I generated a number of test files for the basic drawing system. The tarball is available on my webserver and it also includes the PNG export of all pages generated by the Android app.

The ShapeDatabase.db holds the list of all notes (and a list of their pages) as well as virtual directories and stuff. I deleted everything except the test files.

The next step is to open the SQLite database corresponding to the UID of the note. This then holds the data for all the shapes (and images and text) on each of the pages of that note. Each shape also comes with a transformation matrix (for resize and rotation i guess) plus a lot of flags (if the shape is visible and whatnot).

The binary "point cloud" file then holds the points lists for each drawn shape.

My current code for finding the proper UIDs is the following proof-of-concept mess. I started implementing to do the actual drawing, but got stuck while reverse-engineering the math:

#!/usr/bin/env perl

use v5.38;
use strict;
use warnings;

use DBI;
use DBD::SQLite;
use Data::Dumper;
use JSON::XS;

use Carp;

my $basedir = 'Testfiles';

if(!defined($ARGV[0])) {
    croak("No note name given!");
}

#my ($WIDTH, $HEIGHT) = (XXXXXXX

my $notename = $ARGV[0];

my ($noteuid, $pages) = getUID($basedir, $notename);

print "Note has UID ", $noteuid, " with ", scalar @{$pages}, " pages: 
+", join(', ', @{$pages}), "\n";

my $pagecount = 0;
foreach my $page (@{$pages}) {
    my $pagefname = '' . $notename . '_page' . $pagecount . '.png';
    $pagefname =~ s/\ /_/g;
    renderPage($basedir, $noteuid, $page, $pagefname);
}


sub renderPage($basedir, $noteuid, $page, $pagefname) {
    print "******* $pagefname *******\n";

    my $fname = $basedir . '/' . $noteuid . '.db';
    if(!-f $fname) {
        croak("File $fname not found!");
    }

    my $dbh = DBI->connect("dbi:SQLite:dbname=$fname","","") or croak(
+$!);

    my $selsth = $dbh->prepare("SELECT * FROM NewShapeModel WHERE docu
+mentUniqueId = ? AND pageUniqueId = ? ORDER BY id")
            or croak($dbh->errstr);

    if(!$selsth->execute($noteuid, $page)) {
        print $dbh->errstr, "\n";
        $dbh->disconnect;
        croak("DB Error");
    }

    my @lines;
    while((my $line = $selsth->fetchrow_hashref)) {
        # de-JSON certain fields
        foreach my $key (qw[boundingRect matrixValues shapeLineStyle s
+hapeCreateArgs]) {
            if(!defined($line->{$key}) || $line->{$key} eq '') {
                $line->{$key} = {};
            } else {
                my $tmp = decode_json $line->{$key};
                $line->{$key} = $tmp;
            }

            push @lines, $line;
        }
    }

    #$img = GD::Image->new($self->{width}, $self->{height});
    #$imgblack = $self->{img}->colorAllocate(0, 0, 0);
    #$imgwhite = $self->{img}->colorAllocate(255, 255, 255);




    return;
}


sub getUID($basedir, $name) {
    my $fname = $basedir . '/ShapeDatabase.db';

    if(!-f $fname) {
        croak($fname . " not found");
    }

    my $dbh = DBI->connect("dbi:SQLite:dbname=$fname","","") or croak(
+$!);

    my $selsth = $dbh->prepare("SELECT * FROM NoteModel WHERE title = 
+?")
            or croak($dbh->errstr);

    if(!$selsth->execute($name)) {
        print $dbh->errstr, "\n";
        $dbh->disconnect;
        croak("DB Error");
    }

    my $line = $selsth->fetchrow_hashref;
    $selsth->finish;
    $dbh->disconnect;

    #print Dumper($line);

    if(defined($line) && defined($line->{uniqueId})) {
        my $pages = [];
        if(defined($line->{pageNameList})) {
            my $json = $line->{pageNameList};
            #print "JSON: $json\n";
            my $decoded = decode_json $json;
            #print "Decoded: ", Dumper($decoded);
            if(!defined($decoded->{pageNameList})) {
                print "BLI\n";
            } else {
                print "BLA\n";
                $pages = $decoded->{pageNameList};
            }
        }
        print Dumper($pages);


        return ($line->{uniqueId}, $pages);
    }

    croak("Note not found in ". $fname);
}
[download]

Some observations/best guesses about the points cloud format:

• The last 4 bytes of the file point to the "table of contents", e.g. the list of all lines/shapes.
• The TOC consists of the list of lines/shapes. Each record is the UID, the start offset and the length (in bytes) for the points list.
• The points list consists of 4 flag/filler/color bytes, followed by a number of 16 bytes records holding the points coordinates that go into the matrix calculation. Some of those bytes probably also hold the pen pressure for that line segment.