From 243222067727115019d6e46f26c0e07858e1e76d Mon Sep 17 00:00:00 2001 From: lzt404 <2596933790@qq.com> Date: Fri, 25 Sep 2026 14:13:21 +0800 Subject: [PATCH] feat(usage): update model pricing from external catalogs --- src-tauri/Cargo.lock | 34 ++ src-tauri/Cargo.toml | 2 +- src-tauri/resources/model_prices.json | 8 +- src-tauri/src/main.rs | 3 +- src-tauri/src/usage.rs | 805 ++++++++++++++++++++++---- src/i18n/ja.ts | 19 +- src/i18n/locales/en.ts | 19 +- src/i18n/locales/zh-CN.ts | 19 +- src/pages/UsageRecordsPage.tsx | 187 ++++-- src/styles.css | 54 +- tests/fixtures/operation-feedback.tsx | 10 +- tests/operation-feedback-ui.cjs | 8 +- 12 files changed, 979 insertions(+), 189 deletions(-) diff --git a/src-tauri/Cargo.lock b/src-tauri/Cargo.lock index 5485522..89638c4 100644 --- a/src-tauri/Cargo.lock +++ b/src-tauri/Cargo.lock @@ -92,6 +92,18 @@ dependencies = [ "pin-project-lite", ] +[[package]] +name = "async-compression" +version = "0.4.48" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "fb61aea1a7def73ee7c350a184f0e70b32c182344e2e75bf70c9b621b83417fd" +dependencies = [ + "compression-codecs", + "compression-core", + "pin-project-lite", + "tokio", +] + [[package]] name = "async-executor" version = "1.14.0" @@ -526,6 +538,23 @@ dependencies = [ "memchr", ] +[[package]] +name = "compression-codecs" +version = "0.4.43" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "bef16c47ba2797aa6a909cc37d39911f3a6743811fe7408ac0b0cc0276b656e9" +dependencies = [ + "compression-core", + "flate2", + "memchr", +] + +[[package]] +name = "compression-core" +version = "0.4.33" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "6e8ccc4ea9f6acc32d102c0f6d471d11d913ad15f20c04de743374861fa1d414" + [[package]] name = "concurrent-queue" version = "2.5.0" @@ -4567,12 +4596,17 @@ version = "0.6.11" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "4cfcf7e2740e6fc6d4d688b4ef00650406bb94adf4731e43c096c3a19fe40840" dependencies = [ + "async-compression", "bitflags 2.13.0", "bytes", + "futures-core", "futures-util", "http", "http-body", + "http-body-util", "pin-project-lite", + "tokio", + "tokio-util", "tower", "tower-layer", "tower-service", diff --git a/src-tauri/Cargo.toml b/src-tauri/Cargo.toml index fdbcecd..236ba6b 100644 --- a/src-tauri/Cargo.toml +++ b/src-tauri/Cargo.toml @@ -18,7 +18,7 @@ getrandom = "0.3" json5 = "0.4" notify = "7" regex = "1" -reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls", "socks", "stream"] } +reqwest = { version = "0.12", default-features = false, features = ["gzip", "json", "rustls-tls", "socks", "stream"] } rusqlite = { version = "0.32.1", features = ["bundled"] } semver = "1" serde = { version = "1", features = ["derive"] } diff --git a/src-tauri/resources/model_prices.json b/src-tauri/resources/model_prices.json index 096dc5b..884de06 100644 --- a/src-tauri/resources/model_prices.json +++ b/src-tauri/resources/model_prices.json @@ -1,10 +1,13 @@ { "schemaVersion": 1, - "updatedAt": "2026-09-07", + "updatedAt": "2026-09-24", "models": { "gpt-6-astra": { "inputPer1M": 10.0, "outputPer1M": 50.0, "cacheReadPer1M": 1.0, "cacheCreationPer1M": 12.5 }, + "gpt-6-sol": { "inputPer1M": 2.0, "outputPer1M": 10.0, "cacheReadPer1M": 0.2, "cacheCreationPer1M": 2.5 }, + "gpt-6-luna": { "inputPer1M": 0.1, "outputPer1M": 0.5, "cacheReadPer1M": 0.01, "cacheCreationPer1M": 0.125 }, "gpt-5.6": { "inputPer1M": 4.0, "outputPer1M": 20.0, "cacheReadPer1M": 0.4, "cacheCreationPer1M": 5.0 }, "gpt-5.6-sol": { "inputPer1M": 4.0, "outputPer1M": 20.0, "cacheReadPer1M": 0.4, "cacheCreationPer1M": 5.0 }, + "gpt-5.6-cyber": { "inputPer1M": 12.5, "outputPer1M": 75.0, "cacheReadPer1M": 1.25, "cacheCreationPer1M": 15.625 }, "gpt-5.6-terra": { "inputPer1M": 2.0, "outputPer1M": 12.0, "cacheReadPer1M": 0.2, "cacheCreationPer1M": 2.5 }, "gpt-5.6-luna": { "inputPer1M": 0.2, "outputPer1M": 1.2, "cacheReadPer1M": 0.02, "cacheCreationPer1M": 0.25 }, "gpt-5.5": { "inputPer1M": 5.0, "outputPer1M": 30.0, "cacheReadPer1M": 0.5 }, @@ -41,6 +44,7 @@ "claude-mythos-5-1": { "inputPer1M": 10.0, "outputPer1M": 50.0, "cacheReadPer1M": 0.25, "cacheCreationPer1M": 12.5 }, "claude-fable-5": { "inputPer1M": 10.0, "outputPer1M": 50.0, "cacheReadPer1M": 1.0, "cacheCreationPer1M": 12.5 }, "claude-mythos-5": { "inputPer1M": 10.0, "outputPer1M": 50.0, "cacheReadPer1M": 1.0, "cacheCreationPer1M": 12.5 }, + "claude-opus-5-5": { "inputPer1M": 4.0, "outputPer1M": 20.0, "cacheReadPer1M": 0.2, "cacheCreationPer1M": 5.0 }, "claude-opus-5": { "inputPer1M": 5.0, "outputPer1M": 25.0, "cacheReadPer1M": 0.5, "cacheCreationPer1M": 6.25 }, "claude-opus-4-8": { "inputPer1M": 5.0, "outputPer1M": 25.0, "cacheReadPer1M": 0.5, "cacheCreationPer1M": 6.25 }, "claude-opus-4-7": { "inputPer1M": 5.0, "outputPer1M": 25.0, "cacheReadPer1M": 0.5, "cacheCreationPer1M": 6.25 }, @@ -67,8 +71,10 @@ "deepseek-v4-pro": { "inputPer1M": 1.32, "outputPer1M": 3.96, "cacheReadPer1M": 0.044, "cacheCreationPer1M": 0.0 }, "deepseek-v4-flash": { "inputPer1M": 0.44, "outputPer1M": 1.32, "cacheReadPer1M": 0.014, "cacheCreationPer1M": 0.0 }, "deepseek-v4-flash-vision-exp": { "inputPer1M": 0.44, "outputPer1M": 1.32, "cacheReadPer1M": 0.014, "cacheCreationPer1M": 0.0 }, + "deepseek-flash": { "inputPer1M": 0.3, "outputPer1M": 1.2, "cacheReadPer1M": 0.006, "cacheCreationPer1M": 0.0 }, "deepseek-chat": { "inputPer1M": 0.28, "outputPer1M": 0.42, "cacheReadPer1M": 0.028 }, "deepseek-reasoner": { "inputPer1M": 0.28, "outputPer1M": 0.42, "cacheReadPer1M": 0.028 }, + "grok-4.7": { "inputPer1M": 2.0, "outputPer1M": 6.0, "cacheReadPer1M": 0.5 }, "grok-4.6": { "inputPer1M": 2.0, "outputPer1M": 6.0, "cacheReadPer1M": 0.5 }, "grok-build-0.1": { "inputPer1M": 1.0, "outputPer1M": 2.0, "cacheReadPer1M": 0.2 }, "grok-code-fast-1": { "inputPer1M": 1.0, "outputPer1M": 2.0, "cacheReadPer1M": 0.2 }, diff --git a/src-tauri/src/main.rs b/src-tauri/src/main.rs index 91d5bfb..264e2d1 100644 --- a/src-tauri/src/main.rs +++ b/src-tauri/src/main.rs @@ -2724,7 +2724,8 @@ fn main() { usage::shrink_usage_database, usage::save_usage_model_price, usage::delete_usage_model_price, - usage::sync_usage_model_prices, + usage::preview_usage_model_prices, + usage::apply_usage_model_prices, start_core_process, stop_core_process, restart_core_process, diff --git a/src-tauri/src/usage.rs b/src-tauri/src/usage.rs index 2a9f8cd..770bab7 100644 --- a/src-tauri/src/usage.rs +++ b/src-tauri/src/usage.rs @@ -60,9 +60,21 @@ const SQLITE_BUSY_TIMEOUT_SECONDS: u64 = 5; const BYTES_PER_MB: u64 = 1024 * 1024; const TOKENS_PER_PRICE_UNIT: f64 = 1_000_000.0; const LONG_CONTEXT_INPUT_TOKEN_THRESHOLD: u64 = 272_000; +const GROK_47_LONG_CONTEXT_INPUT_TOKEN_THRESHOLD: u64 = 200_000; +const GROK_47_MODEL_SQL: &str = concat!( + "(LOWER(TRIM(model)) = 'grok-4.7'", + " OR LOWER(TRIM(model)) LIKE 'grok-4.7-%'", + " OR LOWER(TRIM(model)) LIKE '%/grok-4.7'", + " OR LOWER(TRIM(model)) LIKE '%/grok-4.7-%')" +); const BUNDLED_MODEL_PRICE_CATALOG: &str = include_str!