Alexis Palmer
commited on
Commit
·
45dfe77
1
Parent(s):
397ad86
new util version
Browse files
util.py
CHANGED
@@ -27,6 +27,16 @@ def load_raw_text(corpus_directory: str, file_names=None) -> str:
|
|
27 |
corpus += (file_contents + "\n")
|
28 |
return corpus
|
29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
word_regex = r"[\w|\']+"
|
32 |
def tokenize(text):
|
|
|
27 |
corpus += (file_contents + "\n")
|
28 |
return corpus
|
29 |
|
30 |
+
def load_single_raw_text_file(file_name):
|
31 |
+
"""Loads a single text file into one large string"""
|
32 |
+
|
33 |
+
corpus = ""
|
34 |
+
with open(file_name, 'r') as file:
|
35 |
+
file_contents = file.read()
|
36 |
+
corpus += (file_contents + "\n")
|
37 |
+
|
38 |
+
return corpus
|
39 |
+
|
40 |
|
41 |
word_regex = r"[\w|\']+"
|
42 |
def tokenize(text):
|