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

# # Exporting a disclosure file from lcopt

# Note: For this to work you need the latest version of lcopt-dev from the `disclosures` branch on github

# ### Load (or create) an lcopt model

# In[1]:


from lcopt import *


# In[2]:


model = LcoptModel(load=r"files\Example Model.lcopt") # Note: you can download 'Example Model.lcopt' from the files folder on github


# Here's what the model looks like:
# 
# ![Flow chart](images/tea_flow_chart.jpg)

# ### Create the disclosure file and store the filename

# Using `export_disclosure` with no arguments exports the unspecified model (all values 1)
# 
# (Note: `folder_path` optionally chooses a folder to put the files in, otherwise they are saved to the working directory)

# In[3]:


unspecified_file_name = model.export_disclosure(folder_path='files')
unspecified_file_name


# Or you can specify a parameter set by its index:

# In[4]:


ps0_file_name = model.export_disclosure(0, folder_path='files')
ps0_file_name


# or its name:

# In[5]:


ps1_file_name = model.export_disclosure('Milky tea', folder_path='files')
ps1_file_name


# ### Load one of the files to see the format

# In[6]:


import json
from pprint import pprint


# In[7]:


with open(unspecified_file_name, 'r') as j:
    data = json.load(j)


# In[8]:


pprint(data)


# ### Reconstruct the matrices

# Here's what we're expecting:
# 
# ![disclosure_pic.png](images/disclosure_pic.png)

# In[9]:


import numpy as np
import pandas as pd


# Write a helper function to reconstruct a matrix as a numpy array from the json data

# In[10]:


def reconstruct_matrix(matrix_dict):
    m = np.zeros(matrix_dict['shape'])
    for (r, c), v in matrix_dict['data']:
        m[r, c] = v
    
    return m


# Create pandas DataFrames for easy viewing of the resulting matrices

# In[11]:


Af = pd.DataFrame(reconstruct_matrix(data['Af']), index=["({}) {}, {}".format(x['index'], x['name'], x['unit']) for x in data['foreground flows']])


# In[12]:


Ad = pd.DataFrame(reconstruct_matrix(data['Ad']), index=["{} [{}], {},  ({}, {})".format(x['ecoinvent_name'], x['location'],x['unit'], *x['brightway_id']) for x in data['background flows']])


# In[13]:


Bf = pd.DataFrame(reconstruct_matrix(data['Bf']), index=["{}, {}, ({}, {})".format(x['name'], x['unit'], *x['biosphere3_id']) for x in data['foreground emissions']])


# In[14]:


Af


# In[15]:


Ad


# In[16]:


Bf


# ### Reconstruct the data from an lcopt parameter set
# If we want to recreate the results of an lcopt model we can do the same thing with one of the specified files

# In[17]:


with open(ps1_file_name, 'r') as j:
    specified_data = json.load(j)
    
Af_ps1 = pd.DataFrame(reconstruct_matrix(specified_data['Af']), index=["({}) {}, {}".format(x['index'], x['name'], x['unit']) for x in specified_data['foreground flows']])
Ad_ps1 = pd.DataFrame(reconstruct_matrix(specified_data['Ad']), index=["{} [{}], {},  ({}, {})".format(x['ecoinvent_name'], x['location'],x['unit'], *x['brightway_id']) for x in specified_data['background flows']])
Bf_ps1 = pd.DataFrame(reconstruct_matrix(specified_data['Bf']), index=["{}, {}, ({}, {})".format(x['name'], x['unit'], *x['biosphere3_id']) for x in specified_data['foreground emissions']])


# In[18]:


Af_ps1


# In[19]:


Ad_ps1


# In[20]:


Bf_ps1


# This should be all the information we need to recreate the structure of the original foreground model, with the correct links to the background and biosphere databases