15269350331

Committed 27 May 2025 07:35AM UTC coverage: 90.181% (+0.006%) from 90.175%

Build # 15269350331

Build Type

push

github

Committed by

web-flow

Commit Message

chore: make mypy run with `--check-untyped-defs`; fix some errors (#9447)

* chore: make mypy run with --check-untyped-defs; fix some errors

* small fixes

* use HfPipeline

* fix license error

Run Details

11388 of 12628 relevant lines covered (90.18%)

0.9 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

91.07

haystack/components/classifiers/zero_shot_document_classifier.py

# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
#
# SPDX-License-Identifier: Apache-2.0

from typing import Any, Dict, List, Optional

from haystack import Document, component, default_from_dict, default_to_dict
from haystack.lazy_imports import LazyImport
from haystack.utils import ComponentDevice, Secret, deserialize_secrets_inplace
from haystack.utils.hf import deserialize_hf_model_kwargs, resolve_hf_pipeline_kwargs, serialize_hf_model_kwargs

with LazyImport(message="Run 'pip install transformers[torch,sentencepiece]'") as torch_and_transformers_import:
    from transformers import Pipeline as HfPipeline
    from transformers import pipeline


@component
class TransformersZeroShotDocumentClassifier:
    """
    Performs zero-shot classification of documents based on given labels and adds the predicted label to their metadata.

    The component uses a Hugging Face pipeline for zero-shot classification.
    Provide the model and the set of labels to be used for categorization during initialization.
    Additionally, you can configure the component to allow multiple labels to be true.

    Classification is run on the document's content field by default. If you want it to run on another field, set the
    `classification_field` to one of the document's metadata fields.

    Available models for the task of zero-shot-classification include:
        - `valhalla/distilbart-mnli-12-3`
        - `cross-encoder/nli-distilroberta-base`
        - `cross-encoder/nli-deberta-v3-xsmall`

    ### Usage example

    The following is a pipeline that classifies documents based on predefined classification labels
    retrieved from a search pipeline:

    ```python
    from haystack import Document
    from haystack.components.retrievers.in_memory import InMemoryBM25Retriever
    from haystack.document_stores.in_memory import InMemoryDocumentStore
    from haystack.core.pipeline import Pipeline
    from haystack.components.classifiers import TransformersZeroShotDocumentClassifier

    documents = [Document(id="0", content="Today was a nice day!"),
                 Document(id="1", content="Yesterday was a bad day!")]

    document_store = InMemoryDocumentStore()
    retriever = InMemoryBM25Retriever(document_store=document_store)
    document_classifier = TransformersZeroShotDocumentClassifier(
        model="cross-encoder/nli-deberta-v3-xsmall",
        labels=["positive", "negative"],
    )

    document_store.write_documents(documents)

    pipeline = Pipeline()
    pipeline.add_component(instance=retriever, name="retriever")
    pipeline.add_component(instance=document_classifier, name="document_classifier")
    pipeline.connect("retriever", "document_classifier")

    queries = ["How was your day today?", "How was your day yesterday?"]
    expected_predictions = ["positive", "negative"]

    for idx, query in enumerate(queries):
        result = pipeline.run({"retriever": {"query": query, "top_k": 1}})
        assert result["document_classifier"]["documents"][0].to_dict()["id"] == str(idx)
        assert (result["document_classifier"]["documents"][0].to_dict()["classification"]["label"]
                == expected_predictions[idx])
    ```
    """

    def __init__(  # pylint: disable=too-many-positional-arguments
        self,
        model: str,
        labels: List[str],
        multi_label: bool = False,
        classification_field: Optional[str] = None,
        device: Optional[ComponentDevice] = None,
        token: Optional[Secret] = Secret.from_env_var(["HF_API_TOKEN", "HF_TOKEN"], strict=False),
        huggingface_pipeline_kwargs: Optional[Dict[str, Any]] = None,
    ):
        """
        Initializes the TransformersZeroShotDocumentClassifier.

