Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -14,16 +14,7 @@ from langchain.document_loaders import PyPDFLoader, TextLoader, JSONLoader, CSVL
|
|
14 |
import tempfile # μμ νμΌμ μμ±νκΈ° μν λΌμ΄λΈλ¬λ¦¬μ
λλ€.
|
15 |
import os
|
16 |
|
17 |
-
|
18 |
-
print('file -type : ',file.type)
|
19 |
-
if file.type == 'text/plain' :
|
20 |
-
doc_list.extend(get_text_file(file))
|
21 |
-
elif file.type in ['application/octet-stream','application/pdf'] :
|
22 |
-
doc_list.extend(get_pdf_text(file))
|
23 |
-
elif file.type == 'text/csv' :
|
24 |
-
doc_list.extend(get_csv_file(file))
|
25 |
-
elif file.type == 'application/json' :
|
26 |
-
doc_list.extend(get_json_file(file))
|
27 |
|
28 |
# PDF λ¬Έμλ‘λΆν° ν
μ€νΈλ₯Ό μΆμΆνλ ν¨μμ
λλ€.
|
29 |
def get_pdf_text(pdf_docs):
|
@@ -38,17 +29,23 @@ def get_pdf_text(pdf_docs):
|
|
38 |
# κ³Όμ
|
39 |
# μλ ν
μ€νΈ μΆμΆ ν¨μλ₯Ό μμ±
|
40 |
|
41 |
-
def get_text_file(
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
|
|
|
|
|
|
46 |
|
47 |
-
def get_csv_file(
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
|
|
|
|
|
|
52 |
|
53 |
def get_json_file(docs):
|
54 |
text_list = []
|
|
|
14 |
import tempfile # μμ νμΌμ μμ±νκΈ° μν λΌμ΄λΈλ¬λ¦¬μ
λλ€.
|
15 |
import os
|
16 |
|
17 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
|
19 |
# PDF λ¬Έμλ‘λΆν° ν
μ€νΈλ₯Ό μΆμΆνλ ν¨μμ
λλ€.
|
20 |
def get_pdf_text(pdf_docs):
|
|
|
29 |
# κ³Όμ
|
30 |
# μλ ν
μ€νΈ μΆμΆ ν¨μλ₯Ό μμ±
|
31 |
|
32 |
+
def get_text_file(text_docs):
|
33 |
+
temp_dir = tempfile.TemporaryDirectory() # μμ λλ ν 리λ₯Ό μμ±ν©λλ€.
|
34 |
+
temp_filepath = os.path.join(temp_dir.name, text_docs.name) # μμ νμΌ κ²½λ‘λ₯Ό μμ±ν©λλ€.
|
35 |
+
with open(temp_filepath, "wb") as f: # μμ νμΌμ λ°μ΄λ리 μ°κΈ° λͺ¨λλ‘ μ½λλ€.
|
36 |
+
f.write(text_docs.getvalue()) # λ¬Έμμ λ΄μ©μ μμ νμΌμ μλλ€.
|
37 |
+
text_loader = TEXTLoader(temp_file_path)
|
38 |
+
text_doc = loader.load()# ν
μ€νΈλ₯Ό μΆμΆν©λλ€.
|
39 |
+
return text_doc # μΆμΆν ν
μ€νΈλ₯Ό λ°νν©λλ€.
|
40 |
|
41 |
+
def get_csv_file(csv_docs):
|
42 |
+
temp_dir = tempfile.TemporaryDirectory() # μμ λλ ν 리λ₯Ό μμ±ν©λλ€.
|
43 |
+
temp_filepath = os.path.join(temp_dir.name, csv_docs.name) # μμ νμΌ κ²½λ‘λ₯Ό μμ±ν©λλ€.
|
44 |
+
with open(temp_filepath, "wb") as f: # μμ νμΌμ λ°μ΄λ리 μ°κΈ° λͺ¨λλ‘ μ½λλ€.
|
45 |
+
f.write(csv_docs.getvalue()) # λ¬Έμμ λ΄μ©μ μμ νμΌμ μλλ€.
|
46 |
+
csv_loader = CSVLoader(temp_file_path)
|
47 |
+
data = loader.load()# ν
μ€νΈλ₯Ό μΆμΆν©λλ€.
|
48 |
+
return csv_doc # μΆμΆν ν
μ€νΈλ₯Ό λ°νν©λλ€.
|
49 |
|
50 |
def get_json_file(docs):
|
51 |
text_list = []
|