How to parse a JSON string with no root element?

Question

I have a JSON file like below:

{
  "name":"A",
  "age":19
}
{
  "name":"B",
  "age":20
}

So basically the file contains a list of people.

I tried to use json.loads(str_content) in Python 3, but it returned the error of json.decoder.JSONDecodeError: Extra data:.

I checked with an online JSON parser (http://json.parser.online.fr) and it told me the same problem.

How to parse a JSON file without a root element but a list of JSON objects?

The problem is you *don't* have a JSON file; what's creating that file, could it be modified to have a single root array? — jonrsharpe, Nov 22 '18 at 08:40
@jonrsharpe I think the file is created from MongoDB. In fact I god a BSON file, then I converted it to JSON. — mommomonthewind, Nov 22 '18 at 08:43
Is it stored in multi-line json? Like this `{"name":"A","age":19} {"name":"B","age":20}` Each in new line — Srce Cde, Nov 22 '18 at 08:45
But it *isn't* JSON. You can synthesize it into a JSON file by adding a pair of square brackets and a comma between elements. Probably `jq -r .` can do this for you automatically, or if the fragments are one line per dict, you can use [a simple `sed` script.](/q/35021524) — tripleee, Nov 22 '18 at 08:46
Just read it in groups of 4 lines and pass them to `json.loads()` to get one object at a time. — martineau, Nov 22 '18 at 08:46
@Chirag yes, but each JSON object is stored in multi-lines as well. Otherwise I can parse by lines. — mommomonthewind, Nov 22 '18 at 08:46
@martineau each JSON object has a different number of lines (for instance, someone has information of ``weight`` but someone does not) — mommomonthewind, Nov 22 '18 at 08:47
Duplicate? https://stackoverflow.com/questions/41397606/how-to-parse-a-json-file-without-root-element-using-org-json-simple — TZubiri, Nov 22 '18 at 09:08

Guillaume · Answer 1 · 2018-11-22T10:14:25.870

3

The issue is that the string you are trying to parse is not a valid JSON document. It is actually a concatenation of JSON documents. So the simple json.loads() will not work.

You can use instead something based on https://docs.python.org/3/library/json.html#json.JSONDecoder.raw_decode . E.g: (code is a bit ugly but the logic should be clear):

import json

s = """{
  "name":"A",
  "age":19
}
{
  "name":"B",
  "age":20
}"""

def iter_jsons(s):
    decoder = json.JSONDecoder()

    i = 0
    while True:
        doc, i2 = decoder.raw_decode(s[i:].strip())
        yield doc
        if i == i2:
            break
        i= i2

print(list(iter_jsons(s)))

[{'name': 'A', 'age': 19}, {'name': 'B', 'age': 20}]

edited Nov 22 '18 at 10:14

answered Nov 22 '18 at 08:54

Guillaume

5,497
3
24
42

If there are no nested dicts you could simply look for the closing brace and parse up to that point. – tripleee Nov 22 '18 at 09:00
I can't find anything in the json standard that indicates that a JSON document must have a root element. – TZubiri Nov 22 '18 at 09:08
@TomasZubiri It is not explicitly stated as text in https://tools.ietf.org/html/rfc7159#section-2 , but the ABNF indicates that a JSON-text is a SINGLE value – Guillaume Nov 22 '18 at 10:08
2

@TomasZubiri: I believe the standard indicates that a JSON document must either be a single JSON object enclosed in `{`, `}` brackets or a comma delimited list of them enclosed in `[`, `]` brackets. This answer looks like a very clever workaround allowing this non-compliant input to be decoded into the list it should have been in the first place. – martineau Nov 22 '18 at 10:11
@tripleee: that is correct, but I'd rather not assume anything about the content I am trying to parse. – Guillaume Nov 22 '18 at 10:15

score 1 · Answer 2 · answered Jun 25 '19 at 21:29

1

This works like a charm

import json        
json_file = "myfile.json"

objects = json.load(open(json_file))

for person in objects:
    name = person['name']
    print(name)

answered Jun 25 '19 at 21:29

SoSpoon

156
2
6

How to parse a JSON string with no root element?

2 Answers2