ZIm/crates/language_models/src/provider/vercel.rs

use anyhow::{Context as _, Result, anyhow};
use collections::BTreeMap;
use credentials_provider::CredentialsProvider;
use futures::{FutureExt, StreamExt, future::BoxFuture};
use gpui::{AnyView, App, AsyncApp, Context, Entity, Subscription, Task, Window};
use http_client::HttpClient;
use language_model::{
    AuthenticateError, LanguageModel, LanguageModelCompletionError, LanguageModelCompletionEvent,
    LanguageModelId, LanguageModelName, LanguageModelProvider, LanguageModelProviderId,
    LanguageModelProviderName, LanguageModelProviderState, LanguageModelRequest,
    LanguageModelToolChoice, RateLimiter, Role,
};
use menu;
use open_ai::ResponseStreamEvent;
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};
use settings::{Settings, SettingsStore};
use std::sync::Arc;
use strum::IntoEnumIterator;
use vercel::Model;

use ui::{ElevationIndex, List, Tooltip, prelude::*};
use ui_input::SingleLineInput;
use util::ResultExt;

use crate::{AllLanguageModelSettings, ui::InstructionListItem};

const PROVIDER_ID: LanguageModelProviderId = LanguageModelProviderId::new("vercel");
const PROVIDER_NAME: LanguageModelProviderName = LanguageModelProviderName::new("Vercel");

#[derive(Default, Clone, Debug, PartialEq)]
pub struct VercelSettings {
    pub api_url: String,
    pub available_models: Vec<AvailableModel>,
}

#[derive(Clone, Debug, PartialEq, Serialize, Deserialize, JsonSchema)]
pub struct AvailableModel {
    pub name: String,
    pub display_name: Option<String>,
    pub max_tokens: u64,
    pub max_output_tokens: Option<u64>,
    pub max_completion_tokens: Option<u64>,
}

pub struct VercelLanguageModelProvider {
    http_client: Arc<dyn HttpClient>,
    state: gpui::Entity<State>,
}

pub struct State {
    api_key: Option<String>,
    api_key_from_env: bool,
    _subscription: Subscription,
}

const VERCEL_API_KEY_VAR: &str = "VERCEL_API_KEY";

impl State {
    fn is_authenticated(&self) -> bool {
        self.api_key.is_some()
    }

    fn reset_api_key(&self, cx: &mut Context<Self>) -> Task<Result<()>> {
        let credentials_provider = <dyn CredentialsProvider>::global(cx);
        let settings = &AllLanguageModelSettings::get_global(cx).vercel;
        let api_url = if settings.api_url.is_empty() {
            vercel::VERCEL_API_URL.to_string()
        } else {
            settings.api_url.clone()
        };
        cx.spawn(async move |this, cx| {
            credentials_provider
                .delete_credentials(&api_url, &cx)
                .await
                .log_err();
            this.update(cx, |this, cx| {
                this.api_key = None;
                this.api_key_from_env = false;
                cx.notify();
            })
        })
    }

    fn set_api_key(&mut self, api_key: String, cx: &mut Context<Self>) -> Task<Result<()>> {
        let credentials_provider = <dyn CredentialsProvider>::global(cx);
        let settings = &AllLanguageModelSettings::get_global(cx).vercel;
        let api_url = if settings.api_url.is_empty() {
            vercel::VERCEL_API_URL.to_string()
        } else {
            settings.api_url.clone()
        };
        cx.spawn(async move |this, cx| {
            credentials_provider
                .write_credentials(&api_url, "Bearer", api_key.as_bytes(), &cx)
                .await
                .log_err();
            this.update(cx, |this, cx| {
                this.api_key = Some(api_key);
                cx.notify();
            })
        })
    }

    fn authenticate(&self, cx: &mut Context<Self>) -> Task<Result<(), AuthenticateError>> {
        if self.is_authenticated() {
            return Task::ready(Ok(()));
        }

        let credentials_provider = <dyn CredentialsProvider>::global(cx);
        let settings = &AllLanguageModelSettings::get_global(cx).vercel;
        let api_url = if settings.api_url.is_empty() {
            vercel::VERCEL_API_URL.to_string()
        } else {
            settings.api_url.clone()
        };
        cx.spawn(async move |this, cx| {
            let (api_key, from_env) = if let Ok(api_key) = std::env::var(VERCEL_API_KEY_VAR) {
                (api_key, true)
            } else {
                let (_, api_key) = credentials_provider
                    .read_credentials(&api_url, &cx)
                    .await?
                    .ok_or(AuthenticateError::CredentialsNotFound)?;
                (
                    String::from_utf8(api_key).context("invalid {PROVIDER_NAME} API key")?,
                    false,
                )
            };
            this.update(cx, |this, cx| {
                this.api_key = Some(api_key);
                this.api_key_from_env = from_env;
                cx.notify();
            })?;

