Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -50,45 +50,6 @@ api = HfApi()
|
|
50 |
|
51 |
|
52 |
|
53 |
-
def split_python_code_into_chunks(texts, file_paths):
|
54 |
-
chunks = []
|
55 |
-
for text, file_path in zip(texts, file_paths):
|
56 |
-
"""
|
57 |
-
Custom made python code splitter, algorithm iterates through child nodes of ast-tree(max child depth = 2)
|
58 |
-
aims to have full body of methods along signature (+ can handle decorators) in a chunk and adds method specific metadata
|
59 |
-
e.g visbility: public, _internal
|
60 |
-
type: "class", "methods", "command"(CLI commands)
|
61 |
-
source:
|
62 |
-
|
63 |
-
|
64 |
-
with the intend to use a filter when retrieving potentaion useful snippets.
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
"""
|
69 |
-
document_chunks = chunk_python_code_with_metadata(text, file_path)
|
70 |
-
chunks.extend(document_chunks)
|
71 |
-
return chunks
|
72 |
-
|
73 |
-
|
74 |
-
# Split text into chunks
|
75 |
-
def split_into_chunks(texts, references, chunk_size, chunk_overlap):
|
76 |
-
text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
|
77 |
-
chunks = []
|
78 |
-
|
79 |
-
for text, reference in zip(texts, references):
|
80 |
-
chunks.extend([
|
81 |
-
Document(
|
82 |
-
page_content=chunk,
|
83 |
-
metadata={
|
84 |
-
"source": reference,
|
85 |
-
"usage": "doc"
|
86 |
-
}
|
87 |
-
)
|
88 |
-
for chunk in text_splitter.split_text(text)
|
89 |
-
])
|
90 |
-
return chunks
|
91 |
-
|
92 |
|
93 |
# Setup Vectorstore
|
94 |
def embed_documents_into_vectorstore(chunks, model_name, persist_directory):
|
|
|
50 |
|
51 |
|
52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
|
54 |
# Setup Vectorstore
|
55 |
def embed_documents_into_vectorstore(chunks, model_name, persist_directory):
|