        See the Hugging Face [website](https://huggingface.co/models?pipeline_tag=zero-shot-classification&sort=downloads&search=nli)
        for the full list of zero-shot classification models (NLI) models.

        :param model:
            The name or path of a Hugging Face model for zero shot document classification.
        :param labels:
            The set of possible class labels to classify each document into, for example,
            ["positive", "negative"]. The labels depend on the selected model.
        :param multi_label:
            Whether or not multiple candidate labels can be true.
            If `False`, the scores are normalized such that
            the sum of the label likelihoods for each sequence is 1. If `True`, the labels are considered
            independent and probabilities are normalized for each candidate by doing a softmax of the entailment
            score vs. the contradiction score.
        :param classification_field:
            Name of document's meta field to be used for classification.
            If not set, `Document.content` is used by default.
        :param device:
            The device on which the model is loaded. If `None`, the default device is automatically
            selected. If a device/device map is specified in `huggingface_pipeline_kwargs`, it overrides this parameter.
        :param token:
            The Hugging Face token to use as HTTP bearer authorization.
            Check your HF token in your [account settings](https://huggingface.co/settings/tokens).
        :param huggingface_pipeline_kwargs:
            Dictionary containing keyword arguments used to initialize the
            Hugging Face pipeline for text classification.
        """

        torch_and_transformers_import.check()

        self.classification_field = classification_field

        self.token = token
        self.labels = labels
        self.multi_label = multi_label

        huggingface_pipeline_kwargs = resolve_hf_pipeline_kwargs(
            huggingface_pipeline_kwargs=huggingface_pipeline_kwargs or {},
            model=model,
            task="zero-shot-classification",
            supported_tasks=["zero-shot-classification"],
            device=device,
            token=token,
        )

        self.huggingface_pipeline_kwargs = huggingface_pipeline_kwargs
        self.pipeline: Optional[HfPipeline] = None

    def _get_telemetry_data(self) -> Dict[str, Any]:
        """
        Data that is sent to Posthog for usage analytics.
        """
        if isinstance(self.huggingface_pipeline_kwargs["model"], str):
            return {"model": self.huggingface_pipeline_kwargs["model"]}
        return {"model": f"[object of type {type(self.huggingface_pipeline_kwargs['model'])}]"}

    def warm_up(self):
        """
        Initializes the component.
        """
        if self.pipeline is None:
            self.pipeline = pipeline(**self.huggingface_pipeline_kwargs)

    def to_dict(self) -> Dict[str, Any]:
        """
        Serializes the component to a dictionary.

        :returns:
            Dictionary with serialized data.
        """
        serialization_dict = default_to_dict(
            self,
            labels=self.labels,
            model=self.huggingface_pipeline_kwargs["model"],
            huggingface_pipeline_kwargs=self.huggingface_pipeline_kwargs,
            token=self.token.to_dict() if self.token else None,
        )

        huggingface_pipeline_kwargs = serialization_dict["init_parameters"]["huggingface_pipeline_kwargs"]
        huggingface_pipeline_kwargs.pop("token", None)

        serialize_hf_model_kwargs(huggingface_pipeline_kwargs)
        return serialization_dict

    @classmethod
    def from_dict(cls, data: Dict[str, Any]) -> "TransformersZeroShotDocumentClassifier":
        """
        Deserializes the component from a dictionary.

        :param data:
            Dictionary to deserialize from.
        :returns:
            Deserialized component.
        """
        deserialize_secrets_inplace(data["init_parameters"], keys=["token"])
        if data["init_parameters"].get("huggingface_pipeline_kwargs") is not None:
            deserialize_hf_model_kwargs(data["init_parameters"]["huggingface_pipeline_kwargs"])
        return default_from_dict(cls, data)

    @component.output_types(documents=List[Document])
    def run(self, documents: List[Document], batch_size: int = 1):
        """
        Classifies the documents based on the provided labels and adds them to their metadata.

        The classification results are stored in the `classification` dict within
        each document's metadata. If `multi_label` is set to `True`, the scores for each label are available under
        the `details` key within the `classification` dictionary.

