DDingcheol commited on
Commit
8d921b1
ยท
1 Parent(s): 0df3cee

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -2
app.py CHANGED
@@ -25,13 +25,23 @@ def get_pdf_text(pdf_docs):
25
  # ๊ณผ์ œ
26
  # ์•„๋ž˜ ํ…์ŠคํŠธ ์ถ”์ถœ ํ•จ์ˆ˜๋ฅผ ์ž‘์„ฑ
27
  def get_text_file(docs):
28
- text_list = [] # ํ…์ŠคํŠธ๋ฅผ ์ €์žฅํ•  ๋ฆฌ์ŠคํŠธ
 
29
 
 
 
30
  for doc in docs:
31
  content = doc.getvalue().decode("utf-8") # ํŒŒ์ผ ๋‚ด์šฉ์„ ๊ฐ€์ ธ์™€์„œ UTF-8๋กœ ๋””์ฝ”๋”ฉํ•˜์—ฌ ๋ฌธ์ž์—ด๋กœ ๋ณ€ํ™˜
32
  text_list.append(content) # ํ…์ŠคํŠธ๋ฅผ ๋ฆฌ์ŠคํŠธ์— ์ถ”๊ฐ€
33
 
34
- return text_list # ํ…์ŠคํŠธ๊ฐ€ ๋‹ด๊ธด ๋ฆฌ์ŠคํŠธ ๋ฐ˜ํ™˜
 
 
 
 
 
 
 
35
 
36
  def get_csv_file(docs):
37
  pass
 
25
  # ๊ณผ์ œ
26
  # ์•„๋ž˜ ํ…์ŠคํŠธ ์ถ”์ถœ ํ•จ์ˆ˜๋ฅผ ์ž‘์„ฑ
27
  def get_text_file(docs):
28
+ # ํ…์ŠคํŠธ ์ƒ์„ฑ์„ ์œ„ํ•œ Hugging Face์˜ ํ…์ŠคํŠธ ์ƒ์„ฑ ํŒŒ์ดํ”„๋ผ์ธ ์ƒ์„ฑ
29
+ #text_generator = pipeline("text-generation", model="distilgpt2")
30
 
31
+ # ํ…์ŠคํŠธ ํŒŒ์ผ๋กœ๋ถ€ํ„ฐ ํ…์ŠคํŠธ๋ฅผ ์ฝ์–ด์™€ ๋ฆฌ์ŠคํŠธ์— ์ €์žฅํ•ฉ๋‹ˆ๋‹ค.
32
+ text_list = []
33
  for doc in docs:
34
  content = doc.getvalue().decode("utf-8") # ํŒŒ์ผ ๋‚ด์šฉ์„ ๊ฐ€์ ธ์™€์„œ UTF-8๋กœ ๋””์ฝ”๋”ฉํ•˜์—ฌ ๋ฌธ์ž์—ด๋กœ ๋ณ€ํ™˜
35
  text_list.append(content) # ํ…์ŠคํŠธ๋ฅผ ๋ฆฌ์ŠคํŠธ์— ์ถ”๊ฐ€
36
 
37
+ # ๊ฐ ํ…์ŠคํŠธ์— ๋Œ€ํ•ด ํ—ˆ๊น… ํŽ˜์ด์Šค ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ํ…์ŠคํŠธ๋ฅผ ์ƒ์„ฑํ•˜๊ณ  ๊ฒฐ๊ณผ๋ฅผ ์ €์žฅํ•ฉ๋‹ˆ๋‹ค.
38
+ generated_texts = []
39
+ for text in text_list:
40
+ generated_text = text_generator(text, max_length=50, num_return_sequences=1)
41
+ generated_texts.append(generated_text[0]['generated_text'])
42
+
43
+ return generated_texts # ์ƒ์„ฑ๋œ ํ…์ŠคํŠธ ๋ฆฌ์ŠคํŠธ ๋ฐ˜ํ™˜
44
+
45
 
46
  def get_csv_file(docs):
47
  pass