def get_scores(self, terms):
"""creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in the file.
Compute this score for each term in a query and sum all the scores.
Args:
terms (list) : a list of str
Returns:
list : a list of tuples, each containing the filename and its relevancy score
"""
scores = HashTableLinear()
for item in terms:
if item in self.term_freqs:
temp = self.term_freqs[item]
for thing in temp.hash_table:
if thing is not None:
if thing[0] in scores:
scores[thing[0]] += self.get_wf(thing[1])
else:
scores.put(thing[0], self.get_wf(thing[1]))
for item in scores.hash_table:
if item is None:
continue
else:
scores[item[0]] /= self.doc_length[item[0]]
return scores
def count_words(self, filename, words):
""" Count words in a file and store the frequency of each word in the term_freqs hash table.
Words should not contain stopwords. Also store the total count of words contained in the
file in the doc_length hash table.
Args:
filename (str): the file name
words (list): a list of words
"""
# store total count of words in the doc_length hash table
self.doc_length.put(filename, len(words))
# iterate through each word
for word in words:
# calculate frequency of this word in this document
word_frequency = words.count(
word) # returns number of occurences of this word in words
# if word is already in term_freqs
if self.term_freqs.contains(word):
# add new ("doc1", freq) pair to term_freqs[word] (which is the lower hashtable)
self.term_freqs[word][1].put(filename, word_frequency)
# if word is not already in term_freqs
else:
# create new frequency hashtable for each term ("doc1", frequency)
freq_hashtable = HashTable()
freq_hashtable.put(filename, word_frequency)
# put this newly created hash table into term_freqs hash table
self.term_freqs.put(word, freq_hashtable)
def count_words(self, filename, words):
"""count words in a file and store the frequency of each
word in the term_freqs hash table. The keys of the term_freqs hash table shall be
words. The values of the term_freqs hash table shall be hash tables (term_freqs
is a hash table of hash tables). The keys of the hash tables (inner hash table) stored
in the term_freqs shall be file names. The values of the inner hash tables shall be
the frequencies of words. For example, self.term_freqs[word][filename] += 1;
Words should not contain stopwords.
Also store the total count of words contained in the file in the doc_length hash table.
Args:
filename (str) : the file name
words (list) : a list of words
"""
for item in words:
if self.term_freqs.contains(item):
if self.term_freqs[item].contains(filename):
self.term_freqs[item][filename] += 1
else:
self.term_freqs[item].put(filename, 1)
else:
ht = HashTableLinear()
ht.put(filename, 1)
self.term_freqs.put(item, ht)
if self.doc_length.contains(filename):
self.doc_length[filename] += 1
else:
self.doc_length.put(filename, 1)
def search(self, query):
""" Search for the query terms in files
Args:
query (str): query input, "user input goes here"
Returns:
list: a list of files in descending order of relevancy
"""
# parse words
filtered_query = self.parse_words([query])
# remove duplicate words using a hash table
word_table = HashTable()
for word in filtered_query:
word_table.put(word, word)
word_table_keys = word_table.keys()
parsed_query_terms = [] # changes from string to a list
# add all words from hash table to list using keys()
for key in word_table_keys:
parsed_query_terms.append(word_table[key][0])
# pass query terms to get_scores()
tuples = self.get_scores(parsed_query_terms)
# pass resulting list of tuples to rank()
results = self.rank(tuples)
# rank's result will be displayed in descending order on screen
for a_tuple in results:
print(a_tuple[0])
def test_linear2(self):
ht = HashTableLinear()
for i in range(22):
ht.put(chr(i), i)
self.assertEqual(ht.size(), 22)
self.assertTrue(ht.load_factor() <= 0.75)
self.assertTrue(ht.contains(chr(0)))
self.assertTrue(ht.contains(chr(1)))
self.assertTrue(ht.contains(chr(19)))
self.assertFalse(ht.contains(chr(22)))
def test_linear1(self):
ht = HashTableLinear()
for i in range(11):
ht.put(str(i), i)
self.assertEqual(ht.size(), 11)
self.assertTrue(ht.load_factor() <= 0.75)
self.assertTrue(ht.contains('0'))
self.assertTrue(ht.contains('1'))
self.assertTrue(ht.contains('10'))
self.assertFalse(ht.contains('11'))
def get_scores(self, terms):
""" Creates list of scores for each file in corpus.
The score = (weighted frequency / total word count in file)
Compute the score for each term in a query and sum all the scores.
