feat/refactor embedders (#383)

* refactor embedders

* fix mixedbreadai embedder

* create default behavior for embed_query

* fix default behavior of _embed_query

CHANGELOG.md

@@ -1,7 +1,9 @@
-## 0.5.3-dev0
+## 0.5.3-dev1
 
 ### Enhancements
 
+* **Optimize embedder code** - Move duplicate code to base interface, exit early if no elements have text. 
+
 ### Fixes
 
 ## 0.5.2

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "0.5.3-dev0"  # pragma: no cover
+__version__ = "0.5.3-dev1"  # pragma: no cover

unstructured_ingest/embed/azure_openai.py

@@ -44,7 +44,13 @@ def get_async_client(self) -> "AsyncAzureOpenAI":
 class AzureOpenAIEmbeddingEncoder(OpenAIEmbeddingEncoder):
     config: AzureOpenAIEmbeddingConfig
 
+    def get_client(self) -> "AzureOpenAI":
+        return self.config.get_client()
+
 
 @dataclass
 class AsyncAzureOpenAIEmbeddingEncoder(AsyncOpenAIEmbeddingEncoder):
     config: AzureOpenAIEmbeddingConfig
+
+    def get_client(self) -> "AsyncAzureOpenAI":
+        return self.config.get_async_client()

unstructured_ingest/embed/bedrock.py

@@ -15,7 +15,13 @@
 )
 from unstructured_ingest.logger import logger
 from unstructured_ingest.utils.dep_check import requires_dependencies
-from unstructured_ingest.v2.errors import ProviderError, RateLimitError, UserAuthError, UserError
+from unstructured_ingest.v2.errors import (
+    ProviderError,
+    RateLimitError,
+    UserAuthError,
+    UserError,
+    is_internal_error,
+)
 
 if TYPE_CHECKING:
     from botocore.client import BaseClient
@@ -54,6 +60,8 @@ class BedrockEmbeddingConfig(EmbeddingConfig):
     embed_model_name: str = Field(default="amazon.titan-embed-text-v1", alias="model_name")
 
     def wrap_error(self, e: Exception) -> Exception:
+        if is_internal_error(e=e):
+            return e
         from botocore.exceptions import ClientError
 
         if isinstance(e, ClientError):
@@ -148,6 +156,8 @@ def embed_query(self, query: str) -> list[float]:
     def embed_documents(self, elements: list[dict]) -> list[dict]:
         elements = elements.copy()
         elements_with_text = [e for e in elements if e.get("text")]
+        if not elements_with_text:
+            return elements
         embeddings = [self.embed_query(query=e["text"]) for e in elements_with_text]
         for element, embedding in zip(elements_with_text, embeddings):
             element[EMBEDDINGS_KEY] = embedding

unstructured_ingest/embed/huggingface.py

@@ -47,7 +47,7 @@ def get_encoder_kwargs(self) -> dict:
 class HuggingFaceEmbeddingEncoder(BaseEmbeddingEncoder):
     config: HuggingFaceEmbeddingConfig
 
-    def embed_query(self, query: str) -> list[float]:
+    def _embed_query(self, query: str) -> list[float]:
         return self._embed_documents(texts=[query])[0]
 
     def _embed_documents(self, texts: list[str]) -> list[list[float]]:
@@ -58,6 +58,8 @@ def _embed_documents(self, texts: list[str]) -> list[list[float]]:
     def embed_documents(self, elements: list[dict]) -> list[dict]:
         elements = elements.copy()
         elements_with_text = [e for e in elements if e.get("text")]
+        if not elements_with_text:
+            return elements
         embeddings = self._embed_documents([e["text"] for e in elements_with_text])
         for element, embedding in zip(elements_with_text, embeddings):
             element[EMBEDDINGS_KEY] = embedding

unstructured_ingest/embed/interfaces.py

@@ -1,11 +1,12 @@
-import asyncio
-from abc import ABC, abstractmethod
+from abc import ABC
 from dataclasses import dataclass
-from typing import Optional
+from typing import Any, Optional
 
 import numpy as np
 from pydantic import BaseModel, Field
 
+from unstructured_ingest.utils.data_prep import batch_generator
+
 EMBEDDINGS_KEY = "embeddings"
 
 
@@ -50,21 +51,37 @@ def is_unit_vector(self) -> bool:
         exemplary_embedding = self.get_exemplary_embedding()
         return np.isclose(np.linalg.norm(exemplary_embedding), 1.0)
 
-    @abstractmethod
-    def embed_documents(self, elements: list[dict]) -> list[dict]:
-        pass
+    def get_client(self):
+        raise NotImplementedError
 
