highlight extractor

import csv
from tkinter import Tk, Label, Button, Entry, filedialog, StringVar, OptionMenu, Frame, Text, Scrollbar, END
from docx import Document

#Choose the color map for tkinter
TKINTER_COLOR_MAP = {
    "YELLOW": ("yellow",),
    "GREEN": ("light green",),
    "PINK": ("pink",),
    "BLUE": ("light blue",),
    "RED": ("red",),
    "TURQUOISE": ("turquoise",),
    "GRAY": ("gray",)
}

class HighlightExtractorGUI:
    def __init__(self, master):
        """
        Initialize the GUI application and create the necessary components.
        """
        self.master = master
        self.master.title("Highlight Extractor")  # Set the window title
        self.file_path = ""  # To store the selected file path
        self.color_actor_map = {}  # Mapping of colors to actor names
        self.fields = []  # List to store the actor-color fields
        self.row_count = 0  # Keeps track of rows added to the actor-color section

        # Make the window resizable
        self.master.resizable(True, True)
        self.master.grid_rowconfigure(1, weight=1)  # Make row 1 expandable
        self.master.grid_columnconfigure(3, weight=1)  # Make column 3 expandable

        #file selection ui
        Label(master, text="Word File:").grid(row=0, column=0, padx=10, pady=10, sticky="e")
        self.file_label = Label(master, text="No file selected", width=40, anchor="w")
        self.file_label.grid(row=0, column=1, padx=10, pady=10, sticky="w")
        Button(master, text="Browse", command=self.select_file).grid(row=0, column=2, padx=10, pady=10)

        #fields for actor and color
        self.fields_frame = Frame(master)
        self.fields_frame.grid(row=1, column=0, columnspan=3, pady=10, sticky="nsew")
        self.add_actor_color_field()  # Add the initial actor-color field

        #buttons for adding fields, extracting and previewing data, and saving the CSV
        Button(master, text="Add Actor-Color Field", command=self.add_actor_color_field).grid(row=2, column=0, columnspan=3, pady=10)
        Button(master, text="Extract & Preview CSV", command=self.extract_and_preview).grid(row=3, column=0, columnspan=3, pady=10)
        Button(master, text="Save CSV", command=self.save_csv).grid(row=4, column=0, columnspan=3, pady=10)

        #preview section for the CSV
        Label(master, text="CSV Preview:").grid(row=0, column=3, padx=10, pady=10, sticky="w")
        preview_frame = Frame(master)
        preview_frame.grid(row=1, column=3, rowspan=4, padx=10, pady=10, sticky="nsew")

        # Configure the preview area to be scrollable and resizable
        preview_frame.grid_rowconfigure(0, weight=1)
        preview_frame.grid_columnconfigure(0, weight=1)
        self.preview_text = Text(preview_frame, wrap="word")
        self.preview_text.grid(row=0, column=0, sticky="nsew")

        # Scrollbar for the preview text area
        preview_scrollbar = Scrollbar(preview_frame, command=self.preview_text.yview)
        preview_scrollbar.grid(row=0, column=1, sticky="ns")
        self.preview_text.config(yscrollcommand=preview_scrollbar.set)


    def select_file(self):
        """
        Open a file dialog to select a Word document and update the file path.
        """
        self.file_path = filedialog.askopenfilename(filetypes=[("Word Documents", "*.docx")])
        if self.file_path:
            # Display only the filename in the UI
            self.file_label.config(text=self.file_path.split("/")[-1])


    def add_actor_color_field(self):
        """
        Add a new row for mapping an actor to a highlight color.
        """
        color_var = StringVar(value=list(TKINTER_COLOR_MAP.keys())[0])  # Default color
        actor_var = StringVar()
        Label(self.fields_frame, text="Actor:").grid(row=self.row_count, column=0, padx=5, pady=2, sticky="e")
        Entry(self.fields_frame, textvariable=actor_var, width=20).grid(row=self.row_count, column=1, padx=5, pady=2, sticky="w")
        Label(self.fields_frame, text="Color:").grid(row=self.row_count, column=2, padx=5, pady=2, sticky="e")
        OptionMenu(self.fields_frame, color_var, *TKINTER_COLOR_MAP.keys()).grid(row=self.row_count, column=3, padx=5, pady=2, sticky="w")
        self.fields.append((actor_var, color_var))  # Store the actor-color pair
        self.row_count += 1

    def extract_highlighted_text(self):
        if not self.file_path:  # No file selected
            return []

        # Create a mapping of color to actor name
        self.color_actor_map = {color_var.get(): actor_var.get() or "Unknown" for actor_var, color_var in self.fields}

        document = Document(self.file_path) # Load the Word document
        extracted_data = [] # Store the extracted data
        order = 1 # Order of the highlighted text

        for paragraph in document.paragraphs:
            for run in paragraph.runs:
                if run.font.highlight_color: # Check if text is highlighted
                    highlight_color = run.font.highlight_color.name # Get the colour name
                    actor = self.color_actor_map.get(highlight_color, "Unknown") # Get the actor name and match to colour
                    extracted_data.append({"Order": order,
                                           "Actor": actor,
                                           "Color": highlight_color,
                                           "Line": run.text.strip()})  # Store the highlighted text
                    order += 1
        return extracted_data

    def extract_and_preview(self):
        extracted_data = self.extract_highlighted_text()
        if not extracted_data:
            self.preview_text.delete(1.0, END)
            self.preview_text.insert(END, "No highlighted text found or no file selected.")
            return

        # Display the extracted data in the preview text area
        self.preview_text.delete(1.0, END)
        for row in extracted_data:
            text_color = TKINTER_COLOR_MAP.get(row["Color"], ("black",))[0]

            # Insert the order and actor name with styling
            self.preview_text.insert(END, f"Order: {row['Order']}, ", "default")
            self.preview_text.insert(END, f"{row['Actor']}: ", f"actor_color_{row['Color']}")

            # Insert the actor's line with default styling
            self.preview_text.insert(END, f"{row['Line']}\n\n", "default")

            # Configure text tags for foreground and black background
            self.preview_text.tag_configure(
                f"actor_color_{row['Color']}",
                foreground=text_color,
                background="black"
            )

        self.extracted_data = extracted_data  # Save the extracted data for saving to CSV

    def save_csv(self):
        if not hasattr(self, "extracted_data") or not self.extracted_data:
            # If no data is available, display an error message
            self.preview_text.delete(1.0, END)
            self.preview_text.insert(END, "No data to save. Please extract data first.")
            return

        output_file = filedialog.asksaveasfilename(defaultextension=".csv", filetypes=[("CSV Files", "*.csv")])
        if not output_file:
            return  # No file selected

        # Write the extracted data to a CSV file
        with open(output_file, mode="w", newline="", encoding="utf-8") as csv_file:
            writer = csv.DictWriter(csv_file, fieldnames=["Order", "Actor", "Line"])
            writer.writeheader()
            for row in self.extracted_data:
                writer.writerow({"Order": row["Order"], "Actor": row["Actor"], "Line": row["Line"]})

        self.preview_text.delete(1.0, END)
        self.preview_text.insert(END, f"CSV saved successfully at {output_file}.")

if __name__ == "__main__":
    root = Tk()  # Create the main window
    app = HighlightExtractorGUI(root)
    root.mainloop()  # Run the application