Args:
terms (list): a list of strings, raw input string from user query
Returns:
list: a list of tuples, each containing the filename and its relevancy score
"""
# scores = HashMap()
score_table = HashTable(
) # contains tuples of (filename, weighted_frequency)
for query_term in terms:
# fetch a hash table of "term" from self.term_freqs
query_term_table = self.term_freqs[query_term][1]
# for each file in the hash table, add weighted frequency to scores[file]
qt_table_keys = query_term_table.keys()
for key in qt_table_keys: # key is a file name
weighted_frequency = self.get_wf(query_term_table[key][1])
if weighted_frequency != 0:
# if this is the second query_term
if score_table.contains(key):
# new frequency + old frequency
old_freq = score_table[key][1]
updated_freq = weighted_frequency + old_freq
score_table.put(key, updated_freq)
# if score_table[key] is empty, use put (if first query_term)
else:
score_table.put(key, weighted_frequency)
# for each file in scores, do scores[file] /= self.doc_length[file]
score_table_keys = score_table.keys()
score_list = []
for key in score_table_keys: # key is a filename
normalized_score = score_table[key][1] / self.doc_length[key][1]
score_table[key] = normalized_score
score_list.append(score_table[key])
# return scores, which is a list of tuples neglecting terms with frequencies of 0
return score_list
def test_whole_functionality(self):
""" Tests the Separate Chain Hash Table Functionality"""
filename = 'stop_words.txt'
hash_table = HashTableLinear()
hash_table = import_stopwords(filename, hash_table)
self.assertRaises(KeyError, hash_table.get, 'BubbleGum')
self.assertTrue('to' in hash_table)
second_hash = HashTableLinear()
second_hash.put('three', 'three')
third_hash = HashTableLinear()
third_hash.put('three', 'three')
self.assertEqual(second_hash, third_hash)
self.assertNotEqual(hash_table, second_hash)
self.assertNotEqual(hash_table, 5)
expected = "Hash_val = 0: None\n" \
"Hash_val = 1: None\n" \
"Hash_val = 2: None\n" \
"Hash_val = 3: None\n" \
"Hash_val = 4: ('three', 'three')\n" \
"Hash_val = 5: None\n" \
"Hash_val = 6: None\n" \
"Hash_val = 7: None\n" \
"Hash_val = 8: None\n" \
"Hash_val = 9: None\n" \
"Hash_val = 10: None\n"
self.assertEqual(expected, repr(second_hash))
second_hash['four'] = 'four'
self.assertEqual(second_hash['four'], 'four')
second_hash['five'] = 'five'
self.assertEqual(0, hash_table.get('from'))
self.assertFalse(second_hash.contains('p'))
self.assertTrue(second_hash.contains('five'))
second_hash.remove('five')
self.assertFalse(second_hash.contains('five'))
self.assertRaises(KeyError, second_hash.remove, 'p')
self.assertEqual(1, third_hash.size())
self.assertEqual(0, third_hash.collisions())
def test_HashTableLinear(self):
t = HashTableLinear()
self.assertEqual(t.size(), 0)
self.assertFalse(t.contains('us'))
self.assertRaises(KeyError, t.get, 'us')
t.put('us', 'us')
self.assertEqual(t.get('us'), 'us')
self.assertEqual(t['us'], 'us')
self.assertTrue(t.contains('us'))
self.assertFalse(t.contains('say'))
self.assertEqual(t.size(), 1)
self.assertEqual(t.collisions(), 0)
t.put('say', 'say')
self.assertEqual(t.get('say'), 'say')
self.assertTrue(t.contains('say'))
self.assertEqual(t.size(), 2)
self.assertEqual(t.collisions(), 1)
t.remove('say')
self.assertFalse(t.contains('say'))
self.assertTrue(t.contains('us'))
t.remove('us')
self.assertEqual(t.size(), 0)
# print(hash_string('the', 11)) # 'the' = 5
t.put('us', 'us')
t.put('say', 'say')
# self.assertEqual(t.load_factor(), 0.18181818181818182)
t.put('the', 'the')
# t.put(chr(0), chr(0))
# t.put('0', '0')
# print('chr 0', chr(0))
# print('just 0', '0')
# print(type(chr(0)))
# print(type('0'))
# print(hash_string('us', 23)) # 'the' = 5
# print(hash_string('say', 23)) # 'the' = 5
# print(hash_string('the', 23)) # 'the' = 5
# print('from tests', t)
self.assertTrue(t.contains('us'))
self.assertTrue(t.contains('the'))
def test_linear4(self):
ht = HashTableLinear()
for i in range(22):
ht.put(chr(i), i)
self.assertEqual(ht.size(), 22)
self.assertTrue(ht.load_factor() <= 0.75)
self.assertEqual(ht[chr(0)], 0)
self.assertEqual(ht[chr(1)], 1)
self.assertEqual(ht[chr(19)], 19)
self.assertRaises(KeyError, ht.get, 'a')
for i in range(22):
ht.remove(chr(i))
self.assertFalse(ht.contains(chr(0)))
self.assertFalse(ht.contains(chr(1)))
self.assertFalse(ht.contains(chr(19)))
self.assertRaises(KeyError, ht.remove, 'a')
def search(self, query):
""" Search for the query terms in files
Args:
query (str): query input: e.g. "computer science"
Returns:
list: a list of tuples: (files_path_name, score) sorted in
descending order or relevancy excluding files whose relevancy
score is 0.
