modify get_tltschool_cefr_scales.py to callable file

utils/get_tltschool_cefr_scales.py

@@ -1,26 +1,50 @@
-## Functions
-def opentext( file, col, filter ):
+import argparse
+
+def argparse_function():
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument("--input_wavscp_file_path",
+                        default='data/trn/text',
+                        type=str)
+
+    parser.add_argument("--output_cefr_file_path",
+                    default='CEFR_LABELS_PATH/trn_cefr_scores.txt',
+                    type=str)
+
+
+    args = parser.parse_args()
+
+    return args
+
+def open_utt2cefr( file, col, filter ):
     s = {}
     with open(file, "r") as f:
         for l in f.readlines():
             utt_id  = l.split()[col]
             if sum([1 for cefr in filter if cefr.lower() in utt_id.lower()]) == 0:
                 continue
-            utt_id_ = [ str(w).lower() for w in utt_id.split("-")[1].split("_") ]
+            utt_id_ = [ str(w) for w in utt_id.split("-")[1].split("_") ]
             # speakerIp18_A2_002002001008-promptIp18_A2_en_22_20_103
-            if sum([1 for cefr in filter if cefr.lower() in utt_id_]) > 0:
+            if sum([1 for cefr in filter if cefr in utt_id_]) > 0:
                 s[utt_id] = utt_id_[1]
     return s
 
+if __name__ == '__main__':
 
-## CEFR score filter
-_filter = ["a1", "b1", "a2"]
+    # argparse
+    args = argparse_function()
+    input_wavscp_file_path = args.input_wavscp_file_path
+    output_cefr_file_path = args.output_cefr_file_path
 
-## Data
-_data = "/share/nas167/a2y3a1N0n2Yann/speechocean/espnet_amazon/egs/tlt-school/is2021_data-prep-all_baseline/data/train/text"
-data = opentext( _data, 0, _filter )
+    # CEFR score filter
+    CEFR_filter = ["A1", "A2", "B1"]
+    utt2cefr_dict = open_utt2cefr(
+        input_wavscp_file_path,
+        0,
+        CEFR_filter
+    )
 
-## Save
-with open("/share/nas167/a2y3a1N0n2Yann/speechocean/espnet_amazon/egs/tlt-school/is2021_data-prep-all_baseline/data/train/text_cefr", "w") as f:
-    for utt_id, cefr in data.items():
-        f.write("{} {}\n".format(utt_id, cefr))
+    # save
+    with open(output_cefr_file_path, "w") as f:
+        for utt_id, cefr in utt2cefr_dict.items():
+            f.write("{} {}\n".format(utt_id, cefr))