#!/usr/bin/env python
# coding: utf-8

# In[1]:


from dask.distributed import Client, config
client = Client("dask.azure.farsetlabs.org.uk:8786")
client.get_versions(check=True)


# In[8]:


for w,logs in client.get_worker_logs().items():
    for log in logs:
        print(w, ':'.join(log))


# In[7]:


from dask import dataframe
# ('head-1-5-read-parquet-0298763dc1b1d75d8743c178e56ae7cd', 0)
df = dataframe.read_parquet("s3://datasets.farsetlabs.org.uk/parquet_dask_example/",
                           storage_options={'anon': True})
client.publish_dataset(parquet_dask_example=client.persist(df))


# In[ ]:


client.list_datasets()


# In[17]:


df=client.get_dataset('parquet_dask_example')
df['date'] = dataframe.to_datetime(df['date'])
df=client.persist(df)


# In[18]:


df.dtypes


# In[11]:


df.groupby(['date'])


# In[ ]: