Migrate to Pydantic v2 APIs for Python 3.14 support

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "voyageai"
-version = "0.3.8"
+version = "0.3.9"
 description = ""
 authors = ["Yujie Qian <yujieq@voyageai.com>"]
 readme = "README.md"
@@ -16,7 +16,7 @@ numpy = [
 ]
 aiolimiter = "*"
 pillow = "*"
-pydantic = ">=1.10.8"
+pydantic = ">=2.7.4"
 tokenizers = ">=0.14.0"
 langchain-text-splitters = ">=0.3.8"
 ffmpeg-python = "*"

voyageai/client.py

@@ -170,7 +170,7 @@ def multimodal_embed(
                         truncation=truncation,
                         output_dtype=output_dtype,
                         output_dimension=output_dimension,
-                    ).dict(),
+                    ).model_dump(),
                     **self._params,
                 )
 

voyageai/client_async.py

@@ -170,7 +170,7 @@ async def multimodal_embed(
                         truncation=truncation,
                         output_dtype=output_dtype,
                         output_dimension=output_dimension,
-                    ).dict(),
+                    ).model_dump(),
                     **self._params,
                 )
         if response is None:

voyageai/object/multimodal_embeddings.py

@@ -5,16 +5,12 @@
 
 import PIL.Image
 import PIL.ImageFile
+from pydantic import BaseModel, Field, ValidationError
 
 from voyageai import error
 from voyageai.api_resources import VoyageResponse
 from voyageai.video_utils import Video
 
-try:
-    from pydantic import BaseModel, Extra, Field, ValidationError
-except ImportError:
-    from pydantic.v1 import BaseModel, Extra, Field, ValidationError
-
 
 class MultimodalEmbeddingsObject:
     def __init__(self, response: Optional[VoyageResponse] = None):
@@ -51,39 +47,39 @@ class MultimodalInputSegmentText(BaseModel):
     text: str
 
     class Config:
-        extra = Extra.forbid
+        extra = "forbid"
 
 
 class MultimodalInputSegmentImageURL(BaseModel):
     type: Literal[MultimodalInputSegmentType.IMAGE_URL] = MultimodalInputSegmentType.IMAGE_URL
     image_url: str
 
     class Config:
-        extra = Extra.forbid
+        extra = "forbid"
 
 
 class MultimodalInputSegmentImageBase64(BaseModel):
     type: Literal[MultimodalInputSegmentType.IMAGE_BASE64] = MultimodalInputSegmentType.IMAGE_BASE64
     image_base64: str
 
     class Config:
-        extra = Extra.forbid
+        extra = "forbid"
 
 
 class MultimodalInputSegmentVideoURL(BaseModel):
     type: Literal[MultimodalInputSegmentType.VIDEO_URL] = MultimodalInputSegmentType.VIDEO_URL
     video_url: str
 
     class Config:
-        extra = Extra.forbid
+        extra = "forbid"
 
 
 class MultimodalInputSegmentVideoBase64(BaseModel):
     type: Literal[MultimodalInputSegmentType.VIDEO_BASE64] = MultimodalInputSegmentType.VIDEO_BASE64
     video_base64: str
 
     class Config:
-        extra = Extra.forbid
+        extra = "forbid"
 
 
 class MultimodalInput(BaseModel):
@@ -98,7 +94,7 @@ class MultimodalInput(BaseModel):
             ],
             Field(discriminator="type"),
         ]
-    ] = Field(..., min_items=1)
+    ] = Field(..., min_length=1)
 
 
 class MultimodalInputRequest(BaseModel):
@@ -195,7 +191,7 @@ def _process_dict_input(cls, input_data: Dict, idx: int) -> MultimodalInput:
             raise ValueError(f"Input at index {idx} is missing the 'content' field.")
 
         try:
-            return MultimodalInput.parse_obj(input_data)
+            return MultimodalInput.model_validate(input_data)
         except ValidationError as ve:
             raise ValueError(f"Validation error for input at index {idx}: {ve}") from ve