Yehowshua/ZIm
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Replace zed_llm_client with cloud_llm_client (#35309)

Browse source 
This PR replaces the usage of the `zed_llm_client` with the
`cloud_llm_client`.

It was ported into this repo in #35307.

Release Notes:

- N/A
This commit is contained in:
Marshall Bowers 2025-07-29 20:09:14 -04:00 • committed by GitHub
parent 17a0179f0a
commit 7be1f2418d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
47 changed files with 157 additions and 169 deletions
Download patch file Download diff file Expand all files Collapse all files

14
crates/language_models/Cargo.toml
View file

@ -16,18 +16,17 @@ ai_onboarding.workspace = true
anthropic = { workspace = true, features = ["schemars"] }
anyhow.workspace = true
aws-config = { workspace = true, features = ["behavior-version-latest"] }
aws-credential-types = { workspace = true, features = [
"hardcoded-credentials",
] }
aws-credential-types = { workspace = true, features = ["hardcoded-credentials"] }
aws_http_client.workspace = true
bedrock.workspace = true
chrono.workspace = true
client.workspace = true
cloud_llm_client.workspace = true
collections.workspace = true
component.workspace = true
credentials_provider.workspace = true
convert_case.workspace = true
copilot.workspace = true
credentials_provider.workspace = true
deepseek = { workspace = true, features = ["schemars"] }
editor.workspace = true
futures.workspace = true
@ -35,6 +34,7 @@ google_ai = { workspace = true, features = ["schemars"] }
gpui.workspace = true
gpui_tokio.workspace = true
http_client.workspace = true
language.workspace = true
language_model.workspace = true
lmstudio = { workspace = true, features = ["schemars"] }
log.workspace = true
@ -43,8 +43,6 @@ mistral = { workspace = true, features = ["schemars"] }
ollama = { workspace = true, features = ["schemars"] }
open_ai = { workspace = true, features = ["schemars"] }
open_router = { workspace = true, features = ["schemars"] }
vercel = { workspace = true, features = ["schemars"] }
x_ai = { workspace = true, features = ["schemars"] }
partial-json-fixer.workspace = true
proto.workspace = true
release_channel.workspace = true
@ -61,9 +59,9 @@ tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
ui.workspace = true
ui_input.workspace = true
util.workspace = true
vercel = { workspace = true, features = ["schemars"] }
workspace-hack.workspace = true
zed_llm_client.workspace = true
language.workspace = true
x_ai = { workspace = true, features = ["schemars"] }
[dev-dependencies]
editor = { workspace = true, features = ["test-support"] }