("../resources/model_prices.json"); -const MODEL_PRICE_SYNC_URL: &str = - "https://raw.githubusercontent.com/router-for-me/EasyCLIProxyAPI/main/src-tauri/resources/model_prices.json"; +const MODEL_PRICE_SYNC_URL: &str = "https://models.dev/api.json"; +const LITELLM_PRICE_SYNC_URL: &str = + "https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json"; +const MODEL_PRICE_SYNC_MAX_BYTES: usize = 16 * 1024 * 1024; +const MODEL_PRICE_SYNC_ATTEMPTS: usize = 2; +const USAGE_DISPLAY_MODEL_SQL: &str = + "COALESCE(NULLIF(TRIM(alias), ''), NULLIF(TRIM(model), ''), 'unknown')"; static USAGE_QUERY_SLOTS: tokio::sync::Semaphore = tokio::sync::Semaphore::const_new(2); @@ -337,7 +349,6 @@ pub(crate) struct ModelPrice { model: String, prompt: f64, completion: f64, - cache: f64, cache_read: f64, cache_creation: f64, prompt_configured: bool, @@ -405,13 +416,19 @@ struct UsagePriceRow { price: Option, } -#[derive(Default, Serialize)] +#[derive(Serialize)] #[serde(rename_all = "camelCase")] pub(crate) struct ModelPriceSyncResult { imported: usize, - skipped: usize, +} + +#[derive(Serialize)] +#[serde(rename_all = "camelCase")] +pub(crate) struct ModelPriceSyncPreview { + source: String, + source_url: String, + matches: Vec, unmatched: Vec, - used_builtin: bool, } struct UsageCostGroup { @@ -2806,7 +2823,12 @@ fn build_usage_filter(query: &UsageQuery) -> UsageSqlFilter { clauses.push("timestamp_ms <= ?".to_string()); params.push(SqlValue::Integer(end)); } - add_text_filter(&mut clauses, &mut params, "model", query.model.as_deref()); + add_text_filter( + &mut clauses, + &mut params, + USAGE_DISPLAY_MODEL_SQL, + query.model.as_deref(), + ); add_text_filter( &mut clauses, &mut params, @@ -2961,7 +2983,7 @@ fn load_usage_overview( END ELSE local_hour || '-00' END, - COALESCE(NULLIF(TRIM(model), ''), 'unknown'), + {USAGE_DISPLAY_MODEL_SQL}, COUNT(*), COALESCE(SUM(CASE WHEN failed = 0 THEN 1 ELSE 0 END), 0), COALESCE(SUM(CASE WHEN failed != 0 AND canceled = 0 THEN 1 ELSE 0 END), 0), @@ -3079,6 +3101,9 @@ fn load_usage_cost_groups( connection: &Connection, filter: &UsageSqlFilter, ) -> Result, String> { + let long_context_sql = format!( + "input_tokens > {LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} OR (input_tokens > {GROK_47_LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} AND {GROK_47_MODEL_SQL})" + ); let sql = format!( r#" SELECT @@ -3094,10 +3119,10 @@ fn load_usage_cost_groups( COALESCE(SUM(output_tokens), 0), COALESCE(SUM(cache_read_tokens), 0), COALESCE(SUM(cache_creation_tokens), 0), - COALESCE(SUM(CASE WHEN input_tokens > {LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} THEN input_tokens ELSE 0 END), 0), - COALESCE(SUM(CASE WHEN input_tokens > {LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} THEN output_tokens ELSE 0 END), 0), - COALESCE(SUM(CASE WHEN input_tokens > {LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} THEN cache_read_tokens ELSE 0 END), 0), - COALESCE(SUM(CASE WHEN input_tokens > {LONG_CONTEXT_INPUT_TOKEN_THRESHOLD} THEN cache_creation_tokens ELSE 0 END), 0), + COALESCE(SUM(CASE WHEN ({long_context_sql}) THEN input_tokens ELSE 0 END), 0), + COALESCE(SUM(CASE WHEN ({long_context_sql}) THEN output_tokens ELSE 0 END), 0), + COALESCE(SUM(CASE WHEN ({long_context_sql}) THEN cache_read_tokens ELSE 0 END), 0), + COALESCE(SUM(CASE WHEN ({long_context_sql}) THEN cache_creation_tokens ELSE 0 END), 0), COALESCE(SUM(total_tokens), 0) FROM usage_events{} GROUP BY model, alias, service_tier, response_service_tier, executor_type, provider, auth_type @@ -3163,6 +3188,11 @@ fn sum_usage_cost(groups: &[UsageCostGroup], prices: &HashMap f64 { let price = enriched_model_price(model, price); + let long_output_multiplier = if is_model_family(model, "grok-4.7") { + 2.0 + } else { + 1.5 + }; let short_cost = cost_for_token_segment( tokens.input.saturating_sub(tokens.long_input), tokens.output.saturating_sub(tokens.long_output), @@ -3181,10 +3211,13 @@ fn cost_for_price(model: &str, service_tier: &str, tokens: &CostTokens, price: & tokens.long_cache_creation, &price, 2.0, - 1.5, + long_output_multiplier, ); let tier = service_tier.trim().to_ascii_lowercase(); - let multiplier = if tokens.long_input > 0 && matches!