1 files changed, 122 insertions, 0 deletions
diff --git a/utilfn.py b/utilfn.py
new file mode 100644
index 0000000..e062d1b
--- /dev/null
+++ b/utilfn.py
@@ -0,0 +1,122 @@
+#!/usr/bin/env python3
+# A set of functions and types to be included for general use. 
+from enum import Enum
+from typing import * 
+import time
+import modin.pandas as pd
+from dataclasses import dataclass
+
+# Local imports
+from utiltypes import *
+
+
+# Returns a displayable time, use e.g. with print(f"{delta_now(time_begin)}")
+def delta_now(t: float, dp: int = 2) -> float:
+    return round(time.time() - t, dp)
+
+def ask_file(prompt: str = "What file should be opened: ") -> str:
+    filename = input(prompt)
+    return filename
+
+def open_file(file: str) -> pd.DataFrame:
+    if file.endswith(".csv"):
+        return pd.read_csv(file)
+    else:
+        return pd.read_excel(file)
+
+# Returns None if successfully saved, else a str error message.
+# TODO: add error handling.
+def save_df(df: pd.DataFrame, filetype: Filetype, name: str, sheet_name = None):
+    if sheet_name == None:
+        sheet_name = name
+
+    if filetype == Filetype.Excel:
+        df.to_excel(f"{name}.xlsx", sheet_name = f"{sheet_name}", index = False)
+    else:
+        df.to_csv(f"{name}.csv", index = False)
+
+@dataclass
+class PrintOptions:
+    update_user: bool
+    show_timings: bool
+
+def print_noendl(s: str):
+    print(s, end = '', flush = True)
+
+def reprint_noendl(s: str):
+    print(f'\r{s}', end = '', flush = True)
+
+# Please supply your own \n if you need it. 
+def cond_print(print_options: PrintOptions, s: str, timing_str: str = None, suffix = ""):
+    if timing_str and print_options.show_timings: 
+        print(s, timing_str, end = "", flush = True)
+    elif print_options.update_user: 
+        print(s, end = "", flush = True)
+    print(suffix, end = "", flush = True)
+
+@dataclass
+class AskOpenFileReply:
+    df: pd.DataFrame
+    fname: str
+    sheet: str
+    time_to_open_file: float
+
+# Asks for a filename, and opens it; keeps opening until a valid file passed.
+def ask_open_file(show_timings: bool = True) -> AskOpenFileReply:
+    while True:
+        try:
+            file = ask_file()
+            if file.lower().strip() in [ "quit", "exit" ]:
+                print("Quitting...")
+                quit()
+
+            print("Opening file... ", end="", flush=True)
+            # TODO: add in to open a specific sheet; prompt if more than one 
+            # sheet. At present it just opens the first one. Use optparam
+            # sheet_name=, with either str of sheet name or int index (deflt 0)
+            start_time = time.time()
+            df = open_file(file)
+            time_taken = time.time() - start_time
+            if show_timings: print(f" opened in {round(time_taken, 2)}s!")
+            else:            print( " opened!")
+            break
+        except Exception as exc: 
+            print(" File failed to open or could not be found.", flush=True)
+            print(f"Details: {exc}")
+
+    # TODO: make `sheet` sheet value when I add this logic.
+    reply = AskOpenFileReply(df = df, fname = file, sheet = "", time_to_open_file = time_taken) 
+    return reply
+
+def ask_excel_or_csv(given: str = None, prompt: bool = True) -> Filetype:
+    # prompt: if False, throw if given not valid instead of prompting the user.
+    exit_options  = [ "exit", "quit" ]
+    excel_options = [ "xl", "excel", "xlsx", ".xlsx" ]
+    csv_options   = [ "csv", ".csv" ]
+
+    if given:
+        given = given.lower().strip()
+        if given in exit_options:
+            quit()
+        elif given in excel_options:
+            return Filetype.Excel
+        elif given in csv_options:
+            return Filetype.Csv
+        # Do not add an else clause as we need to handle the case when `given`
+        # is `None` in the below prompt.
+
+    if not prompt:
+        # TODO: change this to be a more specific exception type.
+        raise Exception(f"ask_excel_or_csv: passed {given} as parameter with prompt == False")
+
+    answer = input("Should I output as `.xlsx` or `.csv` (`csv` is much faster)? [csv, excel, quit] ")
+    answer = answer.lower().strip()
+    while True:
+        if answer in exit_options:
+            quit()
+        elif answer in csv_options:
+            return Filetype.Csv
+        elif answer in excel_options:
+            return Filetype.Excel
+        else:
+            print("Invalid input - I want one of `csv`, `excel`, or `quit`.")