76
crates/language_models/src/provider/cloud.rs
View file

@ -3,6 +3,13 @@ use anthropic::AnthropicModelMode;
use anyhow::{Context as _, Result, anyhow};
use chrono::{DateTime, Utc};
use client::{Client, ModelRequestUsage, UserStore, zed_urls};
use cloud_llm_client::{
CLIENT_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, CURRENT_PLAN_HEADER_NAME, CompletionBody,
CompletionEvent, CompletionRequestStatus, CountTokensBody, CountTokensResponse,
EXPIRED_LLM_TOKEN_HEADER_NAME, ListModelsResponse, MODEL_REQUESTS_RESOURCE_HEADER_VALUE,
SERVER_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, SUBSCRIPTION_LIMIT_RESOURCE_HEADER_NAME,
TOOL_USE_LIMIT_REACHED_HEADER_NAME, ZED_VERSION_HEADER_NAME,
};
use futures::{
AsyncBufReadExt, FutureExt, Stream, StreamExt, future::BoxFuture, stream::BoxStream,
};
@ -33,13 +40,6 @@ use std::time::Duration;
use thiserror::Error;
use ui::{TintColor, prelude::*};
use util::{ResultExt as _, maybe};
use zed_llm_client::{
CLIENT_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, CURRENT_PLAN_HEADER_NAME, CompletionBody,
CompletionEvent, CompletionRequestStatus, CountTokensBody, CountTokensResponse,
EXPIRED_LLM_TOKEN_HEADER_NAME, ListModelsResponse, MODEL_REQUESTS_RESOURCE_HEADER_VALUE,
SERVER_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, SUBSCRIPTION_LIMIT_RESOURCE_HEADER_NAME,
TOOL_USE_LIMIT_REACHED_HEADER_NAME, ZED_VERSION_HEADER_NAME,
};
use crate::provider::anthropic::{AnthropicEventMapper, count_anthropic_tokens, into_anthropic};
use crate::provider::google::{GoogleEventMapper, into_google};
@ -120,10 +120,10 @@ pub struct State {
user_store: Entity<UserStore>,
status: client::Status,
accept_terms_of_service_task: Option<Task<Result<()>>>,
models: Vec<Arc<zed_llm_client::LanguageModel>>,
default_model: Option<Arc<zed_llm_client::LanguageModel>>,
default_fast_model: Option<Arc<zed_llm_client::LanguageModel>>,
recommended_models: Vec<Arc<zed_llm_client::LanguageModel>>,
models: Vec<Arc<cloud_llm_client::LanguageModel>>,
default_model: Option<Arc<cloud_llm_client::LanguageModel>>,
default_fast_model: Option<Arc<cloud_llm_client::LanguageModel>>,
recommended_models: Vec<Arc<cloud_llm_client::LanguageModel>>,
_fetch_models_task: Task<()>,
_settings_subscription: Subscription,
_llm_token_subscription: Subscription,
@ -238,8 +238,8 @@ impl State {
// Right now we represent thinking variants of models as separate models on the client,
// so we need to insert variants for any model that supports thinking.
if model.supports_thinking {
models.push(Arc::new(zed_llm_client::LanguageModel {
id: zed_llm_client::LanguageModelId(format!("{}-thinking", model.id).into()),
models.push(Arc::new(cloud_llm_client::LanguageModel {
id: cloud_llm_client::LanguageModelId(format!("{}-thinking", model.id).into()),
display_name: format!("{} Thinking", model.display_name),
..model
}));
@ -328,7 +328,7 @@ impl CloudLanguageModelProvider {
fn create_language_model(
&self,
model: Arc<zed_llm_client::LanguageModel>,
model: Arc<cloud_llm_client::LanguageModel>,
llm_api_token: LlmApiToken,
) -> Arc<dyn LanguageModel> {
Arc::new(CloudLanguageModel {
@ -518,7 +518,7 @@ fn render_accept_terms(
pub struct CloudLanguageModel {
id: LanguageModelId,
model: Arc<zed_llm_client::LanguageModel>,
model: Arc<cloud_llm_client::LanguageModel>,
llm_api_token: LlmApiToken,
client: Arc<Client>,
request_limiter: RateLimiter,
@ -611,12 +611,12 @@ impl CloudLanguageModel {
.headers()
.get(CURRENT_PLAN_HEADER_NAME)
.and_then(|plan| plan.to_str().ok())
.and_then(|plan| zed_llm_client::Plan::from_str(plan).ok())
.and_then(|plan| cloud_llm_client::Plan::from_str(plan).ok())
{
let plan = match plan {
zed_llm_client::Plan::ZedFree => Plan::Free,
zed_llm_client::Plan::ZedPro => Plan::ZedPro,
zed_llm_client::Plan::ZedProTrial => Plan::ZedProTrial,
cloud_llm_client::Plan::ZedFree => Plan::Free,
cloud_llm_client::Plan::ZedPro => Plan::ZedPro,
cloud_llm_client::Plan::ZedProTrial => Plan::ZedProTrial,
};
return Err(anyhow!(ModelRequestLimitReachedError { plan }));
}
@ -729,7 +729,7 @@ impl LanguageModel for CloudLanguageModel {
}
fn upstream_provider_id(&self) -> LanguageModelProviderId {
use zed_llm_client::LanguageModelProvider::*;
use cloud_llm_client::LanguageModelProvider::*;
match self.model.provider {
Anthropic => language_model::ANTHROPIC_PROVIDER_ID,
OpenAi => language_model::OPEN_AI_PROVIDER_ID,
@ -738,7 +738,7 @@ impl LanguageModel for CloudLanguageModel {
}
fn upstream_provider_name(&self) -> LanguageModelProviderName {
use zed_llm_client::LanguageModelProvider::*;
use cloud_llm_client::LanguageModelProvider::*;
match self.model.provider {
Anthropic => language_model::ANTHROPIC_PROVIDER_NAME,
OpenAi => language_model::OPEN_AI_PROVIDER_NAME,
@ -772,11 +772,11 @@ impl LanguageModel for CloudLanguageModel {
fn tool_input_format(&self) -> LanguageModelToolSchemaFormat {
