accept multiple articles

app.py

@@ -1,10 +1,11 @@
+import io
 import gradio as gr
 import torch
-from nextus_regressor_class import *
+# from nextus_regressor_class import *
+from nextus_regressor_class1027 import *
 import nltk
 from pprint import pprint
 import pandas as pd
-
 model = NextUsRegressor()
 model.load_state_dict(torch.load("./nextus_regressor1012.pt"))
 model.eval()
@@ -40,26 +41,35 @@ def shap(txt, tok_level):
     # labels = ["+" if s < -1.0*threshold "-" elif s > threshold else " " for s in shapss]
     # print(len(tokens), len(labels))
     # print(list(zip(tokens, labels)))
-    # pprint(list(zip(tokens, shapss)))
+    pprint(list(zip(tokens, shapss)))
     # return str(list(zip(tokens, labels)))
-    return list(zip(tokens, labels))
+    largest_shap = torch.max(y_offs - y_pred).item()
+    largest_shap_span = tokens[torch.argmax(y_offs - y_pred).item()]
+    explanation = "가장 큰 영향을 미친 텍스트는\n'"+ largest_shap_span+ "'\n이며, 해당 텍스트가 없을 경우 Slant 스코어\n" + str(round(y_pred.item(), 4))+ "\n에서\n"+ str(round(largest_shap,4))+ "\n만큼 벗어납니다."
+    return list(zip(tokens, labels)), explanation
     # return txt
 
+
 def parse_file_input(f):
+    # print(f, type(f))
     all_articles = list()
-    for one_file in f:
-        if ".csv" in f:
-            all_articles += pd.read_csv(f).iloc[:, 0].to_list()
-        elif ".xls" in f:
-            all_articles += pd.read_excel(f).iloc[:, 0].to_list()
-        else:
-            pass
+    
+    # with open(f, "r") as fh:
+    if ".csv" in f.name:
+        
+        all_articles += pd.read_csv(f.name).iloc[:, 0].to_list()
+    elif ".xls" in f.name:
+        all_articles += pd.read_excel(f.name).iloc[:, 0].to_list()
+    else:
+        pass
+    # print(len(all_articles))
+    # print(all_articles)
     scores = model(all_articles)
     return scores
 
 demo = gr.Interface(parse_file_input,
                     [
-                        gr.File(file_count="single", file_types=[".csv", ".xls", ".xlsx"], label="기사 파일(csv/excel)을 업로드하세요")
+                        gr.File(file_count="single", file_types=[".csv", ".xls", ".xlsx"], type="file", label="기사 파일(csv/excel)을 업로드하세요")
                         #gr.Textbox(label="기사", lines=30, placeholder="기사를 입력하세요."),
                         # gr.Radio(choices=["sentence", "word"], label="해설 표시 단위", value="sentence", info="문장 단위의 해설은 sentence를, 단어 단위의 해설은 word를 선택하세요.")
                     ],