Jacob wooyakob
wooyakob
/ ca-eateries.sql
Created
October 3, 2025 15:47
SQL++ query for Couchbase that retrieves restaurants (landmark documents) in California with activity type "eat". It ensures only complete records are returned by requiring non-null values for key fields like name, content, price, URL, geo, and phone.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| -- Bucket: travel-sample | |
| -- Scope: inventory | |
| -- Collection: landmark | |
| SELECT | |
| l.name, | |
| l.content, | |
| l.price, | |
| l.url, | |
| l.geo, |
wooyakob
/ restaurant-vector-search.py
Created
October 3, 2025 15:45
Performs a semantic restaurant search using Couchbase’s vector search. The function embeds a user query, retrieves matching documents from a Couchbase search index, and enriches results with Google Maps reverse-geocoded addresses and embed URLs for easy integration into apps.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| def search_restaurants(self, question): | |
| vector = self.embeddings_model.embed_query(question) | |
| try: | |
| search_req = search.SearchRequest.create(search.MatchNoneQuery()).with_vector_search( | |
| VectorSearch.from_vector_query(VectorQuery('embedding', vector, num_candidates=3)) | |
| ) | |
| result = self.scope.search( | |
| self.search_index, | |
| search_req, | |
| SearchOptions(limit=13, fields=["name", "content", "phone", "price", "url", "geo.lat", "geo.lon"]) |
wooyakob
/ cb-property-search.py
Created
October 2, 2025 23:03
Utility function performs a personalized property search using Couchbase’s vector similarity capabilities. It returns the top k most relevant properties for a given query, automatically excluding any properties the user has already saved (using flexible deduplication keys: ID, address, name, price).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| def couchbase_property_search(vector_store, query, k=5, saved_properties=None): | |
| """ | |
| Search properties in Couchbase using vector similarity, excluding saved properties. | |
| - vector_store: CouchbaseSearchVectorStore instance | |
| - query: search string | |
| - k: number of results to return | |
| - saved_properties: list of dicts (already saved by user) | |
| Returns: list of property dicts | |
| """ |
wooyakob
/ audio-vggish.py
Created
October 2, 2025 22:51
Extract audio embeddings to a JSON doc from a .wav file, using Google’s VGGish model, useful for tasks such as audio classification, retrieval, clustering and feature extraction.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import tensorflow as tf | |
| import tensorflow_hub as hub | |
| import numpy as np | |
| import soundfile as sf | |
| # Load VGGish model from TensorFlow Hub | |
| vggish = hub.load('https://tfhub.dev/google/vggish/1') | |
| def load_wav_for_vggish(file_path): | |
| wav, sr = sf.read(file_path) |