mjhjmtu

I am trying to write up a command line interfaces that will removes a particular section / lines of codes within a list of json files. By the way, the json file are located within the sub-folders of the main directory

I am pretty new to this but this is the code that I can come up with so far - find -name "*.json" | xargs sed -i "map" but some of the json files I had, its format is slightly different

So far I am seeing the following 2 formats within my list:

 "tags": , 
 "map": 
 "KPA": 
 "State": True, 
 "namespace": "KPA01"
 
 

or

 "tags": 
 "type": [
 "char"
 ], 
 "dynamic": true
 , 
 "map": 
 "KPA01": 
 "State": True, 
 "namespace": "KPA01"
 
 

and basically, I am trying to omit out the map section that it has, so that it will only display the tags section but the presence of commas and / are making it hard for me.
So my output results should be like this:

 "tags": 

or

 "tags": 
 "type": [
 "char"
 ], 
 "dynamic": true
 

Will this be possible to do so in a command line interface? I heard that jq may be able to do it, however, as I tried executing jq '.map' test.json I am getting parse error: ':' not as part of an object at line 2, column 11 in my terminal. Likewise it also seems to be giving off error if I am using the jq play online..

Any ideas?

First of all, change True to true. As a whole, this works very well:

#!/usr/bin/python
import sys
import json

inputfile = sys.argv[1]
with open(inputfile,'r') as myfile:
 obj = json.loads(myfile.read().replace('True','true'))
 if "map" in obj:
 del obj["map"]
 json.dump(obj,sys.stdout,indent=4,separators=(',',': '))

This writes to standard output.

EDIT: the previous in-place version seemed to be somewhat dangerous. Better do it this way:

#!/usr/bin/python
import sys
import json

inputfile = sys.argv[1]
with open(inputfile,'r') as myfile:
 obj = json.loads(myfile.read().replace('True','true'))
 if "map" in obj:
 del obj["map"]
with open(inputfile,'w') as myfile:
 json.dump(obj,myfile,indent=4,separators=(',',': '))

Because the script is actually aware of what a valid JSON is, it will throw an exception if it encounters invalid code, instead of producing unpredictable output.

This works on python 3, just so you know.

EDIT2:

You can modify the objects in any way you like, the purpose of Json is exactly serialization of objects. Treat them as associative arrays and give them any value you want. For instance, you can do this:

#add a new string on the "ground" level
obj["new_key"]="lol"
#add a new subarray, with properties of different types
obj["this_is_array"]="a": 3, "b": 16, "c": "string", "d": False 
#modify the value of existing field
obj["new_key"]="new value"
#insert into subarray (test if it exists first)
if "this_is_array" in obj:
 obj["this_is_array"]["e"]=42

If you handle the True > true as mentioned elsewhere and get the jq tool, you can just do:

jq 'tags' <infile

For example, after copying one of your examples to my clipboard:

xsel -bo | sed 's/True/true/g' | jq 'tags'

OUTPUT:

 "tags": 
 "type": [
 "char"
 ],
 "dynamic": true
 

I found same error:

parse error: Invalid numeric literal at line 5, column 26

and I'm not used to json but I think True should be in smallcase, as true, so you can run a perl one-liner to fix it and then use jq to filter out the map key, like:

perl -pe 's/(W)T(rue)/$1t$2/g' file1.json | ./jq 'del(.map)'

for first example, that yields:

 "tags": 

and:

perl -pe 's/(W)T(rue)/$1t$2/g' file2.json | ./jq 'del(.map)'

for the second one, that yields:

 "tags": 
 "type": [
 "char"
 ],
 "dynamic": true
 

question is old, however for completeness of options, here's jtc based solution:

bash $ jtc -pw'<map>l+0' input.json

 "tags": 
 "dynamic": true,
 "type": [
 "char"
 ]
 

bash $ 

• it will find all labels "map" and purge'em all (while preserving the rest of json)