Merge pull request #114 from arjbingly/make-config

Ability to create a defaults config file

src/docs/get_started.config.rst

@@ -4,6 +4,14 @@ Configuration
 GRAG gives the user an option to use a config file, in the form of a ``config.ini``.
 The use of a config file streamlines the process of passing arguments to the various components in the code.
 
+Generate Config file
+**********************
+To generate a starters config file with default values, run the below command at the location you want the config file.
+
+``
+python -m grag.components.create_config
+``
+
 File Resolution
 ****************
 GRAG takes the closest ``config.ini`` to the file you run. This enables users to have multiple config files per project,

src/grag/components/create_config.py

@@ -0,0 +1,43 @@
+"""Runnable file for creating a default config.ini file."""
+
+import shutil
+from pathlib import Path
+from typing import Union
+
+import grag.resources
+from importlib_resources import files
+
+
+def create_config(path: Union[str, Path] = '.') -> None:
+    """Create a configuration file if it doesn't exist.
+
+    This function checks for the existence of a 'config.ini' file at the given path.
+    If the file does not exist, it copies a default configuration file from the package's
+    resources to the specified location. If the file already exists, it notifies the user
+    and does not overwrite the existing file.
+
+    Args:
+        path (Union[str, Path]): The directory path where the 'config.ini' should be 
+                                 located. If not specified, defaults to the current 
+                                 directory ('.').
+
+    Returns:
+        None
+
+    Raises:
+        FileNotFoundError: If the default configuration file does not exist.
+        PermissionError: If the process does not have permission to write to the specified
+                         directory.
+    """
+    default_config_path = files(grag.resources).joinpath('default_config.ini')
+    path = Path(path) / 'config.ini'
+    path = path.resolve()
+    if path.exists():
+        print('Config file already exists')
+    else:
+        shutil.copyfile(default_config_path, path, follow_symlinks=True)
+        print(f"Created config file at {path}")
+
+
+if __name__ == '__main__':
+    create_config()

src/grag/resources/default_config.ini

@@ -0,0 +1,64 @@
+; This is the default config.ini file generated by GRAG
+; All values are same as package defaults
+; Values that do not have a default value are commented out
+
+; Make sure to fill in the path configs under data, env, quantize and root (whatever applies)
+
+[llm]
+;model_name : Llama-2-13b-chat
+;quantization : Q5_K_M
+;pipeline : llama_cpp
+device_map : auto
+task : text-generation
+max_new_tokens : 1024
+temperature : 0.1
+n_batch : 1024
+n_ctx : 6000
+n_gpu_layers : -1
+std_out : True
+base_dir : ${root:root_path}/models
+
+[chroma_client]
+host : localhost
+port : 8000
+collection_name : grag
+embedding_type : instructor-embedding
+embedding_model : hkunlp/instructor-xl
+
+[deeplake_client]
+collection_name : grag
+embedding_type : instructor-embedding
+embedding_model : hkunlp/instructor-xl
+store_path : ${data:data_path}/vectordb
+
+[text_splitter]
+chunk_size : 2000
+chunk_overlap : 400
+
+[multivec_retriever]
+store_path : ${data:data_path}/doc_store
+top_k : 3
+id_key : doc_id
+namespace : 71e4b558187b270922923569301f1039
+
+[parse_pdf]
+single_text_out : True
+strategy : hi_res
+infer_table_structure : True
+extract_images : True
+image_output_dir : None
+add_captions_to_text : True
+add_captions_to_blocks : True
+table_as_html : False
+
+;[data]
+;data_path : ${root:root_path}/data
+;
+;[env]
+;env_path : ${root:root_path}/.env
+;
+;[quantize]
+;llama_cpp_path : ${root:root_path}
+;
+;[root]
+;root_path : ~/Capstone_5