dump_sqlite_recorder_file.py

from __future__ import print_function
import pprint
import sqlite3
import sys
from six import PY2, PY3, iteritems

from openmdao.recorders.sqlite_recorder import blob_to_array

if PY2:
    import cPickle as pickle
if PY3:
    import pickle

indent = 4 * ' '


def print_scalar(name, value):
    print(indent, name + ':', value)


def print_header(title, sep):
    print(60 * sep)
    print(title)
    print(60 * sep)


def print_blob(name, blob):
    print(indent, name + ':')
    array = blob_to_array(blob)

    if array.dtype.names:
        for varname in array[0].dtype.names:
            print( indent, indent, varname, array[0][varname] )
    else:
        print(indent, indent, 'None')
    print()


def print_counter(idx, counter):
    print(indent, 'idx: {} counter: {}'.format(idx, counter))

filename = sys.argv[1]

con = sqlite3.connect(filename)
cur = con.cursor()

from six import PY2, PY3

if PY2:
    def pickle_load(pickled_item):
        return pickle.loads(str(pickled_item))
if PY3:
    def pickle_load(pickled_item):
        return pickle.loads(pickled_item)

# Driver metadata
print_header('Driver Metadata', '=')
cur.execute("SELECT model_viewer_data FROM driver_metadata")
for row in cur:
    driver_metadata = pickle_load(row[0])
    print('driver_metadata')
    pprint.pprint(driver_metadata, indent=4)


def print_scaling_factors(scaling_factors, in_out, linear_type):
    """
    Print the names and values of all variables in this vector, one per line.
    """
    if linear_type in scaling_factors[in_out]:
        vector = scaling_factors[in_out][linear_type]
        print(indent, in_out, linear_type)
        if vector:
            for abs_name, view in iteritems(vector._views):
                print(2 * indent, abs_name, view)
        else:
            print(2 * indent, 'None')
        print()

print_header('System Metadata', '=')
cur.execute("SELECT id, scaling_factors FROM system_metadata")
for row in cur:
    id = row[0]
    scaling_factors = pickle_load(row[1])
    print('id = ', id)
    print('scaling_factors')
    for in_out in ['input', 'output', 'residual']:
        for linear_type in ['linear', 'nonlinear']:
            print_scaling_factors(scaling_factors, in_out, linear_type)

print_header('Solver Metadata', '=')
cur.execute("SELECT id, solver_options, solver_class FROM solver_metadata")
for row in cur:
    id = row[0]
    solver_options = pickle_load(row[1])._dict
    solver_class = row[2]
    print('id = ', id)
    print('solver_options = ')
    pprint.pprint(solver_options, indent=16)
    print('solver_class', solver_class)

#  Driver recordings: inputs, outputs, residuals
print_header('Driver Iterations', '=')
cur.execute("SELECT * FROM driver_iterations")
rows = cur.fetchall()

for row in rows:
    idx, counter, iteration_coordinate, timestamp, success, msg, desvars_blob, responses_blob, objectives_blob, \
        constraints_blob, sysincludes_blob = row
    print_header( 'Coord: {}'.format(iteration_coordinate), '-')
    print_counter(idx, counter)
    print_blob('Desvars', desvars_blob )
    print_blob('Responses', responses_blob )
    print_blob('Objectives', objectives_blob )
    print_blob('Constraints', constraints_blob)
    print_blob('Sys Includes', sysincludes_blob)

# Print System recordings: inputs, outputs, residuals
print_header('System Iterations', '=')
cur.execute("SELECT * FROM system_iterations")
rows = cur.fetchall()

for row in rows:
    idx, counter, iteration_coordinate, timestamp, success, msg, inputs_blob , outputs_blob , residuals_blob = row
    print_header('Coord: {}'.format(iteration_coordinate), '-')
    print_counter(idx, counter)
    print_blob('Inputs', inputs_blob )
    print_blob('Outputs', outputs_blob )
    print_blob('Residuals', residuals_blob )

# Print Solver recordings: inputs, outputs, residuals
print_header('Solver Iterations', '=')
cur.execute("SELECT * FROM solver_iterations")
rows = cur.fetchall()

for row in rows:
    idx, counter, iteration_coordinate, timestamp, success, msg, abs_err, rel_err, outputs_blob, residuals_blob = row
    print_header('Coord: {}'.format(iteration_coordinate), '-')
    print_counter(idx, counter)
    print_scalar('abs_err', abs_err )
    print_scalar('rel_err', rel_err )
    print_blob('Outputs', outputs_blob )
    print_blob('Residuals', residuals_blob )

con.close()