Maria@4: # -*- coding: utf-8 -*-
Maria@4: """
Maria@4: Created on Wed May 17 11:35:51 2017
Maria@4: 
Maria@4: @author: mariapanteli
Maria@4: """
Maria@4: import numpy as np
m@84: import pandas as pd
Maria@4: import json
Maria@4: import pysal # before shapely in util_plots
Maria@4: import fiona
m@8: import os
Maria@4: import matplotlib.pyplot as plt
Maria@4: 
Maria@4: 
m@8: DATA_DIR = os.path.join(os.path.dirname(__file__), 'util_data')
m@8: JSON_DB = os.path.join(DATA_DIR, 'countries.json')
m@8: SHAPEFILE = os.path.join(DATA_DIR, 'shapefiles', 'ne_10m_admin_0_countries.shp')
m@8: 
m@8: 
m@8: def neighbors_from_json_file(data_countries, json_DB=JSON_DB):
Maria@4:     neighbors = {}
Maria@4:     with open(json_DB) as json_file:
Maria@4:         countries_dict = json.load(json_file)
Maria@4:     country_names = []
Maria@4:     country_iso = []
Maria@4:     country_borders_iso = []
Maria@4:     for country_info in countries_dict:
Maria@4:         country_names.append(country_info['name']['common'])
Maria@4:         country_iso.append(country_info['cca3'])
Maria@4:         country_borders_iso.append(country_info['borders'])
Maria@4:     # temporary fixes of country names to match json data
Maria@4:     country_names[country_names.index('United States')] = 'United States of America'
Maria@4:     country_names[country_names.index('Tanzania')] = 'United Republic of Tanzania'
Maria@4:     country_names[country_names.index('DR Congo')] = 'Democratic Republic of the Congo'
Maria@4:     country_names[country_names.index('Czechia')] = 'Czech Republic'
Maria@4:     for i, country in enumerate(data_countries):
Maria@4:         neighbors[i] = {} 
Maria@4:         if country in country_names:
Maria@4:             if len(country_borders_iso[country_names.index(country)])>0:
Maria@4:                 # if country has neighbors according to json file
Maria@4:                 neighbors_iso = country_borders_iso[country_names.index(country)]
Maria@4:                 neighbors_names = [country_names[country_iso.index(nn)] for nn in neighbors_iso]
Maria@4:                 for neighbor in neighbors_names:
Maria@4:                     if neighbor in data_countries:
Maria@4:                         neighbor_idx = np.where(data_countries==neighbor)[0][0]
Maria@4:                         neighbors[i][neighbor_idx] = 1.0                
Maria@4:     w = pysal.weights.W(neighbors, id_order=range(len(data_countries)))
Maria@4:     return w
Maria@4: 
Maria@4: 
Maria@4: def get_countries_from_shapefile(shapefile):
Maria@4:     shp = fiona.open(shapefile, 'r')
Maria@4:     countries = []
Maria@4:     if shp[0]["properties"].has_key("ADMIN"):
Maria@4:         country_keyword = "ADMIN"
Maria@4:     elif shp[0]["properties"].has_key("NAME"):
Maria@4:         country_keyword = "NAME"
Maria@4:     else:
Maria@4:         country_keyword = "admin"
Maria@4:     for line in shp:
Maria@4:         countries.append(line["properties"][country_keyword])
Maria@4:     shp.close()
Maria@4:     return countries
Maria@4: 
Maria@4: 
Maria@4: def replace_empty_neighbours_with_KNN(data_countries, w):
m@8:     shapefile = SHAPEFILE
Maria@4:     no_neighbors_idx = w.islands
Maria@4:     knn = 10
Maria@4:     wknn = pysal.knnW_from_shapefile(shapefile, knn)
Maria@4:     knn_countries = get_countries_from_shapefile(shapefile)
Maria@4:     neighbors = w.neighbors
Maria@4:     for nn_idx in no_neighbors_idx:
Maria@4:         country = data_countries[nn_idx]
Maria@4:         print country
Maria@4:         if country not in knn_countries:
Maria@4:             continue
Maria@4:         knn_country_idx = knn_countries.index(country)
Maria@4:         knn_country_neighbors = [knn_countries[nn] for nn in wknn.neighbors[knn_country_idx]]
Maria@4:         for knn_nn in knn_country_neighbors:
Maria@4:             if len(neighbors[nn_idx])>2:
Maria@4:                 continue
Maria@4:             data_country_idx = np.where(data_countries==knn_nn)[0]
Maria@4:             if len(data_country_idx)>0:
Maria@4:                 neighbors[nn_idx][data_country_idx[0]] = 1.0
Maria@4:     w = pysal.weights.W(neighbors, id_order=range(len(data_countries)))
Maria@4:     return w
Maria@4: 
Maria@4: 
Maria@4: def get_neighbors_for_countries_in_dataset(Y):
Maria@4:     # neighbors
Maria@4:     data_countries = np.unique(Y)
Maria@4:     w = neighbors_from_json_file(data_countries)
Maria@4:     w = replace_empty_neighbours_with_KNN(data_countries, w)
Maria@4:     return w, data_countries
Maria@4: 
Maria@4: 
Maria@4: def from_weights_to_dict(w, data_countries):
Maria@4:     w_dict = {}
Maria@4:     for i in w.neighbors:
Maria@4:         w_dict[data_countries[i]] = [data_countries[nn] for nn in w.neighbors[i]]
Maria@4:     return w_dict
Maria@4: 
Maria@4: 
m@77: def get_regions_from_shapefile(shapefile):
m@77:     shp = fiona.open(shapefile, 'r')
m@77:     countries = []
m@77:     regions = []
m@77:     if shp[0]["properties"].has_key("ADMIN"):
m@77:         country_keyword = "ADMIN"
m@77:         region_keyword = "REGION_WB"
m@77:     elif shp[0]["properties"].has_key("NAME"):
m@77:         country_keyword = "NAME"
m@77:         region_keyword = "REGION_WB"
m@77:     else:
m@77:         country_keyword = "admin"
m@77:         region_keyword = "region_wb"
m@77:     for line in shp:
m@77:         #countries.append(line["properties"]["admin"])
m@77:         countries.append(line["properties"][country_keyword])
m@77:         regions.append(line["properties"][region_keyword])
m@77:     shp.close()
m@77:     return countries, regions
m@77: 
m@77: 
m@77: def append_regions(df):
m@77:     countries, regions = get_regions_from_shapefile(SHAPEFILE)
m@77:     if 'French Guiana' not in countries:
m@77:         countries.append('French Guiana')
m@77:         regions.append('Latin America & Caribbean')
m@77:     df_regions = pd.DataFrame({'Country': countries, 'Region': regions})
m@77:     df_append = pd.merge(df, df_regions, how='left', on='Country')
m@77:     return df_append
m@77: