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

# # Generate national and state-level crosswalks
# ## 2000 block group parts to 2010 tracts
# 
# ### NHGIS [block crosswalks](https://www.nhgis.org/user-resources/geographic-crosswalks)
# 
# **James D. Gaboardi, 06/2020**

# In[1]:


get_ipython().run_line_magic('load_ext', 'watermark')
get_ipython().run_line_magic('watermark', '')


# In[2]:


import nhgisxwalk
import inspect
import numpy
import pandas

get_ipython().run_line_magic('load_ext', 'autoreload')
get_ipython().run_line_magic('autoreload', '2')
get_ipython().run_line_magic('watermark', '-w')
get_ipython().run_line_magic('watermark', '-iv')


# ### Source and target years for the crosswalk

# In[3]:


source_year, target_year = "2000", "2010"
gj_src, gj_trg = "GJOIN%s"%source_year, "GJOIN%s"%target_year


# In[4]:


data_in = "../../crosswalks/"
data_tab = "../../tabular_data/"
block_file = "%s_block" % source_year


# ### Source-target building base

# In[5]:


base_xwalk_name = "nhgis_blk%s_blk%s_gj" % (source_year, target_year)
data_types = nhgisxwalk.str_types([gj_src, gj_trg])
from_csv_kws = {"path": data_in, "archived": True, "remove_unpacked": True}
read_csv_kws = {"dtype": data_types}
base_xwalk = nhgisxwalk.xwalk_df_from_csv(
    base_xwalk_name, **from_csv_kws, **read_csv_kws
)
base_xwalk.head()


# ### Source summary data

# In[6]:


base_source_name = "%s/%s.csv" % (block_file, block_file)
base_source_file = "%s%s" % (data_tab, base_source_name)
base_source_file


# ### Convenience code shorthand/lookup

# In[7]:


nhgisxwalk.valid_geo_shorthand(shorthand_name=False)


# ### Set the `nhgisxwalk.GeoCrossWalk` parameters
# ##### see [nhgisxwalk.GeoCrossWalk](https://github.com/jGaboardi/nhgisxwalk/blob/92b4fe55de0a9c53d0315dcda8ec121faaf20aef/nhgisxwalk/geocrosswalk.py#L19) for full details

# In[8]:


nhgisxwalk.desc_code_2000_SF1b


# In[9]:


input_vars = [
    nhgisxwalk.desc_code_2000_SF1b["Persons"]["Total"],
    nhgisxwalk.desc_code_2000_SF1b["Families"]["Total"],
    nhgisxwalk.desc_code_2000_SF1b["Households"]["Total"],
    nhgisxwalk.desc_code_2000_SF1b["Housing Units"]["Total"]
]
input_vars


# In[10]:


input_var_tags = ["pop", "fam", "hh", "hu"]


# In[11]:


xwalk_args = {
    "source_year": source_year,
    "target_year": target_year,
    "source_geo": "bgp",
    "target_geo": "tr",
    "base_source_table": base_source_file,
    "input_var": input_vars,
    "weight_var": input_var_tags,
    "keep_base": False,
    "add_geoid": True
}


# ### Generate data product
# 1. Create a national crosswalk then split by state 
# 2. Write out all products with `README.txt` files

# In[12]:


nhgisxwalk.generate_data_product(base_xwalk, xwalk_args, data_in)


# -----------------------------------------------