extrai.core package¶

Submodules¶

extrai.core.analytics_collector module¶

extrai.core.base_llm_client module¶

extrai.core.batch_models module¶

extrai.core.batch_pipeline module¶

extrai.core.client_rotator module¶

extrai.core.conflict_resolvers module¶

class SimilarityClusterResolver(similarity_threshold: float = 0.6, scorer: ~collections.abc.Callable[[str, str], float] = <function levenshtein_similarity>)[source]¶

Bases: object

Resolves conflicts by clustering values based on string similarity. Useful for filtering out outliers (e.g. “War” vs “Christmas”, “Gifts”).

default_conflict_resolver(path: tuple[str | int, ...], values: list[str | int | float | bool | None], weights: list[float] | None = None) → str | int | float | bool | None[source]¶: Default conflict resolution: if no consensus, omit the field.

levenshtein_similarity(a: str, b: str) → float[source]¶

prefer_most_common_resolver(_path: tuple[str | int, ...], values: list[str | int | float | bool | None], weights: list[float] | None = None) → str | int | float | bool | None[source]¶: Conflict resolution: prefer the most common value. If weights are provided, prefers the value with the highest total weight.

extrai.core.entity_counter module¶

extrai.core.errors module¶

extrai.core.example_json_generator module¶

extrai.core.extraction_config module¶

extrai.core.extraction_context_preparer module¶

extrai.core.extraction_pipeline module¶

extrai.core.extraction_request_factory module¶

extrai.core.hierarchical_extractor module¶

extrai.core.json_consensus module¶

extrai.core.llm_runner module¶

extrai.core.model_registry module¶

extrai.core.model_wrapper_builder module¶

extrai.core.prompt_builder module¶

extrai.core.result_processor module¶

extrai.core.schema_inspector module¶

extrai.core.sqlmodel_generator module¶

extrai.core.workflow_orchestrator module¶

Module contents¶