script.py

# adds collection names in metadatas jsons.

import csv
import os
import json
from Tkinter import Tk
import tkFileDialog
import io

collectionsFileName = "allCollectionNames.csv"
jsonFolder = ""

def run():
	collectionIds = []
	collectionNames = []
	missingCollections = []

	Tk().withdraw() # Remove TKinter default window
	# jsonFolder = tkFileDialog.askdirectory()+"/" # select jsons folder
	jsonFolder = "/Users/bastienGirschig/Downloads/jsonMini 2/"

	print "updating "+jsonFolder+"..."

	# read the collectionNames file, fill collectionIds/collectionNames array with it
	with open(collectionsFileName, "r") as collectionsFile:
		csvReader = csv.reader(collectionsFile)
		for row in csvReader:
			collectionIds.append(row[0])
			collectionNames.append(row[1])

	# loop through json files, and add collectionName item to them
	for fileName in os.listdir(jsonFolder):
		if os.path.splitext(fileName)[1] == ".json":
			modified = False
			with open(jsonFolder+fileName) as jsonFile:
				jsonData = json.load(jsonFile)
				byteify(jsonData)
				if("collectionId" in jsonData.keys()):
					if jsonData["collectionId"] in collectionIds:
						collectionIndex = collectionIds.index(jsonData["collectionId"])
						jsonData["collection"] = {'Id':jsonData["collectionId"], 'name':collectionNames[collectionIndex]}
						del jsonData["collectionId"]
						modified = True
					else:
						if(jsonData["collectionId"] not in missingCollections):
							missingCollections.append(jsonData["collectionId"])
				else:
					print("json file "+fileName+" does not contain a 'collectionId' key")
			
			if(modified):
				modifiedFile = json.dumps(jsonData, ensure_ascii=False, encoding='utf8')
				with io.open(jsonFolder+fileName, 'w', encoding='utf-8') as jsonFile:
					print modifiedFile
					jsonFile.write(modifiedFile)
					print(fileName+" updated")

	if(len(missingCollections)>0):
		print ("\n=====================\nmissing collections:\n")
		for collection in missingCollections: print("\t"+collection)

	print("done")

def byteify(input):
    if isinstance(input, dict):
    	return dict([(byteify(key), byteify(value)) for key, value in input.iteritems()])
    elif isinstance(input, list):
        return [byteify(element) for element in input]
    elif isinstance(input, unicode):
        return input.encode('utf-8')
    else:
        return input

run()