10927734534

Committed 18 Sep 2024 06:09PM UTC coverage: 79.235% (-0.4%) from 79.668%

Build # 10927734534

Build Type

Pull #315

github

Committed by

web-flow

Commit Message

Merge 397ba5241 into f779fa594

Pull Request Pull Request #315: WEIS 1.3.1

Run Details

21 of 180 new or added lines in 4 files covered. (11.67%)

8 existing lines in 4 files now uncovered.

21647 of 27320 relevant lines covered (79.23%)

0.79 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/weis/visualization/utils.py

'''
Various functions for help visualizing WEIS outputs
'''
from weis.aeroelasticse.FileTools import load_yaml
import pandas as pd
import numpy as np
import openmdao.api as om
import glob
import json
import multiprocessing as mp
import plotly.graph_objects as go
import os
import io
import yaml
import re
import socket
from dash import html
from matplotlib.gridspec import GridSpec
import matplotlib.pyplot as plt
import pickle
import raft
from raft.helpers import *
from weis.dtqpy import objective

try:
    import ruamel_yaml as ry
except Exception:
    try:
        import ruamel.yaml as ry
    except Exception:
        raise ImportError('No module named ruamel.yaml or ruamel_yaml')


def checkPort(port, host="0.0.0.0"):
    # check port availability and then close the socket
    sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
    result = False
    try:
        sock.bind((host, port))
        result = True
    except:
        result = False

    sock.close()
    return result


def parse_yaml(file_path):
    '''
    Parse the data contents in dictionary format
    '''
    # print('Reading the input yaml file..')
    try:
        with io.open(file_path, 'r') as stream:
            dict = yaml.safe_load(stream)

        dict['yamlPath'] = file_path
        # print('input file dict:\n', dict)
        return dict

    except FileNotFoundError:
        print('Could not locate the input yaml file..')
        exit()

    except Exception as e:
        print(e)
        exit()


def dict_to_html(data, out_html_list, level):
    '''
    Show the nested dictionary data to html
    '''

    for k1, v1 in data.items():
        if not k1 in ['dirs', 'files']:
            if not isinstance(v1, list) and not isinstance(v1, dict):
                out_html_list.append(html.H6(f'{"---"*level}{k1}: {v1}'))
                continue

            out_html_list.append(html.H6(f'{"---"*level}{k1}'))

        if isinstance(v1, list):
            out_html_list.append(html.Div([
                                    html.H6(f'{"---"*(level+1)}{i}') for i in v1]))

        elif isinstance(v1, dict):
            out_html_list = dict_to_html(v1, out_html_list, level+1)


    return out_html_list


def read_cm(cm_file):
    """
    Function originally from:
    https://github.com/WISDEM/WEIS/blob/main/examples/16_postprocessing/rev_DLCs_WEIS.ipynb

    Parameters
    __________
    cm_file : The file path for case matrix

    Returns
    _______
    cm : The dataframe of case matrix
    dlc_inds : The indices dictionary indicating where corresponding dlc is used for each run
    """
    cm_dict = load_yaml(cm_file, package=1)
    cnames = []
    for c in list(cm_dict.keys()):
        if isinstance(c, ry.comments.CommentedKeySeq):
            cnames.append(tuple(c))
        else:
            cnames.append(c)
    cm = pd.DataFrame(cm_dict, columns=cnames)

    return cm

def parse_contents(data):
    """
    Function from:
    https://github.com/WISDEM/WEIS/blob/main/examples/09_design_of_experiments/postprocess_results.py
    """
    collected_data = {}
    for key in data.keys():
        if key not in collected_data.keys():
            collected_data[key] = []

        for key_idx, _ in enumerate(data[key]):
            if isinstance(data[key][key_idx], int):
                collected_data[key].append(np.array(data[key][key_idx]))
            elif len(data[key][key_idx]) == 1:
                try:
                    collected_data[key].append(np.array(data[key][key_idx][0]))
                except:
                    collected_data[key].append(np.array(data[key][key_idx]))
            else:
                collected_data[key].append(np.array(data[key][key_idx]))

    df = pd.DataFrame.from_dict(collected_data)

    return df


def load_vars_file(fn_vars):
    """
    load a json file of problem variables as output from WEIS (as problem_vars.json)

    parameters:
    -----------
    fn_vars: str
        a filename to read

    returns:
    --------
    vars : dict[dict]
        a dictionary of dictionaries holding the problem_vars from WEIS
    """