            Ok(())
        })
    }
}

impl VercelLanguageModelProvider {
    pub fn new(http_client: Arc<dyn HttpClient>, cx: &mut App) -> Self {
        let state = cx.new(|cx| State {
            api_key: None,
            api_key_from_env: false,
            _subscription: cx.observe_global::<SettingsStore>(|_this: &mut State, cx| {
                cx.notify();
            }),
        });

        Self { http_client, state }
    }

    fn create_language_model(&self, model: vercel::Model) -> Arc<dyn LanguageModel> {
        Arc::new(VercelLanguageModel {
            id: LanguageModelId::from(model.id().to_string()),
            model,
            state: self.state.clone(),
            http_client: self.http_client.clone(),
            request_limiter: RateLimiter::new(4),
        })
    }
}

impl LanguageModelProviderState for VercelLanguageModelProvider {
    type ObservableEntity = State;

    fn observable_entity(&self) -> Option<gpui::Entity<Self::ObservableEntity>> {
        Some(self.state.clone())
    }
}

impl LanguageModelProvider for VercelLanguageModelProvider {
    fn id(&self) -> LanguageModelProviderId {
        PROVIDER_ID
    }

    fn name(&self) -> LanguageModelProviderName {
        PROVIDER_NAME
    }

    fn icon(&self) -> IconName {
        IconName::AiVZero
    }

    fn default_model(&self, _cx: &App) -> Option<Arc<dyn LanguageModel>> {
        Some(self.create_language_model(vercel::Model::default()))
    }

    fn default_fast_model(&self, _cx: &App) -> Option<Arc<dyn LanguageModel>> {
        Some(self.create_language_model(vercel::Model::default_fast()))
    }

    fn provided_models(&self, cx: &App) -> Vec<Arc<dyn LanguageModel>> {
        let mut models = BTreeMap::default();

        for model in vercel::Model::iter() {
            if !matches!(model, vercel::Model::Custom { .. }) {
                models.insert(model.id().to_string(), model);
            }
        }

        for model in &AllLanguageModelSettings::get_global(cx)
            .vercel
            .available_models
        {
            models.insert(
                model.name.clone(),
                vercel::Model::Custom {
                    name: model.name.clone(),
                    display_name: model.display_name.clone(),
                    max_tokens: model.max_tokens,
                    max_output_tokens: model.max_output_tokens,
                    max_completion_tokens: model.max_completion_tokens,
                },
            );
        }

        models
            .into_values()
            .map(|model| self.create_language_model(model))
            .collect()
    }

    fn is_authenticated(&self, cx: &App) -> bool {
        self.state.read(cx).is_authenticated()
    }

    fn authenticate(&self, cx: &mut App) -> Task<Result<(), AuthenticateError>> {
        self.state.update(cx, |state, cx| state.authenticate(cx))
    }

    fn configuration_view(&self, window: &mut Window, cx: &mut App) -> AnyView {
        cx.new(|cx| ConfigurationView::new(self.state.clone(), window, cx))
            .into()
    }

    fn reset_credentials(&self, cx: &mut App) -> Task<Result<()>> {
        self.state.update(cx, |state, cx| state.reset_api_key(cx))
    }
}

pub struct VercelLanguageModel {
    id: LanguageModelId,
    model: vercel::Model,
    state: gpui::Entity<State>,
    http_client: Arc<dyn HttpClient>,
    request_limiter: RateLimiter,
}

impl VercelLanguageModel {
    fn stream_completion(
        &self,
        request: open_ai::Request,
        cx: &AsyncApp,
    ) -> BoxFuture<'static, Result<futures::stream::BoxStream<'static, Result<ResponseStreamEvent>>>>
    {
        let http_client = self.http_client.clone();
        let Ok((api_key, api_url)) = cx.read_entity(&self.state, |state, cx| {
            let settings = &AllLanguageModelSettings::get_global(cx).vercel;
            let api_url = if settings.api_url.is_empty() {
                vercel::VERCEL_API_URL.to_string()
            } else {
                settings.api_url.clone()
            };
            (state.api_key.clone(), api_url)
        }) else {
            return futures::future::ready(Err(anyhow!("App state dropped"))).boxed();
        };

