The initial release of the ham radio QSO elevation profiles Datasette enrichment is up and running!
(In the next few days, I expect there to be several posts on this.)
First, let's talk about the beauty of ChatGPT for personal coding challenges, my personal coding challenges. Over the course of the last week, I've tugged back and forth at the the problem of moving data from the Google Maps Elevation API into a png chart that I thought was going to be provided by the Google Chart API. All of this work taught me a lot, but did not produce useful results.
Yesterday though, I had ChatGPT open seeing what sort of job it could do creating electronic circuit schematics (spoiler: the results aren't pragmatically useful yet, although, they are pretty.) Since I had the app open anyway, I popped over to my ongoing conversation with ChatGPT about my fictional friend Penelope. (You can see her prompt below.) Here's what I jotted down to see if Penelope could solve my charting issues:
How would penelope make a line chart of two dimensional data (distance on the x axis , elevation on the y axis) in python, then create a png file of the chart, then dump out the PNG encoding of the chart so that it could be included in a data:image/png;base64 html tag
ChatGPT came back with an answer a few seconds later that was unlike anything I'd tried so far.
import matplotlib.pyplot as plt # Sample data (distance and elevation) distance = [0, 1, 2, 3, 4, 5] elevation = [100, 120, 110, 130, 125, 140] # Create line chart plt.plot(distance, elevation) plt.xlabel('Distance') plt.ylabel('Elevation') plt.title('Elevation Profile') plt.grid(True) # Save the chart as a PNG file plt.savefig('elevation_chart.png') # Show the chart (optional) # plt.show()
That would get me a chart. But how to get the chart back as png encoded text? ChatGPT wasn't done yet...
import base64 # Read the PNG file as binary with open('elevation_chart.png', 'rb') as file: png_data = file.read() # Encode the PNG image as base64 string encoded_png = base64.b64encode(png_data).decode('utf-8')
and that was that!
It took some tinkering, but I wound up with an enrichment that returned text based png images per row in our Datasette QSO log like the following:
There are definitely scaling issues what could be worked out, but I think the most useful thing would be to simply print a caption on the figure that indicates the angle of the slope over the first two wavelengths, (so in this case, the first 40 meters.)
To wrap up the post, I was curious about the size of png text encoded data vs the actual charts. It turns out that the chart shown above when stored as an image consumes 23kb of disc space. The encoded text when stored in a text file consumes 30kb of disc space, so not a whole lot of overhead. Awesome!
Comments
Post a Comment
Please leave your comments on this topic: