Refactor code to add new API module for utilities

docs/_sidebar.md

@@ -16,6 +16,7 @@
     - [semantix](api/semantix.md)
     - [llms](api/llms.md)
     - [types](api/types.md)
+    - [utils](api/utils.md)
 - Community
     - [Projects](community/projects.md)
     - [Contributors](community/contributors.md)

docs/api/semantix.md

@@ -40,7 +40,7 @@ class Person:
 ## enhance
 
 ```python
-@enhance(meaning: str, llm: BaseLLM, method: str = "Normal", tools: List[Callable] = [], retries=2, **kwargs)
+@enhance(meaning: str, llm: BaseLLM, method: str = "Normal", tools: List[Callable] = [], retries=2, return_additional_info=False **kwargs)
 ```
 
 A decorator to enhance the function with LLM capabilities.
@@ -53,11 +53,13 @@ A decorator to enhance the function with LLM capabilities.
     - The Large Language Model to use.
 - `method` : str, optional
     - The method to use for the enhancement. Default is `"Normal"`.
-    - Options are `"Normal"`, `"Reason"`, `"Chain-of-Thoughts"`, `"ReAct"`, `"Reflection"`.
+    - Options are `"Normal"`, `"Reason"`, `"CoT"`, `"ReAct"`, `"Reflection"`.
 - `tools` : List[Callable | Tool], optional
     - List of tools/functions to be used by the LLM. Default is `[]`.
 - `retries` : int, optional
     - The number of retries to use. Default is `2`.
+- `return_additional_info` : bool, optional
+    - Whether to return additional information in the form of `Output` Object. Default is `False`.
 - `**kwargs`
     - Additional keyword arguments to pass to the LLM.
     - For example, `temperature`, `max_tokens`, etc. The list of arguments depends on the LLM.

docs/api/types.md

@@ -4,13 +4,20 @@ The `semantix.types` module provides a set of classes and functions to work with
 
 ## Image
 
+```python
+class Image:
+    file_path: str
+    quality: str = "low"
+```
+
 The `Image` class is used to represent an image in the library. `Image` class converts the image into a base64 encoded string to be used in the Large Language Model.
 
 !> `pillow` library is required to use the `Image` class. If you don't have it installed, you can install it using the `pip install pillow` or `pip install semantix[image]`.
 
 ### Parameters
 
 - `file_path` (str): The path to the image file.
+- `quality` (str): The quality of the image. Default is `"low"`. Options are `"low"`, `"medium"`, `"high"`.
 
 ### Example
 
@@ -25,6 +32,13 @@ def get_person(img: Semantic[Image, "Image of the Person"]) -> Person:
 
 ## Video
 
+```python
+class Video:
+    file_path: str
+    seconds_per_frame: int = 2
+    quality: str = "low"
+```
+
 The `Video` class is used to represent a video in the library. `Video` class converts the video into a base64 encoded list of frames to be used in the Large Language Model.
 
 !> `opencv-python` library is required to use the `Video` class. If you don't have it installed, you can install it using the `pip install opencv-python-headless` or `pip install semantix[video]`.
@@ -33,6 +47,7 @@ The `Video` class is used to represent a video in the library. `Video` class con
 
 - `file_path` (str): The path to the video file.
 - `seconds_per_frame` (int): The number of seconds per frame to extract from the video. Default is 2.
+- `quality` (str): The quality of the video. Default is `"low"`. Options are `"low"`, `"medium"`, `"high"`.
 
 ### Example
 

docs/api/utils.md

@@ -0,0 +1,63 @@
+# Utilities API Reference
+
+The `semantix.utils` module provides a set of utility functions.
+
+## create_class
+
+```python
+create_class(
+    classname: str, fields: Dict[str, Tuple[Type, Any]], desc: str = ""
+) -> Any
+```
+
+Helper function to create a new dataclass with the given fields. This is helpful when creating classes with a large number of parameters.
+
+### Parameters
+
+- `classname` (str): The name of the class to be created.
+- `fields` (Dict[str, Tuple[Type, Any]): A dictionary containing the fields of the class, where the keys are the field names and the values are tuples containing the field type and default value.
+- `desc` (str, optional): Description of the class. Defaults to "".
+
+## Example
+
+```python
+from semantix.utils import create_class
+
+fields = {
+    "name": (str, ""),
+    "age": (int, 0),
+    "height": (float, 0.0),
+}
+
+Person = create_class("Person", fields, "A class to represent a person.")
+```
+
+## create_enum
+
+```python
+create_enum(
+    classname: str, fields: Dict[str, Any], desc: str = ""
+) -> Enum
+```
+
+Helper function to create a new Enum with the given fields. This is helpful when creating enums with a large number of values.
+
+### Parameters
+
+- `classname` (str): The name of the Enum class.
+- `fields` (Dict[str, Any]): A dictionary containing the field names and values for the Enum.
+- `desc` (str, optional): A description for the Enum. Defaults to "".
+
+## Example
+
+```python
+from semantix.utils import create_enum
+
+colors = {
+    "RED": 1,
+    "GREEN": 2,
+    "BLUE": 3,
+}
+
+Color = create_enum("Color", colors, "An Enum to represent colors.")
+```

docs/building-blocks/large-language-model.md

@@ -80,6 +80,8 @@ To check the supported parameters for the LLMs, you can refer to the documentati
 
