hweller1’s gists

hweller1 / count % nested filter fields

Created November 11, 2025 20:34

	WITH index_fields AS (
	-- Parse each field from the index definitions
	SELECT
	vectorindexes__name as index_name,
	path as field_path,
	numdimensions,
	type,
	CASE
	WHEN numdimensions IS NOT NULL THEN 'vector'
	ELSE 'filter'

hweller1 / normalize_vectors.py

Created August 13, 2025 17:32

Simple script to normalize new vectors or update existing array of floats or binData vectors in MongoDB

	import math
	from pymongo import MongoClient, UpdateOne
	from bson.binary import Binary

	def normalize(vector):
	"""Normalize a vector to unit length."""
	magnitude = math.sqrt(sum(x**2 for x in vector))
	return [x / magnitude for x in vector] if magnitude > 0 else vector

	def normalize_vectors(vectors=None, db_uri="mongodb://localhost:27017", db_name="vectors", collection_name="data", batch_size=1000):

hweller1 / pgvector_filtered_search_query_pattern.py

Created July 10, 2025 19:17

Filtered two stage vector search using pgvector

	`CREATE OR REPLACE FUNCTION two_stage_vector_search_optimized_prefiltered(

	query_embedding vector,
	match_limit int,
	num_candidates int,
	filter_category text DEFAULT NULL::text,
	max_price double precision DEFAULT NULL::double precision
	)
	RETURNS TABLE (
	id int8,

hweller1 / MRL_truncate_view

Created May 22, 2025 23:48

Example view definition for truncating MRL vectors

hweller1 / gist:5174da29e4eb5e5a5230d26f79902aaf

Created October 25, 2024 16:09

bson_sample_dataset_ingest.py

	import pymongo
	import json
	from bson.binary import Binary, BinaryVectorDtype

	connection_str = ""
	client = pymongo.MongoClient(connection_str) # mongodb cluster URI

	db = client['bsontestdb']
	coll = db['embedded_mflix_bson']

hweller1 / openai_query.py

Last active October 23, 2024 20:45

	import os
	import openai

	embedding = openai.Embedding.create(input="test query", model="text-embedding-ada-002").data[0].embedding
	pipeline = [
	{
	"$vectorSearch": {
	"index":'bson_vector_index',
	"path": "plot_embedding",
	"queryVector": embedding,

hweller1 / scalar_quantize_openai.py

Created October 9, 2024 18:08

user space SQ of openAI text-embedding-ada-002 (indexing + querying)

	# deps

	# %pip install pymongo openai numpy

	import numpy as np
	import os
	os.environ["OPENAI_API_KEY"] = '<my-openai-key>'
	import pymongo
	import openai

hweller1 / sample queryString

Created September 20, 2024 14:45

	"knnBeta": {
	"filter": {
	"compound": {
	"must": [
	{
	"queryString": {
	"query": "TEST_NAME_1 AND TEST_NAME_2~2 NOT TEST_NAME_3",
	"defaultPath": "descriptions"
	}
	},

hweller1 / updated_code.py

Last active December 4, 2023 16:49

	searchQ.filter = filter;
	}
	else if(beds \|\| rooms) {
	let filter = { "$and" : []}

	if (beds) {
	filter.$and.push({"beds" : {"$gte" : parseInt(beds) }})
	}
	if (rooms)
	{

hweller1 / gist:be04c89cb21464891ddf1fd6ddcdf6a1

Created December 4, 2023 15:46

sample_query3.py