ai-microcore 3.16.5__tar.gz → 4.0.0.dev2__tar.gz

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ai-microcore
-Version: 3.16.5
+Version: 4.0.0.dev2
 Summary: # Minimalistic Foundation for AI Applications
 Keywords: llm,large language models,ai,similarity search,ai search,gpt,openai
 Author-email: Vitalii Stepanenko <mail@vitalii.in>
@@ -132,6 +132,24 @@ See [transformers installation](https://huggingface.co/docs/transformers/install
     <br>💡 <small>Setting `USE_DOT_ENV` to `false` disables reading configuration files.</small>
 3.  OS environment variables have the lowest priority.
 
+### Vector Databases
+
+Vector database functions are available via `microcore.texts`.
+Default vector database is [Chroma](https://www.trychroma.com/).
+In order to use vector database functions, you need to install the `chromadb` package:
+```bash
+pip install chromadb
+```
+By default, MicroCore will use ChromaDB PersistentClient (if corresponding package is installed).
+Alternatively, you can run Chroma as separate service and configure MicroCore to use HttpClient:
+
+```python
+from microcore import configure
+configure(
+    EMBEDDING_DB_HOST = 'localhost',
+    EMBEDDING_DB_PORT = 8000,
+)
+```
 
 ## 🌟 Core Functions
 

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/README.md

@@ -107,6 +107,24 @@ See [transformers installation](https://huggingface.co/docs/transformers/install
     <br>💡 <small>Setting `USE_DOT_ENV` to `false` disables reading configuration files.</small>
 3.  OS environment variables have the lowest priority.
 
+### Vector Databases
+
+Vector database functions are available via `microcore.texts`.
+Default vector database is [Chroma](https://www.trychroma.com/).
+In order to use vector database functions, you need to install the `chromadb` package:
+```bash
+pip install chromadb
+```
+By default, MicroCore will use ChromaDB PersistentClient (if corresponding package is installed).
+Alternatively, you can run Chroma as separate service and configure MicroCore to use HttpClient:
+
+```python
+from microcore import configure
+configure(
+    EMBEDDING_DB_HOST = 'localhost',
+    EMBEDDING_DB_PORT = 8000,
+)
+```
 
 ## 🌟 Core Functions
 

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/__init__.py

@@ -161,4 +161,4 @@ __all__ = [
     # "wrappers",
 ]
 
-__version__ = "3.16.5"
+__version__ = "4.0.0-dev2"

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/_prepare_llm_args.py

@@ -22,7 +22,7 @@ def prepare_chat_messages(prompt: TPrompt) -> list[dict]:
             dict(role=DEFAULT_MESSAGE_ROLE, content=msg)
             if isinstance(msg, str)
             else (
-                asdict(msg, dict_factory=msg.dict_factory)
+                asdict(msg, dict_factory=msg.DICT_FACTORY)
                 if isinstance(msg, Msg)
                 else msg
             )

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/ai_func/__init__.py

@@ -5,13 +5,23 @@ descr: Allows to describe python functions for LLM
 
 import ast
 import inspect
+from enum import Enum
 from typing import Dict, Any
 import docstring_parser
 from .. import tpl
+from ..utils import dedent
 
 
+class AiFuncSyntax(str, Enum):
+    PYTHONIC: str = "pythonic"
+    JSON: str = "json"
+    DEFAULT: str = str(JSON)
+
+    def __str__(self):
+        return self.value
+
 def func_arg_comments(func):
-    func_source = inspect.getsource(func)
+    func_source = dedent(inspect.getsource(func))
     module = ast.parse(func_source)
     func_def = module.body[0]
 
@@ -60,7 +70,7 @@ def func_metadata(func) -> Dict[str, Any]:
     for name, val in metadata["args"].items():
         val["comment"] = arg_comments[name]
 
-        # Parse docstring
+    # Parse docstring
     parsed_docstring = docstring_parser.parse(inspect.getdoc(func))
 
     # Add descriptions from parsed docstring to parameters
@@ -71,6 +81,17 @@ def func_metadata(func) -> Dict[str, Any]:
     return metadata
 
 
-def describe_ai_func(func):
+def describe_ai_func(func: callable, syntax: AiFuncSyntax | str = None) -> str:
+    """
+    Renders function description for LLM
+    Args:
+        func: callable: function to describe
+        syntax: AiFuncSyntax | str: syntax to use for the description
+                - Use AiFuncSyntax enums to use standard templates (""json", "pythonic")
+                - Use custom template name to use custom template
+    Returns: str: rendered description, part of prompt
+    """
+    syntax = syntax or AiFuncSyntax.DEFAULT
+    tpl_file = f"ai-func.{syntax}.j2" if syntax in AiFuncSyntax else syntax
     metadata = func_metadata(func)
-    return tpl("python_ai_func.j2", **metadata)
+    return tpl(tpl_file, **metadata)

ai_microcore-4.0.0.dev2/microcore/ai_func/ai-func.json.j2

@@ -0,0 +1,11 @@
+# {{ description or name.replace('_', ' ').capitalize() }}
+{
+  "call": "{{ name }}"{% if args %}{{ "," }}{% endif %}
+{%- for k,v in args.items() %}
+  "{{ k }}":
+    {%- if v.type %} <{{ v.type }}>{% endif -%}
+    {%- if v.default != 'NOT_SET' %} (default = {{ v.default }}){%endif-%}
+    {%- if not loop.last -%},{%- endif -%}
+    {%- if v.comment %} {{ v.comment }}{% endif -%}
+{%- endfor -%}
+{{ "\n}" }}

ai_microcore-3.16.5/microcore/ai_func/python_ai_func.j2 → ai_microcore-4.0.0.dev2/microcore/ai_func/ai-func.pythonic.j2

