please
Browse files
app.py
CHANGED
@@ -30,19 +30,47 @@ def get_pdf_text(pdf_docs):
|
|
30 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
31 |
|
32 |
def get_text_file(docs):
|
33 |
-
|
34 |
-
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
|
37 |
def get_csv_file(docs):
|
38 |
-
|
39 |
-
|
40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
|
42 |
def get_json_file(docs):
|
43 |
-
|
44 |
-
|
45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
|
48 |
# ๋ฌธ์๋ค์ ์ฒ๋ฆฌํ์ฌ ํ
์คํธ ์ฒญํฌ๋ก ๋๋๋ ํจ์์
๋๋ค.
|
|
|
30 |
# ์๋ ํ
์คํธ ์ถ์ถ ํจ์๋ฅผ ์์ฑ
|
31 |
|
32 |
def get_text_file(docs):
|
33 |
+
def get_text_file(docs):
|
34 |
+
temp_dir = tempfile.TemporaryDirectory() # ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์์ฑํฉ๋๋ค.
|
35 |
+
temp_filepath = os.path.join(temp_dir.name, "text_file.txt") # ์์ ํ์ผ ๊ฒฝ๋ก๋ฅผ ์์ฑํฉ๋๋ค.
|
36 |
+
with open(temp_filepath, "w", encoding="utf-8") as f: # ์์ ํ์ผ์ ํ
์คํธ ์ฐ๊ธฐ ๋ชจ๋๋ก ์ฝ๋๋ค.
|
37 |
+
f.write(docs) # ํ
์คํธ ๋ฌธ์์ ๋ด์ฉ์ ์์ ํ์ผ์ ์๋๋ค.
|
38 |
+
with open(temp_filepath, "r", encoding="utf-8") as f: # ์์ ํ์ผ์ ํ
์คํธ ์ฝ๊ธฐ ๋ชจ๋๋ก ์ฝ๋๋ค.
|
39 |
+
text_content = f.read() # ํ
์คํธ ํ์ผ์ ๋ด์ฉ์ ์ฝ์ต๋๋ค.
|
40 |
+
return text_content # ์ถ์ถํ ํ
์คํธ๋ฅผ ๋ฐํํฉ๋๋ค.
|
41 |
+
|
42 |
|
43 |
def get_csv_file(docs):
|
44 |
+
temp_dir = tempfile.TemporaryDirectory() # ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์์ฑํฉ๋๋ค.
|
45 |
+
temp_filepath = os.path.join(temp_dir.name, "csv_file.csv") # ์์ ํ์ผ ๊ฒฝ๋ก๋ฅผ ์์ฑํฉ๋๋ค.
|
46 |
+
|
47 |
+
# ํ
์คํธ ๋ฐ์ดํฐ๋ฅผ CSV ํ์ผ์ ์ฐ๊ธฐ
|
48 |
+
with open(temp_filepath, "w", newline="", encoding="utf-8") as csv_file:
|
49 |
+
csv_writer = csv.writer(csv_file)
|
50 |
+
for line in docs:
|
51 |
+
# ์ฌ๊ธฐ์ line์ CSV ํ์ผ์ ํ ํ์ ๋ํ๋
๋๋ค.
|
52 |
+
csv_writer.writerow(line.split(',')) # ์์: ์ผํ๋ก ๊ตฌ๋ถ๋ ๋ฐ์ดํฐ๋ฅผ ์ฌ์ฉ
|
53 |
+
|
54 |
+
# CSV ํ์ผ์์ ๋ฐ์ดํฐ ์ฝ๊ธฐ
|
55 |
+
with open(temp_filepath, "r", encoding="utf-8") as csv_file:
|
56 |
+
csv_reader = csv.reader(csv_file)
|
57 |
+
csv_data = [row for row in csv_reader]
|
58 |
+
|
59 |
+
return csv_data # ์ฝ์ด์จ CSV ๋ฐ์ดํฐ๋ฅผ ๋ฐํํฉ๋๋ค.
|
60 |
|
61 |
def get_json_file(docs):
|
62 |
+
temp_dir = tempfile.TemporaryDirectory() # ์์ ๋๋ ํ ๋ฆฌ๋ฅผ ์์ฑํฉ๋๋ค.
|
63 |
+
temp_filepath = os.path.join(temp_dir.name, "json_file.json") # ์์ ํ์ผ ๊ฒฝ๋ก๋ฅผ ์์ฑํฉ๋๋ค.
|
64 |
+
|
65 |
+
# ํ
์คํธ ๋ฐ์ดํฐ๋ฅผ JSON ํ์ผ์ ์ฐ๊ธฐ
|
66 |
+
with open(temp_filepath, "w", encoding="utf-8") as json_file:
|
67 |
+
json.dump(docs, json_file, indent=2) # ๋ฐ์ดํฐ๋ฅผ JSON ํ์ผ์ ์ฐ๊ธฐ
|
68 |
+
|
69 |
+
# JSON ํ์ผ์์ ๋ฐ์ดํฐ ์ฝ๊ธฐ
|
70 |
+
with open(temp_filepath, "r", encoding="utf-8") as json_file:
|
71 |
+
json_data = json.load(json_file)
|
72 |
+
|
73 |
+
return json_data # ์ฝ์ด์จ JSON ๋ฐ์ดํฐ๋ฅผ ๋ฐํํฉ๋๋ค.
|
74 |
|
75 |
|
76 |
# ๋ฌธ์๋ค์ ์ฒ๋ฆฌํ์ฌ ํ
์คํธ ์ฒญํฌ๋ก ๋๋๋ ํจ์์
๋๋ค.
|