SETUP — 3 STEPS
1. Drag the "PDF Search Viewer" element onto your page.
  900 x 700 px is a good starting size.

2. Fill in "PDF URL" — a Bubble file upload (FileUploader A's value), a database
  field, or any public PDF link.

3. Preview the page. Type a word in the built-in search box and press Enter.

OPTIONS

PDF URL              The PDF to display. Required.
Render quality       1 = fast, 1.4 = balanced (default), 2-3 = sharp but slower.
Enable OCR fallback  ON = pages with no text layer are read with OCR.
                    Turn OFF if you only ever handle digital PDFs and want
                    every search to be instant.
OCR language code    Tesseract language code. eng (default), tur, deu, fra, spa,
                    ita, por, rus, nld, pol and many more. For a document in two
                    languages combine them with a plus sign, e.g. eng+tur.
Highlight color      Colour of a match.
Active highlight     Colour of the match you are currently on.
Show controls        OFF hides the built-in bar if you prefer your own buttons.
Background color     Colour behind the page.

USING YOUR OWN SEARCH BOX

Set "Show controls" to OFF, then in a workflow:
 Element Actions > PDF Search Viewer A > Search        (takes the search term)
 Element Actions > PDF Search Viewer A > Next match
 Element Actions > PDF Search Viewer A > Previous match
 Element Actions > PDF Search Viewer A > Go to page    (takes a page number)
 Element Actions > PDF Search Viewer A > Clear search

To show progress, put a Text element with:
 PDF Search Viewer A's current_match / PDF Search Viewer A's match_count

EVENTS

 "the document is loaded"  — the PDF has finished opening
 "the search finishes"     — a search completed with at least one match
 "no results found"        — a search completed with zero matches
 "the page changes"        — the visible page changed
 "an error occurs"         — read the reason from the last_error state

HOW THE OCR FALLBACK WORKS

For each page the plugin first searches the PDF's text layer. If the term is not
found there and OCR is enabled, that single page is rendered and read with the
Tesseract engine, then searched again. Pages that already contain the term are
never sent through OCR, so digital documents stay fast.

This matters for scanned documents and for forms printed from a web page: those
files often contain a little text (the values someone typed) while the printed
labels are part of a background image. A text-layer-only search finds the values
but misses every label.

NOTES

- The PDF must be reachable by the visitor's browser. Files uploaded to Bubble
 work out of the box. If an external PDF stays blank, that server is most likely
 missing CORS headers — check the last_error state.
- The OCR engine (about 2 MB) is downloaded the first time it is needed, then
 cached by the browser.
- OCR accuracy depends on scan quality. Clean 300 dpi scans read very well;
 low-resolution photographs read less well.
- Raise "Render quality" if highlights look slightly offset on text-layer pages.

DEMO
Live demo:   https://emrejsx.bubbleapps.io/version-test/pdfsearch?debug_mode=true
Demo editor: https://bubble.io/page?id=emrejsx&tab=Design&name=pdfsearch&type=page&elements=bTGOu