(tier.as_str(), "priority" | "fast") { + let multiplier = if tokens.long_input > 0 + && matches!(tier.as_str(), "priority" | "fast") + && !is_model_family(model, "gpt-6") + { 1.0 } else { match tier.as_str() { @@ -3253,7 +3286,6 @@ fn parse_model_price_catalog( model: model.trim().to_string(), prompt: entry.input_per_1_m, completion: entry.output_per_1_m, - cache: cache_read, cache_read, cache_creation, prompt_configured: true, @@ -3270,6 +3302,113 @@ fn parse_model_price_catalog( Ok(prices) } +fn parse_models_dev_prices( + content: &str, + updated_at_ms: i64, +) -> Result, String> { + let catalog: Value = serde_json::from_str(content) + .map_err(|error| format!("解析 models.dev 价格失败: {error}"))?; + let providers = catalog + .get("providers") + .and_then(Value::as_object) + .or_else(|| catalog.as_object()) + .ok_or("models.dev 价格目录格式无效")?; + let mut prices = Vec::new(); + for (provider, entry) in providers { + let Some(models) = entry + .get("models") + .and_then(Value::as_object) + else { + continue; + }; + for (model_id, entry) in models { + let Some(cost) = entry.get("cost") else { + continue; + }; + let (Some(prompt), Some(completion)) = ( + cost.get("input").and_then(Value::as_f64), + cost.get("output").and_then(Value::as_f64), + ) else { + continue; + }; + let cache_read = cost.get("cache_read").and_then(Value::as_f64); + let cache_creation = cost.get("cache_write").and_then(Value::as_f64); + let price = ModelPrice { + model: model_id.trim().to_string(), + prompt, + completion, + cache_read: cache_read.unwrap_or(0.0), + cache_creation: cache_creation.unwrap_or(0.0), + prompt_configured: true, + completion_configured: true, + cache_read_configured: cache_read.is_some(), + cache_creation_configured: cache_creation.is_some(), + source: "models.dev".to_string(), + source_model_id: format!("{provider}/{model_id}"), + updated_at_ms, + }; + validate_model_price(&price)?; + prices.push(price); + } + } + if prices.is_empty() { + return Err("models.dev 没有可用的模型价格".to_string()); + } + Ok(prices) +} + +fn parse_litellm_prices(content: &str, updated_at_ms: i64) -> Result, String> { + let catalog: Value = serde_json::from_str(content) + .map_err(|error| format!("解析 LiteLLM 价格失败: {error}"))?; + let models = catalog.as_object().ok_or("LiteLLM 价格目录格式无效")?; + let mut prices = Vec::new(); + for (model_id, entry) in models { + let mode = entry.get("mode").and_then(Value::as_str).unwrap_or_default(); + if !matches!(mode, "chat" | "completion" | "responses") { + continue; + } + if let Some(modalities) = entry.get("supported_output_modalities").and_then(Value::as_array) { + if !modalities.iter().any(|value| value.as_str() == Some("text")) { + continue; + } + } + let (Some(prompt), Some(completion)) = ( + entry.get("input_cost_per_token").and_then(Value::as_f64), + entry.get("output_cost_per_token").and_then(Value::as_f64), + ) else { + continue; + }; + let cache_read = entry + .get("cache_read_input_token_cost") + .and_then(Value::as_f64) + .or_else(|| entry.get("input_cost_per_token_cache_hit").and_then(Value::as_f64)); + let cache_creation = entry + .get("cache_creation_input_token_cost") + .and_then(Value::as_f64); + let