adm_ai_qdrant_api

Package for interacting with Qdrant vector database Handles point creation, updates, and queries

Types

r_collection_detail_type

Type for collection detailed info

Definition:

type r_collection_detail_type is record (
  status varchar2(100 char),
  optimizer_status varchar2(100 char),
  points_count number,
  segments_count number
);

Fields:

Name	Type	Description
status	varchar2(100 char)	-
optimizer_status	varchar2(100 char)	-
points_count	number	-
segments_count	number	-

---

Functions and Procedures

upsert_point

Store a single vector point in Qdrant

Signature:

procedure upsert_point (
  p_collection_name in varchar2,
  p_point_id in varchar2,
  p_vector in json_array_t,
  p_payload in json_object_t default null
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_point_id	in	varchar2	Unique identifier for the point
p_vector	in	json_array_t	JSON array containing the vector embeddings
p_payload	in	json_object_t default null	Optional JSON object with metadata

---

upsert_points

Store multiple vector points in Qdrant in a single request

Signature:

procedure upsert_points (
  p_collection_name in varchar2,
  p_points_array in json_array_t
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_points_array	in	json_array_t	JSON array of point objects (each with id, vector, and optional payload)

---

delete_point

Delete a point from Qdrant

Signature:

procedure delete_point (
  p_collection_name in varchar2,
  p_point_id in varchar2
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_point_id	in	varchar2	ID of the point to delete

---

search_similar

Search for similar vectors in Qdrant

Signature:

function search_similar (
  p_collection_name in varchar2,
  p_vector in json_array_t,
  p_limit in number default 10,
  p_score_threshold in number default 0.5
) return json_array_t;

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_vector	in	json_array_t	Query vector
p_limit	in	number default 10	Maximum number of results to return
p_score_threshold	in	number default 0.5	Minimum similarity score threshold

Returns: json_array_t - JSON array of search results

---

scroll_all_point_ids

Scroll through all points in a Qdrant collection

Signature:

function scroll_all_point_ids (
  p_collection_name in varchar2,
  p_limit in number default 100
) return apex_t_varchar2;

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_limit	in	number default 100	Number of points to retrieve per scroll request (default: 100)

Returns: apex_t_varchar2 - JSON array of all point IDs in the collection

---

delete_points

Delete multiple points from Qdrant in a single request

Signature:

procedure delete_points (
  p_collection_name in varchar2,
  p_point_ids in apex_t_varchar2
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the Qdrant collection
p_point_ids	in	apex_t_varchar2	Array of point IDs to delete

---

sync_with_db

Synchronize Qdrant collection with Oracle DB

• Removes points from Qdrant that don’t exist in adm_ai_rag_chunks
• Sets embedded_at to NULL for chunks in DB that don’t exist in Qdrant

Signature:

procedure sync_with_db (
  p_rag_collection_id in number
);

Parameters:

Name	Direction	Type	Description
p_rag_collection_id	in	number	ID of the RAG collection to sync

---

get_all_collections

Get all collections from Qdrant

Signature:

function get_all_collections return t_collection_names pipelined;

Returns: t_collection_names pipelined - Pipelined table of collection names

---

create_collection

Create a new collection in Qdrant

Signature:

procedure create_collection (
  p_collection_name in varchar2,
  p_vector_size in number default 1024,
  p_distance in varchar2 default 'Cosine'
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name for the new collection
p_vector_size	in	number default 1024	Size of the vector embeddings (default: 1024)
p_distance	in	varchar2 default 'Cosine'	Distance metric for similarity (default: ‘Cosine’)

---

get_collection_info

Get detailed information about a specific collection

Signature:

function get_collection_info (
  p_collection_name in varchar2
) return r_collection_detail_type;

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the collection to query

Returns: r_collection_detail_type - Collection statistics (status, optimizer_status, points_count, segments_count)

---

delete_collection

Delete a collection from Qdrant

Signature:

procedure delete_collection (
  p_collection_name in varchar2
);

Parameters:

Name	Direction	Type	Description
p_collection_name	in	varchar2	Name of the collection to delete

---