        let future = self.request_limiter.stream(async move {
            let Some(api_key) = api_key else {
                return Err(LanguageModelCompletionError::NoApiKey {
                    provider: PROVIDER_NAME,
                });
            };
            let request =
                open_ai::stream_completion(http_client.as_ref(), &api_url, &api_key, request);
            let response = request.await?;
            Ok(response)
        });

        async move { Ok(future.await?.boxed()) }.boxed()
    }
}

impl LanguageModel for VercelLanguageModel {
    fn id(&self) -> LanguageModelId {
        self.id.clone()
    }

    fn name(&self) -> LanguageModelName {
        LanguageModelName::from(self.model.display_name().to_string())
    }

    fn provider_id(&self) -> LanguageModelProviderId {
        PROVIDER_ID
    }

    fn provider_name(&self) -> LanguageModelProviderName {
        PROVIDER_NAME
    }

    fn supports_tools(&self) -> bool {
        true
    }

    fn supports_images(&self) -> bool {
        true
    }

    fn supports_tool_choice(&self, choice: LanguageModelToolChoice) -> bool {
        match choice {
            LanguageModelToolChoice::Auto
            | LanguageModelToolChoice::Any
            | LanguageModelToolChoice::None => true,
        }
    }

    fn telemetry_id(&self) -> String {
        format!("vercel/{}", self.model.id())
    }

    fn max_token_count(&self) -> u64 {
        self.model.max_token_count()
    }

    fn max_output_tokens(&self) -> Option<u64> {
        self.model.max_output_tokens()
    }

    fn count_tokens(
        &self,
        request: LanguageModelRequest,
        cx: &App,
    ) -> BoxFuture<'static, Result<u64>> {
        count_vercel_tokens(request, self.model.clone(), cx)
    }

    fn stream_completion(
        &self,
        request: LanguageModelRequest,
        cx: &AsyncApp,
    ) -> BoxFuture<
        'static,
        Result<
            futures::stream::BoxStream<
                'static,
                Result<LanguageModelCompletionEvent, LanguageModelCompletionError>,
            >,
            LanguageModelCompletionError,
        >,
    > {
        let request = crate::provider::open_ai::into_open_ai(
            request,
            self.model.id(),
            self.model.supports_parallel_tool_calls(),
            self.model.supports_prompt_cache_key(),
            self.max_output_tokens(),
            None,
        );
        let completions = self.stream_completion(request, cx);
        async move {
            let mapper = crate::provider::open_ai::OpenAiEventMapper::new();
            Ok(mapper.map_stream(completions.await?).boxed())
        }
        .boxed()
    }
}

pub fn count_vercel_tokens(
    request: LanguageModelRequest,
    model: Model,
    cx: &App,
) -> BoxFuture<'static, Result<u64>> {
    cx.background_spawn(async move {
        let messages = request
            .messages
            .into_iter()
            .map(|message| tiktoken_rs::ChatCompletionRequestMessage {
                role: match message.role {
                    Role::User => "user".into(),
                    Role::Assistant => "assistant".into(),
                    Role::System => "system".into(),
                },
                content: Some(message.string_contents()),
                name: None,
                function_call: None,
            })
            .collect::<Vec<_>>();

        match model {
            Model::Custom { max_tokens, .. } => {
                let model = if max_tokens >= 100_000 {
                    // If the max tokens is 100k or more, it is likely the o200k_base tokenizer from gpt4o
                    "gpt-4o"
                } else {
                    // Otherwise fallback to gpt-4, since only cl100k_base and o200k_base are
                    // supported with this tiktoken method
                    "gpt-4"
                };
                tiktoken_rs::num_tokens_from_messages(model, &messages)
            }
            // Map Vercel models to appropriate OpenAI models for token counting
            // since Vercel uses OpenAI-compatible API
            Model::VZeroOnePointFiveMedium => {
                // Vercel v0 is similar to GPT-4o, so use gpt-4o for token counting
                tiktoken_rs::num_tokens_from_messages("gpt-4o", &messages)
            }
        }
        .map(|tokens| tokens as u64)
    })
    .boxed()
}

struct ConfigurationView {
    api_key_editor: Entity<SingleLineInput>,
    state: gpui::Entity<State>,
    load_credentials_task: Option<Task<()>>,
}

impl ConfigurationView {
    fn new(state: gpui::Entity<State>, window: &mut Window, cx: &mut Context<Self>) -> Self {
        let api_key_editor = cx.new(|cx| {
            SingleLineInput::new(
                window,
                cx,
                "v1:0000000000000000000000000000000000000000000000000",
            )
            .label("API key")
        });

        cx.observe(&state, |_, _, cx| {
            cx.notify();
        })
        .detach();

        let load_credentials_task = Some(cx.spawn_in(window, {
            let state = state.clone();
            async move |this, cx| {
                if let Some(task) = state
                    .update(cx, |state, cx| state.authenticate(cx))
                    .log_err()
                {
                    // We don't log an error, because "not signed in" is also an error.
                    let _ = task.await;
                }
                this.update(cx, |this, cx| {
                    this.load_credentials_task = None;
                    cx.notify();
                })
                .log_err();
            }
        }));

