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

use anyhow::{Context as _, Result, anyhow};
use collections::{BTreeMap, HashMap};
use credentials_provider::CredentialsProvider;
use editor::{Editor, EditorElement, EditorStyle};
use futures::Stream;
use futures::{FutureExt, StreamExt, future::BoxFuture, stream::BoxStream};
use gpui::{
    AnyView, AppContext as _, AsyncApp, Entity, FontStyle, Subscription, Task, TextStyle,
    WhiteSpace,
};
use http_client::HttpClient;
use language_model::{
    AuthenticateError, LanguageModel, LanguageModelCompletionError, LanguageModelCompletionEvent,
    LanguageModelId, LanguageModelName, LanguageModelProvider, LanguageModelProviderId,
    LanguageModelProviderName, LanguageModelProviderState, LanguageModelRequest,
    LanguageModelToolChoice, LanguageModelToolResultContent, LanguageModelToolUse, MessageContent,
    RateLimiter, Role, StopReason, TokenUsage,
};
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};
use settings::{Settings, SettingsStore};
use std::pin::Pin;
use std::str::FromStr;
use std::sync::Arc;
use theme::ThemeSettings;
use ui::{Icon, IconName, List, prelude::*};
use util::ResultExt;

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

const PROVIDER_ID: LanguageModelProviderId = LanguageModelProviderId::new("deepseek");
const PROVIDER_NAME: LanguageModelProviderName = LanguageModelProviderName::new("DeepSeek");
const DEEPSEEK_API_KEY_VAR: &str = "DEEPSEEK_API_KEY";

#[derive(Default)]
struct RawToolCall {
    id: String,
    name: String,
    arguments: String,
}

#[derive(Default, Clone, Debug, PartialEq)]
pub struct DeepSeekSettings {
    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 struct DeepSeekLanguageModelProvider {
    http_client: Arc<dyn HttpClient>,
    state: Entity<State>,
}

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

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 api_url = AllLanguageModelSettings::get_global(cx)
            .deepseek
            .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 api_url = AllLanguageModelSettings::get_global(cx)
            .deepseek
            .api_url
            .clone();
        cx.spawn(async move |this, cx| {
            credentials_provider
                .write_credentials(&api_url, "Bearer", api_key.as_bytes(), &cx)
                .await?;
            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 api_url = AllLanguageModelSettings::get_global(cx)
            .deepseek
            .api_url
            .clone();
        cx.spawn(async move |this, cx| {
            let (api_key, from_env) = if let Ok(api_key) = std::env::var(DEEPSEEK_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 DeepSeekLanguageModelProvider {
    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: deepseek::Model) -> Arc<dyn LanguageModel> {
        Arc::new(DeepSeekLanguageModel {
            id: LanguageModelId::from(model.id().to_string()),
            model,
            state: self.state.clone(),
            http_client: self.http_client.clone(),
            request_limiter: RateLimiter::new(4),
        }) as Arc<dyn LanguageModel>
    }
}

impl LanguageModelProviderState for DeepSeekLanguageModelProvider {
    type ObservableEntity = State;

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

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

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

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

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

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

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

        models.insert("deepseek-chat", deepseek::Model::Chat);
        models.insert("deepseek-reasoner", deepseek::Model::Reasoner);

        for available_model in AllLanguageModelSettings::get_global(cx)
            .deepseek
            .available_models
            .iter()
        {
            models.insert(
                &available_model.name,
                deepseek::Model::Custom {
                    name: available_model.name.clone(),
                    display_name: available_model.display_name.clone(),
                    max_tokens: available_model.max_tokens,
                    max_output_tokens: available_model.max_output_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 DeepSeekLanguageModel {
    id: LanguageModelId,
    model: deepseek::Model,
    state: Entity<State>,
    http_client: Arc<dyn HttpClient>,
    request_limiter: RateLimiter,
}

impl DeepSeekLanguageModel {
    fn stream_completion(
        &self,
        request: deepseek::Request,
        cx: &AsyncApp,
    ) -> BoxFuture<'static, Result<BoxStream<'static, Result<deepseek::StreamResponse>>>> {
        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).deepseek;
            (state.api_key.clone(), settings.api_url.clone())
        }) else {
            return futures::future::ready(Err(anyhow!("App state dropped"))).boxed();
        };

        let future = self.request_limiter.stream(async move {
            let api_key = api_key.context("Missing DeepSeek API Key")?;
            let request =
                deepseek::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 DeepSeekLanguageModel {
    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_tool_choice(&self, _choice: LanguageModelToolChoice) -> bool {
        true
    }