 ## How to create a custom LLM?
 
+!> If your model follows a different message template, its little tricky. Check one of the in-built LLMs to understand the implementation.
+
 Each and every in-built LLM classes in Semantix are subclasses of the `BaseLLM` class. You can create your own custom LLM by subclassing the `BaseLLM` class and implementing the necessary methods.
 
 For example, to create a custom LLM that always returns a name of the model, you can do the following:

semantix/decorators.py

@@ -34,7 +34,7 @@ def enhance(
         method (str, optional): The enhancement method to be applied. Defaults to "Normal". Options are: "Normal", "Reason", "CoT", "ReAct", "Reflection".
         tools (List[Union[Callable, Tool]], optional): A list of functions or Tool objects that the LLM can use. Defaults to [].
         retries (int, optional): The number of retry attempts for LLM operations. Defaults to 2.
-        return_output_obj (bool, optional): Whether to return the output and additional information. Defaults to False.
+        return_additional_info (bool, optional): Whether to return the output and additional information. Defaults to False.
         **kwargs (dict): Additional keyword arguments to be passed to the LLM.
 
     Returns:
@@ -49,7 +49,7 @@ def enhance(
     @enhance(
         meaning="Summarize text",
         model=my_llm_instance,
-        method="Chain-of-Thoughts",
+        method="CoT",
         temperature=0.7
     )
     def summarize_text(text: str) -> str:

try.ipynb

@@ -18,6 +18,13 @@
         "</div>"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Initialize the LLM"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 1,
@@ -45,6 +52,26 @@
         "# os.environ['ANTHROPIC_API_KEY'] = getpass('Enter your Anthropic API key: ') # Uncomment this line if you want to use the Anthropic API"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from semantix.llms import OpenAI\n",
+        "# from semantix.llms import Anthropic # Uncomment this line if you want to use the Anthropic API\n",
+        "\n",
+        "llm = OpenAI()\n",
+        "# llm = Anthropic() # Uncomment this line if you want to use the Anthropic API"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Define the Structured Classes, Enums and Functions"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 4,
@@ -58,11 +85,6 @@
         "from typing import List\n",
         "\n",
         "from semantix import Semantic, enhance\n",
-        "from semantix.llms import OpenAI\n",
-        "# from semantix.llms import Anthropic # Uncomment this line if you want to use the Anthropic API\n",
-        "\n",
-        "llm = OpenAI()\n",
-        "# llm = Anthropic() # Uncomment this line if you want to use the Anthropic API\n",
         "\n",
         "\n",
         "class Personality(Enum):\n",
@@ -84,10 +106,11 @@
         "\n",
         "@dataclass\n",
         "class Person:\n",
+        "    \"\"\"Person Information\"\"\"\n",
         "    full_name: str\n",
-        "    yod: Semantic[int, \"Year of Death\"]\n",
-        "    personality: Semantic[Personality, \"Personality of the Person\"]\n",
-        "    life_works: Semantic[List[LifeWork], \"Life's Works of the Person\"]\n",
+        "    yod: Semantic[int, \"Year of Death\"] # type: ignore\n",
+        "    personality: Semantic[Personality, \"Personality of the Person\"] # type: ignore\n",
+        "    life_works: Semantic[List[LifeWork], \"Life's Works of the Person\"] # type: ignore\n",
         "\n",
         "    def __repr__(self) -> str:\n",
         "        repr_str = (\n",
@@ -101,11 +124,18 @@
         "\n",
         "@enhance(\"Get Person Information use common knowledge\", llm)\n",
         "def get_person_info(\n",
-        "    name: Semantic[str, \"Name of the Person\"],\n",
+        "    name: Semantic[str, \"Name of the Person\"], # type: ignore\n",
         ") -> Person:\n",
         "    ..."
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Run the functions as you would normally"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 5,
@@ -200,6 +230,13 @@
         "get_person_info(name=\"Martin Luther King Jr.\")"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### Multimodal Support"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 7,
@@ -277,7 +314,7 @@
         "\n",
         "@enhance(\"Get Person Information use common knowledge\", llm, method=\"Reason\")\n",
         "def get_person_info(\n",
-        "    img: Semantic[Image, \"Image of a Person\"],\n",
+        "    img: Semantic[Image, \"Image of a Person\"], # type: ignore\n",
         ") -> Person: ..."
       ]
     },