"""
terms = self.parse_words([query])
cleaned_terms = []
hash_terms = HashTableLinear()
for term in terms:
if not hash_terms.contains(term):
cleaned_terms.append(term)
hash_terms.put(term, term)
scores = self.get_scores(cleaned_terms)
scores = self.rank(scores)
return scores
def get_scores(self, terms):
"""creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in the file.
Compute this score for each term in a query and sum all the scores.
Args:
terms (list) : a list of str
Returns:
list : a list of tuples, each containing the filename and its relevancy score
"""
scores = HashTableLinear()
for term in terms:
if self.term_freqs.contains(term):
hashtable = self.term_freqs[term]
for file1 in hashtable.slots:
if file1 != None and hashtable.contains(file1[0]):
if scores.contains(file1[0]):
key, val = scores.remove(file1[0])
scores.put(file1[0], val + self.get_wf(file1[1]))
else:
scores.put(file1[0], self.get_wf(file1[1]))
for file1 in scores.slots:
if file1 is not None:
key, val = scores.remove(file1[0])
val /= self.doc_length.get(file1[0])
scores.put(file1[0], val)
return scores
def count_words(self, file_path_name, words):
""" Count words in a file and store the frequency of each word in the
term_freqs hash table. The keys of the term_freqs hash table shall
be words. The values of the term_freqs hash table shall be hash
tables (term_freqs is a hash table of hash tables). The keys of
the hash tables (inner hash table) stored in the term_freqs shall
be file names. Values of inner hash tables shall be the frequencies
of words.
Args:
file_path_name (str): the file name
words (list): a list of words
"""
self.doc_length.put(file_path_name, len(words))
while len(words) > 0:
current_word = words[0]
word_freq = 0
word_freq = words.count(current_word)
try:
while True:
words.remove(current_word)
except ValueError:
pass
# If the word already in term_freqs, retrieve the doc freq table
# otherwise, create a new hash table
if current_word in self.term_freqs:
freq_hash = self.term_freqs.get(current_word)
else:
freq_hash = HashTableLinear()
freq_hash.put(file_path_name, word_freq)
self.term_freqs.put(current_word, freq_hash)
class SearchEngine:
"""class for SearchEngine
Attributes:
directory (str) : a directory name
stopwords (HashMap) : a hash table containing stopwords
doc_length (HashMap) : a hash table containing the total number of words in each
document
term_freqs (HashMap) : a hash table of hash tables for each term. Each hash table
contains the frequency of the term in documents
(document names are the keys and the frequencies are the values)
file_list : a list of files
"""
def __init__(self, directory, stopwords):
self.doc_length = HashTableLinear(
) # Replace HashMap() with your hash table.
self.term_freqs = HashTableLinear()
self.stopwords = stopwords
self.index_files(directory)
def __repr__(self):
return "doc_length: %s, term_freqs: %s, stopwords: %s"\
% (self.doc_length, self.term_freqs, self.stopwords)
def __eq__(self, other):
return isinstance(other, SearchEngine) and self.doc_length == other.doc_length\
and self.term_freqs == other.term_freqs
def read_file(self, infile):
"""A helper function to read a file
Args:
infile (str) : the path to a file
Returns:
list : a list of str read from a file
"""
file_list = []
with open(infile, 'r') as i:
for line in i.readlines():
file_list.append(line)
return file_list
def parse_words(self, lines):
"""split strings into words
Convert words to lower cases and remove new line chars.
Exclude stopwords.
Args:
lines (list) : a list of strings
Returns:
list : a list of words
"""
word_list = []
for item in lines:
split_arr = item.split(' ')
for word in split_arr:
if word != '\n':
word_list.append(word.lower())
final_arr = []
for item in word_list:
item = item.replace("(", '')
item = item.replace(")", '')
item = item.replace(".", '')
item = item.replace(",", '')
final_arr.append(item.rstrip())
final_arr = self.exclude_stopwords(final_arr)
return final_arr
def exclude_stopwords(self, terms):
"""exclude stopwords from the list of terms
Args:
terms (list) :
Returns:
list : a list of str with stopwords removed
"""
new_terms = []
for item in terms:
if item in self.stopwords:
continue
else:
new_terms.append(item)
return new_terms
def count_words(self, filename, words):
"""count words in a file and store the frequency of each
word in the term_freqs hash table. The keys of the term_freqs hash table shall be
words. The values of the term_freqs hash table shall be hash tables (term_freqs
is a hash table of hash tables). The keys of the hash tables (inner hash table) stored
in the term_freqs shall be file names. The values of the inner hash tables shall be
the frequencies of words. For example, self.term_freqs[word][filename] += 1;
Words should not contain stopwords.
Also store the total count of words contained in the file in the doc_length hash table.