@@ -1,4 +1,4 @@
-# {{ description }}
+# {{ description or name.replace('_', ' ').capitalize() }}
 {{ name }}(
 {%- for name,v in args.items() -%}
 {{ "\n\t" }}{{ name }}

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/configuration.py

@@ -369,6 +369,10 @@ class Config(LLMConfig):
 
     EMBEDDING_DB_ALLOW_DUPLICATES: bool = from_env(dtype=bool, default=False)
 
+    EMBEDDING_DB_HOST: str = from_env(default=None)
+
+    EMBEDDING_DB_PORT: str = from_env(default=None)
+
     DEFAULT_ENCODING: str = from_env("utf-8")
     """Used in file system operations, utf-8 by default"""
 

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/embedding_db/__init__.py

@@ -1,5 +1,4 @@
 import logging
-import sys
 from abc import ABC, abstractmethod
 from dataclasses import dataclass
 
@@ -8,6 +7,8 @@ import tiktoken
 from ..utils import ExtendedString
 
 
+INT32_MAX = 2**31 - 1  # 2147483647
+
 class SearchResults(list):
     def fit_to_token_size(
         self,
@@ -115,7 +116,7 @@ class AbstractEmbeddingDB(ABC):
         **kwargs,
     ) -> SearchResults | list[str | SearchResult]:
         return self.search(
-            collection, query, n_results=sys.maxsize - 1, where=where, **kwargs
+            collection, query, n_results=INT32_MAX, where=where, **kwargs
         )
 
     @abstractmethod

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/embedding_db/chromadb.py

@@ -1,3 +1,4 @@
+import logging
 from dataclasses import dataclass
 import uuid
 
@@ -16,10 +17,22 @@ class ChromaEmbeddingDB(AbstractEmbeddingDB):
     client: chromadb.Client = None
 
     def __post_init__(self):
-        self.client = chromadb.PersistentClient(
-            path=f"{self.config.STORAGE_PATH}/{self.config.EMBEDDING_DB_FOLDER}",
-            settings=Settings(anonymized_telemetry=False),
-        )
+        if self.config.EMBEDDING_DB_HOST:
+            logging.info(
+                "Connecting to ChromaDB at %s:%s",
+                self.config.EMBEDDING_DB_HOST,
+                self.config.EMBEDDING_DB_PORT
+            )
+            self.client = chromadb.HttpClient(
+                host=self.config.EMBEDDING_DB_HOST,
+                port=self.config.EMBEDDING_DB_PORT or 8000,
+                settings=Settings(anonymized_telemetry=False),
+            )
+        else:
+            self.client = chromadb.PersistentClient(
+                path=f"{self.config.STORAGE_PATH}/{self.config.EMBEDDING_DB_FOLDER}",
+                settings=Settings(anonymized_telemetry=False),
+            )
         self.embedding_function = (
             self.config.EMBEDDING_DB_FUNCTION
             or embedding_functions.DefaultEmbeddingFunction()

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/llm/shared.py

@@ -21,4 +21,5 @@ def prepare_callbacks(config: Config, args, set_stream: bool = True) -> list[cal
             callbacks.append(cb)
     if set_stream and "stream" not in args:
         args["stream"] = bool(callbacks)
-    return callbacks
+
+    return callbacks

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/logging.py

@@ -41,6 +41,20 @@ def _format_request_log_str(prompt, **kwargs) -> str:
         )
         if out.endswith("\n"):
             out = out[:-1]
+    if LoggingConfig.STRIP_REQUEST_LINES:
+        start_lines, end_lines = LoggingConfig.STRIP_REQUEST_LINES
+        max_lines = start_lines + end_lines
+        lines = out.split("\n")
+        if len(lines) > max_lines:
+            out = "\n".join(
+                lines[:start_lines]
+                + [
+                    f"{LoggingConfig.INDENT}{Fore.YELLOW}"
+                    f"...(output was truncated)..."
+                    f"{LoggingConfig.PROMPT_COLOR}"
+                ]
+                + (lines[-end_lines:] if end_lines else [])
+            )
     return out
 
 
@@ -72,6 +86,7 @@ class LoggingConfig:
     OUTPUT_METHOD: callable = print
     REQUEST_FORMATTER: callable = _format_request_log_str
     RESPONSE_FORMATTER: callable = _format_response_log_str
+    STRIP_REQUEST_LINES: tuple[int, int] | None = [40, 15]
 
 
 def _log_request(prompt, **kwargs):

{ai_microcore-3.16.5 → ai_microcore-4.0.0.dev2}/microcore/message_types.py

@@ -2,6 +2,7 @@
 
 from enum import Enum
 from dataclasses import dataclass, field
+from typing import ClassVar
 
 
 class Role(str, Enum):
@@ -9,16 +10,20 @@ class Role(str, Enum):
     USER = "user"
     ASSISTANT = "assistant"
 
+    def __str__(self):
+        return self.value
+
 
 DEFAULT_MESSAGE_ROLE = Role.USER
 
 
 @dataclass
 class Msg:
-    dict_factory = dict
     role: str = field(default=DEFAULT_MESSAGE_ROLE)
     content: str = field(default="")
 
+    DICT_FACTORY: ClassVar = dict
+
     def __str__(self):
         return str(self.content)
 
@@ -49,10 +54,11 @@ class PartialMsg(AssistantMsg):
         is_partial = True
         """Custom dictionary class to handle additional properties"""
 
-    dict_factory = _PartialMsgDict
     placeholder = "<|placeholder|>"
     variants_splitter = "<|or|>"
 
+    DICT_FACTORY: ClassVar = _PartialMsgDict
+
     @staticmethod
     def split_prefix_and_suffixes(content: str):
         parts = content.split(PartialMsg.placeholder)