Metadata for a text chunk, tracking byte positions, indices, and page range.
Fields
:byte_start- Start byte offset in the original content:byte_end- End byte offset in the original content:token_count- Optional number of tokens in the chunk:chunk_index- Zero-indexed position of this chunk:total_chunks- Total number of chunks:first_page- Optional first page number covered by this chunk:last_page- Optional last page number covered by this chunk:heading_context- Optional heading hierarchy for this chunk's section
Summary
Types
@type heading_context() :: %{headings: [heading_level()]}
@type heading_level() :: %{level: non_neg_integer(), text: String.t()}
@type t() :: %Kreuzberg.ChunkMetadata{ byte_end: non_neg_integer(), byte_start: non_neg_integer(), chunk_index: non_neg_integer(), first_page: non_neg_integer() | nil, heading_context: heading_context() | nil, last_page: non_neg_integer() | nil, token_count: non_neg_integer() | nil, total_chunks: non_neg_integer() }