# coding: utf-8
import pandas as pd
import numpy as np
import re

# read BSRN meta data file
df = pd.read_csv('bsrn_stations.tab', '\t')
staff = pd.read_csv('bsrn_staffs.tab', '\t')
mapping = pd.read_csv('colnames-mapping-BG.csv', index_col=0)

staff.index = staff['First name'] + ' ' + staff['Last name']


candidate_patt = re.compile(r'\s*BSRN Candidate\.\s*')
email_patt = re.compile(r'\s*\(.+@.+\).*')


def add_comment(d, val):
    if 'Unsorted Comments' in d:
        d['Unsorted Comments'] += '; ' + val
    else:
        d['Unsorted Comments'] = val


def create_dict(s):
    """Transform a *Comment* string into a :class:`dict`"""
    items = re.split(r';\s*', s)
    d = {}
    for item in items:
        try:
            key, val = item.split(': ', 1)
        except ValueError:
            add_comment(d, item)
        else:
            candidate = candidate_patt.search(key)
            if candidate:
                add_comment(d, 'BSRN Candidate')
                key = candidate_patt.sub('', key)
            d[key] = val
    if 'Station scientist' in d or '. Station scientist' in d:
        scientist = email_patt.sub(
            '', d.get('Station scientist', d.get('. Station scientist')))
        try:
            row = staff.loc[scientist]
        except KeyError:
            pass
        else:
            d.update(row.to_dict())
    return d


dicts = df.Comment.apply(create_dict).values.tolist()
comments = pd.DataFrame(dicts)
joined = df.join(comments)

# Manual fixes of some typos in the comments
if 'Surface type' in joined:
    joined.loc[joined['Surface type'].notnull(), 'Surface Type'] = joined.loc[
        joined['Surface type'].notnull(), 'Surface type']
    del joined['Surface type']

if 'BSRN station no' in joined:
    joined.loc[joined['BSRN station no'].notnull(), 'BSRN Station no'] = joined.loc[
        joined['BSRN station no'].notnull(), 'BSRN station no']
    del joined['BSRN station no']

if '. Station scientist' in joined:
    joined.loc[joined['. Station scientist'].notnull(), 'Station scientist'] = joined.loc[
        joined['. Station scientist'].notnull(), '. Station scientist']
    del joined['. Station scientist']

if 'Plain terrain, Soil ' in joined:
    joined.loc[joined['Plain terrain, Soil '].notnull(),
               'Plain terrain, Soil'] = joined.loc[
        joined['Plain terrain, Soil '].notnull(), 'Plain terrain, Soil ']
    del joined['Plain terrain, Soil ']

if 'Plain terrain, Soil type' in joined:
    joined.loc[joined['Plain terrain, Soil type'].notnull(),
               'Plain terrain, Soil'] = joined.loc[
        joined['Plain terrain, Soil type'].notnull(),
        'Plain terrain, Soil type']
    del joined['Plain terrain, Soil type']

if 'Station no' in joined:
    joined.loc[joined['Station no'].notnull(), 'BSRN Station no'] = joined.loc[
        joined['Station no'].notnull(), 'Station no']
    del joined['Station no']

insert_station_no_fill = False
if 'BSRN Station no' in joined:
    try:
        joined['BSRN Station no'] = joined['BSRN Station no'].astype(int)
    except ValueError:
        joined['BSRN Station no'] = joined['BSRN Station no'].fillna(
            '-9999').astype(int)
        insert_station_no_fill = True


joined.index = pd.Index(np.arange(1, len(joined) + 1), name='stationid')
ds = joined.to_xarray()

for key in list(ds.variables):
    var = ds[key]
    var.attrs['bsrn_name'] = key
    try:
        row = mapping.loc[key]
    except KeyError:
        pass
    else:
        for attr, val in row[row.notnull()].items():
            if attr != 'name':
                var.attrs[attr] = val
        ds = ds.rename({key: row['name']})

new_names = {}

patt = re.compile(r'[_\/\-\s]')
for key, var in ds.variables.items():
    if var.attrs['bsrn_name'] not in df:
        var.attrs['derived_from'] = 'Comment'
    else:
        var.attrs['derived_from'] = var.attrs['bsrn_name']
    new_names[key] = patt.sub('_', key.lower()).replace(',', '')


if insert_station_no_fill:
    ds['BSRN Station no'].encoding['_FillValue'] = -9999



ds = ds.rename(new_names)
ds.to_netcdf('BSRN_Metadaten_Geographie.nc')