Metadata for a chunk including source document references.
| Name | Type | Description | Notes |
|---|---|---|---|
| polygons | List[PolygonReference] | List of bounding boxes in the source document for the chunk, potentially from multiple areas of multiple pages. | [optional] |
| s3_urls | List[str] | Ordered s3:// URIs to visual assets for this chunk. Single-element for standard IMAGE/TABLE/HTML chunks, multi-element for multi-page single-chunk ingestion. | [optional] |
| summary | str | LLM-generated summary of the chunk content. Used for TABLE and HTML chunks to enrich embedding text, and for JSON/YAML chunks (with summarize_for_embedding) as the sole dense embedding text. | [optional] |
| summarize_for_embedding | bool | When True, this chunk's dense embedding is built from its LLM-generated summary (see summary) instead of its raw content. Set for parsed JSON/YAML single chunks so noisy structured text does not dominate the vector; the raw content is still kept for display and sparse (keyword) retrieval. Enrichment generates the summary when this is set and summary is empty. | [optional] [default to False] |
| extracted_text_s3_uri | str | S3 URI to extracted PDF text used for LLM grounding during enrichment | [optional] |
| secondary_taxonomy | ImageTaxonomy | [optional] | |
| sheet_name | str | Worksheet name this chunk was extracted from (XLSX only) | [optional] |
| block_type | str | XLSXParser block type (e.g. table, calculation_block, chart_anchor) | [optional] |
| source_uri | str | Cell range URI reference in the source workbook (XLSX only) | [optional] |
| enriched_html | str | Rendered HTML for non-table XLSX chunks (tables use render_html as content) | [optional] |
| cell_range | str | Cell address range, e.g. 'A1:D10' (XLSX only) | [optional] |
| dependency_summary | Dict[str, object] | Upstream/downstream/cross-sheet cell references for audit reasoning (XLSX only) | [optional] |
| formulas | List[Dict[str, str]] | Formula cells in this chunk as [{address, formula, value}] (XLSX only) | [optional] |
| key_cells | List[str] | Notable output/header cells as A1 refs, e.g. 'Sheet1!A1' (XLSX only) | [optional] |
| named_ranges | List[str] | Names of named ranges overlapping this chunk (XLSX only) | [optional] |
from ksapi.models.chunk_metadata import ChunkMetadata
# TODO update the JSON string below
json = "{}"
# create an instance of ChunkMetadata from a JSON string
chunk_metadata_instance = ChunkMetadata.from_json(json)
# print the JSON string representation of the object
print(ChunkMetadata.to_json())
# convert the object into a dict
chunk_metadata_dict = chunk_metadata_instance.to_dict()
# create an instance of ChunkMetadata from a dict
chunk_metadata_from_dict = ChunkMetadata.from_dict(chunk_metadata_dict)