Args:
filename (str) : the file name
words (list) : a list of words
"""
for item in words:
if self.term_freqs.contains(item):
if self.term_freqs[item].contains(filename):
self.term_freqs[item][filename] += 1
else:
self.term_freqs[item].put(filename, 1)
else:
ht = HashTableLinear()
ht.put(filename, 1)
self.term_freqs.put(item, ht)
if self.doc_length.contains(filename):
self.doc_length[filename] += 1
else:
self.doc_length.put(filename, 1)
def index_files(self, directory):
"""index all text files in a given directory
Args:
directory (str) : the path of a directory
"""
dir_list = os.listdir(directory)
for item in dir_list:
if os.path.isfile(os.path.join(directory, item)) is True:
parts = os.path.splitext(item)
if parts[1] == '.txt':
line_arr = self.read_file(os.path.join(directory, item))
words = self.parse_words(line_arr)
self.count_words(os.path.join(directory, item), words)
def get_wf(self, tf):
"""computes the weighted frequency
Args:
tf (float) : term frequency
Returns:
float : the weighted frequency
"""
if tf > 0:
wf = 1 + math.log(tf)
else:
wf = 0
return wf
def get_scores(self, terms):
"""creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in the file.
Compute this score for each term in a query and sum all the scores.
Args:
terms (list) : a list of str
Returns:
list : a list of tuples, each containing the filename and its relevancy score
"""
scores = HashTableLinear()
for item in terms:
if item in self.term_freqs:
temp = self.term_freqs[item]
for thing in temp.hash_table:
if thing is not None:
if thing[0] in scores:
scores[thing[0]] += self.get_wf(thing[1])
else:
scores.put(thing[0], self.get_wf(thing[1]))
for item in scores.hash_table:
if item is None:
continue
else:
scores[item[0]] /= self.doc_length[item[0]]
return scores
def rank(self, scores):
"""ranks files in the descending order of relevancy
Args:
scores(list) : a list of tuples: (filename, score)
Returns:
list : a list of tuples: (filename, score) sorted in descending order of relevancy
"""
new_scores = list(filter(None, scores))
sorted_scores = sorted(new_scores, key=lambda x: x[1], reverse=True)
return sorted_scores
def search(self, query):
""" search for the query terms in files
Args:
query (str) : query input
Returns:
list : list of files in descending order or relevancy
"""
scores = self.get_scores(query).hash_table
new_scores = self.rank(scores)
score_arr = []
for item in new_scores:
if item is not None:
score_arr.append(item)
print(item)
return score_arr
class SearchEngine:
""" Search engine class to build an inverted index of documents stored
in a specified directory and provides a functionality to search
documents with query terms.
Attributes:
directory (str): a directory name
stopwords (HashMap): a hash table containing stop words
doc_length (HashMap): a hash table containing the total number of
words in each document
term_freqs (HashMap): a hash table of hash tables for each term.
Each hash table contains the frequency of
the term in documents (document names are
the keys and the frequencies of the values)
"""
def __init__(self, directory, stopwords):
""" Initialize the data structure by taking a directory name and a
hash table containing stopwords.
Args:
directory (str): a directory name
stopwords (HashMap): a hash table containing stopwords
"""
self.doc_length = HashTableLinear()
self.term_freqs = HashTableLinear()
self.stopwords = stopwords
self.file_list = []
self.index_files(directory)
def __eq__(self, other):
""" Compares the data structure to other"""
return isinstance(other, SearchEngine) and \
self.doc_length == other.doc_length and \
self.term_freqs == other.term_freqs and \
self.stopwords == other.stopwords
def __repr__(self):
""" How the data structure is represented"""
return "SearchEngine Instance:\n" + str(self.term_freqs)
def read_file(self, infile):
""" Reads all words contained in the file except for stop words
Args:
infile (str): the path to a file
Returns:
list: a list of str read from a file
"""
with open(infile, 'r') as file:
str_list = file.readlines()
for index, line in enumerate(str_list):
str_list[index] = line.rstrip('\n')
return str_list
def parse_words(self, lines):
""" Splits strings into words by spaces, converts words to lower
cases, and removes newline chars, parentheses, brackets, and
punctuations
Args:
lines (list): a list of strings
Returns:
list: a list of words
"""
words = []
for each in lines:
line_list = each.split()
for index, word in enumerate(line_list):
line_list[index] = re.sub('[\W_]+', '', word)
words += line_list
for index, word in enumerate(words):
words[index] = word.lower()
words = self.exclude_stopwords(words)
return words
def exclude_stopwords(self, terms):
""" Exclude stopwords from the list of terms
Args:
terms (list): list of terms to be cleaned of stop words
Returns:
list: a list of str with stopwords removed
"""
out_list = []
for each in terms:
if each not in self.stopwords:
out_list.append(each)
return out_list
def count_words(self, file_path_name, words):
""" Count words in a file and store the frequency of each word in the
term_freqs hash table. The keys of the term_freqs hash table shall
be words. The values of the term_freqs hash table shall be hash
tables (term_freqs is a hash table of hash tables). The keys of
the hash tables (inner hash table) stored in the term_freqs shall
be file names. Values of inner hash tables shall be the frequencies
of words.
