Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -25,13 +25,32 @@ def get_pdf_text(pdf_docs):
|
|
| 25 |
# ๊ณผ์
|
| 26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
| 27 |
def get_text_file(docs):
|
| 28 |
-
|
|
|
|
| 29 |
|
| 30 |
def get_csv_file(docs):
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
def get_json_file(docs):
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
|
| 37 |
# ๋ฌธ์๋ค์ ์ฒ๋ฆฌํ์ฌ ํ
์คํธ ์ฒญํฌ๋ก ๋๋๋ ํจ์์
๋๋ค.
|
|
|
|
| 25 |
# ๊ณผ์
|
| 26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
| 27 |
def get_text_file(docs):
|
| 28 |
+
text = file.getvalue().decode("utf-8") # ํ
์คํธ ํ์ผ์ ๋ฌธ์์ด๋ก ๋์ฝ๋ฉํฉ๋๋ค.
|
| 29 |
+
return [text] # ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ๋ด์ ๋ฐํํฉ๋๋ค.
|
| 30 |
|
| 31 |
def get_csv_file(docs):
|
| 32 |
+
import pandas as pd
|
| 33 |
+
|
| 34 |
+
# CSV ํ์ผ์ Pandas DataFrame์ผ๋ก ์ฝ์ต๋๋ค.
|
| 35 |
+
df = pd.read_csv(file)
|
| 36 |
+
|
| 37 |
+
# ์ฌ๊ธฐ์์ ์ํ๋ ์ด์ ์ ํํ๊ฑฐ๋ ์ ์ฒด ํ์ผ์ ์ฝ์ด์ฌ ์ ์์ต๋๋ค.
|
| 38 |
+
# ์๋ฅผ ๋ค์ด, 'text_column' ์ด์์ ํ
์คํธ๋ฅผ ์ถ์ถํ๋ ๋ฐฉ๋ฒ์ ๋ค์๊ณผ ๊ฐ์ต๋๋ค.
|
| 39 |
+
texts = df['text_column'].tolist()
|
| 40 |
+
|
| 41 |
+
return texts # ํ
์คํธ ๋ฆฌ์คํธ๋ฅผ ๋ฐํํฉ๋๋ค.
|
| 42 |
|
| 43 |
def get_json_file(docs):
|
| 44 |
+
import json
|
| 45 |
+
|
| 46 |
+
# JSON ํ์ผ์ ๋์ฝ๋ฉํ์ฌ ๋ฐ์ดํฐ๋ฅผ ์ถ์ถํฉ๋๋ค.
|
| 47 |
+
data = json.load(file)
|
| 48 |
+
|
| 49 |
+
# ์ฌ๊ธฐ์์ ์ ์ ํ ๋ฐฉ๋ฒ์ผ๋ก JSON ๋ฐ์ดํฐ์์ ํ
์คํธ๋ฅผ ์ถ์ถํฉ๋๋ค.
|
| 50 |
+
# ์๋ฅผ ๋ค์ด, 'text' ํค์ ํด๋นํ๋ ๊ฐ์ ์ถ์ถํ๋ ๋ฐฉ๋ฒ์ ๋ค์๊ณผ ๊ฐ์ต๋๋ค.
|
| 51 |
+
texts = [item['text'] for item in data]
|
| 52 |
+
|
| 53 |
+
return texts
|
| 54 |
|
| 55 |
|
| 56 |
# ๋ฌธ์๋ค์ ์ฒ๋ฆฌํ์ฌ ํ
์คํธ ์ฒญํฌ๋ก ๋๋๋ ํจ์์
๋๋ค.
|