Golang, image manipulation, postgresql

This is a teledentistry program. Patients send us images, our software finds the area of interest. You will be writing the program (/usr/local/bin/zoomer) that digitally zooms in on the area of interest in the picture and stores it in an s3 bucket. There will be a configuration file (in the home directory of the user the program runs as) which contains the s3 path to put the images into. This configuration file also has database credentials and a "context percentage" parameter and a "dimming percentage" parameter. You will use viper to read it. There will also be AWS credentials in the user's home directory so that you can use the ordinary AWS golang APIs to access files from s3. The program you will be writing can be invoked in one of three ways: /usr/local/bin/zoomer /usr/local/bin/zoomer -a /usr/local/bin/zoomer -i {image_id} If it is the first version (with no arguments given), then it begins a transaction and runs select image_id from images_received where consensus_created and not cropped_version_created order by random() for update skip locked and iterates through that list of images, stopping after 59 seconds have elapsed. If it is the second version (with -a) then it starts a transaction and runs the same query but doesn't stop after 59 seconds -- it just keeps going until it is finished. If it is the third version, it just operates on that one image id. (But will still need to start a transaction and do a Select ... For Update... ) If that image_id appears in the Consensus_Mirrorless table, then the final image is the same as the source image. If the image_id is not in the Consensus_Mirrorless table, it will be in the Consensus_Mirror_positions table, where the top_left, left_edge, bottom_edge and right_edge columns will tell you the rectangle that you are supposed to be extracting. The Images_Received table has a column s3_path; that's where you can get the original image. The config file has a "cropped_images_bucket" -- you will be creating a file s3://$cropped_images_bucket/$image_id The config file has a "context percentage" parameter. If the context percentage parameter is 0, then you simply crop the original image (top_left, left_edge, bottom_edge, right_edge) and put it into s3://$cropped_images_bucket/$image_id If the context percentage parameter is not zero, then you are going to have a slightly larger image (the width will be Width * (1 + 2 * Context_Percentage) pixels wide and height will be Height * ( 1 + 2 * Context_Percentage) high. Suppose (top_left, left_edge, bottom_edge, right_edge) = (100, 100, 200, 200) and context percentage is 25. Then the image you create will be 150x150; it will basically be the original image from (75, 75) to (225, 225), but any pixels with 25 pixels of the edge will be dimmed. See the attached image for the kind of effect we are trying to get, but keep in mind I just did it by hand so the pixel counts aren't quite right. You dim pixels by the "dimming percentage" from the config file. If the context border would exceed the dimension of the original image, then the context border shrinks down to fit. After the image has been created, it updates the Images_Received table and sets cropped_version_created = true, cropped_version_creation_timestamp = timenow(), cropped_version_s3_path = s3://$cropped_images_bucket/$image_id At the program's completion, it commits the transaction. SQL table samples... create table images_received ( image_id uuid primary key, s3_path varchar not null, -- several columns hidden here because they aren't relevant cropped_version_created boolean default false, cropped_version_creation_timestamp timestamp, cropped_version_s3_path varchar ); create table consensus_mirror_positions ( image_id uuid references images_received, top_edge int, left_edge int, bottom_edge int, right_edge int, width int, height int ); create table consensus_mirrorless ( image_id uuid references images_received );