Commit
ยท
8d921b1
1
Parent(s):
0df3cee
Update app.py
Browse files
app.py
CHANGED
@@ -25,13 +25,23 @@ def get_pdf_text(pdf_docs):
|
|
25 |
# ๊ณผ์
|
26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
27 |
def get_text_file(docs):
|
28 |
-
|
|
|
29 |
|
|
|
|
|
30 |
for doc in docs:
|
31 |
content = doc.getvalue().decode("utf-8") # ํ์ผ ๋ด์ฉ์ ๊ฐ์ ธ์์ UTF-8๋ก ๋์ฝ๋ฉํ์ฌ ๋ฌธ์์ด๋ก ๋ณํ
|
32 |
text_list.append(content) # ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ์ถ๊ฐ
|
33 |
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
36 |
def get_csv_file(docs):
|
37 |
pass
|
|
|
25 |
# ๊ณผ์
|
26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
27 |
def get_text_file(docs):
|
28 |
+
# ํ
์คํธ ์์ฑ์ ์ํ Hugging Face์ ํ
์คํธ ์์ฑ ํ์ดํ๋ผ์ธ ์์ฑ
|
29 |
+
#text_generator = pipeline("text-generation", model="distilgpt2")
|
30 |
|
31 |
+
# ํ
์คํธ ํ์ผ๋ก๋ถํฐ ํ
์คํธ๋ฅผ ์ฝ์ด์ ๋ฆฌ์คํธ์ ์ ์ฅํฉ๋๋ค.
|
32 |
+
text_list = []
|
33 |
for doc in docs:
|
34 |
content = doc.getvalue().decode("utf-8") # ํ์ผ ๋ด์ฉ์ ๊ฐ์ ธ์์ UTF-8๋ก ๋์ฝ๋ฉํ์ฌ ๋ฌธ์์ด๋ก ๋ณํ
|
35 |
text_list.append(content) # ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ์ถ๊ฐ
|
36 |
|
37 |
+
# ๊ฐ ํ
์คํธ์ ๋ํด ํ๊น
ํ์ด์ค ๋ชจ๋ธ์ ์ฌ์ฉํ์ฌ ํ
์คํธ๋ฅผ ์์ฑํ๊ณ ๊ฒฐ๊ณผ๋ฅผ ์ ์ฅํฉ๋๋ค.
|
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 # ์์ฑ๋ ํ
์คํธ ๋ฆฌ์คํธ ๋ฐํ
|
44 |
+
|
45 |
|
46 |
def get_csv_file(docs):
|
47 |
pass
|