metadata_filtering.py

from llama_index.core.vector_stores import (
    MetadataFilters,
    MetadataFilter,
    FilterOperator,
    FilterCondition,
)

def build_filters_from_query(query: str):
    terms = [t.lower() for t in query.split() if len(t) > 2]

    if not terms:
        return None

    return MetadataFilters(
        filters=[
            MetadataFilter(
                key="excerpt_keywords",
                operator=FilterOperator.TEXT_MATCH,
                value=term,
            )
            for term in terms
        ],
        condition=FilterCondition.OR
    )