如何将CSV文件转换为多行JSON？

Here’s my code, really simple stuff…

import csv
import json

csvfile = open('file.csv', 'r')
jsonfile = open('file.json', 'w')

fieldnames = ("FirstName","LastName","IDNumber","Message")
reader = csv.DictReader( csvfile, fieldnames)
out = json.dumps( [ row for row in reader ] )
jsonfile.write(out)

Declare some field names, the reader uses CSV to read the file, and the filed names to dump the file to a JSON format. Here’s the problem…

Each record in the CSV file is on a different row. I want the JSON output to be the same way. The problem is it dumps it all on one giant, long line.

I’ve tried using something like for line in csvfile: and then running my code below that with reader = csv.DictReader( line, fieldnames) which loops through each line, but it does the entire file on one line, then loops through the entire file on another line… continues until it runs out of lines.

Any suggestions for correcting this?

Edit: To clarify, currently I have: (every record on line 1)

[{"FirstName":"John","LastName":"Doe","IDNumber":"123","Message":"None"},{"FirstName":"George","LastName":"Washington","IDNumber":"001","Message":"Something"}]

What I’m looking for: (2 records on 2 lines)

{"FirstName":"John","LastName":"Doe","IDNumber":"123","Message":"None"}
{"FirstName":"George","LastName":"Washington","IDNumber":"001","Message":"Something"}

Not each individual field indented/on a separate line, but each record on it’s own line.

Some sample input.

"John","Doe","001","Message1"
"George","Washington","002","Message2"

The problem with your desired output is that it is not valid json document,; it’s a stream of json documents!

That’s okay, if its what you need, but that means that for each document you want in your output, you’ll have to call json.dumps.

Since the newline you want separating your documents is not contained in those documents, you’re on the hook for supplying it yourself. So we just need to pull the loop out of the call to json.dump and interpose newlines for each document written.

import csv
import json

csvfile = open('file.csv', 'r')
jsonfile = open('file.json', 'w')

fieldnames = ("FirstName","LastName","IDNumber","Message")
reader = csv.DictReader( csvfile, fieldnames)
for row in reader:
    json.dump(row, jsonfile)
    jsonfile.write('\n')

You can use Pandas DataFrame to achieve this, with the following Example:

import pandas as pd
csv_file = pd.DataFrame(pd.read_csv("path/to/file.csv", sep = ",", header = 0, index_col = False))
csv_file.to_json("/path/to/new/file.json", orient = "records", date_format = "epoch", double_precision = 10, force_ascii = True, date_unit = "ms", default_handler = None)

I took @SingleNegationElimination’s response and simplified it into a three-liner that can be used in a pipeline:

import csv
import json
import sys

for row in csv.DictReader(sys.stdin):
    json.dump(row, sys.stdout)
    sys.stdout.write('\n')

import csv
import json

file = 'csv_file_name.csv'
json_file = 'output_file_name.json'

#Read CSV File
def read_CSV(file, json_file):
    csv_rows = []
    with open(file) as csvfile:
        reader = csv.DictReader(csvfile)
        field = reader.fieldnames
        for row in reader:
            csv_rows.extend([{field[i]:row[field[i]] for i in range(len(field))}])
        convert_write_json(csv_rows, json_file)

#Convert csv data into json
def convert_write_json(data, json_file):
    with open(json_file, "w") as f:
        f.write(json.dumps(data, sort_keys=False, indent=4, separators=(',', ': '))) #for pretty
        f.write(json.dumps(data))


read_CSV(file,json_file)

Documentation of json.dumps()

You can try this

import csvmapper

# how does the object look
mapper = csvmapper.DictMapper([ 
  [ 
     { 'name' : 'FirstName'},
     { 'name' : 'LastName' },
     { 'name' : 'IDNumber', 'type':'int' },
     { 'name' : 'Messages' }
  ]
 ])

# parser instance
parser = csvmapper.CSVParser('sample.csv', mapper)
# conversion service
converter = csvmapper.JSONConverter(parser)

print converter.doConvert(pretty=True)

Edit:

Simpler approach

import csvmapper

fields = ('FirstName', 'LastName', 'IDNumber', 'Messages')
parser = CSVParser('sample.csv', csvmapper.FieldMapper(fields))

converter = csvmapper.JSONConverter(parser)

print converter.doConvert(pretty=True)

Add the indent parameter to json.dumps

 data = {'this': ['has', 'some', 'things'],
         'in': {'it': 'with', 'some': 'more'}}
 print(json.dumps(data, indent=4))

Also note that, you can simply use json.dump with the open jsonfile:

json.dump(data, jsonfile)

I see this is old but I needed the code from SingleNegationElimination however I had issue with the data containing non utf-8 characters. These appeared in fields I was not overly concerned with so I chose to ignore them. However that took some effort. I am new to python so with some trial and error I got it to work. The code is a copy of SingleNegationElimination with the extra handling of utf-8. I tried to do it with https://docs.python.org/2.7/library/csv.html but in the end gave up. The below code worked.

import csv, json

csvfile = open('file.csv', 'r')
jsonfile = open('file.json', 'w')

fieldnames = ("Scope","Comment","OOS Code","In RMF","Code","Status","Name","Sub Code","CAT","LOB","Description","Owner","Manager","Platform Owner")
reader = csv.DictReader(csvfile , fieldnames)

code = ''
for row in reader:
    try:
        print('+' + row['Code'])
        for key in row:
            row[key] = row[key].decode('utf-8', 'ignore').encode('utf-8')      
        json.dump(row, jsonfile)
        jsonfile.write('\n')
    except:
        print('-' + row['Code'])
        raise

How about using Pandas to read the csv file into a DataFrame (pd.read_csv), then manipulating the columns if you want (dropping them or updating values) and finally converting the DataFrame back to JSON (pd.DataFrame.to_json).

Note: I haven’t checked how efficient this will be but this is definitely one of the easiest ways to manipulate and convert a large csv to json.

As slight improvement to @MONTYHS answer, iterating through a tup of fieldnames:

import csv
import json

csvfilename = 'filename.csv'
jsonfilename = csvfilename.split('.')[0] + '.json'
csvfile = open(csvfilename, 'r')
jsonfile = open(jsonfilename, 'w')
reader = csv.DictReader(csvfile)

fieldnames = ('FirstName', 'LastName', 'IDNumber', 'Message')

output = []

for each in reader:
  row = {}
  for field in fieldnames:
    row[field] = each[field]
output.append(row)

json.dump(output, jsonfile, indent=2, sort_keys=True)

import csv
import json
csvfile = csv.DictReader('filename.csv', 'r'))
output =[]
for each in csvfile:
    row ={}
    row['FirstName'] = each['FirstName']
    row['LastName']  = each['LastName']
    row['IDNumber']  = each ['IDNumber']
    row['Message']   = each['Message']
    output.append(row)
json.dump(output,open('filename.json','w'),indent=4,sort_keys=False)