gllm-inference-binary 0.5.17__cp313-cp313-macosx_13_0_x86_64.whl → 0.5.18__cp313-cp313-macosx_13_0_x86_64.whl

gllm_inference/em_invoker/langchain/em_invoker_embeddings.pyi

@@ -1,14 +1,23 @@
 from gllm_inference.em_invoker.em_invoker import BaseEMInvoker as BaseEMInvoker
-from gllm_inference.schema import Vector as Vector
+from gllm_inference.schema import Attachment as Attachment, Vector as Vector
+from gllm_inference.utils.io_utils import DEFAULT_BASE64_ALLOWED_MIMETYPES as DEFAULT_BASE64_ALLOWED_MIMETYPES, base64_to_bytes as base64_to_bytes
 from langchain_core.embeddings import Embeddings
 from pydantic import BaseModel
 from typing import Any
 
 class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
-    """An adapter class that enables an `EMInvoker` to be used as a LangChain `Embeddings`.
+    '''An adapter class that enables an `EMInvoker` to be used as a LangChain `Embeddings`.
 
     Attributes:
         em_invoker (BaseEMInvoker): The `EMInvoker` instance to be interacted with.
+        use_base64 (bool):
+            Whether to apply strict base64 encoding to the input.
+            1, If `True`, only inputs with specific MIME types (e.g. images,
+              audio, and video) will be converted into base64 strings before being sent.
+            2. If `False`, each input is treated as a raw string.
+
+            This ensures "strict" handling: base64 encoding is not applied
+            universally, but only when required for those MIME types.
 
     Usage example:
     ```python
@@ -18,8 +27,9 @@ class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
     em_invoker = OpenAIEMInvoker(...)
     embeddings = EMInvokerEmbeddings(em_invoker=em_invoker)
     ```
-    """
+    '''
     em_invoker: BaseEMInvoker
+    use_base64: bool
     async def aembed_documents(self, texts: list[str], **kwargs: Any) -> list[Vector]:
         """Asynchronously embed documents using the `EMInvoker`.
 
@@ -29,6 +39,9 @@ class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
 
         Returns:
             list[Vector]: List of embeddings, one for each text.
+
+        Raises:
+            ValueError: If `texts` is not a list of strings.
         """
     async def aembed_query(self, text: str, **kwargs: Any) -> Vector:
         """Asynchronously embed query using the `EMInvoker`.
@@ -39,6 +52,9 @@ class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
 
         Returns:
             Vector: Embeddings for the text.
+
+        Raises:
+            ValueError: If `text` is not a string.
         """
     def embed_documents(self, texts: list[str], **kwargs: Any) -> list[Vector]:
         """Embed documents using the `EMInvoker`.
@@ -49,6 +65,9 @@ class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
 
         Returns:
             list[Vector]: List of embeddings, one for each text.
+
+        Raises:
+            ValueError: If `texts` is not a list of strings.
         """
     def embed_query(self, text: str, **kwargs: Any) -> Vector:
         """Embed query using the `EMInvoker`.
@@ -59,4 +78,7 @@ class EMInvokerEmbeddings(BaseModel, Embeddings, arbitrary_types_allowed=True):
 
         Returns:
             Vector: Embeddings for the text.
+
+        Raises:
+            ValueError: If `text` is not a string.
         """

gllm_inference/utils/__init__.pyi

@@ -1,4 +1,5 @@
+from gllm_inference.utils.io_utils import base64_to_bytes as base64_to_bytes
 from gllm_inference.utils.langchain import load_langchain_model as load_langchain_model, parse_model_data as parse_model_data
 from gllm_inference.utils.validation import validate_string_enum as validate_string_enum
 
-__all__ = ['load_langchain_model', 'parse_model_data', 'validate_string_enum']
+__all__ = ['base64_to_bytes', 'load_langchain_model', 'parse_model_data', 'validate_string_enum']

gllm_inference/utils/io_utils.pyi

@@ -0,0 +1,26 @@
+from _typeshed import Incomplete
+
+logger: Incomplete
+DEFAULT_BASE64_ALLOWED_MIMETYPES: Incomplete
+
+def base64_to_bytes(value: str, *, allowed_mimetypes: tuple[str, ...] | None = ...) -> str | bytes:
+    '''Decode a base64 string to bytes based on allowed MIME type.
+
+    The conversion steps are as follows:
+        1. The function first attempts to decode the given string from base64.
+        2. If decoding succeeds, it checks the MIME type of the decoded content.
+        3. When the MIME type matches one of the allowed patterns (e.g., ``"image/*"``),
+            the raw bytes are returned. Otherwise, the original string is returned unchanged.
+
+    Args:
+        value (str): Input data to decode.
+        allowed_mimetypes (tuple[str, ...], optional): MIME type prefixes that are allowed
+            to be decoded into bytes. Defaults to ("image/*", "audio/*", "video/*").
+
+    Returns:
+        str | bytes: Base64-encoded string or raw bytes if MIME type is allowed;
+            otherwise returns original string.
+
+    Raises:
+        ValueError: If the input is not a string.
+    '''

gllm_inference.pyi

@@ -117,4 +117,6 @@ import pathlib
 import pathlib.Path
 import filetype
 import magic
+import binascii
+import fnmatch
 import importlib