-    @abstractmethod
-    def embed_query(self, query: str) -> list[float]:
-        pass
+    def embed_batch(self, client: Any, batch: list[str]) -> list[list[float]]:
+        raise NotImplementedError
 
-    def _embed_documents(self, elements: list[str]) -> list[list[float]]:
-        results = []
-        for text in elements:
-            response = self.embed_query(query=text)
-            results.append(response)
+    def embed_documents(self, elements: list[dict]) -> list[dict]:
+        client = self.get_client()
+        elements = elements.copy()
+        elements_with_text = [e for e in elements if e.get("text")]
+        texts = [e["text"] for e in elements_with_text]
+        embeddings = []
+        try:
+            for batch in batch_generator(texts, batch_size=self.config.batch_size or len(texts)):
+                embeddings = self.embed_batch(client=client, batch=batch)
+                embeddings.extend(embeddings)
+        except Exception as e:
+            raise self.wrap_error(e=e)
+        for element, embedding in zip(elements_with_text, embeddings):
+            element[EMBEDDINGS_KEY] = embedding
+        return elements
+
+    def _embed_query(self, query: str) -> list[float]:
+        client = self.get_client()
+        return self.embed_batch(client=client, batch=[query])[0]
 
-        return results
+    def embed_query(self, query: str) -> list[float]:
+        try:
+            return self._embed_query(query=query)
+        except Exception as e:
+            raise self.wrap_error(e=e)
 
 
 @dataclass
@@ -88,14 +105,35 @@ async def is_unit_vector(self) -> bool:
         exemplary_embedding = await self.get_exemplary_embedding()
         return np.isclose(np.linalg.norm(exemplary_embedding), 1.0)
 
-    @abstractmethod
+    def get_client(self):
+        raise NotImplementedError
+
+    async def embed_batch(self, client: Any, batch: list[str]) -> list[list[float]]:
+        raise NotImplementedError
+
     async def embed_documents(self, elements: list[dict]) -> list[dict]:
-        pass
+        client = self.get_client()
+        elements = elements.copy()
+        elements_with_text = [e for e in elements if e.get("text")]
+        texts = [e["text"] for e in elements_with_text]
+        embeddings = []
+        try:
+            for batch in batch_generator(texts, batch_size=self.config.batch_size or len(texts)):
+                embeddings = await self.embed_batch(client=client, batch=batch)
+                embeddings.extend(embeddings)
+        except Exception as e:
+            raise self.wrap_error(e=e)
+        for element, embedding in zip(elements_with_text, embeddings):
+            element[EMBEDDINGS_KEY] = embedding
+        return elements
+
+    async def _embed_query(self, query: str) -> list[float]:
+        client = self.get_client()
+        embeddings = await self.embed_batch(client=client, batch=[query])
+        return embeddings[0]
 
-    @abstractmethod
     async def embed_query(self, query: str) -> list[float]:
-        pass
-
-    async def _embed_documents(self, elements: list[str]) -> list[list[float]]:
-        results = await asyncio.gather(*[self.embed_query(query=text) for text in elements])
-        return results
+        try:
+            return await self._embed_query(query=query)
+        except Exception as e:
+            raise self.wrap_error(e=e)

unstructured_ingest/embed/mixedbreadai.py

@@ -1,17 +1,14 @@
-import asyncio
 import os
 from dataclasses import dataclass
 from typing import TYPE_CHECKING
 
 from pydantic import Field, SecretStr
 
 from unstructured_ingest.embed.interfaces import (
-    EMBEDDINGS_KEY,
     AsyncBaseEmbeddingEncoder,
     BaseEmbeddingEncoder,
     EmbeddingConfig,
 )
-from unstructured_ingest.utils.data_prep import batch_generator
 from unstructured_ingest.utils.dep_check import requires_dependencies
 
 USER_AGENT = "@mixedbread-ai/unstructured"
@@ -85,7 +82,7 @@ class MixedbreadAIEmbeddingEncoder(BaseEmbeddingEncoder):
 
     def get_exemplary_embedding(self) -> list[float]:
         """Get an exemplary embedding to determine dimensions and unit vector status."""
-        return self._embed(["Q"])[0]
+        return self.embed_query(query="Q")
 
     @requires_dependencies(
         ["mixedbread_ai"],
@@ -100,59 +97,19 @@ def get_request_options(self) -> "RequestOptions":
             additional_headers={"User-Agent": USER_AGENT},
         )
 
