Format a string to a proper JSON object

Question

I have a string (from an API call) that looks something like this:

val=

{input:a,matches:[{in:["w","x","y","z"],output:{num1:0d-2,num2:7.0d-1}},
{in:["w","x"],output:{num1:0d-2,num2:8.0d-1}}]}

I need to do temp=json.loads(val) ; but the problem is that the string is not a valid JSON. The keys and values do not have the quotes around them. I tried explicitly putting the quotes and that worked.

How can I programatically include the quotes for such a string before reading it as a JSON?

Also, how can I replace the numbers scientific notations with decimals? eg. 0d-2 becomes "0" and 8.0d-1 becomes "0.8"?

Answer 1

You could catch anything thats a string with regex and replace it accordingly.

Assuming your strings that need quotes:

• start with a letter
• can have numbers at the end
• never start with numbers
• never have numbers or special characters in between them

This would be a regex code to catch them:

([a-z]*\d*):

You can try it out here . Or learn more about regex here .

Let's do it in python:

import re

# catch a string in json

json_string = '{input:a,matches:[{in:["w","x","y","z"],output:{num1:0d-2,num2:7.0d-1}},
{in:["w","x"],output:{num1:0d-2,num2:8.0d-1}}]}' # note the single quotes!

# search the strings according to our rule
string_search = re.search('([a-z]*\d*):', json_string)

# extract the first capture group; so everything we matched in brackets
# this is to exclude the colon at the end from the found string as
# we don't want to enquote the colons as well
extracted_strings = string_search.group(1)

This is a solution in case you will build a loop later. However if you just want to catch all possible strings in python as a list you can do simply the following instead:

import re

# catch ALL strings in json

json_string = '{input:a,matches:[{in:["w","x","y","z"],output:{num1:0d-2,num2:7.0d-1}},
{in:["w","x"],output:{num1:0d-2,num2:8.0d-1}}]}' # note the single quotes!
extract_all_strings = re.findall(r'([a-z]*\d*):', json_string)
# note that this by default catches only our capture group in brackets
# so no extra step required

This was about basically regex and finding everything.

With these basics you could either use re.sub to replace everything with itself just in quotes, or generate a list of replacements to verify first that everything went right (probably somethign you'd rather want to do with this maybe a little bit unstable approach) like this . Note that this is why I made this kind of comprehensive answer instead of just pointing you to a "re.sub" one-liner.

You can apporach your scientific number notation problem accordingly.