import json

import pandas as pd

pod_infos = json.load(open('kube_pod_info_asset_snapshot.json'))
node_infos = json.load(open('kube_node_info_asset_snapshot.json'))

pods_data = []
nodes_data = []
for pod in pod_infos['data']['result']:
    kube_pod_info_data = pod['metric']
    cluster = kube_pod_info_data['cluster']
    if 'shoot/' in cluster or 'seed/' in cluster:
        continue
    if 'node' not in kube_pod_info_data:
        continue
    node = kube_pod_info_data['node']
    pod = kube_pod_info_data['pod']
    namespace = kube_pod_info_data['namespace']
    result = {
        "pod": pod, "namespace": namespace,
        "node": node, "cluster": cluster
    }
    pods_data.append(result)

for node in node_infos['data']['result']:
    kube_node_info_data = node['metric']
    cluster = kube_node_info_data['cluster']
    if 'shoot/' in cluster or 'seed/' in cluster:
        continue
    if 'node' not in kube_node_info_data:
        continue
    node = kube_node_info_data['node']
    result = {
        "node": node, "cluster": cluster,
        # TODO(chenfei): to investigate why there is no this field
        'node_instance_id': kube_node_info_data.get('provider_id', '')
    }
    nodes_data.append(result)

pd_a = pd.DataFrame(pods_data)
pd_b = pd.DataFrame(nodes_data)

result = pd.merge(pd_a, pd_b, left_on=('cluster', 'node'),
                  right_on=('cluster', 'node'), how='left')
result.to_json('pod_node_scheduled_data.json', orient="records")