{gllm_inference_binary-0.5.17.dist-info → gllm_inference_binary-0.5.18.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gllm-inference-binary
-Version: 0.5.17
+Version: 0.5.18
 Summary: A library containing components related to model inferences in Gen AI applications.
 Author: Henry Wicaksono
 Author-email: henry.wicaksono@gdplabs.id

{gllm_inference_binary-0.5.17.dist-info → gllm_inference_binary-0.5.18.dist-info}/RECORD

@@ -15,7 +15,7 @@ gllm_inference/em_invoker/bedrock_em_invoker.pyi,sha256=KdX1PMWPfrmFIrSbf8Y6jkDu
 gllm_inference/em_invoker/em_invoker.pyi,sha256=dgIeAIetQcBmohwYwgo1vNw7YNO_3DQCobUaabBtf7g,5043
 gllm_inference/em_invoker/google_em_invoker.pyi,sha256=oDS4dXBcLg59ePeiLTwdl09927oJNZ_ykIe0n6Ba8gU,6557
 gllm_inference/em_invoker/langchain/__init__.pyi,sha256=VYGKE5OgU0my1RlhgzkU_A7-GLGnUDDnNFuctuRwILE,148
-gllm_inference/em_invoker/langchain/em_invoker_embeddings.pyi,sha256=6nASLqi0FXCpqyYPl7kM3g7hAW-xS5ZwsS3GFudns98,2347
+gllm_inference/em_invoker/langchain/em_invoker_embeddings.pyi,sha256=VU3-Vhb9BCDhJo8PPdWHe2rBEOCs_HMXT6ZaWwjUzZE,3304
 gllm_inference/em_invoker/langchain_em_invoker.pyi,sha256=HuQD5Do4jwqKoMMgMjgZkic2L21n2ayJewIce09bZ3M,3163
 gllm_inference/em_invoker/openai_compatible_em_invoker.pyi,sha256=GudWfL7QCAKLInMw94jTNogbyELlD9tDbrDErHB4RRI,5360
 gllm_inference/em_invoker/openai_em_invoker.pyi,sha256=vsfEmDNvwrlBhDxqsCKyMpMZbl_FaQUWHEgQc9yeo14,4656
@@ -91,12 +91,13 @@ gllm_inference/schema/token_usage.pyi,sha256=1GTQVORV0dBNmD_jix8aVaUqxMKFF04KpLP
 gllm_inference/schema/tool_call.pyi,sha256=zQaVxCnkVxOfOEhBidqohU85gb4PRwnwBiygKaunamk,389
 gllm_inference/schema/tool_result.pyi,sha256=cAG7TVtB4IWJPt8XBBbB92cuY1ZsX9M276bN9aqjcvM,276
 gllm_inference/schema/type_alias.pyi,sha256=cQRlLT5vz9YV50n9x5BmufgqUz0UByInpmAvHSv2WTY,719
-gllm_inference/utils/__init__.pyi,sha256=npmBmmlBv7cPHMg1hdL3S2_RelD6vk_LhCsGELhN_7s,295
+gllm_inference/utils/__init__.pyi,sha256=mDJ2gLSeQzm-79Tov5-dhrMNaDqgcD1VVzDYAWvIRqA,391
+gllm_inference/utils/io_utils.pyi,sha256=7kUTacHAVRYoemFUOjCH7-Qmw-YsQGd6rGYxjf_qmtw,1094
 gllm_inference/utils/langchain.pyi,sha256=VluQiHkGigDdqLUbhB6vnXiISCP5hHqV0qokYY6dC1A,1164
 gllm_inference/utils/validation.pyi,sha256=toxBtRp-VItC_X7sNi-GDd7sjibBdWMrR0q01OI2D7k,385
 gllm_inference.build/.gitignore,sha256=aEiIwOuxfzdCmLZe4oB1JsBmCUxwG8x-u-HBCV9JT8E,1
-gllm_inference.cpython-313-darwin.so,sha256=z2SVGeHahBPIH7z1PGOu8ZGK9YQrzgcEdvvjzRqr9cg,4152112
-gllm_inference.pyi,sha256=BVwSps0XHsSJ4zwbhDLQ-R5gar7s7oQD9o4FVN-Z1x0,3605
-gllm_inference_binary-0.5.17.dist-info/METADATA,sha256=XT6EE6ZK5-7X1bv4wjRjo93u_EvxAXd0cZKnin-Y54A,4608
-gllm_inference_binary-0.5.17.dist-info/WHEEL,sha256=PCOZcL_jcbAVhuFR5ylE4Mr-7HPGHAcfJk9OpuMh7RQ,107
-gllm_inference_binary-0.5.17.dist-info/RECORD,,
+gllm_inference.cpython-313-darwin.so,sha256=KSsTBbpELj6uU0DXWlgq-KpEzG-i5rpnYNjjuNmQI-k,4190792
+gllm_inference.pyi,sha256=uxl1voKdn19LurAHKEZLWbq9ryPO4UkJ1Nk1MM8IL34,3636
+gllm_inference_binary-0.5.18.dist-info/METADATA,sha256=Heo4b0XS1gQgOtbWPzIYPTic4-esBmsLMJ2LO68SP3U,4608
+gllm_inference_binary-0.5.18.dist-info/WHEEL,sha256=PCOZcL_jcbAVhuFR5ylE4Mr-7HPGHAcfJk9OpuMh7RQ,107
+gllm_inference_binary-0.5.18.dist-info/RECORD,,