#!/usr/bin/python3

import sys
import nltk

#-------------------------------------------------------------------------------
stmts = [
  "disassemble 10 first instructions at main",
  "pet 10 cats and 2 rabbits",
  "what's the weather like in california",
  "Diaphora 3.0 checkpoint at end of March 2023"
]

INTERESTING = [
  "JJ", "CD", "NN"
]

#-------------------------------------------------------------------------------
def summarize(sentence):
  tokens = nltk.word_tokenize(sentence)
  tagged = nltk.pos_tag(tokens)
  entities = nltk.chunk.ne_chunk(tagged)
  
  summary = []
  for ent in entities:
    for inte in INTERESTING:
      if ent[1].startswith(inte):
        summary.append(ent)

  print(">Sentence:", repr(sentence))
  print(">Summary :", list(summary))
  print(">Entities:", list(entities))
  print()

#-------------------------------------------------------------------------------
def main():
  for stmt in stmts:
    summarize(stmt)

if __name__ == "__main__":
  main()