Commit
ยท
f7e4221
1
Parent(s):
19f561f
Update app.py
Browse files
app.py
CHANGED
@@ -26,22 +26,13 @@ def get_pdf_text(pdf_docs):
|
|
26 |
# ๊ณผ์
|
27 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
28 |
def get_text_file(docs):
|
29 |
-
# ํ
์คํธ ์์ฑ์ ์ํ Hugging Face์ ํ
์คํธ ์์ฑ ํ์ดํ๋ผ์ธ ์์ฑ
|
30 |
-
text_generator = pipeline("text-generation", model="distilgpt2")
|
31 |
-
|
32 |
# ํ
์คํธ ํ์ผ๋ก๋ถํฐ ํ
์คํธ๋ฅผ ์ฝ์ด์ ๋ฆฌ์คํธ์ ์ ์ฅํฉ๋๋ค.
|
33 |
text_list = []
|
34 |
for doc in docs:
|
35 |
content = doc.getvalue().decode("utf-8") # ํ์ผ ๋ด์ฉ์ ๊ฐ์ ธ์์ UTF-8๋ก ๋์ฝ๋ฉํ์ฌ ๋ฌธ์์ด๋ก ๋ณํ
|
36 |
text_list.append(content) # ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ์ถ๊ฐ
|
37 |
|
38 |
-
#
|
39 |
-
generated_texts = []
|
40 |
-
for text in text_list:
|
41 |
-
generated_text = text_generator(text, max_length=50, num_return_sequences=1)
|
42 |
-
generated_texts.append(generated_text[0]['generated_text'])
|
43 |
-
|
44 |
-
return generated_texts # ์์ฑ๋ ํ
์คํธ ๋ฆฌ์คํธ ๋ฐํ
|
45 |
|
46 |
|
47 |
def get_csv_file(docs):
|
|
|
26 |
# ๊ณผ์
|
27 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
28 |
def get_text_file(docs):
|
|
|
|
|
|
|
29 |
# ํ
์คํธ ํ์ผ๋ก๋ถํฐ ํ
์คํธ๋ฅผ ์ฝ์ด์ ๋ฆฌ์คํธ์ ์ ์ฅํฉ๋๋ค.
|
30 |
text_list = []
|
31 |
for doc in docs:
|
32 |
content = doc.getvalue().decode("utf-8") # ํ์ผ ๋ด์ฉ์ ๊ฐ์ ธ์์ UTF-8๋ก ๋์ฝ๋ฉํ์ฌ ๋ฌธ์์ด๋ก ๋ณํ
|
33 |
text_list.append(content) # ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ์ถ๊ฐ
|
34 |
|
35 |
+
return text_list # ์ถ์ถ๋ ํ
์คํธ ๋ฆฌ์คํธ ๋ฐํ
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
|
37 |
|
38 |
def get_csv_file(docs):
|