-    def _embed(self, texts: list[str]) -> list[list[float]]:
-        """
-        Embed a list of texts using the Mixedbread AI API.
-
-        Args:
-            texts (list[str]): List of texts to embed.
-
-        Returns:
-            list[list[float]]: List of embeddings.
-        """
-
-        responses = []
-        client = self.config.get_client()
-        for batch in batch_generator(texts, batch_size=self.config.batch_size or len(texts)):
-            response = client.embeddings(
-                model=self.config.embedder_model_name,
-                normalized=True,
-                encoding_format=ENCODING_FORMAT,
-                truncation_strategy=TRUNCATION_STRATEGY,
-                request_options=self.get_request_options(),
-                input=batch,
-            )
-            responses.append(response)
-        return [item.embedding for response in responses for item in response.data]
-
-    def embed_documents(self, elements: list[dict]) -> list[dict]:
-        """
-        Embed a list of document elements.
-
-        Args:
-            elements (list[Element]): List of document elements.
-
-        Returns:
-            list[Element]: Elements with embeddings.
-        """
-        elements = elements.copy()
-        elements_with_text = [e for e in elements if e.get("text")]
-        embeddings = self._embed([e["text"] for e in elements_with_text])
-        for element, embedding in zip(elements_with_text, embeddings):
-            element[EMBEDDINGS_KEY] = embedding
-        return elements
-
-    def embed_query(self, query: str) -> list[float]:
-        """
-        Embed a query string.
-
-        Args:
-            query (str): Query string to embed.
-
-        Returns:
-            list[float]: Embedding of the query.
-        """
-        return self._embed([query])[0]
+    def get_client(self) -> "MixedbreadAI":
+        return self.config.get_client()
+
+    def embed_batch(self, client: "MixedbreadAI", batch: list[str]) -> list[list[float]]:
+        response = client.embeddings(
+            model=self.config.embedder_model_name,
+            normalized=True,
+            encoding_format=ENCODING_FORMAT,
+            truncation_strategy=TRUNCATION_STRATEGY,
+            request_options=self.get_request_options(),
+            input=batch,
+        )
+        return [datum.embedding for datum in response.data]
 
 
 @dataclass
@@ -162,8 +119,7 @@ class AsyncMixedbreadAIEmbeddingEncoder(AsyncBaseEmbeddingEncoder):
 
     async def get_exemplary_embedding(self) -> list[float]:
         """Get an exemplary embedding to determine dimensions and unit vector status."""
-        embedding = await self._embed(["Q"])
-        return embedding[0]
+        return await self.embed_query(query="Q")
 
     @requires_dependencies(
         ["mixedbread_ai"],
@@ -178,58 +134,16 @@ def get_request_options(self) -> "RequestOptions":
             additional_headers={"User-Agent": USER_AGENT},
         )
 
-    async def _embed(self, texts: list[str]) -> list[list[float]]:
-        """
-        Embed a list of texts using the Mixedbread AI API.
-
-        Args:
-            texts (list[str]): List of texts to embed.
-
-        Returns:
-            list[list[float]]: List of embeddings.
-        """
-        client = self.config.get_async_client()
-        tasks = []
-        for batch in batch_generator(texts, batch_size=self.config.batch_size or len(texts)):
-            tasks.append(
-                client.embeddings(
-                    model=self.config.embedder_model_name,
-                    normalized=True,
-                    encoding_format=ENCODING_FORMAT,
-                    truncation_strategy=TRUNCATION_STRATEGY,
-                    request_options=self.get_request_options(),
-                    input=batch,
-                )
-            )
-        responses = await asyncio.gather(*tasks)
-        return [item.embedding for response in responses for item in response.data]
-
-    async def embed_documents(self, elements: list[dict]) -> list[dict]:
-        """
-        Embed a list of document elements.
-
-        Args:
-            elements (list[Element]): List of document elements.
-
-        Returns:
-            list[Element]: Elements with embeddings.
-        """
-        elements = elements.copy()
-        elements_with_text = [e for e in elements if e.get("text")]
-        embeddings = await self._embed([e["text"] for e in elements_with_text])
-        for element, embedding in zip(elements_with_text, embeddings):
-            element[EMBEDDINGS_KEY] = embedding
-        return elements
-
-    async def embed_query(self, query: str) -> list[float]:
-        """
-        Embed a query string.
-
-        Args:
-            query (str): Query string to embed.
-
-        Returns:
-            list[float]: Embedding of the query.
-        """
-        embedding = await self._embed([query])
-        return embedding[0]
+    def get_client(self) -> "AsyncMixedbreadAI":
+        return self.config.get_async_client()
+
+    async def embed_batch(self, client: "AsyncMixedbreadAI", batch: list[str]) -> list[list[float]]:
+        response = await client.embeddings(
+            model=self.config.embedder_model_name,
+            normalized=True,
+            encoding_format=ENCODING_FORMAT,
+            truncation_strategy=TRUNCATION_STRATEGY,
+            request_options=self.get_request_options(),
+            input=batch,
+        )
+        return [datum.embedding for datum in response.data]