veil.config

class veil.config.ApiServerConfig(host: str = '0.0.0.0', port: int = 8000, reload: bool = False)

Bases: object

Parameters:

• host (str)

• port (int)

• reload (bool)

host: str = '0.0.0.0'

port: int = 8000

reload: bool = False

class veil.config.BaseEntityDetectorConfig(priority=<factory>, hierarchy_position=0)

Bases: BasePolyConfig

Marker base for entity detector configuration objects.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

classmethod get_type()

hierarchy_position: int = 0

priority: Dict[str, int]

class veil.config.BaseEntityResolverConfig

Bases: BasePolyConfig

Marker base for entity resolver configuration objects.

classmethod get_type()

class veil.config.DataHandlerConfig(input_path=None, output_path=None, text_field='text', ground_truth_field=None, doc_id_field=None, text_encoding='utf-8')

Bases: object

Configuration for input/output data handling.

Parameters:

• input_path (str)

• output_path (str)

• text_field (str)

• ground_truth_field (str | None)

• doc_id_field (str | None)

• text_encoding (str)

doc_id_field: str | None = None

ground_truth_field: str | None = None

input_path: str = None

output_path: str = None

text_encoding: str = 'utf-8'

text_field: str = 'text'

class veil.config.EmbeddingsEntityResolverConfig(threshold=0.82, context_chars=0)

Bases: BaseEntityResolverConfig

Configuration for an embeddings-based entity resolver.

For v1 we keep options minimal and local-only.

Parameters:

• threshold (float)

• context_chars (int)

context_chars: int = 0

classmethod get_type()

threshold: float = 0.82

class veil.config.EvaluatorConfig(enabled=True, strict_type=True, report_exact=True, report_exact_by_id=True, report_iou_thresholds=<factory>, use_hungarian_id_mapping=False)

Bases: object

Configuration for the evaluator.

Parameters:

• enabled (bool)

• strict_type (bool)

• report_exact (bool)

• report_exact_by_id (bool)

• report_iou_thresholds (List[float])

• use_hungarian_id_mapping (bool)

enabled: bool = True

report_exact: bool = True

report_exact_by_id: bool = True

report_iou_thresholds: List[float]

strict_type: bool = True

use_hungarian_id_mapping: bool = False

class veil.config.GlinerEntityDetectorConfig(priority=<factory>, hierarchy_position=0, labels=<factory>, model='urchade/gliner_multi-v2.1', cuda_device=0, threshold=0.6, batch_size=8, max_length=384, chunk_overlap=50, nms_iou_threshold=0.8, min_span_chars=3, max_span_chars=80, top_k_per_chunk=100)

Bases: BaseEntityDetectorConfig

Configuration for the Gliner entity detector integration.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

• labels (List[str])

• model (str)

• cuda_device (int)

• threshold (float)

• batch_size (int)

• max_length (int)

• chunk_overlap (int)

• nms_iou_threshold (float)

• min_span_chars (int)

• max_span_chars (int)

• top_k_per_chunk (int)

batch_size: int = 8

chunk_overlap: int = 50

cuda_device: int = 0

classmethod get_type()

labels: List[str]

max_length: int = 384

max_span_chars: int = 80

min_span_chars: int = 3

model: str = 'urchade/gliner_multi-v2.1'

nms_iou_threshold: float = 0.8

threshold: float = 0.6

top_k_per_chunk: int = 100

class veil.config.HostedMaskerApiEntityDetectorConfig(priority=<factory>, hierarchy_position=0, api_url='', headers=<factory>, model='', system_prompt='You are a helpful assistant that masks sensitive information.', max_tokens=4000, top_p=1, top_k=40, presence_penalty=0, frequency_penalty=0, temperature=0.6, timeout=30, retries=2, retry_backoff_base=0.5, retry_on_truncation=True, chunk_on_truncation=True, chunk_char_limit=4000, truncation_min_fraction=0.6)

Bases: MaskerApiEntityDetectorConfig

Configuration for the hosted Masker API entity detector integration.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

• api_url (str)

• headers (Dict[str, str])

• model (str)

• system_prompt (str)

• max_tokens (int)

• top_p (float)

• top_k (int)

• presence_penalty (float)

• frequency_penalty (float)

• temperature (float)

• timeout (float)

• retries (int)

• retry_backoff_base (float)

• retry_on_truncation (bool)

• chunk_on_truncation (bool)

• chunk_char_limit (int)

• truncation_min_fraction (float)

classmethod get_type()

class veil.config.MaskerApiEntityDetectorConfig(priority=<factory>, hierarchy_position=0, api_url='', headers=<factory>, model='', system_prompt='You are a helpful assistant that masks sensitive information.', max_tokens=4000, top_p=1, top_k=40, presence_penalty=0, frequency_penalty=0, temperature=0.6, timeout=30, retries=2, retry_backoff_base=0.5, retry_on_truncation=True, chunk_on_truncation=True, chunk_char_limit=4000, truncation_min_fraction=0.6)