    rawvars = load_yaml(fn_vars)
    vars = {}
    for k, v in rawvars.items():
        for (_, v2) in v:
            for k3, v3 in v2.items():
                if k3 in ["lower", "upper"]:
                    v2[k3] = float(v3)
                if k3 == "val":
                    v2[k3] = np.array(v3)
        vars[k] = dict(v)
    return vars


def compare_om_data(
    dataOM_1,
    dataOM_2,
    fn_1="data 1",
    fn_2="data 2",
    verbose=False,
):
    """
    compare openmdao data dictionaries to find the in-common (and not) keys

    args:
        dataOM_1: dict
            an openmdao data dictionary
        dataOM_2: dict
            an openmdao data dictionary
        fn_1: str (optional)
            display name for the first data dictionary
        fn_2: str (optional)
            display name for the second data dictionary
        verbose : bool (optional, default: False)
            if we want to print what's happening

    returns:
        keys_all: set
            intersection (i.e. common) keys between the two OM data dictionaries
        diff_keys_12: set
            directional difference of keys between first and second OM data dicts
        diff_keys_21: set
            directional difference of keys between second and first OM data dicts
    """

    diff_keys_12 = set(dataOM_1).difference(dataOM_2)
    diff_keys_21 = set(dataOM_2).difference(dataOM_1)
    if len(diff_keys_12):
        if verbose:
            print(f"the following keys are only in {fn_1}:")
    for key_m in diff_keys_12:
        if verbose:
            print(f"\t{key_m}")
    if len(diff_keys_21):
        if verbose:
            print(f"the following keys are only in {fn_2}:")
    for key_m in diff_keys_21:
        if verbose:
            print(f"\t{key_m}")
    keys_all = set(dataOM_1).intersection(dataOM_2)
    if verbose:
        print(f"the following keys are in both {fn_1} and {fn_2}:")
    for key_m in keys_all:
        if verbose:
            print(f"\t{key_m}")

    return keys_all, diff_keys_12, diff_keys_21

def load_OMsql(
    log,
    parse_multi=False,
    meta=None,
    verbose=False,
):
    """
    load the openmdao sql file produced by a WEIS run into a dictionary

    parameters:
    -----------
        log : str
            filename of the log sql database that should be loaded
        parse_multi : bool
            switch to turn on rank/iteration parsing and storage
        meta : str
            filename of the meta log sql database that should be loaded
        verbose : bool (optional, default: False)
            if we want to print what's happening

    returns:
        rec_data: dict
            dictionary of the data recorded by openMDAO

    """

    # heads-up print
    if verbose:
        print(f"loading {log}")

    # create an openmdao reader for recorded output data
    cr = om.CaseReader(log, metadata_filename=meta)

    # create a dict for output data that's been recorded
    rec_data = {}
    # loop over the cases
    for case in cr.get_cases("driver"):
        if parse_multi:
            rankNo = case.name.split(":")[0]
            assert rankNo.startswith("rank")
            rankNo = int(rankNo[4:])
            iterNo = int(case.name.split("|")[-1])

        # for each key in the outputs
        for key in case.outputs.keys():

            if key not in rec_data:
                # if this key isn't present, create a new list
                rec_data[key] = []

            if hasattr(case[key], '__len__') and len(case[key]) != 1:
                # convert to a numpy array if possible and add the data to the list
                rec_data[key].append(np.array(case[key]))
            else:
                rec_data[key].append(case[key])

        if parse_multi:
            # add rank/iter metadata
            for key in ["rank", "iter"]:  # for each key in the outputs
                if key not in rec_data:  # if this key isn't present, create a new list
                    rec_data[key] = []
            rec_data["rank"].append(rankNo)
            rec_data["iter"].append(iterNo)

    return rec_data  # return the output


def load_OMsql_multi(
    log_fmt,
    meta_in=None,
    process_multi = True,
    verbose=False,
):
    """
    load the multi-processor openmdao sql files produced by WEIS into a dict

    parameters:
    -----------
        log_fmt : str
            format string for the process-wise WEIS/OM log files
        meta_in : str (optional, default: None)
            filename string of the meta log file (will override automatic discovery)
        post_multi : bool (optional, default: True)
            postprocess in parallel using the multiprocessing library
        verbose : bool (optional, default: False)
            if we want to print what's happening

    returns:
    --------
        data_dict : dict
            dictionary of all the datapoints extracted from the WEIS/OM log files
    """

    # use glob to find the logs that match the format string
    opt_logs = sorted(
        glob.glob(log_fmt),
        key = lambda v : int(v.split("_")[-1])
            if (v.split("_")[-1] != "meta")
            else 1e8,
    )
    if len(opt_logs) < 1:
        raise FileExistsError("No output logs to postprocess!")

