Semantic Index (#10329)

This introduces semantic indexing in Zed based on chunking text from
files in the developer's workspace and creating vector embeddings using
an embedding model. As part of this, we've created an embeddings
provider trait that allows us to work with OpenAI, a local Ollama model,
or a Zed hosted embedding.

The semantic index is built by breaking down text for known
(programming) languages into manageable chunks that are smaller than the
max token size. Each chunk is then fed to a language model to create a
high dimensional vector which is then normalized to a unit vector to
allow fast comparison with other vectors with a simple dot product.
Alongside the vector, we store the path of the file and the range within
the document where the vector was sourced from.

Zed will soon grok contextual similarity across different text snippets,
allowing for natural language search beyond keyword matching. This is
being put together both for human-based search as well as providing
results to Large Language Models to allow them to refine how they help
developers.

Remaining todo:

* [x] Change `provider` to `model` within the zed hosted embeddings
database (as its currently a combo of the provider and the model in one
name)


Release Notes:

- N/A

---------

Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Conrad Irwin <conrad@zed.dev>
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
Co-authored-by: Antonio <antonio@zed.dev>

crates/semantic_index/src/embedding/cloud.rs

@@ -0,0 +1,88 @@
+use crate::{Embedding, EmbeddingProvider, TextToEmbed};
+use anyhow::{anyhow, Context, Result};
+use client::{proto, Client};
+use collections::HashMap;
+use futures::{future::BoxFuture, FutureExt};
+use std::sync::Arc;
+
+pub struct CloudEmbeddingProvider {
+    model: String,
+    client: Arc<Client>,
+}
+
+impl CloudEmbeddingProvider {
+    pub fn new(client: Arc<Client>) -> Self {
+        Self {
+            model: "openai/text-embedding-3-small".into(),
+            client,
+        }
+    }
+}
+
+impl EmbeddingProvider for CloudEmbeddingProvider {
+    fn embed<'a>(&'a self, texts: &'a [TextToEmbed<'a>]) -> BoxFuture<'a, Result<Vec<Embedding>>> {
+        // First, fetch any embeddings that are cached based on the requested texts' digests
+        // Then compute any embeddings that are missing.
+        async move {
+            let cached_embeddings = self.client.request(proto::GetCachedEmbeddings {
+                model: self.model.clone(),
+                digests: texts
+                    .iter()
+                    .map(|to_embed| to_embed.digest.to_vec())
+                    .collect(),
+            });
+            let mut embeddings = cached_embeddings
+                .await
+                .context("failed to fetch cached embeddings via cloud model")?
+                .embeddings
+                .into_iter()
+                .map(|embedding| {
+                    let digest: [u8; 32] = embedding
+                        .digest
+                        .try_into()
+                        .map_err(|_| anyhow!("invalid digest for cached embedding"))?;
+                    Ok((digest, embedding.dimensions))
+                })
+                .collect::<Result<HashMap<_, _>>>()?;
+
+            let compute_embeddings_request = proto::ComputeEmbeddings {
+                model: self.model.clone(),
+                texts: texts
+                    .iter()
+                    .filter_map(|to_embed| {
+                        if embeddings.contains_key(&to_embed.digest) {
+                            None
+                        } else {
+                            Some(to_embed.text.to_string())
+                        }
+                    })
+                    .collect(),
+            };
+            if !compute_embeddings_request.texts.is_empty() {
+                let missing_embeddings = self.client.request(compute_embeddings_request).await?;
+                for embedding in missing_embeddings.embeddings {
+                    let digest: [u8; 32] = embedding
+                        .digest
+                        .try_into()
+                        .map_err(|_| anyhow!("invalid digest for cached embedding"))?;
+                    embeddings.insert(digest, embedding.dimensions);
+                }
+            }
+
+            texts
+                .iter()
+                .map(|to_embed| {
+                    let dimensions = embeddings.remove(&to_embed.digest).with_context(|| {
+                        format!("server did not return an embedding for {:?}", to_embed)
+                    })?;
+                    Ok(Embedding::new(dimensions))
+                })
+                .collect()
+        }
+        .boxed()
+    }
+
+    fn batch_size(&self) -> usize {
+        2048
+    }
+}