        :param documents:
            Documents to process.
        :param batch_size:
            Batch size used for processing the content in each document.
        :returns:
            A dictionary with the following key:
            - `documents`: A list of documents with an added metadata field called `classification`.
        """

        if self.pipeline is None:
            raise RuntimeError(
                "The component TransformerZeroShotDocumentClassifier wasn't warmed up. "
                "Run 'warm_up()' before calling 'run()'."
            )

        if not isinstance(documents, list) or documents and not isinstance(documents[0], Document):
            raise TypeError(
                "TransformerZeroShotDocumentClassifier expects a list of documents as input. "
                "In case you want to classify and route a text, please use the TransformersZeroShotTextRouter."
            )

        invalid_doc_ids = []

        for doc in documents:
            if self.classification_field is not None and self.classification_field not in doc.meta:
                invalid_doc_ids.append(doc.id)

        if invalid_doc_ids:
            raise ValueError(
                f"The following documents do not have the classification field '{self.classification_field}': "
                f"{', '.join(invalid_doc_ids)}"
            )

        texts = [
            (doc.content if self.classification_field is None else doc.meta[self.classification_field])
            for doc in documents
        ]

        predictions = self.pipeline(texts, self.labels, multi_label=self.multi_label, batch_size=batch_size)

        for prediction, document in zip(predictions, documents):
            formatted_prediction = {
                "label": prediction["labels"][0],
                "score": prediction["scores"][0],
                "details": dict(zip(prediction["labels"], prediction["scores"])),
            }
            document.meta["classification"] = formatted_prediction

