Update app.py
Browse files
app.py
CHANGED
|
@@ -23,9 +23,10 @@ def extract_text(url: str, language: str = "ja", length: int = 150000):
|
|
| 23 |
pdf_file_path = os.path.join("/app/downloads", os.path.basename(url))
|
| 24 |
with open(pdf_file_path, "wb") as pdf_file:
|
| 25 |
pdf_file.write(response.content)
|
| 26 |
-
loader = UnstructuredPDFLoader(pdf_file_path)
|
| 27 |
-
docs = loader.load()
|
| 28 |
-
text_content = docs[0].page_content
|
|
|
|
| 29 |
else:
|
| 30 |
# それ以外の場合
|
| 31 |
loader = WebBaseLoader(url)
|
|
|
|
| 23 |
pdf_file_path = os.path.join("/app/downloads", os.path.basename(url))
|
| 24 |
with open(pdf_file_path, "wb") as pdf_file:
|
| 25 |
pdf_file.write(response.content)
|
| 26 |
+
# loader = UnstructuredPDFLoader(pdf_file_path)
|
| 27 |
+
# docs = loader.load()
|
| 28 |
+
# text_content = docs[0].page_content
|
| 29 |
+
text_content = pdf_file_path
|
| 30 |
else:
|
| 31 |
# それ以外の場合
|
| 32 |
loader = WebBaseLoader(url)
|