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

    fn telemetry_id(&self) -> String {
        format!("deepseek/{}", 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>> {
        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<_>>();

            tiktoken_rs::num_tokens_from_messages("gpt-4", &messages).map(|tokens| tokens as u64)
        })
        .boxed()
    }

    fn stream_completion(
        &self,
        request: LanguageModelRequest,
        cx: &AsyncApp,
    ) -> BoxFuture<
        'static,
        Result<
            BoxStream<'static, Result<LanguageModelCompletionEvent, LanguageModelCompletionError>>,
            LanguageModelCompletionError,
        >,
    > {
        let request = into_deepseek(request, &self.model, self.max_output_tokens());
        let stream = self.stream_completion(request, cx);

        async move {
            let mapper = DeepSeekEventMapper::new();
            Ok(mapper.map_stream(stream.await?).boxed())
        }
        .boxed()
    }
}

pub fn into_deepseek(
    request: LanguageModelRequest,
    model: &deepseek::Model,
    max_output_tokens: Option<u64>,
) -> deepseek::Request {
    let is_reasoner = *model == deepseek::Model::Reasoner;

    let mut messages = Vec::new();
    for message in request.messages {
        for content in message.content {
            match content {
                MessageContent::Text(text) => messages.push(match message.role {
                    Role::User => deepseek::RequestMessage::User { content: text },
                    Role::Assistant => deepseek::RequestMessage::Assistant {
                        content: Some(text),
                        tool_calls: Vec::new(),
                    },
                    Role::System => deepseek::RequestMessage::System { content: text },
                }),
                MessageContent::Thinking { .. } => {}
                MessageContent::RedactedThinking(_) => {}
                MessageContent::Image(_) => {}
                MessageContent::ToolUse(tool_use) => {
                    let tool_call = deepseek::ToolCall {
                        id: tool_use.id.to_string(),
                        content: deepseek::ToolCallContent::Function {
                            function: deepseek::FunctionContent {
                                name: tool_use.name.to_string(),
                                arguments: serde_json::to_string(&tool_use.input)
                                    .unwrap_or_default(),
                            },
                        },
                    };

                    if let Some(deepseek::RequestMessage::Assistant { tool_calls, .. }) =
                        messages.last_mut()
                    {
                        tool_calls.push(tool_call);
                    } else {
                        messages.push(deepseek::RequestMessage::Assistant {
                            content: None,
                            tool_calls: vec![tool_call],
                        });
                    }
                }
                MessageContent::ToolResult(tool_result) => {
                    match &tool_result.content {
                        LanguageModelToolResultContent::Text(text) => {
                            messages.push(deepseek::RequestMessage::Tool {
                                content: text.to_string(),
                                tool_call_id: tool_result.tool_use_id.to_string(),
                            });
                        }
                        LanguageModelToolResultContent::Image(_) => {}
                    };
                }
            }
        }
    }

    deepseek::Request {
        model: model.id().to_string(),
        messages,
        stream: true,
        max_tokens: max_output_tokens,
        temperature: if is_reasoner {
            None
        } else {
            request.temperature
        },
        response_format: None,
        tools: request
            .tools
            .into_iter()
            .map(|tool| deepseek::ToolDefinition::Function {
                function: deepseek::FunctionDefinition {
                    name: tool.name,
                    description: Some(tool.description),
                    parameters: Some(tool.input_schema),
                },
            })
            .collect(),
    }
}

pub struct DeepSeekEventMapper {
    tool_calls_by_index: HashMap<usize, RawToolCall>,
}

impl DeepSeekEventMapper {
    pub fn new() -> Self {
        Self {
            tool_calls_by_index: HashMap::default(),
        }
    }

    pub fn map_stream(
        mut self,
        events: Pin<Box<dyn Send + Stream<Item = Result<deepseek::StreamResponse>>>>,
    ) -> impl Stream<Item = Result<LanguageModelCompletionEvent, LanguageModelCompletionError>>
    {
        events.flat_map(move |event| {
            futures::stream::iter(match event {
                Ok(event) => self.map_event(event),
                Err(error) => vec![Err(LanguageModelCompletionError::from(error))],
            })
        })
    }

    pub fn map_event(
        &mut self,
        event: deepseek::StreamResponse,
    ) -> Vec<Result<LanguageModelCompletionEvent, LanguageModelCompletionError>> {
        let Some(choice) = event.choices.first() else {
            return vec![Err(LanguageModelCompletionError::from(anyhow!(
                "Response contained no choices"
            )))];
        };

        let mut events = Vec::new();
        if let Some(content) = choice.delta.content.clone() {
            events.push(Ok(LanguageModelCompletionEvent::Text(content)));
        }

        if let Some(reasoning_content) = choice.delta.reasoning_content.clone() {
            events.push(Ok(LanguageModelCompletionEvent::Thinking {
                text: reasoning_content,
                signature: None,
            }));
        }

        if let Some(tool_calls) = choice.delta.tool_calls.as_ref() {
            for tool_call in tool_calls {
                let entry = self.tool_calls_by_index.entry(tool_call.index).or_default();

                if let Some(tool_id) = tool_call.id.clone() {
                    entry.id = tool_id;
                }

                if let Some(function) = tool_call.function.as_ref() {
                    if let Some(name) = function.name.clone() {
                        entry.name = name;
                    }

                    if let Some(arguments) = function.arguments.clone() {
                        entry.arguments.push_str(&arguments);
                    }
                }
            }
        }

