import shutil

import attr

import pandas as pd

from aochildesnouns.co_occurrence import collect_left_and_right_co_occurrences

from aochildesnouns.targets import make_targets

from aochildesnouns import configs

from aochildesnouns.measure import measure_dvs

from aochildesnouns.pre_processing import prepare_data

from aochildesnouns.params import Conditions

if configs.Dirs.tmp.exists():

shutil.rmtree(configs.Dirs.tmp)

configs.Dirs.tmp.mkdir()

df_rows = []

row_id = 0

for params in Conditions.all(): # each param holds information about IVs in a single condition

# load spacy-processed data to analyze, and targets

age2doc = prepare_data(params)

targets_exp, targets_ctl = make_targets(params, age2doc)

# for each target condition (experimental vs control)

for targets_control, targets in zip([False, True],

[targets_exp, targets_ctl]):

# for each age

for age, doc in sorted(age2doc.items(), key=lambda i: i[0]):

# for each direction (left, right, both)

for direction in configs.Conditions.directions:

# update + print IV realizations in current condition

print()

params = attr.evolve(params,

age=age,

targets_control=targets_control,

direction=direction,

)

print(params)

# get co-occurrence data

# warning: do not call this function until attr.evolve(params) has been called

co_data = collect_left_and_right_co_occurrences(doc, targets, params)

# measure Dvs

data: dict = measure_dvs(params, co_data) # a full set of dvs for a single condition

row_id += 1

# add info about condition to df row - don't use magic methods which are uninfluenced attr.evolve()

ivs = {ivn: attr.asdict(params)[ivn] for ivn in Conditions.ivs}

data.update(ivs)

row = pd.DataFrame(data=data, index=[row_id])

# collect each row here

df_rows.append(row)

df = pd.concat(df_rows, axis=0)

df.to_csv(configs.Dirs.results / 'results.csv', index=False)