Args:
file_path_name (str): the file name
words (list): a list of words
"""
self.doc_length.put(file_path_name, len(words))
while len(words) > 0:
current_word = words[0]
word_freq = 0
word_freq = words.count(current_word)
try:
while True:
words.remove(current_word)
except ValueError:
pass
# If the word already in term_freqs, retrieve the doc freq table
# otherwise, create a new hash table
if current_word in self.term_freqs:
freq_hash = self.term_freqs.get(current_word)
else:
freq_hash = HashTableLinear()
freq_hash.put(file_path_name, word_freq)
self.term_freqs.put(current_word, freq_hash)
def index_files(self, directory):
""" Processes a directory and makes an index of all the files
Args:
directory (str): the directory being indexed
"""
dir_list = os.listdir(directory)
full_dir_list = []
for index, item in enumerate(dir_list):
full_dir_list.append(os.path.join(directory, item))
file_list = []
for index, item in enumerate(full_dir_list):
if os.path.isfile(item):
parts = os.path.splitext(item)
if parts[1] == '.txt':
file_list.append(full_dir_list[index])
# The list of txt files in directory is now in file_list
self.file_list = file_list
for file in file_list:
str_list = self.read_file(file)
words = self.parse_words(str_list)
self.count_words(file, words)
def get_wf(self, tf):
""" computes the weighted frequency
Args:
tf (float): term frequency
Returns:
float: the weighted frequency
"""
if tf > 0:
wf = 1 + math.log(tf)
else:
wf = 0
return wf
def get_scores(self, terms):
""" Creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in file.
Args:
terms (list): a list of str
Returns:
list: a list of tuples, each containing the file_path_name
and its relevancy score
"""
scores = HashTableLinear()
for term in terms:
word_hash_table = self.term_freqs.get(term)
for file in self.file_list:
if word_hash_table.contains(file):
if scores.contains(file):
scores[file] += self.get_wf(word_hash_table[file])
else:
scores[file] = self.get_wf(word_hash_table[file])
score_list = []
for file in self.file_list:
if scores.contains(file) and scores[file] > 0:
norm_score = scores[file] / self.doc_length[file]
score_list.append((file, norm_score))
return score_list
def rank(self, scores):
""" Ranks files in the descending order of relevancy
Args:
scores (list): a list of tuples: (file_path_name, score)
Returns:
list: a list of tuples (file_path_name, score) sorted in
descending order of relevancy
"""
for pos in range(1, len(scores)):
cur_pos = pos
while cur_pos > 0 and scores[cur_pos - 1][1] < scores[cur_pos][1]:
temp = scores[cur_pos - 1]
scores[cur_pos - 1] = scores[cur_pos]
scores[cur_pos] = temp
# Reverse the order so its descending
# scores.reverse()
return scores
def search(self, query):
""" Search for the query terms in files
Args:
query (str): query input: e.g. "computer science"
Returns:
list: a list of tuples: (files_path_name, score) sorted in
descending order or relevancy excluding files whose relevancy
score is 0.
"""
terms = self.parse_words([query])
cleaned_terms = []
hash_terms = HashTableLinear()
for term in terms:
if not hash_terms.contains(term):
cleaned_terms.append(term)
hash_terms.put(term, term)
scores = self.get_scores(cleaned_terms)
scores = self.rank(scores)
return scores
def print_nice_results(self, scores):
""" Takes the output of scores method and makes the results more
presentable to look at.
Args:
scores (list): list of tuples that is output from self.scores
"""
for each in scores:
print(each[0] + str(each[1]))
class SearchEngine:
""" Builds and maintains an inverted index of documents stored in a specified directory and
provides a functionality to search documents with query terms
Attributes:
directory (str): a directory name
stopwords (HashTable): contains stopwords
doc_length (HashTable): contains number of words in each document
doc_freqs (HashTable): contains number of documents containing the term for each term
term_freqs (HashTable): hash table of hash tables for each term, each hash table contains
the frequency of the term in documents (document names are the keys
and the frequencies are the values)
"""
def __init__(self, directory, stopwords):
self.doc_length = HashTable()
self.doc_freqs = HashTable() #this will not be used in this assignment
self.term_freqs = HashTable()
self.stopwords = stopwords
self.index_files(directory)
# PREPROCESSING ================================================================================
def read_file(self, infile):
""" A helper function to read a file
Args:
infile (str): the path to a file
Returns:
list: a list of strings read from a file
"""
# open file
with open(infile, "r") as filepointer:
lines = filepointer.readlines(
) # looks like ["line 1 here", "line 2 here"]
return lines
def parse_words(self, lines):
""" Split strings into words, convert words to lower cases and remove newline characters,
exclude stopwords
Args:
lines (list): a list of lists of strings
Returns:
list: a list of words
"""
raw_words = []
for line in lines:
split_line = line.split(
" ") # split line looks like ["line", "1", "here"]
raw_words.extend(split_line)
# create new list with all words that aren't stop words
filtered_words = [
word.rstrip().lower() for word in raw_words
if word not in self.stopwords
]
return filtered_words
def count_words(self, filename, words):
""" Count words in a file and store the frequency of each word in the term_freqs hash table.