        return {"documents": documents}

1	# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
2	#
3	# SPDX-License-Identifier: Apache-2.0
4
5	from typing import Any, Dict, List, Optional	1✔
6
7	from haystack import Document, component, default_from_dict, default_to_dict	1✔
8	from haystack.lazy_imports import LazyImport	1✔
9	from haystack.utils import ComponentDevice, Secret, deserialize_secrets_inplace	1✔
10	from haystack.utils.hf import deserialize_hf_model_kwargs, resolve_hf_pipeline_kwargs, serialize_hf_model_kwargs	1✔
11
12	with LazyImport(message="Run 'pip install transformers[torch,sentencepiece]'") as torch_and_transformers_import:	1✔
13	from transformers import Pipeline as HfPipeline	1✔
14	from transformers import pipeline	1✔
15
16
17	@component	1✔
18	class TransformersZeroShotDocumentClassifier:	1✔
19	"""
20	Performs zero-shot classification of documents based on given labels and adds the predicted label to their metadata.
21
22	The component uses a Hugging Face pipeline for zero-shot classification.
23	Provide the model and the set of labels to be used for categorization during initialization.
24	Additionally, you can configure the component to allow multiple labels to be true.
25
26	Classification is run on the document's content field by default. If you want it to run on another field, set the
27	`classification_field` to one of the document's metadata fields.
28
29	Available models for the task of zero-shot-classification include:
30	- `valhalla/distilbart-mnli-12-3`
31	- `cross-encoder/nli-distilroberta-base`
32	- `cross-encoder/nli-deberta-v3-xsmall`
33
34	### Usage example
35
36	The following is a pipeline that classifies documents based on predefined classification labels
37	retrieved from a search pipeline:
38
39	```python
40	from haystack import Document
41	from haystack.components.retrievers.in_memory import InMemoryBM25Retriever
42	from haystack.document_stores.in_memory import InMemoryDocumentStore
43	from haystack.core.pipeline import Pipeline
44	from haystack.components.classifiers import TransformersZeroShotDocumentClassifier
45
46	documents = [Document(id="0", content="Today was a nice day!"),
47	Document(id="1", content="Yesterday was a bad day!")]
48
49	document_store = InMemoryDocumentStore()
50	retriever = InMemoryBM25Retriever(document_store=document_store)
51	document_classifier = TransformersZeroShotDocumentClassifier(
52	model="cross-encoder/nli-deberta-v3-xsmall",
53	labels=["positive", "negative"],
54	)
55
56	document_store.write_documents(documents)
57
58	pipeline = Pipeline()
59	pipeline.add_component(instance=retriever, name="retriever")
60	pipeline.add_component(instance=document_classifier, name="document_classifier")
61	pipeline.connect("retriever", "document_classifier")
62
63	queries = ["How was your day today?", "How was your day yesterday?"]
64	expected_predictions = ["positive", "negative"]
65
66	for idx, query in enumerate(queries):
67	result = pipeline.run({"retriever": {"query": query, "top_k": 1}})
68	assert result["document_classifier"]["documents"][0].to_dict()["id"] == str(idx)
69	assert (result["document_classifier"]["documents"][0].to_dict()["classification"]["label"]
70	== expected_predictions[idx])
71	```
72	"""
73
74	def __init__( # pylint: disable=too-many-positional-arguments	1✔
75	self,
76	model: str,
77	labels: List[str],
78	multi_label: bool = False,
79	classification_field: Optional[str] = None,
80	device: Optional[ComponentDevice] = None,
81	token: Optional[Secret] = Secret.from_env_var(["HF_API_TOKEN", "HF_TOKEN"], strict=False),
82	huggingface_pipeline_kwargs: Optional[Dict[str, Any]] = None,
83	):
84	"""
85	Initializes the TransformersZeroShotDocumentClassifier.
86
87	See the Hugging Face [website](https://huggingface.co/models?pipeline_tag=zero-shot-classification&sort=downloads&search=nli)
88	for the full list of zero-shot classification models (NLI) models.
89
90	:param model:
91	The name or path of a Hugging Face model for zero shot document classification.
92	:param labels:
93	The set of possible class labels to classify each document into, for example,
94	["positive", "negative"]. The labels depend on the selected model.
95	:param multi_label:
96	Whether or not multiple candidate labels can be true.
97	If `False`, the scores are normalized such that
98	the sum of the label likelihoods for each sequence is 1. If `True`, the labels are considered
99	independent and probabilities are normalized for each candidate by doing a softmax of the entailment
100	score vs. the contradiction score.
101	:param classification_field:
102	Name of document's meta field to be used for classification.
103	If not set, `Document.content` is used by default.
104	:param device:
105	The device on which the model is loaded. If `None`, the default device is automatically
106	selected. If a device/device map is specified in `huggingface_pipeline_kwargs`, it overrides this parameter.
107	:param token:
108	The Hugging Face token to use as HTTP bearer authorization.
