pub struct SearchObject<'a> {
pub obj_tp: char,
pub obj_id: &'a str,
pub content_type: Option<&'a str>,
pub owner_tag: Option<&'a str>,
pub visibility: Option<char>,
pub root_id: Option<&'a str>,
pub created_at: Option<Timestamp>,
pub fts_cl: bool,
}Expand description
The object a set of SearchParts belongs to. The ACL columns are
denormalised mirrors of the source row so the search query can pre-filter
in SQL.
Fields§
§obj_tp: char'F' file, 'D' deep document part, 'A' action, 'P' profile.
obj_id: &'a strfile_id / action_id / id_tag; for 'D' the container file_id.
content_type: Option<&'a str>§owner_tag: Option<&'a str>§visibility: Option<char>None: Direct, P: Public, V: Verified, 2: 2nd degree, F: Follower, C: Connected
root_id: Option<&'a str>§created_at: Option<Timestamp>§fts_cl: boolWhich of the two FTS indexes these rows belong to. false (the default)
keeps the plain-text extract alongside an external-content index; true
stores no text at all and indexes the body into a contentless one —
matching and ranking unchanged, but no result snippets.
Decided per tenant from search.store_text, not per write: flipping it
moves an object’s rows between two physically separate indexes, so it only
takes effect through a full reindex.