generate_csv.py

import csv
import os
import matplotlib.pyplot as plt
import tkinter as tk

class CEvent_Categorical:
    def __init__(self, label, onset, offset):
        self.label = label
        self.onset = onset
        self.offset = offset


def generate_csv(input_file):
    """ Takes an input CSV file generated by datavyu and generates a formatted CSV file
    Args:
        input_file::str
                Path to CSV file generated by datavyu
    Returns:
        None
    """
    output_rows = []
    output_filename = input_file.split(".")[0] + "_FORMATTED.csv"

    # Open a datavyu CSV file for reading
    with open(input_file, "r") as csv_file:
        reader = csv.reader(csv_file, delimiter=",")
        labels = next(reader)
        names = []
        index = 0
        while(labels[index] != ""):
                names.append(labels[index].split(".")[0])
                # Increment by 4 in order to get to next name (there are 4 columns per name)
                index += 4
        for row in reader:
                # Set index to 1 because we're looking at the 'onset time' field for each name
                index = 1
                while(index < len(row)):
                        if(row[index] != ""):
                                # Divide by 1000 to get data in seconds
                                # Divide index for 'names' list by 4 to get appropriate name for data
                                output_row = [float(row[index])/1000, float(row[index+1])/1000, names[index/4]]
                                output_rows.append(output_row)
                        index += 4
        # Sort the rows of the CSV file by 'onset time' field
        output_rows = sorted(output_rows, key=lambda row: row[1])

    # Open a CSV file to write formatted data 
    with open(output_filename, "wb") as output_csv_file:
        writer = csv.writer(output_csv_file, delimiter=",")
        writer.writerows(output_rows)


def generate_csvs(directory):
        """ Takes a directory of subjects and generates a formatted CSV file for each datavyu CSV file
        Args:
            directory::str
                Path to directory of subject directories
        Returns:
            None
        """
        for subject in os.listdir(directory):
                try:
                        path = "{0}\\{1}\\Data\\Activity Coding\\{2}_EXP.csv".format(directory, subject, subject.upper())
                        if(os.path.exists(path)):
                                generate_csv(path)
                except IOError as e:
                        print("Error opening CSV file: " + str(e))

def generate_plot(subject, formatted_file, title, timepoints=None, objects=None, fixed=False, activity=True):
        """ Generates plot for specified subject, C_Event file, time range, and objects
        Args:
                subject::str
                        Subject whose data is being examined
                formatted_file::str
                        Path to C_Event file in CSV format for given subject
                timepoints::tuple
                        Tuple of float values where timepoints[0] is the start point and timepoints[1] is the end point
                objects::list
                        List of strings of objects to include in plot.  If None is given, all objects are included
        Returns:
                None
        """
        labels = set()
        label_indices = dict()
        c_events = []
        colors = ["#ff0000", "#ffd700", "#40e0d0", "#0000ff", "#00ff00", "#ffc0cb", "#468499", "#800080",
                "#808080", "#000000", "#ffc3a0", "#ff7f50", "#bada55", "#afeeee", "#cbbeb5", "#53454C",
                "#01B9FE", "#1C6406", "#98B686", "#B6C050", "#C87F35", "#2B106A", "#D330BC"]

        if(fixed):
                fixed_labels =  ["BB", "BERT", "BK_CLR", "BK_LS", "CAR", "COW_B", "COW_W", "DOG",
                        "DOLL", "EGG", "FOOT", "HAND", "HAT", "OTH", "PENG", "PIG_L", "PIG_S", "RING",
                        "RTL", "SHEEP_L", "SHEEP_S", "TRUCK", "MULTI"]

                labels = set(fixed_labels)
                for i, label in enumerate(fixed_labels):
                        label_indices[label] = i

        with open(formatted_file, "r") as csvfile:
                reader = csv.reader(csvfile, delimiter=",")         
                for row in reader:
                        if((not objects) or (objects and row[2] in objects)):
                                c_event = CEvent_Categorical(row[2], float(row[0]), float(row[1]))
                                c_events.append(c_event)
                                if(not fixed):
                                        labels.add(c_event.label)
                                        if(not c_event.label in label_indices):
                                                label_indices[c_event.label] = len(labels)-1