    # remove the "meta" log from the collection
    meta_found = None
    for idx, log in enumerate(opt_logs):
        if "meta" in log:
            meta_found = log  # save the meta file
            opt_logs.pop(idx)  # remove the meta log from the list
            break

    # handle meta logfile discovery... not sure what it actually does
    if meta_in is not None:
        meta = meta_in  # if a meta is given, override
    elif meta_found is not None:
        meta = meta_found  # if a meta is not given but one is found, use that
    else:
        meta = None  # otherwise, run without a meta

    # extract the ranks from the sql files
    sql_ranks = [ol.split("_")[-1] for ol in opt_logs]

    # run multiprocessing
    if process_multi:
        cores = mp.cpu_count()
        pool = mp.Pool(min(len(opt_logs), cores))

        # load sql file
        outdata = pool.starmap(load_OMsql, [(log, True, meta, verbose) for log in opt_logs])
        pool.close()
        pool.join()
    else: # no multiprocessing
        outdata = [load_OMsql(log, parse_multi=True, verbose=verbose, meta=meta) for log in opt_logs]

    # create a dictionary and turn it into a dataframe for convenience
    collected_data = {}
    ndarray_keys = []
    for sql_rank, data in zip(sql_ranks, outdata):
        for key in data.keys():
            if key not in collected_data.keys():
                collected_data[key] = []
            if key == "rank": # adjust the rank based on sql file rank
                data[key] = [int(sql_rank) for _ in data[key]]
            for idx_key, _ in enumerate(data[key]):
                if isinstance(data[key][idx_key], int):
                    collected_data[key].append(int(np.array(data[key][idx_key])))
                elif isinstance(data[key][idx_key], float):
                    collected_data[key].append(float(np.array(data[key][idx_key])))
                elif len(data[key][idx_key]) == 1:
                    collected_data[key].append(float(np.array(data[key][idx_key])))
                    # try:
                    #     collected_data[key].append(np.array(data[key][idx_key][0]))
                    # except:
                    #     collected_data[key].append(np.array(data[key][idx_key]))
                else:
                    collected_data[key].append(np.array(data[key][idx_key]).tolist())
                    ndarray_keys.append(key)
    df = pd.DataFrame(collected_data)

    # return a dictionary of the data that was extracted
    return df.to_dict(orient="list")


def consolidate_multi(
    dataOMmulti,
    vars_dict,
    feas_tol=1e-5,
):
    """
    load the multi-processor openmdao sql files and squash them to the
    per-iteration best-feasible result

    parameters:
    -----------
        dataOMmulti : dict
            dictionary of all the datapoints extracted from the multiprocess
            WEIS/OM log files
        vars_dict:
            experiment design variables to be analyzed
        feas_tol : float (optional)
            tolerance for feasibility analysis
    returns:
    --------
        dataOMbest_DE : dict
            dictionary of the per-iteration best-feasible simulations
    """
    objective_name = list(vars_dict["objectives"].values())[0]["name"]

    dfOMmulti = pd.DataFrame(dataOMmulti)
    tfeas, cfeas = get_feasible_iterations(dataOMmulti, vars_dict, feas_tol=feas_tol)

    dfOMmulti = dfOMmulti[tfeas].reset_index()

    dataOMbest_DE = dfOMmulti.groupby("iter").apply(
        lambda grp : grp.loc[grp[objective_name].idxmin()],
        include_groups=False,
    ).to_dict()

    for key in dataOMbest_DE.keys():
        dataOMbest_DE[key] = np.array(list(dataOMbest_DE[key].values()))

    return dataOMbest_DE


def get_feasible_iterations(
    dataOM,
    vars_dict,
    feas_tol=1e-5,
):
    """
    get iteration-wise total and per-constraint feasibility from an experiment

    args:
        dataOM: dict
            openmdao data dictionary
        vars_dict:
            experiment design variables for checking

    returns:
        total_feasibility: np.ndarray[bool]
            iteration-wise total feasibility indications
        feasibility_constraintwise: dict[np.ndarray[bool]]
            dictionary to map from constraint names to iteration-wise feasibility indications for that constraint
    """