        if let Some(usage) = event.usage {
            events.push(Ok(LanguageModelCompletionEvent::UsageUpdate(TokenUsage {
                input_tokens: usage.prompt_tokens,
                output_tokens: usage.completion_tokens,
                cache_creation_input_tokens: 0,
                cache_read_input_tokens: 0,
            })));
        }

        match choice.finish_reason.as_deref() {
            Some("stop") => {
                events.push(Ok(LanguageModelCompletionEvent::Stop(StopReason::EndTurn)));
            }
            Some("tool_calls") => {
                events.extend(self.tool_calls_by_index.drain().map(|(_, tool_call)| {
                    match serde_json::Value::from_str(&tool_call.arguments) {
                        Ok(input) => Ok(LanguageModelCompletionEvent::ToolUse(
                            LanguageModelToolUse {
                                id: tool_call.id.clone().into(),
                                name: tool_call.name.as_str().into(),
                                is_input_complete: true,
                                input,
                                raw_input: tool_call.arguments.clone(),
                            },
                        )),
                        Err(error) => Ok(LanguageModelCompletionEvent::ToolUseJsonParseError {
                            id: tool_call.id.clone().into(),
                            tool_name: tool_call.name.as_str().into(),
                            raw_input: tool_call.arguments.into(),
                            json_parse_error: error.to_string(),
                        }),
                    }
                }));

                events.push(Ok(LanguageModelCompletionEvent::Stop(StopReason::ToolUse)));
            }
            Some(stop_reason) => {
                log::error!("Unexpected DeepSeek stop_reason: {stop_reason:?}",);
                events.push(Ok(LanguageModelCompletionEvent::Stop(StopReason::EndTurn)));
            }
            None => {}
        }

        events
    }
}

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

impl ConfigurationView {
    fn new(state: Entity<State>, window: &mut Window, cx: &mut Context<Self>) -> Self {
        let api_key_editor = cx.new(|cx| {
            let mut editor = Editor::single_line(window, cx);
            editor.set_placeholder_text("sk-00000000000000000000000000000000", cx);
            editor
        });

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

        let load_credentials_task = Some(cx.spawn({
            let state = state.clone();
            async move |this, cx| {
                if let Some(task) = state
                    .update(cx, |state, cx| state.authenticate(cx))
                    .log_err()
                {
                    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).text(cx);
        if api_key.is_empty() {
            return;
        }

        let state = self.state.clone();
        cx.spawn(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, |editor, cx| editor.set_text("", window, cx));

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

        cx.notify();
    }

    fn render_api_key_editor(&self, cx: &mut Context<Self>) -> impl IntoElement {
        let settings = ThemeSettings::get_global(cx);
        let text_style = TextStyle {
            color: cx.theme().colors().text,
            font_family: settings.ui_font.family.clone(),
            font_features: settings.ui_font.features.clone(),
            font_fallbacks: settings.ui_font.fallbacks.clone(),
            font_size: rems(0.875).into(),
            font_weight: settings.ui_font.weight,
            font_style: FontStyle::Normal,
            line_height: relative(1.3),
            background_color: None,
            underline: None,
            strikethrough: None,
            white_space: WhiteSpace::Normal,
            ..Default::default()
        };
        EditorElement::new(
            &self.api_key_editor,
            EditorStyle {
                background: cx.theme().colors().editor_background,
                local_player: cx.theme().players().local(),
                text: text_style,
                ..Default::default()
            },
        )
    }

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

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

        if self.load_credentials_task.is_some() {
            div().child(Label::new("Loading credentials...")).into_any()
        } else if self.should_render_editor(cx) {
            v_flex()
                .size_full()
                .on_action(cx.listener(Self::save_api_key))
                .child(Label::new("To use DeepSeek in Zed, you need an API key:"))
                .child(
                    List::new()
                        .child(InstructionListItem::new(
                            "Get your API key from the",
                            Some("DeepSeek console"),
                            Some("https://platform.deepseek.com/api_keys"),
                        ))
                        .child(InstructionListItem::text_only(
                            "Paste your API key below and hit enter to start using the assistant",
                        )),
                )
                .child(
                    h_flex()
                        .w_full()
                        .my_2()
                        .px_2()
                        .py_1()
                        .bg(cx.theme().colors().editor_background)
                        .border_1()
                        .border_color(cx.theme().colors().border)
                        .rounded_sm()
                        .child(self.render_api_key_editor(cx)),
                )
                .child(
                    Label::new(format!(
                        "Or set the {} environment variable.",
                        DEEPSEEK_API_KEY_VAR
                    ))
                    .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 {}", DEEPSEEK_API_KEY_VAR)
                        } else {
                            "API key configured".to_string()
                        })),
                )
                .child(
                    Button::new("reset-key", "Reset Key")
                        .label_size(LabelSize::Small)
                        .icon(Some(IconName::Trash))
                        .icon_size(IconSize::Small)
                        .icon_position(IconPosition::Start)
                        .disabled(env_var_set)
                        .on_click(
                            cx.listener(|this, _, window, cx| this.reset_api_key(window, cx)),
                        ),
                )
                .into_any()
        }
    }
}