import os
from typing import Any, Callable, Dict, List, Optional
from pydantic.v1 import PrivateAttr
from redisvl.utils.vectorize.base import BaseVectorizer
[docs]
class CustomTextVectorizer(BaseVectorizer):
"""The CustomTextVectorizer class wraps user-defined embeding methods to create
embeddings for text data.
This vectorizer is designed to accept a provided callable text vectorizer and
provides a class definition to allow for compatibility with RedisVL.
The vectorizer may support both synchronous and asynchronous operations which
allows for batch processing of texts, but at a minimum only syncronous embedding
is required to satisfy the 'embed()' method.
.. code-block:: python
# Synchronous embedding of a single text
vectorizer = CustomTextVectorizer(
embed = my_vectorizer.generate_embedding
)
embedding = vectorizer.embed("Hello, world!")
# Asynchronous batch embedding of multiple texts
embeddings = await vectorizer.aembed_many(
["Hello, world!", "How are you?"],
batch_size=2
)
"""
_embed_func: Callable = PrivateAttr()
_embed_many_func: Optional[Callable] = PrivateAttr()
_aembed_func: Optional[Callable] = PrivateAttr()
_aembed_many_func: Optional[Callable] = PrivateAttr()
def __init__(
self,
embed: Callable,
embed_many: Optional[Callable] = None,
aembed: Optional[Callable] = None,
aembed_many: Optional[Callable] = None,
):
"""Initialize the Custom vectorizer.
Args:
embed (Callable): a Callable function that accepts a string object and returns a list of floats.
embed_many (Optional[Callable)]: a Callable function that accepts a list of string objects and returns a list containing lists of floats. Defaults to None.
aembed (Optional[Callable]): an asyncronous Callable function that accepts a string object and returns a lists of floats. Defaults to None.
aembed_many (Optional[Callable]): an asyncronous Callable function that accepts a list of string objects and returns a list containing lists of floats. Defaults to None.
Raises:
ValueError if any of the provided functions accept or return incorrect types.
TypeError if any of the provided functions are not Callable objects.
"""
self._validate_embed(embed)
self._embed_func = embed
if embed_many:
self._validate_embed_many(embed_many)
self._embed_many_func = embed_many
if aembed:
self._validate_aembed(aembed)
self._aembed_func = aembed
if aembed_many:
self._validate_aembed_many(aembed_many)
self._aembed_many_func = aembed_many
super().__init__(model=self.type, dims=self._set_model_dims())
def _validate_embed(self, func: Callable):
"""calls the func with dummy input and validates that it returns a vector"""
try:
test_str = "this is a test sentence"
candidate_vector = func(test_str)
if type(candidate_vector) != list or type(candidate_vector[0]) != float:
raise ValueError(
f"Candidate function for embed() does not have the correct return type. Please provide a function with with return type List[float]"
)
except TypeError:
raise TypeError(f"{func} is not a callable object")
def _validate_embed_many(self, func: Callable):
"""calls the func with dummy input and validates that it returns a list of vectors"""
try:
test_strs = ["first test sentence", "second test sentence"]
candidate_vectors = func(test_strs)
if (
type(candidate_vectors) != list
or type(candidate_vectors[0]) != list
or type(candidate_vectors[0][0]) != float
):
raise ValueError(
f"Candidate function for embed_many does not have the correct return type. Please provide a function with with return type List[List[float]]"
)
except TypeError:
raise TypeError(f"{func} is not a callable object")
def _validate_aembed(self, func: Callable):
"""calls the func with dummy input and validates that it returns a vector"""
import asyncio
try:
test_str = "this is a test sentence"
loop = asyncio.get_event_loop()
candidate_vector = loop.run_until_complete(func(test_str))
if type(candidate_vector) != list or type(candidate_vector[0]) != float:
raise ValueError(
f"Candidate function for aembed() does not have the correct return type. Please provide a function with with return type List[float]"
)
except TypeError:
raise TypeError(f"{func} is not a callable object")
def _validate_aembed_many(self, func: Callable):
"""calls the func with dummy input and validates that it returns a list of vectors"""
import asyncio
try:
test_strs = ["first test sentence", "second test sentence"]
loop = asyncio.get_event_loop()
candidate_vectors = loop.run_until_complete(func(test_strs))
if (
type(candidate_vectors) != list
or type(candidate_vectors[0]) != list
or type(candidate_vectors[0][0]) != float
):
raise ValueError(
f"Candidate function for aembed_many does not have the correct return type. Please provide a function with with return type List[List[float]]"
)
except TypeError:
raise TypeError(f"{func} is not a callable object")
def _set_model_dims(self) -> int:
try:
test_string = "dimension test"
embedding = self._embed_func(test_string)
except Exception as e: # pylint: disable=broad-except
raise ValueError(
f"Error in checking model dimensions. Attempted to embed '{test_string}'. :{str(e)}"
)
return len(embedding)
[docs]
def embed(
self,
text: str,
preprocess: Optional[Callable] = None,
as_buffer: bool = False,
**kwargs,
) -> List[float]:
"""Embed a chunk of text using the provided function.
