utils.py

import nltk
import pickle
import gzip
import re
import numpy as np

nltk.download('stopwords')
from nltk.corpus import stopwords

# Paths for all resources for the bot.
RESOURCE_PATH = {
    'INTENT_RECOGNIZER': 'intent_recognizer.pkl',
    'TAG_CLASSIFIER': 'tag_classifier.pkl',
    'TFIDF_VECTORIZER': 'tfidf_vectorizer.pkl',
    'THREAD_EMBEDDINGS_FOLDER': 'thread_embeddings_by_tags',
    'WORD_EMBEDDINGS': 'word_embeddings.tsv.gz',
}



def text_prepare(text):
    """Performs tokenization and simple preprocessing."""
    
    replace_by_space_re = re.compile('[/(){}\[\]\|@,;]')
    bad_symbols_re = re.compile('[^0-9a-z #+_]')
    stopwords_set = set(stopwords.words('english'))

    text = text.lower()
    text = replace_by_space_re.sub(' ', text)
    text = bad_symbols_re.sub('', text)
    text = ' '.join([x for x in text.split() if x and x not in stopwords_set])

    return text.strip()


def load_embeddings(embeddings_path):
    """Loads pre-trained word embeddings from tsv file.

    Args:
      embeddings_path - path to the embeddings file.

    Returns:
      embeddings - dict mapping words to vectors;
      embeddings_dim - dimension of the vectors.
    """
    
    # Hint: you have already implemented a similar routine in the 3rd assignment.
    # Note that here you also need to know the dimension of the loaded embeddings.
    # When you load the embeddings, use numpy.float32 type as dtype

    ########################
    #### YOUR CODE HERE ####
    ########################

    embeddings = dict()
    text = gzip.open(embeddings_path,'rb').read().split("\n")
    for t in text:
      if len(t) < 1 : continue
      w = t.split("\t")
      embeddings[w[0]] = list(map(float,w[1:]))
    return embeddings, len(w)-1
    
def question_to_vec(question, embeddings, dim=100):
    """
        question: a string
        embeddings: dict where the key is a word and a value is its' embedding
        dim: size of the representation

        result: vector representation for the question
    """
    ######################################
    ######### YOUR CODE HERE #############
    ######################################
    result = np.zeros((dim),dtype=np.float32)
    res_list = []
    for token in question.split():
        if token in embeddings:
            res_list.append(embeddings[token])
    if len(res_list) > 0 :
        result = np.mean(np.array(res_list), axis=0)
    return result

def unpickle_file(filename):
    """Returns the result of unpickling the file content."""
    with open(filename, 'rb') as f:
        return pickle.load(f)