match self.model.provider {
zed_llm_client::LanguageModelProvider::Anthropic
| zed_llm_client::LanguageModelProvider::OpenAi => {
cloud_llm_client::LanguageModelProvider::Anthropic
| cloud_llm_client::LanguageModelProvider::OpenAi => {
LanguageModelToolSchemaFormat::JsonSchema
}
zed_llm_client::LanguageModelProvider::Google => {
cloud_llm_client::LanguageModelProvider::Google => {
LanguageModelToolSchemaFormat::JsonSchemaSubset
}
}
@ -795,15 +795,15 @@ impl LanguageModel for CloudLanguageModel {
fn cache_configuration(&self) -> Option<LanguageModelCacheConfiguration> {
match &self.model.provider {
zed_llm_client::LanguageModelProvider::Anthropic => {
cloud_llm_client::LanguageModelProvider::Anthropic => {
Some(LanguageModelCacheConfiguration {
min_total_token: 2_048,
should_speculate: true,
max_cache_anchors: 4,
})
}
zed_llm_client::LanguageModelProvider::OpenAi
| zed_llm_client::LanguageModelProvider::Google => None,
cloud_llm_client::LanguageModelProvider::OpenAi
| cloud_llm_client::LanguageModelProvider::Google => None,
}
}
@ -813,15 +813,17 @@ impl LanguageModel for CloudLanguageModel {
cx: &App,
) -> BoxFuture<'static, Result<u64>> {
match self.model.provider {
zed_llm_client::LanguageModelProvider::Anthropic => count_anthropic_tokens(request, cx),
zed_llm_client::LanguageModelProvider::OpenAi => {
cloud_llm_client::LanguageModelProvider::Anthropic => {
count_anthropic_tokens(request, cx)
}
cloud_llm_client::LanguageModelProvider::OpenAi => {
let model = match open_ai::Model::from_id(&self.model.id.0) {
Ok(model) => model,
Err(err) => return async move { Err(anyhow!(err)) }.boxed(),
};
count_open_ai_tokens(request, model, cx)
}
zed_llm_client::LanguageModelProvider::Google => {
cloud_llm_client::LanguageModelProvider::Google => {
let client = self.client.clone();
let llm_api_token = self.llm_api_token.clone();
let model_id = self.model.id.to_string();
@ -832,7 +834,7 @@ impl LanguageModel for CloudLanguageModel {
let token = llm_api_token.acquire(&client).await?;
let request_body = CountTokensBody {
provider: zed_llm_client::LanguageModelProvider::Google,
provider: cloud_llm_client::LanguageModelProvider::Google,
model: model_id,
provider_request: serde_json::to_value(&google_ai::CountTokensRequest {
generate_content_request,
@ -893,7 +895,7 @@ impl LanguageModel for CloudLanguageModel {
let app_version = cx.update(|cx| AppVersion::global(cx)).ok();
let thinking_allowed = request.thinking_allowed;
match self.model.provider {
zed_llm_client::LanguageModelProvider::Anthropic => {
cloud_llm_client::LanguageModelProvider::Anthropic => {
let request = into_anthropic(
request,
self.model.id.to_string(),
@ -924,7 +926,7 @@ impl LanguageModel for CloudLanguageModel {
prompt_id,
intent,
mode,
provider: zed_llm_client::LanguageModelProvider::Anthropic,
provider: cloud_llm_client::LanguageModelProvider::Anthropic,
model: request.model.clone(),
provider_request: serde_json::to_value(&request)
.map_err(|e| anyhow!(e))?,
@ -948,7 +950,7 @@ impl LanguageModel for CloudLanguageModel {
});
async move { Ok(future.await?.boxed()) }.boxed()
}
zed_llm_client::LanguageModelProvider::OpenAi => {
cloud_llm_client::LanguageModelProvider::OpenAi => {
let client = self.client.clone();
let model = match open_ai::Model::from_id(&self.model.id.0) {
Ok(model) => model,
@ -976,7 +978,7 @@ impl LanguageModel for CloudLanguageModel {
prompt_id,
intent,
mode,
provider: zed_llm_client::LanguageModelProvider::OpenAi,
provider: cloud_llm_client::LanguageModelProvider::OpenAi,
model: request.model.clone(),
provider_request: serde_json::to_value(&request)
.map_err(|e| anyhow!(e))?,
@ -996,7 +998,7 @@ impl LanguageModel for CloudLanguageModel {
});
async move { Ok(future.await?.boxed()) }.boxed()
}
zed_llm_client::LanguageModelProvider::Google => {
cloud_llm_client::LanguageModelProvider::Google => {
let client = self.client.clone();
let request =
into_google(request, self.model.id.to_string(), GoogleModelMode::Default);
@ -1016,7 +1018,7 @@ impl LanguageModel for CloudLanguageModel {
prompt_id,
intent,
mode,
provider: zed_llm_client::LanguageModelProvider::Google,
provider: cloud_llm_client::LanguageModelProvider::Google,
model: request.model.model_id.clone(),
provider_request: serde_json::to_value(&request)
.map_err(|e| anyhow!(e))?,

2
crates/language_models/src/provider/copilot_chat.rs
View file

@ -3,6 +3,7 @@ use std::str::FromStr as _;
use std::sync::Arc;
use anyhow::{Result, anyhow};
use cloud_llm_client::CompletionIntent;
use collections::HashMap;
use copilot::copilot_chat::{
ChatMessage, ChatMessageContent, ChatMessagePart, CopilotChat, ImageUrl,
@ -30,7 +31,6 @@ use settings::SettingsStore;
use std::time::Duration;
use ui::prelude::*;
use util::debug_panic;
use zed_llm_client::CompletionIntent;
use super::anthropic::count_anthropic_tokens;
use super::google::count_google_tokens;