Words should not contain stopwords. Also store the total count of words contained in the
file in the doc_length hash table.
Args:
filename (str): the file name
words (list): a list of words
"""
# store total count of words in the doc_length hash table
self.doc_length.put(filename, len(words))
# iterate through each word
for word in words:
# calculate frequency of this word in this document
word_frequency = words.count(
word) # returns number of occurences of this word in words
# if word is already in term_freqs
if self.term_freqs.contains(word):
# add new ("doc1", freq) pair to term_freqs[word] (which is the lower hashtable)
self.term_freqs[word][1].put(filename, word_frequency)
# if word is not already in term_freqs
else:
# create new frequency hashtable for each term ("doc1", frequency)
freq_hashtable = HashTable()
freq_hashtable.put(filename, word_frequency)
# put this newly created hash table into term_freqs hash table
self.term_freqs.put(word, freq_hashtable)
def index_files(self, directory):
""" Index all text files in a given directory
Args:
directory (str) : the path of a directory
"""
# get a list of files in the directory
file_list = os.listdir(directory)
# for each item in file_list, item is a filename
for item in file_list:
# construct full path of each file
path = os.path.join(directory, item)
# if item is not a file, skip it
if not os.path.isfile(path) or item == "stop_words.txt":
continue
# split path into file extension and the rest
parts = os.path.splitext(
item) # maybe change item stuff here to path
# only process text files
if parts[1] == ".txt":
# process it
item_lines = self.read_file(path)
item_words = self.parse_words(item_lines)
self.count_words(path, item_words)
# SEARCHING ====================================================================================
def get_wf(self, term_frequency):
""" Computes the weighted frequency
Args:
term_frequency (float): term frequency
Returns:
float: the weighted frequency
"""
if term_frequency > 0:
weighted_freq = 1 + math.log(term_frequency)
else:
weighted_freq = 0
return weighted_freq
def get_scores(self, terms):
""" Creates list of scores for each file in corpus.
The score = (weighted frequency / total word count in file)
Compute the score for each term in a query and sum all the scores.
Args:
terms (list): a list of strings, raw input string from user query
Returns:
list: a list of tuples, each containing the filename and its relevancy score
"""
# scores = HashMap()
score_table = HashTable(
) # contains tuples of (filename, weighted_frequency)
for query_term in terms:
# fetch a hash table of "term" from self.term_freqs
query_term_table = self.term_freqs[query_term][1]
# for each file in the hash table, add weighted frequency to scores[file]
qt_table_keys = query_term_table.keys()
for key in qt_table_keys: # key is a file name
weighted_frequency = self.get_wf(query_term_table[key][1])
if weighted_frequency != 0:
# if this is the second query_term
if score_table.contains(key):
# new frequency + old frequency
old_freq = score_table[key][1]
updated_freq = weighted_frequency + old_freq
score_table.put(key, updated_freq)
# if score_table[key] is empty, use put (if first query_term)
else:
score_table.put(key, weighted_frequency)
# for each file in scores, do scores[file] /= self.doc_length[file]
score_table_keys = score_table.keys()
score_list = []
for key in score_table_keys: # key is a filename
normalized_score = score_table[key][1] / self.doc_length[key][1]
score_table[key] = normalized_score
score_list.append(score_table[key])
# return scores, which is a list of tuples neglecting terms with frequencies of 0
return score_list
def rank(self, scores):
""" Ranks files in the descending order of relevancy
Args:
scores (list): list of tuples of (filename, score)
Returns:
list: a list of filenames sorted in descending order of relevancy
"""
return sorted(scores, key=lambda x: x[1], reverse=True)
def search(self, query):
""" Search for the query terms in files
Args:
query (str): query input, "user input goes here"
Returns:
list: a list of files in descending order of relevancy
"""
# parse words
filtered_query = self.parse_words([query])
# remove duplicate words using a hash table
word_table = HashTable()
for word in filtered_query:
word_table.put(word, word)
word_table_keys = word_table.keys()
parsed_query_terms = [] # changes from string to a list
# add all words from hash table to list using keys()
for key in word_table_keys:
parsed_query_terms.append(word_table[key][0])
# pass query terms to get_scores()
tuples = self.get_scores(parsed_query_terms)
# pass resulting list of tuples to rank()
results = self.rank(tuples)
# rank's result will be displayed in descending order on screen
for a_tuple in results:
print(a_tuple[0])
def test_basic(self):
""" Tests basic functionality"""
hash_table = HashTableLinear()
hash_table.put('unless', 'unless')
self.assertTrue(hash_table.contains('unless'))
hash_table.put('every', 'every')
hash_table.put('being', 'being')
hash_table.put('elsewhere', 'elsewhere')
hash_table.put('nothing', 'nothing')