    # assert len(vars_dict["objectives"].values()) == 1, "can't handle multi-objective... yet. -cfrontin"
    objective_name = list(vars_dict["objectives"].values())[0]["name"]

    feasibility_constraintwise = dict()
    total_feasibility = np.ones_like(np.array(dataOM[objective_name]).reshape(-1,1), dtype=bool)
    for k, v in vars_dict["constraints"].items():
        feasibility = np.ones_like(dataOM[objective_name], dtype=bool).reshape(-1, 1)
        values = np.array(dataOM[v["name"]])
        if len(values.shape) == 1:
            values = values.reshape(-1,1)
        if v.get("upper") is not None:
            feasibility = np.logical_and(feasibility, np.all(np.less_equal(values, (1+feas_tol)*v["upper"]), axis=1).reshape(-1, 1))
        if v.get("lower") is not None:
            feasibility = np.logical_and(feasibility, np.all(np.greater_equal(values, (1-feas_tol)*v["lower"]), axis=1).reshape(-1, 1))
        feasibility_constraintwise[v["name"]] = feasibility
        total_feasibility = np.logical_and(total_feasibility, feasibility)
    return total_feasibility, feasibility_constraintwise


def verify_vars(
    vars_1,
    *vars_i,
):
    """
    verifies format of DVs, constraints, objective variable file
    guarantees a list of experiments has the same variables
    adjusts unbounded constraints
    returns verified list of vars
    """

    for vars_2 in vars_i:
        if vars_2 is not None:
            for k0 in set(vars_1.keys()).union(vars_2):
                assert k0 in vars_1
                assert k0 in vars_2
                for k1 in set(vars_1[k0].keys()).union(vars_2[k0].keys()):
                    assert k1 in vars_1[k0]
                    assert k1 in vars_2[k0]
                    for k2 in set(vars_1[k0][k1].keys()).union(vars_2[k0][k1].keys()):
                        assert k2 in vars_1[k0][k1]
                        assert k2 in vars_2[k0][k1]
                        if k2 == "val":
                            continue
                        if isinstance(vars_1[k0][k1][k2], str):
                            assert vars_1[k0][k1][k2] == vars_2[k0][k1][k2]
                        elif vars_1[k0][k1][k2] is not None:
                            assert np.all(np.isclose(vars_1[k0][k1][k2], vars_2[k0][k1][k2]))
                        else:
                            assert (vars_1[k0][k1][k2] is None) and (vars_2[k0][k1][k2] is None)

    vars_unified = vars_1.copy()
    for k0 in vars_unified.keys():
        for k1 in vars_unified[k0].keys():
            if (vars_unified[k0][k1].get("lower") is not None) and (vars_unified[k0][k1].get("lower") < -1e28):
                vars_unified[k0][k1]["lower"] = -np.inf
            if (vars_unified[k0][k1].get("upper") is not None) and (vars_unified[k0][k1].get("upper") > 1e28):
                vars_unified[k0][k1]["upper"] = np.inf

    return vars_unified


def prettyprint_variables(
    keys_all,
    keys_obj,
    keys_DV,
    keys_constr,
):
    """
    print the variables we have with a prefix showing whether they are an
    objective variable (**), design variabie (--), constraint (<>), or unknown
    (??)
    """

    # print them nicely
    print()
    [
        print(
            f"** {key}"
            if key in keys_obj
            else f"-- {key}" if key in keys_DV else f"<> {key}" if key in keys_constr else f"?? {key}"
        )
        for key in keys_all
    ]
    print()

def read_per_iteration(iteration, stats_paths):

    stats_path_matched = [x for x in stats_paths if f'iteration_{iteration}' in x][0]
    iteration_path = '/'.join(stats_path_matched.split('/')[:-1])
    stats = pd.read_pickle(stats_path_matched)
    # dels = pd.read_pickle(iteration_path+'/DELs.p')
    # fst_vt = pd.read_pickle(iteration_path+'/fst_vt.p')
    print('iteration path with ', iteration, ': ', stats_path_matched)

    return stats, iteration_path


def get_timeseries_data(run_num, stats, iteration_path):

    stats = stats.reset_index()     # make 'index' column that has elements of 'IEA_22_Semi_00, ...'
    filename = stats.loc[run_num, 'index'].to_string()      # filenames are not same - stats: IEA_22_Semi_83 / timeseries/: IEA_22_Semi_0_83.p
    if filename.split('_')[-1].startswith('0'):
        filename = ('_'.join(filename.split('_')[:-1])+'_0_'+filename.split('_')[-1][1:]+'.p').strip()
    else:
        filename = ('_'.join(filename.split('_')[:-1])+'_0_'+filename.split('_')[-1]+'.p').strip()

