上传 json 文件以创建单个 pandas dataframe

Question

I have to upload from a pc folder lots of json files representing each one a row of the final dataframe I have to build. Moreover, only 4 keys of the json files(nested dictionaries) have to appear in the dataframe.

Here I show you the working script that already does exactly what I need, but just for one json file:

with open('2020-03-02-10-43-08-9148.json') as inf:
    j = json.load(inf)
    log_cols = ["spotName", "CurrentCurve", "VoltageCurve"]
    data = [j['Message']['WeldLog'][col] for col in log_cols] + [j["TimeStamp"]]
    col_names = ["SpotName", "CurrentCurve", "VoltageCurve", "TimeStamp"]
    df = pd.DataFrame([data], columns=col_names)

Then I very naively tried to add some code in order to extend the script to all the json files in my folder, but it doesn't work. When I print the dataframe it seems to be composed just by one row(a single json file)!

import pandas as pd
import json 
import io
import glob



for json_file in glob.glob("*.json"): #Assuming that json files and .py file are in the same directory
    
    with open(json_file) as inf:
        
        j = json.load(inf)
        log_cols = ["spotName", "CurrentCurve", "VoltageCurve"]
        data = [j['Message']['WeldLog'][col] for col in log_cols] + [j["TimeStamp"]]
        col_names = ["SpotName", "CurrentCurve", "VoltageCurve", "TimeStamp"] #Names of the df columns
        
          
df = pd.DataFrame([data], columns=col_names)
        

Lastly, if can be helpful, I leave you here an example of the json file structure:

{
    "Name": "WeldLog",
    "WeldTimer": "SCC005R01",
    "TimeStamp": "2019-11-07T12:29:01",
    "OutputFormat": "JSON",
    "Message": {
        "WeldLog": {
            "dateTime": "2019-10-23T18:30:31.8",
            "iActual1": 0.00,
            "iActual2": 7.98,
            "iActual3": 0.00,
            "partIdentString": "",
            "pha1": 0.00,
            "pha2": 32.24,
            "pha3": 0.00,
            "progNo": 49,
            "spotName": "60090_0_00",
            "timerName": "SCC005R01",
            "currentActualValue": 7.97,
            "currentFactor": 0,
            "currentReferenceValue": 7.82,
            "iDemand1": 3.00,
            "iDemand2": 7.80,
            "iDemand3": 3.00,
            "electrodeNo": 1,
            "iDemandStd": 9.00,
            "energyActualValue": 5159.189,
            "energyRefValue": 5301.22,
            "contactWaitTime": null,
            "monitorMode": 1,
            "monitorState": 0,
            "powerActualValue": 12514.94,
            "powerRefValue": 13288.9,
            "powerState": 0,
            "resistanceActualValue": 192,
            "resistanceRefValue": 209,
            "protRecord_ID": 670196.0,
            "uipActualValue": 86,
            "uipRefValue": 0,
            "uirExpulsionTime": 0,
            "voltageActualValue": 1.57,
            "voltageRefValue": 1.70,
            "wear": 1.00,
            "tipDressCounter": 26,
            "weldSpotCustDataP16_1": 0,
            "weldSpotCustDataP16_2": 0,
            "weldSpotCustDataP16_3": 0,
            "weldSpotCustDataP16_4": 0,
            "CurrentCurve": null,
            "VoltageCurve": [1, 2, 3, 4, 8],
            "ForceCurve": [6, 7, 8, 3, 6, 9, 6]
        }
    }
}

Answer 1

The reason is that data is override in every loop. You need to store them globally.

data = []

for json_file in glob.glob("*.json"): #Assuming that json files and .py file are in the same directory
    with open(json_file) as inf:
        j = json.load(inf)
        log_cols = ["spotName", "CurrentCurve", "VoltageCurve"]
        data.append([j['Message']['WeldLog'][col] for col in log_cols] + [j["TimeStamp"]])

col_names = ["SpotName", "CurrentCurve", "VoltageCurve", "TimeStamp"] #Names of the df columns

df = pd.DataFrame(data, columns=col_names)