109	Check your HF token in your [account settings](https://huggingface.co/settings/tokens).
110	:param huggingface_pipeline_kwargs:
111	Dictionary containing keyword arguments used to initialize the
112	Hugging Face pipeline for text classification.
113	"""
114
115	torch_and_transformers_import.check()	1✔
116
117	self.classification_field = classification_field	1✔
118
119	self.token = token	1✔
120	self.labels = labels	1✔
121	self.multi_label = multi_label	1✔
122
123	huggingface_pipeline_kwargs = resolve_hf_pipeline_kwargs(	1✔
124	huggingface_pipeline_kwargs=huggingface_pipeline_kwargs or {},
125	model=model,
126	task="zero-shot-classification",
127	supported_tasks=["zero-shot-classification"],
128	device=device,
129	token=token,
130	)
131
132	self.huggingface_pipeline_kwargs = huggingface_pipeline_kwargs	1✔
133	self.pipeline: Optional[HfPipeline] = None	1✔
134
135	def _get_telemetry_data(self) -> Dict[str, Any]:	1✔
136	"""
137	Data that is sent to Posthog for usage analytics.
138	"""
139	if isinstance(self.huggingface_pipeline_kwargs["model"], str):	×
140	return {"model": self.huggingface_pipeline_kwargs["model"]}	×
141	return {"model": f"[object of type {type(self.huggingface_pipeline_kwargs['model'])}]"}	×
142
143	def warm_up(self):	1✔
144	"""
145	Initializes the component.
146	"""
147	if self.pipeline is None:	1✔
148	self.pipeline = pipeline(**self.huggingface_pipeline_kwargs)	1✔
149
150	def to_dict(self) -> Dict[str, Any]:	1✔
151	"""
152	Serializes the component to a dictionary.
153
154	:returns:
155	Dictionary with serialized data.
156	"""
157	serialization_dict = default_to_dict(	1✔
158	self,
159	labels=self.labels,
160	model=self.huggingface_pipeline_kwargs["model"],
161	huggingface_pipeline_kwargs=self.huggingface_pipeline_kwargs,
162	token=self.token.to_dict() if self.token else None,
163	)
164
165	huggingface_pipeline_kwargs = serialization_dict["init_parameters"]["huggingface_pipeline_kwargs"]	1✔
166	huggingface_pipeline_kwargs.pop("token", None)	1✔
167
168	serialize_hf_model_kwargs(huggingface_pipeline_kwargs)	1✔
169	return serialization_dict	1✔
170
171	@classmethod	1✔
172	def from_dict(cls, data: Dict[str, Any]) -> "TransformersZeroShotDocumentClassifier":	1✔
173	"""
174	Deserializes the component from a dictionary.
175
176	:param data:
177	Dictionary to deserialize from.
178	:returns:
179	Deserialized component.
180	"""
181	deserialize_secrets_inplace(data["init_parameters"], keys=["token"])	1✔
182	if data["init_parameters"].get("huggingface_pipeline_kwargs") is not None:	1✔
183	deserialize_hf_model_kwargs(data["init_parameters"]["huggingface_pipeline_kwargs"])	1✔
184	return default_from_dict(cls, data)	1✔
185
186	@component.output_types(documents=List[Document])	1✔
187	def run(self, documents: List[Document], batch_size: int = 1):	1✔
188	"""
189	Classifies the documents based on the provided labels and adds them to their metadata.
190
191	The classification results are stored in the `classification` dict within
192	each document's metadata. If `multi_label` is set to `True`, the scores for each label are available under
193	the `details` key within the `classification` dictionary.
194
195	:param documents:
196	Documents to process.
197	:param batch_size:
198	Batch size used for processing the content in each document.
199	:returns:
200	A dictionary with the following key:
201	- `documents`: A list of documents with an added metadata field called `classification`.
202	"""
203
204	if self.pipeline is None:	1✔
205	raise RuntimeError(	1✔
206	"The component TransformerZeroShotDocumentClassifier wasn't warmed up. "
207	"Run 'warm_up()' before calling 'run()'."
208	)
209
210	if not isinstance(documents, list) or documents and not isinstance(documents[0], Document):	1✔
211	raise TypeError(	1✔
212	"TransformerZeroShotDocumentClassifier expects a list of documents as input. "
213	"In case you want to classify and route a text, please use the TransformersZeroShotTextRouter."
214	)
215
216	invalid_doc_ids = []	1✔
217
218	for doc in documents:	1✔
219	if self.classification_field is not None and self.classification_field not in doc.meta:	1✔
220	invalid_doc_ids.append(doc.id)	×
221
222	if invalid_doc_ids:	1✔
223	raise ValueError(	×
224	f"The following documents do not have the classification field '{self.classification_field}': "
225	f"{', '.join(invalid_doc_ids)}"
226	)
227
228	texts = [	1✔
229	(doc.content if self.classification_field is None else doc.meta[self.classification_field])
230	for doc in documents
231	]
232
233	predictions = self.pipeline(texts, self.labels, multi_label=self.multi_label, batch_size=batch_size)	1✔
234
235	for prediction, document in zip(predictions, documents):	1✔
236	formatted_prediction = {	1✔
237	"label": prediction["labels"][0],
238	"score": prediction["scores"][0],
239	"details": dict(zip(prediction["labels"], prediction["scores"])),
240	}
241	document.meta["classification"] = formatted_prediction	1✔
242
243	return {"documents": documents}	1✔

deepset-ai / haystack / 15269350331

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous