Update app.py
Browse files
app.py
CHANGED
@@ -14,7 +14,6 @@ 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 |
# PDF λ¬Έμλ‘λΆν° ν
μ€νΈλ₯Ό μΆμΆνλ ν¨μμ
λλ€.
|
19 |
def get_pdf_text(pdf_docs):
|
20 |
temp_dir = tempfile.TemporaryDirectory() # μμ λλ ν 리λ₯Ό μμ±ν©λλ€.
|
@@ -25,20 +24,39 @@ def get_pdf_text(pdf_docs):
|
|
25 |
pdf_doc = pdf_loader.load() # ν
μ€νΈλ₯Ό μΆμΆν©λλ€.
|
26 |
return pdf_doc # μΆμΆν ν
μ€νΈλ₯Ό λ°νν©λλ€.
|
27 |
|
28 |
-
# κ³Όμ
|
29 |
-
# μλ ν
μ€νΈ μΆμΆ ν¨μλ₯Ό μμ±
|
30 |
|
31 |
def get_text_file(docs):
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
|
35 |
def get_csv_file(docs):
|
36 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
|
38 |
def get_json_file(docs):
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
|
41 |
-
|
42 |
# λ¬Έμλ€μ μ²λ¦¬νμ¬ ν
μ€νΈ μ²ν¬λ‘ λλλ ν¨μμ
λλ€.
|
43 |
def get_text_chunks(documents):
|
44 |
text_splitter = RecursiveCharacterTextSplitter(
|
|
|
14 |
import tempfile # μμ νμΌμ μμ±νκΈ° μν λΌμ΄λΈλ¬λ¦¬μ
λλ€.
|
15 |
import os
|
16 |
|
|
|
17 |
# PDF λ¬Έμλ‘λΆν° ν
μ€νΈλ₯Ό μΆμΆνλ ν¨μμ
λλ€.
|
18 |
def get_pdf_text(pdf_docs):
|
19 |
temp_dir = tempfile.TemporaryDirectory() # μμ λλ ν 리λ₯Ό μμ±ν©λλ€.
|
|
|
24 |
pdf_doc = pdf_loader.load() # ν
μ€νΈλ₯Ό μΆμΆν©λλ€.
|
25 |
return pdf_doc # μΆμΆν ν
μ€νΈλ₯Ό λ°νν©λλ€.
|
26 |
|
|
|
|
|
27 |
|
28 |
def get_text_file(docs):
|
29 |
+
temp_dir = tempfile.TemporaryDirectory()
|
30 |
+
temp_filepath = os.path.join(temp_dir.name, docs.name)
|
31 |
+
with open(temp_filepath, "wb") as f:
|
32 |
+
f.write(docs.getvalue())
|
33 |
+
text_loader = TextLoader(temp_filepath)
|
34 |
+
text_doc = text_loader.load()
|
35 |
+
return text_doc
|
36 |
|
37 |
|
38 |
def get_csv_file(docs):
|
39 |
+
temp_dir = tempfile.TemporaryDirectory()
|
40 |
+
temp_filepath = os.path.join(temp_dir.name, docs.name)
|
41 |
+
with open(temp_filepath, "wb") as f:
|
42 |
+
f.write(docs.getvalue())
|
43 |
+
csv_loader = CSVLoader(temp_filepath)
|
44 |
+
csv_doc = csv_loader.load()
|
45 |
+
return csv_doc
|
46 |
|
47 |
def get_json_file(docs):
|
48 |
+
temp_dir = tempfile.TemporaryDirectory()
|
49 |
+
temp_filepath = os.path.join(temp_dir.name, docs.name)
|
50 |
+
with open(temp_filepath, "wb") as f:
|
51 |
+
f.write(docs.getvalue())
|
52 |
+
json_loader = JSONLoader(temp_filepath,
|
53 |
+
jq_schema='.scans[].relationships',
|
54 |
+
text_content=False)
|
55 |
+
|
56 |
+
json_doc = json_loader.load()
|
57 |
+
# print('json_doc = ',json_doc)
|
58 |
+
return json_doc
|
59 |
|
|
|
60 |
# λ¬Έμλ€μ μ²λ¦¬νμ¬ ν
μ€νΈ μ²ν¬λ‘ λλλ ν¨μμ
λλ€.
|
61 |
def get_text_chunks(documents):
|
62 |
text_splitter = RecursiveCharacterTextSplitter(
|