JSON import

[rburghol]

Objectives:

• [x] Dynamically load a model from json into a model domain.
  • [x] Automatic at runtime based on hdf5 name? See #65
  • [ ] CMD with hsp2 import_json [json file name]
• [x] Make mode where it fails when unknown json components are found
• [x] Make mode where unknown json components are loaded as constant = 0.0
• [x] Which json library
• Generate objects:
  • Can create objects in python and export:
  • Can generate json from R with objects: riverseg_facilities.R
• To Create a facility WD model for testing see: https://github.com/HARPgroup/vahydro/issues/844
• Example load model JSON from vahydro:
  • Original https://github.com/HARPgroup/HSPsquared/blob/specom/tests/testcbp/HSP2results/json.py
  • Recent version for facility: https://github.com/HARPgroup/HSPsquared/blob/spects/tests/testcbp/HSP2results/load_facility_json.py
  • Load a river segment: https://github.com/HARPgroup/HSPsquared/blob/spects/tests/testcbp/HSP2results/load_test_riverseg.py

    
    import json
    import requests
    from requests.auth import HTTPBasicAuth
    import csv
    import pandas as pd

src_json_node = '...d.dh/node/62' el_pid = 4723109 json_url = src_json_node + "/" + str(el_pid)

authentication using rest un and pw

jfile = open("/var/www/python/auth.private") jj = json.load(jfile) rest_uname = jj[0]['rest_uname'] rest_pass = jj[0]['rest_pw'] basic = HTTPBasicAuth(rest_uname, rest_pass )

Opening JSON file

jraw = requests.get(json_url, auth=basic) model_json = jraw.content.decode('utf-8')

returns JSON object as

a dictionary

model_data = json.loads(model_json) mlist = list(model_data)

now extract the model using the name

model_data= model_data[mlist[0]] for k in model_data.keys(): print(k)

[jdkleiner]

@rburghol What format is your "/var/www/python/auth.private" file in? So I can test it on my end :)

I suppose we could consider storing the auth info in a single location like "/var/www/auth/auth.private" since the same login info can be used for REST calls from both R and python (we'd probably need to store the file in a format (JSON?) that makes it easy to read the rest_uname and rest_pw to variables in both R and python environments)

[rburghol]

@jdkleiner Just wanted to get this in writing -- the auth.private file used here is JSON, so, that makes it an ideal candidate moving forward for a cross-platform single file.

[jdkleiner]

Excellent. FWIW I ran some tests using an auth.private file in the following format:

{
    "auths": {
        "rest_uname": "un_XXXX",
        "rest_pw": "pwd_XXXX"
    },
    "src_json_node": "XXXX/d.dh/node/62"
}