        fig = plt.figure()
        fig.set_size_inches(18.5, 5.5, forward=True)
        ax = fig.add_subplot(111)
        
        label_positions = [0.1*i for i in label_indices.values()]
        plt.yticks(label_positions, labels)

        plt.title(title)

        plt.xlabel("Time (s)")
        if(timepoints):
                plt.xlim(timepoints[0], timepoints[1])

        for point in c_events:
                ax.barh(0.1*label_indices[point.label], point.offset-point.onset, 0.05, 
                        left=point.onset, align="center", color=colors[label_indices[point.label]])
        plt.show()


def plot(subject, start_time=None, end_time=None, objects_list="", fixed=False, variable="Activity"):
        """ Wrapper function for generate_plot().  Parses input and calls generate_plot()
        Args:
                subject::str
                        Subject whose data is being examined
                start_time::str
                        String that represents start time for plot.  Cast to float
                end_time::str
                        String that represents end time for plot.  Cast to float
                objects_list::str
                        String of comma-separated objects.  If only whitespace is given, all objects are included
        Returns:
                None
        """
        # Testing for whitespace-only data in objects_list
        objects_list = objects_list.rstrip()
        if(objects_list == ""):
                objects = None
        else:
                objects = objects_list.split(",")

        if(variable == "Activity"):
                path = "Q:\\PSY-LAB\\Suanda\\Rollins\\9mos\\{0}\\Data\\Activity Coding\\{1}_EXP_FORMATTED.csv".format(subject, subject.upper())
                title = "Activity Coding - " + subject
        elif(variable == "Reference"):
                path = "Q:\\PSY-LAB\\Suanda\\Rollins\\9mos\\{0}\\Data\\Reference Coding\\{1}_cevent_ref.csv".format(subject, subject.upper())
                title = "Reference Coding - " + subject
        else:
                return False

        # Try to specify timepoints using start_time and end_time, exits "try" block if either start_time or end_time weren't specified/aren't proper numbers
        try:
                generate_plot(subject,
                        path,
                        title,
                        (float(start_time), float(end_time)),
                        objects,
                        fixed=fixed)
        except ValueError:
                generate_plot(subject,
                        path,
                        title,
                        None,
                        objects,
                        fixed=fixed)


def main():
        window = tk.Tk()
        window.title("Plotting")
        window.geometry("350x250")

        tk.Label(window, text="Subject ID").grid(row=0)
        subject = tk.Entry(window, width=20)
        subject.grid(row=0, column=1)

        tk.Label(window, text="Start Time").grid(row=1)
        start_time = tk.Entry(window, width=20)
        start_time.grid(row=1, column=1)

        tk.Label(window, text="End Time").grid(row=2)
        end_time = tk.Entry(window, width=20)
        end_time.grid(row=2, column=1)

        tk.Label(window, text="Objects (comma-separated)").grid(row=3)
        objects = tk.Text(window, height=3, width=20)
        objects.grid(row=3, column=1)

        choices = {"Auto", "Fixed"}
        var = tk.StringVar(window)
        var.set("Auto")
        tk.Label(window, text="Label Ordering").grid(row=4)
        dropdown = tk.OptionMenu(window, var, *choices)
        dropdown.grid(row=4, column=1)

        types = {"Activity", "Reference"}
        types_var = tk.StringVar(window)
        types_var.set("Activity")
        tk.Label(window, text="Variable").grid(row=5)
        types_dropdown = tk.OptionMenu(window, types_var, *types)
        types_dropdown.grid(row=5, column=1)

        padding = tk.Frame(window, height=25)
        padding.grid(row=6)

        button = tk.Button(window, text="Plot", command=lambda: plot(subject.get(), start_time.get(), end_time.get(), objects.get("1.0", tk.END), (var.get() == "Fixed"), types_var.get()))
        button.grid(row=7, column=1)
        window.mainloop()

if(__name__ == "__main__"):
        main()