[Instagram Web Retriever] Instagram web content retriever to json object #instagram #web #retriever #bash #script #curl #json

instagram-web-retriever.sh

#!/bin/bash
CUSTOMER_NAME=dove
URL_INSTA="https://www.instagram.com/"
PIC_SIZE="640x640"

# Fill custerlist.txt with customer ID (one by line)

# create empty final json file

while read CUSTOMER_NAME
do
#	sleep 60	
	mkdir -p epresspack-instagram-feed/${CUSTOMER_NAME}
	echo -e "[" > epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	echo -e "\n" >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json 
	
	curl -L -X GET ${URL_INSTA}${CUSTOMER_NAME} > ${CUSTOMER_NAME}-list
	cat ${CUSTOMER_NAME}-list| grep -o -P '(?<=sharedData = ).*(?=</script>)' > ${CUSTOMER_NAME}-object.json
	cat ${CUSTOMER_NAME}-object.json | jq .entry_data.ProfilePage[].graphql.user[] > ${CUSTOMER_NAME}-object2.json
	cat ${CUSTOMER_NAME}-object2.json | jq .edges[].node.id | tr -d '"'> ${CUSTOMER_NAME}-postsid.txt
	echo "$(sort -u ${CUSTOMER_NAME}-postsid.txt)" > ${CUSTOMER_NAME}-postsid.txt
	
	while read ID
	do
		echo $ID
		cat ${CUSTOMER_NAME}-object.json | jq --arg ID "$ID" '.entry_data.ProfilePage[].graphql.user[]' > ${CUSTOMER_NAME}-object2.json
		cat ${CUSTOMER_NAME}-object2.json | jq --arg ID "$ID" '.edges[]|select(.node.id == $ID)'|sed 's/^{//' > ${CUSTOMER_NAME}-object-results.json
		sed -i '1 i\{' ${CUSTOMER_NAME}-object-results.json
		SHORTCODE=$(cat ${CUSTOMER_NAME}-object-results.json|jq .node.shortcode)
#		PIC_LINK=$(cat ${CUSTOMER_NAME}-object-results.json|jq -c '.node.thumbnail_resources[]|select(.config_width == 640)|select(.config_height == 640)|.src')
		PIC_LINK=$(cat ${CUSTOMER_NAME}-object-results.json|jq -c '.node.thumbnail_src')
		TIMESTAMP=$(cat ${CUSTOMER_NAME}-object-results.json|jq -c '.node.taken_at_timestamp')
		# CLEAN VARS
		SHORTCODE=$(echo $SHORTCODE|tr -d '"')
		echo -e "  {" >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
		echo -e "   \"instaLink\": \"${URL_INSTA}/p/${SHORTCODE}\"," >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
		echo -e "   \"instaPic\": $PIC_LINK," >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
		echo -e "   \"instaId\": \"$ID\"," >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
		echo -e "   \"timestamp\": \"$TIMESTAMP\"" >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
		echo -e "  }," >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	done <${CUSTOMER_NAME}-postsid.txt

	sed -i '$d' epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	echo -e "  }" >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	echo -e "]" >> epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	mv epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}-temp.json
	cat epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}-temp.json | jq '.|=sort_by(.timestamp) | reverse' > epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}.json
	rm -rf epresspack-instagram-feed/${CUSTOMER_NAME}/${CUSTOMER_NAME}-temp.json
	rm -rf ${CUSTOMER_NAME}-list ${CUSTOMER_NAME}-object.json ${CUSTOMER_NAME}-object-results.json ${CUSTOMER_NAME}-postsid.txt ${CUSTOMER_NAME}-object2.json

done < customerlist.txt