hash_table.put('hereby', 'hereby')
hash_table.put('latter', 'latter')
hash_table.put('and', 'and')
hash_table.put('afterwards', 'afterwards')
hash_table.put('say', 'say')
hash_table.put('very', 'very')
hash_table.put('few', 'few')
hash_table.put('well', 'well')
hash_table.put('various', 'various')
hash_table.put('make', 'make')
hash_table.put('regarding', 'regarding')
hash_table.put('take', 'take')
hash_table.put('give', 'give')
hash_table.put('whole', 'whole')
hash_table.put('i', 'i')
hash_table.put('against', 'against')
hash_table.put('can', 'can')
hash_table.get('every')
hash_table.get('being')
hash_table.get('elsewhere')
hash_table.get('nothing')
hash_table.get('hereby')
hash_table.get('latter')
hash_table.get('and')
hash_table.get('afterwards')
hash_table.get('say')
hash_table.get('very')
hash_table.get('few')
hash_table.get('well')
hash_table.get('various')
hash_table.get('make')
hash_table.get('regarding')
hash_table.get('take')
hash_table.get('give')
hash_table.get('whole')
hash_table.get('i')
hash_table.get('against')
hash_table.get('can')
class SearchEngine:
"""
doc_length: the number of words contained in document contained in a hashtable
term_freqs: a hashtable containing each word in text file and an inner hashtable containing the frequency count
stopwords: a hashtable containing stop words
"""
def __init__(self, directory, stopwords):
self.doc_length = HashTableLinear()
self.term_freqs = HashTableLinear()
self.stopwords = stopwords
self.index_files(directory)
def read_file(self, infile):
"""A helper function to read a file
Args:
infile (str) : the path to a file
Returns:
list : a list of str read from a file
"""
with open(infile, 'r') as fi:
strings = fi.read()
strings = strings.split()
return strings
def parse_words(self, lines):
"""split strings into words
Convert words to lower cases and remove new line chars.
Exclude stopwords.
Args:
lines (list) : a list of strings
Returns:
list : a list of words
"""
list1 = []
for line in lines:
words = line.lower()
words = words.split()
for word in words:
list1.append(word)
list1 = self.exclude_stopwords(list1)
return list1
def exclude_stopwords(self, terms):
"""exclude stopwords from the list of terms
Args:
terms (list) :
Returns:
list : a list of str with stopwords removed
"""
list1 = []
for i in terms:
if not self.stopwords.contains(i):
list1.append(i)
return list1
def count_words(self, filename, words):
"""
Args:
filename (str) : the file name
words (list) : a list of words
"""
for word in words:
if word not in self.term_freqs:
self.term_freqs[word] = HashTableLinear()
self.term_freqs[word][filename] = 1
else:
if filename not in self.term_freqs[word]:
self.term_freqs[word][filename] = 1
else:
self.term_freqs[word][filename] += 1
self.doc_length.put(filename, len(words))
def index_files(self, directory):
"""index all text files in a given directory
Args:
directory (str) : the path of a directory
"""
file_list = os.listdir(directory)
for item in file_list:
val = os.path.join(directory, item)
if os.path.isfile(val):
parts = os.path.splitext(val)
if parts[1] == '.txt':
words1 = self.read_file(val)
words2 = self.parse_words(words1)
self.count_words(val, words2)
def get_wf(self, tf):
"""comptes the weighted frequency
Args:
tf (float) : term frequency
Returns:
float : the weighted frequency
"""
if tf > 0:
wf = 1 + math.log(tf)
else:
wf = 0
return wf
def get_scores(self, terms):
"""creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in the file.
Compute this score for each term in a query and sum all the scores.
Args:
terms (list) : a list of str
Returns:
list : a list of tuples, each containing the filename and its relevancy score
"""
scores = HashTableLinear()
for term in terms:
if self.term_freqs.contains(term):
hashtable = self.term_freqs[term]
for file1 in hashtable.slots:
if file1 != None and hashtable.contains(file1[0]):
if scores.contains(file1[0]):
key, val = scores.remove(file1[0])
scores.put(file1[0], val + self.get_wf(file1[1]))
else:
scores.put(file1[0], self.get_wf(file1[1]))
for file1 in scores.slots:
if file1 is not None:
key, val = scores.remove(file1[0])
val /= self.doc_length.get(file1[0])
scores.put(file1[0], val)
return scores
def rank(self, scores):
"""ranks files in the descending order of relevancy
Args:
scores(list) : a list of tuples: (filename, score)
Returns:
list : a list of tuples: (filename, score) sorted in descending order of relevancy
"""
no_none = []
for val in scores.slots:
if val != None:
no_none.append(val)
no_none = sorted(no_none, key=lambda x: x[1], reverse=True)
new_line_out = ''
for i in no_none:
new_line_out += i[0] + '\n'
return new_line_out
def search(self, query):
""" search for the query terms in files
Args:
query (str) : query input
Returns:
list : list of files in descending order or relevancy
"""
list1 = self.parse_words([query])
scores = self.get_scores(list1)
print(scores)
return self.rank(scores)
class SearchEngine:
"""
Builds and maintains an inverted index of documents stored in a specified directory and
provides a functionality to search documents with query terms.
