metametrik2sql

#!/usr/bin/env python

import json
import psycopg2
import sys

def tuple_from_frag(frag, keys):
    deref = lambda i: frag[i]
    return tuple(map(deref, keys))

def get_var(frag):
    return tuple_from_frag(
        frag, 
        ["variable_name", "variable_description"]
        )

def get_model(frag):
    return tuple_from_frag(
        frag,
        ["model_type", "sample_size", "model_name", "goodness_of_fit"]
        )

def get_effect(frag):
    return tuple_from_frag(
        frag,
        ["independent_variable", "model_name", "standard_error", "significance"]
        )

def check_effect(e, models, ivs):
    iv_names = [iv[0] for iv in ivs]
    model_names = [m[2] for m in models]

    assert e[0] in iv_names
    assert e[1] in model_names

def save(dv, ivs, models, effects):
    db = psycopg2.connect(database="metametrik")

    c = db.cursor()

    def save_var(var):
        sql = 'insert into var (name, description) values (%s, %s) returning var_id'
        c.execute(sql, var)
        return (var[0], c.fetchone()[0])

    def save_result_set(dv_id):
        sql = 'insert into result_set (dependent_var) values (%s) returning rs_id'
        c.execute(sql, (dv_id,))
        return c.fetchone()[0]

    def save_model(model):
        # fixme
        x = (model[2], model[1], model[2], model[3])
        sql = 'insert into model (name, sample_size, description, r_squared) values (%s, %s, %s, %s) returning model_id'
        c.execute(sql, x)
        return (model[2], c.fetchone()[0])

    def save_effect(e):
        sql = 'insert into effect (model_id, var_id, coefficient, standard_error) values (%s, %s, %s, %s)'
        c.execute(sql, e)

    _, dv_id = save_var(dv)
    save_result_set(dv_id)

    variables = dict([ save_var(var) for var in ivs ])

    model_data = dict([ save_model(m) for m in models ])

    for effect in effects:
        e = (model_data[effect[1]], variables[effect[0]], effect[2], effect[3])
        save_effect(e)

    

    db.commit()

def run():
    data = json.load(sys.stdin)["regression_result_set"]

    dv = get_var(data["dependent_variable"])

    models = [ get_model(m) for m in data["models"] ]

    ivs = [ get_var(iv) for iv in data["independent_variables"] ]

    effects = [ get_effect(e) for e in data["effects"] ]
    [ check_effect(e, models, ivs) for e in effects ]

    save(dv, ivs, models, effects)

if __name__ == '__main__':
    run()