replace funny os.sep joins with os.path.join for consistency. (#1506)

Author: erip

torchtext/datasets/amazonreviewfull.py

@@ -24,8 +24,8 @@
 _PATH = 'amazon_review_full_csv.tar.gz'
 
 _EXTRACTED_FILES = {
-    'train': f'{os.sep}'.join(['amazon_review_full_csv', 'train.csv']),
-    'test': f'{os.sep}'.join(['amazon_review_full_csv', 'test.csv']),
+    'train': os.path.join('amazon_review_full_csv', 'train.csv'),
+    'test': os.path.join('amazon_review_full_csv', 'test.csv'),
 }
 
 _EXTRACTED_FILES_MD5 = {

torchtext/datasets/amazonreviewpolarity.py

@@ -23,8 +23,8 @@
 _PATH = 'amazon_review_polarity_csv.tar.gz'
 
 _EXTRACTED_FILES = {
-    'train': f'{os.sep}'.join(['amazon_review_polarity_csv', 'train.csv']),
-    'test': f'{os.sep}'.join(['amazon_review_polarity_csv', 'test.csv']),
+    'train': os.path.join('amazon_review_polarity_csv', 'train.csv'),
+    'test': os.path.join('amazon_review_polarity_csv', 'test.csv'),
 }
 
 

torchtext/datasets/sogounews.py

@@ -24,8 +24,8 @@
 _PATH = 'sogou_news_csv.tar.gz'
 
 _EXTRACTED_FILES = {
-    'train': f'{os.sep}'.join(['sogou_news_csv', 'train.csv']),
-    'test': f'{os.sep}'.join(['sogou_news_csv', 'test.csv']),
+    'train': os.path.join('sogou_news_csv', 'train.csv'),
+    'test': os.path.join('sogou_news_csv', 'test.csv'),
 }
 
 _EXTRACTED_FILES_MD5 = {

torchtext/experimental/datasets/sst2.py

@@ -29,9 +29,9 @@
 _PATH = "SST-2.zip"
 
 _EXTRACTED_FILES = {
-    "train": f"{os.sep}".join([_PATH, "SST-2", "train.tsv"]),
-    "dev": f"{os.sep}".join([_PATH, "SST-2", "dev.tsv"]),
-    "test": f"{os.sep}".join([_PATH, "SST-2", "test.tsv"]),
+    "train": os.path.join(_PATH, "SST-2", "train.tsv"),
+    "dev": os.path.join(_PATH, "SST-2", "dev.tsv"),
+    "test": os.path.join(_PATH, "SST-2", "test.tsv"),
 }
 
 _EXTRACTED_FILES_MD5 = {