Update app.py
Browse files
app.py
CHANGED
|
@@ -25,26 +25,14 @@ def get_pdf_text(pdf_docs):
|
|
| 25 |
# ๊ณผ์
|
| 26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
| 27 |
|
| 28 |
-
def
|
| 29 |
-
# ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์์ฑํฉ๋๋ค.
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
with open(temp_filepath, "w", encoding="utf-8") as f:
|
| 37 |
-
f.write(text_docs.getvalue())
|
| 38 |
-
|
| 39 |
-
# ์์ ํ์ผ์์ ํ
์คํธ๋ฅผ ์ฝ์ด์ต๋๋ค.
|
| 40 |
-
with open(temp_filepath, "r", encoding="utf-8") as f:
|
| 41 |
-
text_content = f.read()
|
| 42 |
-
|
| 43 |
-
# ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์๋์ผ๋ก ์ ๋ฆฌํฉ๋๋ค.
|
| 44 |
-
temp_dir.cleanup()
|
| 45 |
-
|
| 46 |
-
return [text_content] # ์ถ์ถํ ํ
์คํธ๋ฅผ ๋ฆฌ์คํธ์ ๋ด์ ๋ฐํํฉ๋๋ค.
|
| 47 |
-
|
| 48 |
|
| 49 |
|
| 50 |
|
|
|
|
| 25 |
# ๊ณผ์
|
| 26 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
| 27 |
|
| 28 |
+
def get_pdf_text(text_docs):
|
| 29 |
+
temp_dir = tempfile.TemporaryDirectory() # ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์์ฑํฉ๋๋ค.
|
| 30 |
+
temp_filepath = os.path.join(temp_dir.name, text_docs.name) # ์์ ํ์ผ ๊ฒฝ๋ก๋ฅผ ์์ฑํฉ๋๋ค.
|
| 31 |
+
with open(temp_filepath, "wb") as f: # ์์ ํ์ผ์ ๋ฐ์ด๋๋ฆฌ ์ฐ๊ธฐ ๋ชจ๋๋ก ์ฝ๋๋ค.
|
| 32 |
+
f.write(text_docs.getvalue()) # Text ๋ฌธ์์ ๋ด์ฉ์ ์์ ํ์ผ์ ์๋๋ค.
|
| 33 |
+
text_loader = TextLoader(temp_filepath) # TextLoader๋ฅผ ์ฌ์ฉํด PDF๋ฅผ ๋ก๋ํฉ๋๋ค.
|
| 34 |
+
text_doc = text_loader.load() # ํ
์คํธ๋ฅผ ์ถ์ถํฉ๋๋ค.
|
| 35 |
+
return text_doc # ์ถ์ถํ ํ
์คํธ๋ฅผ ๋ฐํํฉ๋๋ค.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
|
| 37 |
|
| 38 |
|