delete instead of archive, batch OpenAI requests (#129)

* delete instead of archive

* vectorscale optional

* break up large openai requests

* Delete core/src/transformers/debug-test.py

core/src/transformers/http_handler.rs

@@ -1,8 +1,7 @@
-use anyhow::Result;
-
 use crate::transformers::types::{
     EmbeddingPayload, EmbeddingRequest, EmbeddingResponse, Inputs, PairedEmbeddings,
 };
+use anyhow::Result;
 pub async fn handle_response<T: for<'de> serde::Deserialize<'de>>(
     resp: reqwest::Response,
     method: &'static str,
@@ -26,22 +25,47 @@ pub async fn openai_embedding_request(
     timeout: i32,
 ) -> Result<Vec<Vec<f64>>> {
     let client = reqwest::Client::new();
-    let mut req = client
-        .post(request.url)
-        .timeout(std::time::Duration::from_secs(timeout as u64))
-        .json::<EmbeddingPayload>(&request.payload)
-        .header("Content-Type", "application/json");
-    if let Some(key) = request.api_key {
-        req = req.header("Authorization", format!("Bearer {}", key));
+
+    // openai request size limit is 2048 inputs
+    let number_inputs = request.payload.input.len();
+    let todo_requests: Vec<EmbeddingPayload> = if number_inputs > 2048 {
+        split_vector(request.payload.input, 2048)
+            .iter()
+            .map(|chunk| EmbeddingPayload {
+                input: chunk.clone(),
+                model: request.payload.model.clone(),
+            })
+            .collect()
+    } else {
+        vec![request.payload]
+    };
+
+    let mut all_embeddings: Vec<Vec<f64>> = Vec::with_capacity(number_inputs);
+
+    for request_payload in todo_requests.iter() {
+        let mut req = client
+            .post(&request.url)
+            .timeout(std::time::Duration::from_secs(timeout as u64))
+            .json::<EmbeddingPayload>(request_payload)
+            .header("Content-Type", "application/json");
+        if let Some(key) = request.api_key.as_ref() {
+            req = req.header("Authorization", format!("Bearer {}", key));
+        }
+        let resp = req.send().await?;
+        let embedding_resp: EmbeddingResponse =
+            handle_response::<EmbeddingResponse>(resp, "embeddings").await?;
+        let embeddings: Vec<Vec<f64>> = embedding_resp
+            .data
+            .iter()
+            .map(|d| d.embedding.clone())
+            .collect();
+        all_embeddings.extend(embeddings);
     }
-    let resp = req.send().await?;
-    let embedding_resp = handle_response::<EmbeddingResponse>(resp, "embeddings").await?;
-    let embeddings = embedding_resp
-        .data
-        .iter()
-        .map(|d| d.embedding.clone())
-        .collect();
-    Ok(embeddings)
+    Ok(all_embeddings)
+}
+
+fn split_vector(vec: Vec<String>, chunk_size: usize) -> Vec<Vec<String>> {
+    vec.chunks(chunk_size).map(|chunk| chunk.to_vec()).collect()
 }
 
 // merges the vec of inputs with the embedding responses

extension/Trunk.toml

@@ -6,7 +6,7 @@ description = "The simplest way to orchestrate vector search on Postgres."
 homepage = "https://github.com/tembo-io/pg_vectorize"
 documentation = "https://github.com/tembo-io/pg_vectorize"
 categories = ["orchestration", "machine_learning"]
-version = "0.16.0"
+version = "0.17.0"
 
 [build]
 postgres_version = "15"

extension/src/transformers/mod.rs

@@ -43,8 +43,13 @@ pub fn transform(input: &str, transformer: &Model, api_key: Option<String>) -> V
                 input: vec![input.to_string()],
                 model: transformer.name.to_string(),
             };
+
+            let url = match guc::get_guc(guc::VectorizeGuc::OpenAIServiceUrl) {
+                Some(k) => k,
+                None => OPENAI_BASE_URL.to_string(),
+            };
             EmbeddingRequest {
-                url: format!("{OPENAI_BASE_URL}/embeddings"),
+                url: format!("{url}/embeddings"),
                 payload: embedding_request,
                 api_key: Some(api_key.to_string()),
             }

extension/src/workers/mod.rs

@@ -51,7 +51,7 @@ pub async fn run_worker(
 
     // delete message from queue
     if delete_it {
-        match queue.archive(queue_name, msg_id).await {
+        match queue.delete(queue_name, msg_id).await {
             Ok(_) => {
                 info!("pg-vectorize: deleted message: {}", msg_id);
             }

extension/vectorize.control

@@ -4,4 +4,4 @@ module_pathname = '$libdir/vectorize'
 relocatable = false
 superuser = true
 schema = 'vectorize'
-requires = 'pg_cron,pgmq,vector,vectorscale'
+requires = 'pg_cron,pgmq,vector'