Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -498,30 +498,27 @@ def initialize():
|
|
| 498 |
|
| 499 |
download_gitlab_project_by_version()
|
| 500 |
#download_gitlab_repo()
|
| 501 |
-
|
| 502 |
-
|
| 503 |
-
|
| 504 |
doc_partial_paths = ['docs/source/setup/']
|
| 505 |
doc_file_paths = ['docs/source/usage/lib.rst']
|
| 506 |
|
| 507 |
-
|
| 508 |
-
|
| 509 |
|
| 510 |
|
| 511 |
doc_files, doc_file_references = process_directory(DATA_DIR,doc_partial_paths, doc_file_paths)
|
| 512 |
print("LEEEEEEEEEEEENGTH of doc_files: ", len(doc_files))
|
| 513 |
-
#code_files, code_file_references = process_directory5(DATA_DIR, code_partial_paths, code_file_path)
|
| 514 |
|
| 515 |
-
|
| 516 |
-
|
| 517 |
-
#code_chunks = split_pythoncode_into_chunks(code_files, code_file_references, 1500, 0)
|
| 518 |
doc_chunks = split_into_chunks(doc_files, doc_file_references, CHUNK_SIZE, CHUNK_OVERLAP)
|
| 519 |
|
| 520 |
-
|
| 521 |
-
|
| 522 |
|
| 523 |
docstore = setup_vectorstore(doc_chunks, EMBEDDING_MODEL_NAME, PERSIST_DOC_DIRECTORY)
|
| 524 |
-
|
| 525 |
|
| 526 |
llm = setup_llm(LLM_MODEL_NAME, LLM_TEMPERATURE, GROQ_API_KEY)
|
| 527 |
|
|
|
|
| 498 |
|
| 499 |
download_gitlab_project_by_version()
|
| 500 |
#download_gitlab_repo()
|
| 501 |
+
code_partial_paths = ['kadi_apy/lib/']
|
| 502 |
+
code_file_paths = []
|
| 503 |
+
doc_partial_paths = []
|
| 504 |
doc_partial_paths = ['docs/source/setup/']
|
| 505 |
doc_file_paths = ['docs/source/usage/lib.rst']
|
| 506 |
|
| 507 |
+
code_files, code_file_references = process_directory(DATA_DIR,code_partial_paths, code_file_paths)
|
| 508 |
+
print("LEEEEEEEEEEEENGTH of code_files: ", len(code_files))
|
| 509 |
|
| 510 |
|
| 511 |
doc_files, doc_file_references = process_directory(DATA_DIR,doc_partial_paths, doc_file_paths)
|
| 512 |
print("LEEEEEEEEEEEENGTH of doc_files: ", len(doc_files))
|
|
|
|
| 513 |
|
| 514 |
+
code_chunks = split_pythoncode_into_chunks(code_files, code_file_references, 1500, 0)
|
|
|
|
|
|
|
| 515 |
doc_chunks = split_into_chunks(doc_files, doc_file_references, CHUNK_SIZE, CHUNK_OVERLAP)
|
| 516 |
|
| 517 |
+
print(f"Total number of code_chunks: {len(code_chunks)}")
|
| 518 |
+
print(f"Total number of doc_chunks: {len(doc_chunks)}")
|
| 519 |
|
| 520 |
docstore = setup_vectorstore(doc_chunks, EMBEDDING_MODEL_NAME, PERSIST_DOC_DIRECTORY)
|
| 521 |
+
codestore = setup_vectorstore(code_chunks, EMBEDDING_MODEL_NAME, PERSIST_CODE_DIRECTORY)
|
| 522 |
|
| 523 |
llm = setup_llm(LLM_MODEL_NAME, LLM_TEMPERATURE, GROQ_API_KEY)
|
| 524 |
|