fix_notes.py

#!/usr/bin/python
import MySQLdb
from PIL import Image, ImageEnhance, ImageStat
import json
import MySQLdb
import numpy as np
import tensorflow as tf
from tensorflow import keras
import PIL
import urllib.request
import regionRoutine
import cv2 as cv
import csv

url = 'http://pad.crc.nd.edu'
dest = './temp.png'
# Training Image Set properties
IMG_SHAPE = (227, 227,3)
HEIGHT_INPUT, WIDTH_INPUT, DEPTH = IMG_SHAPE
NUM_CHANNELS = DEPTH; 
NUM_CLASSES = 4

# set lite model
model_file = '/var/www/html/joomla/neuralnetworks/tf_lite/fhi360_conc_large_lite/1.0/fhi360_conc_large_1_21.tflite'

conc = [100,80,50,20]

# pls
coefficients_file = "pls_coefficients_ios.csv"

# get coeff file
coeff = {}
with open(coefficients_file) as csvcoeffs:
  csvcoeffreader = csv.reader(csvcoeffs)
  i=0
  for row in csvcoeffreader:
    elmts = []
    for j in range(1,len(row)):
      elmts.append(float(row[j]))
    coeff[row[0]] = elmts

#### PLS version of concentration
def get_pls_quantity(file_url, coeff, drug):
    urllib.request.urlretrieve(url + file_url, dest)

    img = cv.imread(dest)
    #print("OK",row[5],row[9])
        # get pixel analysis
    f = {}
    f = regionRoutine.fullRoutine(img, regionRoutine.intFind.findMaxIntensitiesFiltered, f, True, 10)

    # drug?
    # continue if no coefficients
    if drug not in coeff:
        return 0.

    drug_coeff = coeff[drug] #coeff['amoxicillin'] #

    # start with offst
    pls_concentration = drug_coeff[0]

    coeff_index = 1

    for letter in ['A','B','C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L']:
        for region in range(10):
            for color_letter in ['R', 'G', 'B']:
                pixval = f[letter + str(region + 1) + '-' + color_letter]
                pls_concentration += float(pixval) * drug_coeff[coeff_index]
                coeff_index += 1

    return pls_concentration

#### NN version of concentration
def get_nn_quantity(file_url):
    urllib.request.urlretrieve(url + file_url, dest)

    #time.sleep(1)
    #img = cv.imread(dest)
    #print("OK",row[5],row[9])
        # get pixel analysis
    #Load png file using the PIL library
    img = PIL.Image.open(dest)

    #crop out active area
    img = img.crop((71, 359, 71+636, 359+490))

    #resize
    img = img.resize((227,227), PIL.Image.ANTIALIAS)

    #reshape the image as numpy
    im = np.asarray(img).flatten().reshape(1, HEIGHT_INPUT, WIDTH_INPUT, DEPTH).astype(np.float32)
    #print("shape/type:", im.shape, im.dtype)


    # Load the TFLite model and allocate tensors.
    interpreter =  tf.contrib.lite.Interpreter(model_path=model_file)
    interpreter.allocate_tensors()

    # Get input and output tensors.
    input_details = interpreter.get_input_details()
    output_details = interpreter.get_output_details()
    #print("input", input_details[0])

    # Test the model on random input data.
    input_shape = input_details[0]['shape']
    input_data = np.array(np.random.random_sample(input_shape), dtype=np.float32)
    interpreter.set_tensor(input_details[0]['index'], im)

    # predict
    interpreter.invoke()

    # The function `get_tensor()` returns a copy of the tensor data.
    # Use `tensor()` in order to get a pointer to the tensor.
    output_data = interpreter.get_tensor(output_details[0]['index'])

    concentration = conc[np.argmax(output_data[0])]
    confidence = output_data[0][np.argmax(output_data[0])]

    return concentration
    #print("Result:", conc[np.argmax(output_data[0])])

# setup query
QUERY1 = 'SELECT `sample_id`,`notes`,`id`,`processed_file_location` FROM `card` WHERE `category`="FHI2022" AND `notes` LIKE "%Predicted drug = %" AND `notes` NOT LIKE "%Notes version%"'
# AND `notes` LIKE "%(pls%" 

#get database credentials
with open('credentials.txt') as f:
    line = f.readline()
    split = line.split(",")
f.close()

#open database
db = MySQLdb.connect(host="localhost", # your host, usually localhost
                     user=split[0], # your username
                      passwd=split[1], # your password
                      db="pad") # name of the data base

# you must create a Cursor object. It will let
#  you execute all the queries you need
cur = db.cursor()

# get list of samples
cur.execute(QUERY1)

count = 0

doBreak = True

# loop through CSV
# loop
for row in cur.fetchall() :
    #print(row[1])
    count += 1
    notes_split = row[1].split(',')
    pred = notes_split[0][17:].split('(')
    pred_drug = pred[0].strip()
    pred_conf = pred[1][:-1]
    quant = notes_split[1].strip()
    quant_nn = notes_split[2].strip()[:-1]
    quant_pls = notes_split[3][5:-2]
    #print("4", notes_split[4])
    #print("5", notes_split[5])
    if quant_pls != "":
        other = notes_split[5].split('.')
    else:
        other = notes_split[4].split('.')
    safe = other[0].strip()[10:]
    user = (other[1].strip())[6:]
    nn = (other[2].strip())[12:]

    #print(pred_drug, pred_conf, quant, quant_nn, quant_pls, safe, user, nn)
    json_string = {}
    json_string["Predicted drug"] = pred_drug
    json_string["User"] = user
    json_string["App type"] = "Android"
    if quant_nn !="":
        json_string["Quantity NN"] = float(quant_nn)
    else:
        json_string["Quantity NN"] = get_nn_quantity(row[3])
    json_string["Prediction score"] = float(pred_conf)
    if quant_pls != "":
        json_string["Quantity PLS"] = float(quant_pls)
    else:
        json_string["Quantity PLS"] = round(get_pls_quantity(row[3], coeff, pred_drug.lower()), 1)
    json_string["Notes version"] = 0
    json_string["Neural net"] = nn
    json_string["Notes"] = ""

    QUERY1 =  'UPDATE `card` SET `notes` = \'%s\' WHERE `id`=%s' % \
            (json.dumps(json_string), row[2])

    print(QUERY1)

    cur.execute(QUERY1)

    # commit your changes
    db.commit()
    #break

    
print("Count",count)

# Close all cursors
cur.close()
# Close all databases
db.close()