Args:
text (str): Chunk of text to embed.
preprocess (Optional[Callable], optional): Optional preprocessing callable to
perform before vectorization. Defaults to None.
as_buffer (bool, optional): Whether to convert the raw embedding
to a byte string. Defaults to False.
Returns:
List[float]: Embedding.
Raises:
TypeError: If the wrong input type is passed in for the text.
"""
if not isinstance(text, str):
raise TypeError("Must pass in a str value to embed.")
if preprocess:
text = preprocess(text)
else:
result = self._embed_func(text, **kwargs)
return self._process_embedding(result, as_buffer)
[docs]
def embed_many(
self,
texts: List[str],
preprocess: Optional[Callable] = None,
batch_size: int = 10,
as_buffer: bool = False,
**kwargs,
) -> List[List[float]]:
"""Embed many chunks of texts using the provided function.
Args:
texts (List[str]): List of text chunks to embed.
preprocess (Optional[Callable], optional): Optional preprocessing
callable to perform before vectorization. Defaults to None.
batch_size (int, optional): Batch size of texts to use when creating
embeddings. Defaults to 10.
as_buffer (bool, optional): Whether to convert the raw embedding
to a byte string. Defaults to False.
Returns:
List[List[float]]: List of embeddings.
Raises:
TypeError: If the wrong input type is passed in for the text.
NotImplementedError: if embed_many was not passed to constructor.
"""
if not isinstance(texts, list):
raise TypeError("Must pass in a list of str values to embed.")
if len(texts) > 0 and not isinstance(texts[0], str):
raise TypeError("Must pass in a list of str values to embed.")
if not self._embed_many_func:
raise NotImplementedError
embeddings: List = []
for batch in self.batchify(texts, batch_size, preprocess):
results = self._embed_many_func(batch, **kwargs)
embeddings += [self._process_embedding(r, as_buffer) for r in results]
return embeddings
[docs]
async def aembed(
self,
text: str,
preprocess: Optional[Callable] = None,
as_buffer: bool = False,
**kwargs,
) -> List[float]:
"""Asynchronously embed a chunk of text.
Args:
text (str): Chunk of text to embed.
preprocess (Optional[Callable], optional): Optional preprocessing callable to
perform before vectorization. Defaults to None.
as_buffer (bool, optional): Whether to convert the raw embedding
to a byte string. Defaults to False.
Returns:
List[float]: Embedding.
Raises:
TypeError: If the wrong input type is passed in for the text.
NotImplementedError: if aembed was not passed to constructor.
"""
if not isinstance(text, str):
raise TypeError("Must pass in a str value to embed.")
if not self._aembed_func:
raise NotImplementedError
if preprocess:
text = preprocess(text)
else:
result = await self._aembed_func(text, **kwargs)
return self._process_embedding(result, as_buffer)
[docs]
async def aembed_many(
self,
texts: List[str],
preprocess: Optional[Callable] = None,
batch_size: int = 1000,
as_buffer: bool = False,
**kwargs,
) -> List[List[float]]:
"""Asynchronously embed many chunks of texts.
Args:
texts (List[str]): List of text chunks to embed.
preprocess (Optional[Callable], optional): Optional preprocessing callable to
perform before vectorization. Defaults to None.
batch_size (int, optional): Batch size of texts to use when creating
embeddings. Defaults to 10.
as_buffer (bool, optional): Whether to convert the raw embedding
to a byte string. Defaults to False.
Returns:
List[List[float]]: List of embeddings.
Raises:
TypeError: If the wrong input type is passed in for the text.
NotImplementedError: If aembed_many was not passed to constructor.
"""
if not isinstance(texts, list):
raise TypeError("Must pass in a list of str values to embed.")
if len(texts) > 0 and not isinstance(texts[0], str):
raise TypeError("Must pass in a list of str values to embed.")
if not self._aembed_many_func:
raise NotImplementedError
embeddings: List = []
for batch in self.batchify(texts, batch_size, preprocess):
results = await self._aembed_many_func(batch, **kwargs)
embeddings += [self._process_embedding(r, as_buffer) for r in results]
return embeddings
@property
def type(self) -> str:
return "custom"
