Method remove_document

nodedb/src/engine/sparse/sparse_vector.rs:169–253

Remove a document from the sparse vector index.

(&self, collection: &str, doc_id: &str)

    /// Remove a document from the sparse vector index.
    pub fn remove_document(&self, collection: &str, doc_id: &str) -> crate::Result<()> {
        let write_txn = self.db.begin_write().map_err(|e| crate::Error::Storage {
            engine: "sparse_vector".into(),
            detail: format!("write txn: {e}"),
        })?;
        {
            let mut postings_table =
                write_txn
                    .open_table(SPARSE_POSTINGS)
                    .map_err(|e| crate::Error::Storage {
                        engine: "sparse_vector".into(),
                        detail: format!("open postings: {e}"),
                    })?;
            let mut norms_table =
                write_txn
                    .open_table(SPARSE_NORMS)
                    .map_err(|e| crate::Error::Storage {
                        engine: "sparse_vector".into(),
                        detail: format!("open norms: {e}"),
                    })?;

            // Scan all posting lists for this collection and remove doc entries.
            // This is O(vocabulary) but removal is infrequent.
            let prefix = format!("{collection}:");
            let end = format!("{collection}:\u{ffff}");
            let keys_to_update: Vec<(String, Vec<SparsePosting>)> = {
                let range = postings_table
                    .range(prefix.as_str()..end.as_str())
                    .map_err(|e| crate::Error::Storage {
                        engine: "sparse_vector".into(),
                        detail: format!("range: {e}"),
                    })?;
                let mut updates = Vec::new();
                for entry in range {
                    let entry = entry.map_err(|e| crate::Error::Storage {
                        engine: "sparse_vector".into(),
                        detail: format!("entry: {e}"),
                    })?;
                    let key = entry.0.value().to_string();
                    let postings: Vec<SparsePosting> =
                        zerompk::from_msgpack(entry.1.value()).unwrap_or_default();
                    if postings.iter().any(|p| p.doc_id == doc_id) {
                        let filtered: Vec<SparsePosting> = postings
                            .into_iter()
                            .filter(|p| p.doc_id != doc_id)
                            .collect();
                        updates.push((key, filtered));
                    }
                }
                updates
            };

            for (key, postings) in keys_to_update {
                if postings.is_empty() {
                    postings_table
                        .remove(key.as_str())
                        .map_err(|e| crate::Error::Storage {
                            engine: "sparse_vector".into(),