Added support for float16 and bfloat16. Renamed vector_convert_type to vector_as_type.

Author: marcobambini

API.md

@@ -163,22 +163,23 @@ SELECT vector_cleanup('documents', 'embedding');
 
 ---
 
-## `vector_convert_f32(value)`
+## `vector_as_f32(value)`
 
-## `vector_convert_f16(value)`
+## `vector_as_f16(value)`
 
-## `vector_convert_bf16(value)`
+## `vector_as_bf16(value)`
 
-## `vector_convert_i8(value)`
+## `vector_as_i8(value)`
 
-## `vector_convert_u8(value)`
+## `vector_as_u8(value)`
 
 **Returns:** `BLOB`
 
 **Description:**
 Encodes a vector into the required internal BLOB format to ensure correct storage and compatibility with the system’s vector representation.
+A real conversion is performed ONLY in case of JSON input. When input is a BLOB, it is assumed to be already properly formatted.
 
-Functions in the `vector_convert_` family should be used in all `INSERT`, `UPDATE`, and `DELETE` statements to properly format vector values. However, they are *not* required when specifying input vectors for the `vector_full_scan` or `vector_quantize_scan` virtual tables.
+Functions in the `vector_as_` family should be used in all `INSERT`, `UPDATE`, and `DELETE` statements to properly format vector values. However, they are *not* required when specifying input vectors for the `vector_full_scan` or `vector_quantize_scan` virtual tables.
 
 **Parameters:**
 
@@ -193,10 +194,10 @@ Functions in the `vector_convert_` family should be used in all `INSERT`, `UPDAT
 
 ```sql
 -- Insert a Float32 vector using JSON
-INSERT INTO documents(embedding) VALUES(vector_convert_f32('[0.1, 0.2, 0.3]'));
+INSERT INTO documents(embedding) VALUES(vector_as_f32('[0.1, 0.2, 0.3]'));
 
 -- Insert a UInt8 vector using raw BLOB (ensure correct formatting!)
-INSERT INTO compressed_vectors(embedding) VALUES(vector_convert_u8(X'010203'));
+INSERT INTO compressed_vectors(embedding) VALUES(vector_as_u8(X'010203'));
 ```
 
 ---
@@ -219,7 +220,7 @@ Performs a brute-force nearest neighbor search using the given vector. Despite i
 
 ```sql
 SELECT rowid, distance
-FROM vector_full_scan('documents', 'embedding', vector_convert_f32('[0.1, 0.2, 0.3]'), 5);
+FROM vector_full_scan('documents', 'embedding', vector_as_f32('[0.1, 0.2, 0.3]'), 5);
 ```
 
 ---
@@ -250,7 +251,7 @@ You **must run `vector_quantize()`** before using `vector_quantize_scan()` and w
 
 ```sql
 SELECT rowid, distance
-FROM vector_quantize_scan('documents', 'embedding', vector_convert_f32('[0.1, 0.2, 0.3]'), 10);
+FROM vector_quantize_scan('documents', 'embedding', vector_as_f32('[0.1, 0.2, 0.3]'), 10);
 ```
 
 ---

README.md

@@ -66,7 +66,7 @@ CREATE TABLE images (
 INSERT INTO images (embedding, label) VALUES (?, 'cat');
 
 -- Insert a JSON vector (Float32, 384 dimensions)
-INSERT INTO images (embedding, label) VALUES (vector_convert_f32('[0.3, 1.0, 0.9, 3.2, 1.4,...]'), 'dog');
+INSERT INTO images (embedding, label) VALUES (vector_as_f32('[0.3, 1.0, 0.9, 3.2, 1.4,...]'), 'dog');
 
 -- Initialize the vector. By default, the distance function is L2.
 -- To use a different metric, specify one of the following options:

examples/semantic_search/semantic_search.py

@@ -124,7 +124,7 @@ def index_file(self, filepath: str) -> int:
             embedding_json = json.dumps(embedding.tolist())
 
             cursor.execute(
-                "INSERT INTO documents (filepath, content, embedding) VALUES (?, ?, vector_convert_f32(?))",
+                "INSERT INTO documents (filepath, content, embedding) VALUES (?, ?, vector_as_f32(?))",
                 (filepath, chunk, embedding_json)
             )
             chunk_count += 1
@@ -168,7 +168,7 @@ def search(self, query: str, limit: int = 3) -> Tuple[float, List[Tuple[str, str
         cursor.execute("""
             SELECT d.id, d.filepath, d.content, v.distance
             FROM documents AS d
-                JOIN vector_quantize_scan('documents', 'embedding', vector_convert_f32(?), ?) AS v
+                JOIN vector_quantize_scan('documents', 'embedding', vector_as_f32(?), ?) AS v
                 ON d.id = v.rowid;
         """, (query_json, limit))
         elapsed_ms = round((time.time() - start_time) * 1000, 2)