Bases: BaseEntityDetectorConfig

Marker base for masking-API entity detector configuration objects.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

• api_url (str)

• headers (Dict[str, str])

• model (str)

• system_prompt (str)

• max_tokens (int)

• top_p (float)

• top_k (int)

• presence_penalty (float)

• frequency_penalty (float)

• temperature (float)

• timeout (float)

• retries (int)

• retry_backoff_base (float)

• retry_on_truncation (bool)

• chunk_on_truncation (bool)

• chunk_char_limit (int)

• truncation_min_fraction (float)

api_url: str = ''

chunk_char_limit: int = 4000

chunk_on_truncation: bool = True

frequency_penalty: float = 0

classmethod get_type()

headers: Dict[str, str]

max_tokens: int = 4000

model: str = ''

presence_penalty: float = 0

retries: int = 2

retry_backoff_base: float = 0.5

retry_on_truncation: bool = True

system_prompt: str = 'You are a helpful assistant that masks sensitive information.'

temperature: float = 0.6

timeout: float = 30

top_k: int = 40

top_p: float = 1

truncation_min_fraction: float = 0.6

class veil.config.MaskerConfig(method='ENTITY_TAG')

Bases: object

Configuration for the masker.

Parameters:

method (str)

method: str = 'ENTITY_TAG'

class veil.config.MetricStoreConfig(enabled: bool = False, output_dir: str = 'veil_runs', save_config_json: bool = True)

Bases: object

Parameters:

• enabled (bool)

• output_dir (str)

• save_config_json (bool)

enabled: bool = False

output_dir: str = 'veil_runs'

save_config_json: bool = True

class veil.config.OverlapResolverConfig(iou_threshold=0.75, overlap_policy='same_type_only')

Bases: object

Configuration for overlap resolution between spans from different detectors.

IoU threshold controls when two spans are considered conflicting: - If IoU > threshold, they conflict and the lower-priority span is dropped. - With threshold=0.0, any real overlap is considered a conflict.

Parameters:

• iou_threshold (float)

• overlap_policy (str)

iou_threshold: float = 0.75

overlap_policy: str = 'same_type_only'

class veil.config.PipelineConfig(mode='offline', datahandler=None, api_server=None, entity_detectors=<factory>, entity_resolvers=None, metric_store=<factory>, masker=<factory>, overlap_resolver=<factory>, evaluator=None, concurrency=1, log_level='INFO')

Bases: object

Root configuration dataclass for a Veil masking run.

Parameters:

• mode (str)

• datahandler (DataHandlerConfig | None)

• api_server (ApiServerConfig | None)

• entity_detectors (List[BaseEntityDetectorConfig])

• entity_resolvers (List[BaseEntityResolverConfig] | None)

• metric_store (MetricStoreConfig)

• masker (MaskerConfig)

• overlap_resolver (OverlapResolverConfig)

• evaluator (EvaluatorConfig | None)

• concurrency (int)

• log_level (str)

api_server: ApiServerConfig | None = None

concurrency: int = 1

classmethod create_from_cli_args()

Return one or many PipelineConfig instances based on CLI/YAML combos.

datahandler: DataHandlerConfig | None = None

entity_detectors: List[BaseEntityDetectorConfig]

entity_resolvers: List[BaseEntityResolverConfig] | None = None

evaluator: EvaluatorConfig | None = None

log_level: str = 'INFO'

masker: MaskerConfig

metric_store: MetricStoreConfig

mode: str = 'offline'

overlap_resolver: OverlapResolverConfig

class veil.config.RegexEntityDetectorConfig(priority=<factory>, hierarchy_position=0, enable_validation=True, min_confidence=0.0, preserve_format=True, case_sensitive=False)

Bases: BaseEntityDetectorConfig

Configuration wrapper for Veil’s RegexEntityDetector engine.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

• enable_validation (bool)

• min_confidence (float)

• preserve_format (bool)

• case_sensitive (bool)

case_sensitive: bool = False

enable_validation: bool = True

classmethod get_type()

min_confidence: float = 0.0

preserve_format: bool = True

class veil.config.SpacyEntityDetectorConfig(priority=<factory>, hierarchy_position=0, model='es_core_news_sm', cuda_device=0)

Bases: BaseEntityDetectorConfig

Configuration for the Spacy entity detector integration.

Parameters:

• priority (Dict[str, int])

• hierarchy_position (int)

• model (str)

• cuda_device (int)

cuda_device: int = 0

classmethod get_type()

model: str = 'es_core_news_sm'