Google Cloud Vision OCR JSON Response Converter

gcvocr.sh

# this script is updated in order to support for multi-threaded applications. 
# $3 is a random md5 hash or a uuid.uuid4() object that enables gcvocr.sh to be used without having two overlapping files named temp.json
#!/bin/bash
#cd ~

echo '{"requests":[{"image":{"content":"' > ./$3.json
openssl base64 -in $1 | cat >> ./$3.json
echo '"},
"features":{"type":"TEXT_DETECTION","maxResults":2048}}]}' >> ./$3.json
curl -k -s -H "Content-Type: application/json" https://vision.googleapis.com/v1/images:annotate?key=$2 --data-binary @./$3.json > "$1.json"
rm ./$3.json

import uuid
import os
random_number = uuid.uuid4()
IMAGE_PATH = "path/to/image"
api_key = ""
os.system('./gcvocr.sh {IMAGE_PATH} "{api_key}" {random_number}')

Single Threaded Mode

In single threaded mode as shown above we can go ahead and convert the file with little to no problem.

Multi-Threaded Mode

In order to go ahead and run the script in the Multi-Threaded Mode we can go ahead and do the following bash command to generate a hash that is random in order to go ahead and run the script from the terminal.

./gcvocr.sh ./program_work/27b6db7ac84649e57ed059d0c5b8ebd7/final_images/1.jpeg "xxxxxxxxxxxxxxxxxxxxx" "$RANDOM"