        Self {
            api_key_editor,
            state,
            load_credentials_task,
        }
    }

    fn save_api_key(&mut self, _: &menu::Confirm, window: &mut Window, cx: &mut Context<Self>) {
        let api_key = self
            .api_key_editor
            .read(cx)
            .editor()
            .read(cx)
            .text(cx)
            .trim()
            .to_string();

        // Don't proceed if no API key is provided and we're not authenticated
        if api_key.is_empty() && !self.state.read(cx).is_authenticated() {
            return;
        }

        let state = self.state.clone();
        cx.spawn_in(window, async move |_, cx| {
            state
                .update(cx, |state, cx| state.set_api_key(api_key, cx))?
                .await
        })
        .detach_and_log_err(cx);

        cx.notify();
    }

    fn reset_api_key(&mut self, window: &mut Window, cx: &mut Context<Self>) {
        self.api_key_editor.update(cx, |input, cx| {
            input.editor.update(cx, |editor, cx| {
                editor.set_text("", window, cx);
            });
        });

        let state = self.state.clone();
        cx.spawn_in(window, async move |_, cx| {
            state.update(cx, |state, cx| state.reset_api_key(cx))?.await
        })
        .detach_and_log_err(cx);

        cx.notify();
    }

    fn should_render_editor(&self, cx: &mut Context<Self>) -> bool {
        !self.state.read(cx).is_authenticated()
    }
}

impl Render for ConfigurationView {
    fn render(&mut self, _: &mut Window, cx: &mut Context<Self>) -> impl IntoElement {
        let env_var_set = self.state.read(cx).api_key_from_env;

        let api_key_section = if self.should_render_editor(cx) {
            v_flex()
                .on_action(cx.listener(Self::save_api_key))
                .child(Label::new("To use Zed's agent with Vercel v0, you need to add an API key. Follow these steps:"))
                .child(
                    List::new()
                        .child(InstructionListItem::new(
                            "Create one by visiting",
                            Some("Vercel v0's console"),
                            Some("https://v0.dev/chat/settings/keys"),
                        ))
                        .child(InstructionListItem::text_only(
                            "Paste your API key below and hit enter to start using the agent",
                        )),
                )
                .child(self.api_key_editor.clone())
                .child(
                    Label::new(format!(
                        "You can also assign the {VERCEL_API_KEY_VAR} environment variable and restart Zed."
                    ))
                    .size(LabelSize::Small)
                    .color(Color::Muted),
                )
                .child(
                    Label::new("Note that Vercel v0 is a custom OpenAI-compatible provider.")
                        .size(LabelSize::Small)
                        .color(Color::Muted),
                )
                .into_any()
        } else {
            h_flex()
                .mt_1()
                .p_1()
                .justify_between()
                .rounded_md()
                .border_1()
                .border_color(cx.theme().colors().border)
                .bg(cx.theme().colors().background)
                .child(
                    h_flex()
                        .gap_1()
                        .child(Icon::new(IconName::Check).color(Color::Success))
                        .child(Label::new(if env_var_set {
                            format!("API key set in {VERCEL_API_KEY_VAR} environment variable.")
                        } else {
                            "API key configured.".to_string()
                        })),
                )
                .child(
                    Button::new("reset-api-key", "Reset API Key")
                        .label_size(LabelSize::Small)
                        .icon(IconName::Undo)
                        .icon_size(IconSize::Small)
                        .icon_position(IconPosition::Start)
                        .layer(ElevationIndex::ModalSurface)
                        .when(env_var_set, |this| {
                            this.tooltip(Tooltip::text(format!("To reset your API key, unset the {VERCEL_API_KEY_VAR} environment variable.")))
                        })
                        .on_click(cx.listener(|this, _, window, cx| this.reset_api_key(window, cx))),
                )
                .into_any()
        };

        if self.load_credentials_task.is_some() {
            div().child(Label::new("Loading credentials…")).into_any()
        } else {
            v_flex().size_full().child(api_key_section).into_any()
        }
    }
}