#!/usr/bin/env python3

#######################################################################
# map our sw collection to frequency lists
#
# Author: Garry Morrison
# Date: 2015-03-25
# Update:
# Copyright: closed for now
#
# Usage: ./play_with_sw_to_freq.py
#
#######################################################################


import sys
import glob
import os

from the_semantic_db_code import *
from the_semantic_db_functions import *
from the_semantic_db_processor import *

C = context_list("sw files to frequency lists")


# the .sw directory:
sw_file_dir = "sw-examples"
# check it exists:
if not os.path.exists(sw_file_dir):
  print(sw_file_dir + " not found. Exiting ...")
  sys.exit(0)

# NB: we need "dir\file" instead of "dir/file" for our "file != dest_file" line to work.
dest_file = "sw-examples\sw-files-to-frequency-lists.sw"

size_threshold = 10000  # start with 10K


for file in glob.glob(sw_file_dir + "/*.sw"):
  if file != dest_file:
    size = os.path.getsize(file)
    if size <= size_threshold:                    # only work with small files for now.

      working_context = context_list("")          # reset the context_list for each file

#      print("file:",file) 
      base = os.path.basename(file)
      print("loading sw file:",base)
      x = ket("sw file: " + base)
      print("x:",x)

      load_sw(working_context,file)
      C.learn("kets",x,working_context.multiverse_to_freq_list())


# code to save the results:
save_sw(C,dest_file)