    # visualization_demo/openfast_runs/rank_0/iteration_0/timeseries/IEA_22_Semi_0_0.p
    timeseries_path = '/'.join([iteration_path, 'timeseries', filename])
    timeseries_data = pd.read_pickle(timeseries_path)

    return filename, timeseries_data


def empty_figure():
    '''
    Draw empty figure showing nothing once initialized
    '''
    fig = go.Figure(go.Scatter(x=[], y=[]))
    fig.update_layout(template=None)
    fig.update_xaxes(showgrid=False, showticklabels=False, zeroline=False)
    fig.update_yaxes(showgrid=False, showticklabels=False, zeroline=False)

    return fig


def toggle(click, is_open):
    if click:
        return not is_open
    return is_open


def store_dataframes(var_files):
    dfs = []
    for idx, file_path in var_files.items():
        if file_path == 'None':
            dfs.append({idx: []})
            continue
        df = pd.read_csv(file_path, skiprows=[0,1,2,3,4,5,7], sep='\s+')
        dfs.append({idx: df.to_dict('records')})

    return dfs


def get_file_info(file_path):
    file_name = file_path.split('/')[-1]
    file_abs_path = os.path.abspath(file_path)
    file_size = round(os.path.getsize(file_path) / (1024**2), 2)
    creation_time = os.path.getctime(file_path)
    modification_time = os.path.getmtime(file_path)

    file_info = {
        'file_name': file_name,
        'file_abs_path': file_abs_path,
        'file_size': file_size,
        'creation_time': creation_time,
        'modification_time': modification_time
    }

    return file_info


def find_file_path_from_tree(nested_dict, filename, prepath=()):
    # Works for multi-keyed files
    # Sample outputs: ('outputDirStructure', 'sample_test') ('outputDirStructure', 'sample_multi')
    for k, v in nested_dict.items():
        path = prepath + (k,)
        if v == filename:
            yield path + (v, )
        elif isinstance(v, list) and filename in v:
            yield path + (filename, )
        elif hasattr(v, 'items'):
            yield from find_file_path_from_tree(v, filename, path)

def find_iterations(nested_dict, prepath=()):
    for k, v in nested_dict.items():
        path = prepath + (k,)
        if 'iteration' in k:
            yield int(re.findall(r'\d+', k)[0])
        elif hasattr(v, 'items'):
            yield from find_iterations(v, path)


def update_yaml(input_dict, yaml_filepath):
    with open(yaml_filepath, 'w') as outfile:
        yaml.dump(input_dict, outfile, default_flow_style=False)


def read_cost_variables(labels, refturb_variables):
    # Read tcc cost-related variables from CSV file

    cost_matrix = [['Main Turbine Components', 'Cost']]

    for l in labels:
        cost_matrix.append([l, eval(refturb_variables[f'tcc.{l}_cost']['values'])[0]])

    return cost_matrix


def generate_raft_img(raft_design_dir, plot_dir, log_data):
    '''
    Temporary function to visualize raft 3d plot using matplotlib.
    TODO: to build interactive 3d plot using plotly
    '''
    n_plots = len(os.listdir(raft_design_dir))
    print('n_plots: ', n_plots)
    os.makedirs(plot_dir,exist_ok=True)

    opt_outs = {}
    opt_outs['max_pitch'] = np.squeeze(np.array(log_data['raft.Max_PtfmPitch']))

    for i_plot in range(n_plots):
        # Set up subplots
        fig = plt.figure()
        fig.patch.set_facecolor('white')
        ax = plt.axes(projection='3d')

        with open(os.path.join(raft_design_dir,f'raft_design_{i_plot}.pkl'),'rb') as f:
            design = pickle.load(f)

        # TODO: Found typo on gamma value at 1_raft_opt example
        if design['turbine']['tower']['gamma'] == np.array([0.]):
            design['turbine']['tower']['gamma'] = 0.0       # Change it from array([0.])

        # set up the model
        model1 = raft.Model(design)
        model1.analyzeUnloaded(
            ballast= False,
            heave_tol = 1.0
            )

        model1.fowtList[0].r6[4] = np.radians(opt_outs['max_pitch'][i_plot])

        _, ax = model1.plot(ax=ax)

        ax.azim = -88.63636363636361
        ax.elev = 27.662337662337674
        ax.set_xlim3d((-110.90447789470043, 102.92063063344857))
        ax.set_ylim3d((64.47420067304586, 311.37818252335893))
        ax.set_zlim3d((-88.43591080818854, -57.499893019459606))

        image_filename = os.path.join(plot_dir,f'ptfm_{i_plot}.png')
        plt.savefig(image_filename, bbox_inches='tight')
        print('saved ', image_filename)

WISDEM / WEIS / 10927734534

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous