Use tool calling instead of XML parsing to generate edit operations (#15385)

Release Notes:

- N/A

---------

Co-authored-by: Nathan <nathan@zed.dev>

crates/proto/proto/zed.proto

@@ -194,8 +194,12 @@ message Envelope {
 
         JoinHostedProject join_hosted_project = 164;
 
-        QueryLanguageModel query_language_model = 224;
-        QueryLanguageModelResponse query_language_model_response = 225; // current max
+        CompleteWithLanguageModel complete_with_language_model = 226;
+        CompleteWithLanguageModelResponse complete_with_language_model_response = 227;
+        StreamCompleteWithLanguageModel stream_complete_with_language_model = 228;
+        StreamCompleteWithLanguageModelResponse stream_complete_with_language_model_response = 229;
+        CountLanguageModelTokens count_language_model_tokens = 230;
+        CountLanguageModelTokensResponse count_language_model_tokens_response = 231; // current max
         GetCachedEmbeddings get_cached_embeddings = 189;
         GetCachedEmbeddingsResponse get_cached_embeddings_response = 190;
         ComputeEmbeddings compute_embeddings = 191;
@@ -267,6 +271,7 @@ message Envelope {
 
     reserved 158 to 161;
     reserved 166 to 169;
+    reserved 224 to 225;
 }
 
 // Messages
@@ -2050,10 +2055,31 @@ enum LanguageModelRole {
     reserved 3;
 }
 
-message QueryLanguageModel {
+message CompleteWithLanguageModel {
     LanguageModelProvider provider = 1;
-    LanguageModelRequestKind kind = 2;
-    string request = 3;
+    string request = 2;
+}
+
+message CompleteWithLanguageModelResponse {
+    string completion = 1;
+}
+
+message StreamCompleteWithLanguageModel {
+    LanguageModelProvider provider = 1;
+    string request = 2;
+}
+
+message StreamCompleteWithLanguageModelResponse {
+    string event = 1;
+}
+
+message CountLanguageModelTokens {
+    LanguageModelProvider provider = 1;
+    string request = 2;
+}
+
+message CountLanguageModelTokensResponse {
+    uint32 token_count = 1;
 }
 
 enum LanguageModelProvider {
@@ -2062,15 +2088,6 @@ enum LanguageModelProvider {
     Google = 2;
 }
 
-enum LanguageModelRequestKind {
-    Complete = 0;
-    CountTokens = 1;
-}
-
-message QueryLanguageModelResponse {
-    string response = 1;
-}
-
 message GetCachedEmbeddings {
     string model = 1;
     repeated bytes digests = 2;

crates/proto/src/proto.rs

@@ -294,8 +294,12 @@ messages!(
     (PrepareRename, Background),
     (PrepareRenameResponse, Background),
     (ProjectEntryResponse, Foreground),
-    (QueryLanguageModel, Background),
-    (QueryLanguageModelResponse, Background),
+    (CompleteWithLanguageModel, Background),
+    (CompleteWithLanguageModelResponse, Background),
+    (StreamCompleteWithLanguageModel, Background),
+    (StreamCompleteWithLanguageModelResponse, Background),
+    (CountLanguageModelTokens, Background),
+    (CountLanguageModelTokensResponse, Background),
     (RefreshInlayHints, Foreground),
     (RejoinChannelBuffers, Foreground),
     (RejoinChannelBuffersResponse, Foreground),
@@ -463,7 +467,12 @@ request_messages!(
     (PerformRename, PerformRenameResponse),
     (Ping, Ack),
     (PrepareRename, PrepareRenameResponse),
-    (QueryLanguageModel, QueryLanguageModelResponse),
+    (CompleteWithLanguageModel, CompleteWithLanguageModelResponse),
+    (
+        StreamCompleteWithLanguageModel,
+        StreamCompleteWithLanguageModelResponse
+    ),
+    (CountLanguageModelTokens, CountLanguageModelTokensResponse),
     (RefreshInlayHints, Ack),
     (RejoinChannelBuffers, RejoinChannelBuffersResponse),
     (RejoinRoom, RejoinRoomResponse),