PDFJS 4.x. Adds also accesibility

3f01771

Not a lot since we have so far next/prev. But it does work. Tab selection. Passes the tests.

DiegoPino self-assigned this Apr 4, 2024

DiegoPino commented Apr 4, 2024

View reviewed changes

DiegoPino added 19 commits April 4, 2024 17:20

Reads better

6967224

IIIF Content Search API + small type on JS (doc)

21999c8

Adds differentiated processor settings for X/Y (OCR), Time based, and…

26a9a7f

… pure text to the Content Search API @alliomeria (will explain tomorrow)

Just a doc type from olde times

f952eb7

What was this? Gosh

3e7adee

Now this is good. New JMESPATH for Subtitles/VTT

6bce0cf

+ Plus a whole veggie garden of code to support VTT

Include Audio in the JMESPATH selector

996c6f0

Update Mirador to 3.3.0 (uff.. who knows right) and allow 4.0.0-alpha

c52c098

@alliomeria 4.0.0 alpha-2 at least allows an user to pick a subtitle, does not enable one by default though.

Fixes race conditon where a Mirador WindowID is set, but does not exi…

19e5da4

…st yet So we can still dispatch an event, but not from our Window Dom, but from the parent/initialized Archipelago DIV (will it work? who knows ... but probably was not working before anyways hahaha)

Damn annotations. Need to exist inside the canvas, but not at the int…

589d585

…ernal items level. So basically a new JMESPATH selector (wish i had better things to do on a weekend)

More updates. Almost there with the results for time fragments

d043074

Adds schema variables for Content Search API differentiated queries/p…

c4dfed9

…rocessors

Adds option so one could return either as target the original Annotat…

ec63f0c

…ion OR a Canvas as target. None of these work with Mirador sadly (at least so far)

Robuster interface

011bc4e

Wrong Schema

c521d6d

Reduce the size of the H4/headings on search results/mirador 4

d7d6434

This kinda adds option for before and after

bcd2833

.. but i have not yet implemented the extra "hits" structure for V1 or "Annotations" for V2 that allow the before and after to be used. @alliomeria will explain during our call

Gosh. Don't accumulate file UUIDs. BC a VTT is a single sequence

6f7b0ce

A single file_uuid per hit.

DiegoPino changed the title ~~ISSUE-436: PDFJS 4.x~~ ISSUE-436: PDFJS 4.x + VTT Content Search API Apr 30, 2024

DiegoPino changed the title ~~ISSUE-436: PDFJS 4.x + VTT Content Search API~~ ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 Apr 30, 2024

DiegoPino added 12 commits May 14, 2024 21:39

don't round miliseconds so much W3C specs allow a .

caf2ce9

see https://www.w3.org/TR/media-frags/#mf-advanced

Haha! Mirador 4 hacked (more like redux-saga-ed) with time fragments

fb97129

@alliomeria you will have a great presentation. I'm not sure about OR (me) though!

Fix to this very complex filter. This one is a clockwork

bc4a293

Adds Webannotation controller for OCR driven annotations + Annotoriou…

3c0da34

…s on single images Not ready. Lot's to do still. This is 🌰

Adds Pako so we can zip/deflate/inflate things via JS

5e7c752

Dispatch image changes ... new little thing for Annotation based Views

50a0195

The controller was not right. Add the proper data. Needs cleanup

0943175

Allow Exposed filters. But avoid if already set same "key" via a Cont…

bc06a4e

…extual one Also, for contextual ones, use the "exception" value instead of NULL So we can avoid passing "NULL" and end being not validated because of that. $executable->exposed_data = $filters; was the trick ..

initialize the $filters

66da54d

Merge branch 'ISSUE-449' into ISSUE-436

ccf22e0

Cool. Overrides big error on Ajax/Views Drupal + annotations full wor…

7ce6d4f

…king

DiegoPino changed the title ~~ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2~~ ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML May 27, 2024

DiegoPino added 5 commits May 29, 2024 09:57

Make sure <a> link around image exists before trying to intercept it

5147a2e

First pass on Plain Text search

7e89605

So far so good. But the Actual Search needs adapting for non OCR so we return a proper annotation

Fully functional Plaintext Content Search API response

0de131a

@alliomeria need to share the IIIF Manifest that drives this. Very similar to VTT

Fixes Canvas not being selected before sending the Annotation

bc5c3a6

When multiple canvas, videos,audios, text are all present in the Content Search Results

I know the delay is large (1.5 seconds) but really a video/update mig…

b46c6a2

…ht take that time @alliomeria i think i got it this time. But i can't say for sure. At least nothing breaks and the canvas swapping does work

DiegoPino added this to the 1.4.0 milestone Jun 19, 2024

DiegoPino merged commit 6cac775 into 1.4.0 Jun 19, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML #437

ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML #437

DiegoPino commented Apr 4, 2024 •

edited

Loading

DiegoPino Apr 4, 2024 •

edited

Loading

DiegoPino commented Jun 19, 2024

ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML #437

ISSUE-436: PDFJS 4.x + VTT Content Search API + Mirador 4.0 alpha 2 + ML #437

Conversation

DiegoPino commented Apr 4, 2024 • edited Loading

DiegoPino Apr 4, 2024 • edited Loading

Choose a reason for hiding this comment

DiegoPino commented Jun 19, 2024

DiegoPino commented Apr 4, 2024 •

edited

Loading

DiegoPino Apr 4, 2024 •

edited

Loading