Attributes:
directory (str) : a directory name
stopwords (HashTableLinear) : a hash table containing stopwords
doc_length (HashTableLinear) : a hash table containing the total number of words in each
document
doc_freqs (HashTableLinear) : a hash table containing the number of documents containing the
term for each term
term_freqs (HashTableLinear) : a hash table of hash tables for each term. Each hash table
contains the frequency of the term in documents (document names are the keys and the
frequencies are the values)
"""
def __init__(self, directory, stopwords=[]):
self.doc_length = HashTableLinear(
) # Replace HashTableLinear() with your hash table.
self.doc_freqs = HashTableLinear(
) # this will not be used in this assignment
self.term_freqs = HashTableLinear()
self.stopwords = stopwords
self.index_files(directory)
self.directory = directory
def read_file(self, infile):
"""A helper function to read a file
Args:
infile (str) : the path to a file
Returns:
list : a list of str read from a file
"""
with open(infile, 'r') as reader:
return_list = reader.readlines()
return return_list
def parse_words(self, lines):
"""split strings into words
Convert words to lower cases and remove new line chars.
Exclude stopwords.
Args:
lines (list) : a list of strings
Returns:
list : a list of words
"""
return_list = []
for i in lines:
i = i.lower()
i = i.split()
for j in i:
j = j.replace('\n', '')
if j not in self.stopwords:
return_list.append(j)
return return_list
def count_words(self, filename, words):
"""count words in a file and store the frequency of each
word in the term_freqs hash table. Words should not contain stopwords.
Also store the total count of words contained in the file
in the doc_length hash table.
Args:
filename (str) : the file name
words (list) : a list of words
"""
#file_lines = self.read_file(filename)
#str_list = self.parse_words(file_lines)
for i in words:
self.term_freqs.put(i, HashTableLinear())
self.doc_freqs.put(filename, 0)
for i in words:
if i in self.term_freqs and filename in self.term_freqs[i]:
self.term_freqs[i][filename] = self.term_freqs[i][filename] + 1
elif i and i in self.term_freqs:
self.term_freqs[i].put(filename, 1)
elif i:
self.term_freqs.put(i, HashTableLinear())
self.term_freqs[i].put(filename, 1)
self.doc_freqs[filename] = len(words)
def index_files(self, directory):
"""index all text files in a given directory
Args:
directory (str) : the path of a directory
"""
txt_list = os.listdir(directory)
for i in txt_list:
if ".txt" in i:
j = self.parse_words(self.read_file(os.path.join(directory,
i)))
self.count_words(i, j)
def get_weighted_freq(self, total_freq):
"""comptes the weighted frequency
Args:
tf (float) : term frequency
Returns:
float : the weighted frequency
"""
if total_freq > 0:
weighted_freq = 1 + math.log(total_freq)
else:
weighted_freq = 0
return weighted_freq
def get_scores(self, terms):
"""creates a list of scores for each file in corpus
The score = weighted frequency / the total word count in the file.
Compute this score for each term in a query and sum all the scores.
Args:
terms (list) : a list of str
Returns:
list : a list of tuples, each containing the filename and its relevancy score
"""
scores = ([], [])
for term in terms:
if term not in self.term_freqs:
raise ValueError
for i in self.term_freqs[term].hash_table:
if i and i.key not in scores[0]:
file = i.key
freq = i.data
scores[0].append(file)
freq = self.get_weighted_freq(freq) / self.doc_freqs[file]
scores[1].append(freq)
elif i and i.key in scores[0]:
file = i.key
freq = i.data
freq = self.get_weighted_freq(freq) / self.doc_freqs[file]
ind = scores[0].index(i.key)
scores[1][ind] += freq
return scores
def rank(self, scores):
"""ranks files in the descending order of relevancy
Args:
scores(list) : a list of tuples: (filename, score)
Returns:
list : a list of tuples: (filename, score) sorted in descending order of relevancy
"""
for i in range(
len(scores[0]) - 1, 0,
-1): # extra credit implementation of sorting algorithm.
for j in range(len(scores[0]) - 1):
if scores[1][j] < scores[1][j + 1]:
temp = scores[0][j]
temp2 = scores[1][j]
scores[0][j] = scores[0][j + 1]
scores[1][j] = scores[1][j + 1]
scores[0][j + 1] = temp
scores[1][j + 1] = temp2
return scores
def search(self, query):
"""
Searches query and returns stuff.
:param query:
:return:
"""
temp_list = query.lower()
temp_list = temp_list.split()
query_list = self.parse_words(temp_list)
scores_tuple = self.get_scores(query_list)
ranked_scores = self.rank(scores_tuple)
str = ""
for i in ranked_scores[0]:
if i:
str += "%s/%s \n" % (self.directory, i)
return str