Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -52,6 +52,31 @@ login(HF_TOKEN)
|
|
52 |
api = HfApi()
|
53 |
|
54 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
def rag_workflow(query):
|
56 |
"""
|
57 |
RAGChain class to perform the complete RAG workflow.
|
@@ -98,32 +123,14 @@ def rag_workflow(query):
|
|
98 |
|
99 |
return response
|
100 |
|
101 |
-
|
102 |
-
def initialize():
|
103 |
-
global vectorstore, chunks, llm
|
104 |
-
|
105 |
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
doc_texts, doc_references = extract_repo_files(DATA_DIR, [], [])
|
111 |
-
|
112 |
-
print("LEEEEEEEEEEEENGTH of code_texts: ", len(code_texts))
|
113 |
-
print("LEEEEEEEEEEEENGTH of doc_files: ", len(doc_texts))
|
114 |
-
|
115 |
-
code_chunks = chunk_pythoncode_and_add_metadata(code_texts, code_references)
|
116 |
-
doc_chunks = chunk_text_and_add_metadata(doc_texts, doc_references, CHUNK_SIZE, CHUNK_OVERLAP)
|
117 |
-
|
118 |
-
print(f"Total number of code_chunks: {len(code_chunks)}")
|
119 |
-
print(f"Total number of doc_chunks: {len(doc_chunks)}")
|
120 |
|
121 |
-
|
122 |
-
llm = get_groq_llm(LLM_MODEL_NAME, LLM_MODEL_TEMPERATURE, GROQ_API_KEY)
|
123 |
|
124 |
-
from langchain_community.document_loaders import TextLoader
|
125 |
-
|
126 |
-
initialize()
|
127 |
|
128 |
|
129 |
# Gradio utils
|
@@ -140,14 +147,6 @@ def add_text(history, text):
|
|
140 |
|
141 |
import gradio as gr
|
142 |
|
143 |
-
|
144 |
-
def bot_kadi(history):
|
145 |
-
user_query = history[-1][0]
|
146 |
-
response = rag_workflow(user_query)
|
147 |
-
history[-1] = (user_query, response)
|
148 |
-
|
149 |
-
yield history
|
150 |
-
|
151 |
def main():
|
152 |
with gr.Blocks() as demo:
|
153 |
gr.Markdown("## KadiAPY - AI Coding-Assistant")
|
|
|
52 |
api = HfApi()
|
53 |
|
54 |
|
55 |
+
def initialize():
|
56 |
+
global vectorstore, chunks, llm
|
57 |
+
|
58 |
+
|
59 |
+
download_gitlab_repo_to_hfspace(GITLAB_API_URL, GITLAB_PROJECT_ID, GITLAB_PROJECT_VERSION)
|
60 |
+
|
61 |
+
code_texts, code_references = extract_repo_files(DATA_DIR, ['kadi_apy'], [])
|
62 |
+
#doc_texts, doc_references = extract_files_and_filepath_from_dir(DATA_DIR, ['docs/source/'], [])
|
63 |
+
doc_texts, doc_references = extract_repo_files(DATA_DIR, [], [])
|
64 |
+
|
65 |
+
print("LEEEEEEEEEEEENGTH of code_texts: ", len(code_texts))
|
66 |
+
print("LEEEEEEEEEEEENGTH of doc_files: ", len(doc_texts))
|
67 |
+
|
68 |
+
code_chunks = chunk_pythoncode_and_add_metadata(code_texts, code_references)
|
69 |
+
doc_chunks = chunk_text_and_add_metadata(doc_texts, doc_references, CHUNK_SIZE, CHUNK_OVERLAP)
|
70 |
+
|
71 |
+
print(f"Total number of code_chunks: {len(code_chunks)}")
|
72 |
+
print(f"Total number of doc_chunks: {len(doc_chunks)}")
|
73 |
+
|
74 |
+
vectorstore = setup_vectorstore(doc_chunks + code_chunks, EMBEDDING_MODEL_NAME, VECTORSTORE_DIRECTORY)
|
75 |
+
llm = get_groq_llm(LLM_MODEL_NAME, LLM_MODEL_TEMPERATURE, GROQ_API_KEY)
|
76 |
+
|
77 |
+
|
78 |
+
initialize()
|
79 |
+
|
80 |
def rag_workflow(query):
|
81 |
"""
|
82 |
RAGChain class to perform the complete RAG workflow.
|
|
|
123 |
|
124 |
return response
|
125 |
|
|
|
|
|
|
|
|
|
126 |
|
127 |
+
def bot_kadi(history):
|
128 |
+
user_query = history[-1][0]
|
129 |
+
response = rag_workflow(user_query)
|
130 |
+
history[-1] = (user_query, response)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
|
132 |
+
yield history
|
|
|
133 |
|
|
|
|
|
|
|
134 |
|
135 |
|
136 |
# Gradio utils
|
|
|
147 |
|
148 |
import gradio as gr
|
149 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
150 |
def main():
|
151 |
with gr.Blocks() as demo:
|
152 |
gr.Markdown("## KadiAPY - AI Coding-Assistant")
|