price = ModelPrice { + model: model_id.trim().to_string(), + prompt: prompt * TOKENS_PER_PRICE_UNIT, + completion: completion * TOKENS_PER_PRICE_UNIT, + cache_read: cache_read.unwrap_or(0.0) * TOKENS_PER_PRICE_UNIT, + cache_creation: cache_creation.unwrap_or(0.0) * TOKENS_PER_PRICE_UNIT, + prompt_configured: true, + completion_configured: true, + cache_read_configured: cache_read.is_some(), + cache_creation_configured: cache_creation.is_some(), + source: "litellm-sync".to_string(), + source_model_id: model_id.clone(), + updated_at_ms, + }; + if validate_model_price(&price).is_ok() { + prices.push(price); + } + } + if prices.is_empty() { + return Err("LiteLLM 没有可用的模型价格".to_string()); + } + Ok(prices) +} + fn find_model_price<'a>( prices: &'a HashMap, model: &str, @@ -3331,21 +3470,6 @@ fn enriched_model_price(model: &str, price: &ModelPrice) -> ModelPrice { price.completion = official.completion; } } - if !price.cache_read_configured && price.cache_read <= 0.0 { - price.cache_read = if price.cache > 0.0 { - price.cache - } else { - price.prompt * 0.1 - }; - } - if !price.cache_creation_configured && price.cache_creation <= 0.0 { - price.cache_creation = price.prompt - * if is_model_family(model, "gpt-5.6") { - 1.25 - } else { - 1.0 - }; - } price } @@ -3363,7 +3487,8 @@ fn is_model_family(model: &str, family: &str) -> bool { fn service_tier_multiplier(model: &str) -> f64 { if is_model_family(model, "gpt-5.5") { 2.5 - } else if is_model_family(model, "gpt-5.6") + } else if is_model_family(model, "gpt-6") + || is_model_family(model, "gpt-5.6") || is_model_family(model, "gpt-5.4") || is_model_family(model, "gpt-5.4-mini") || is_model_family(model, "gpt-5.3-codex") @@ -3379,7 +3504,7 @@ fn load_model_prices(connection: &Connection) -> Result Result Result<(), String> { for value in [ price.prompt, price.completion, - price.cache, price.cache_read, price.cache_creation, ] { @@ -3448,16 +3571,16 @@ fn upsert_model_price(connection: &Connection, price: &ModelPrice) -> Result<(), .execute( r#" INSERT INTO model_prices ( - model, prompt_per_1m, completion_per_1m, cache_per_1m, + model, prompt_per_1m, completion_per_1m, cache_read_per_1m, cache_creation_per_1m, prompt_configured, completion_configured, cache_read_configured, cache_creation_configured, source, source_model_id, updated_at_ms - ) VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13) + ) VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12) ON CONFLICT(model) DO UPDATE SET prompt_per_1m = excluded.prompt_per_1m, completion_per_1m = excluded.completion_per_1m, - cache_per_1m = excluded.cache_per_1m, + cache_per_1m = 0, cache_read_per_1m = excluded.cache_read_per_1m, cache_creation_per_1m = excluded.cache_creation_per_1m, prompt_configured = excluded.prompt_configured, @@ -3472,7 +3595,6 @@ fn upsert_model_price(connection: &Connection, price: &ModelPrice) -> Result<(), price.model.trim(), price.prompt, price.completion, - price.cache, price.cache_read, price.cache_creation, price.prompt_configured, @@ -3534,7 +3656,7 @@ fn load_usage_pricing(connection: &Connection, query: &UsageQuery) -> Result Result Result<(), String .await } +fn sync_price_for_model<'a>(model: &str, prices: &'a [ModelPrice]) -> Option<&'a ModelPrice> { + let model = model.trim(); + if model.is_empty() { + return None; + } + let exact = prices + .iter() + .filter(|price| price.source_model_id.eq_ignore_ascii_case(model)) + .collect::>(); + if exact.len() == 1 { + return Some(exact[0]); + } + let tail = model.rsplit('/').next().unwrap_or(model); + let family = tail.to_ascii_lowercase(); + let mut candidates = prices + .iter() + .filter(|price| price.model.eq_ignore_ascii_case(tail)) + .collect::>(); + if candidates.is_empty() { + let normalized = canonical_model_tail(tail); + candidates = prices + .iter() + .filter(|price| canonical_model_tail(&price.model) == normalized) + .collect(); + } + let provider = if family.starts_with("gpt-") + || family.starts_with("chatgpt-") + || family.starts_with("o1-") + || family.starts_with("o3-") + || family.starts_with("o4-") + { + Some("openai") + } else if family.starts_with("claude-") { + Some("anthropic") + } else if family.starts_with("gemini-") { + Some("google") + } else if family.starts_with("deepseek-") { + Some("deepseek") + } else if family.starts_with("grok-") { + Some("xai") + } else { + None + }; + if let Some(provider) = provider { + let official = candidates + .iter() + .filter(|price| price.source_model_id.starts_with(&format!("{provider}/"))) + .collect::>(); + if official.len() == 1 { + return Some(official[0]); + } + } + (candidates.len() == 1).then(|| candidates[0]) +} + +fn build_model_price_preview( + models: &[(String, String)], + remote_prices: &[ModelPrice], + source: &str, + source_url: &str, +) -> ModelPriceSyncPreview { + let mut matches = Vec::new(); + let mut unmatched = Vec::new(); + for (model, alias) in models { + let matched = [model.as_str(), alias.as_str()] + .into_iter() + .find_map(|candidate| sync_price_for_model(candidate, remote_prices)); + if let Some(price) = matched { + let mut price = price.clone(); + price.model = model.clone(); + matches.push(price); + } else { + unmatched.push(model.clone()); + } + } + ModelPriceSyncPreview { + source: source.to_string(), + source_url: source_url.to_string(), + matches, + unmatched, + } +} + +fn collect_price_preview_models( + connection: &Connection, + query: &UsageQuery, +) -> Result, String> { + let mut models = std::collections::BTreeMap::::new(); + for price in load_model_prices(connection)?.values() { + models.insert(price.model.to_ascii_lowercase(), (price.model.clone(), String::new())); + } + let filter = build_usage_filter(query); + for group in load_usage_cost_groups(connection, &filter)? { + let entry = models + .entry(group.model.to_ascii_lowercase()) + .or_insert_with(|| (group.model, String::new())); + if entry.1.is_empty() && !group.alias.trim().is_empty() { + entry.1 = group.alias; + } + } + Ok(models.into_values().collect()) +} + +async fn fetch_model_price_catalog_once( + client: &reqwest::Client, + source_name: &str, + source_url: &str, +) -> Result { + let mut response = client + .get(source_url) + .send() + .await + .map_err(|error| (true, format!("连接 {source_name} 失败: {error}")))?; + let status = response.status(); + if !status.is_success() { + let retryable = status.is_server_error() || status == reqwest::StatusCode::TOO_MANY_REQUESTS; + return Err((retryable, format!("读取 {source_name} 价格失败: HTTP {status}"))); + } + if response.content_length().is_some_and(|length| length > MODEL_PRICE_SYNC_MAX_BYTES as u64) { + return Err((false, format!("{source_name} 价格目录过大"))); + } + let mut content = Vec::new(); + while let Some(chunk) = response + .chunk() + .await + .map_err(|error| (true, format!("读取 {source_name} 响应失败: {error}")))? + { + if content.len().saturating_add(chunk.len()) > MODEL_PRICE_SYNC_MAX_BYTES { + return Err((false, format!("{source_name} 价格目录过大"))); + } + content.extend_from_slice(&chunk); + } + String::from_utf8(content) + .map_err(|error| (true, format!("{source_name} 响应不是 UTF-8: {error}"))) +} + +async fn fetch_model_price_catalog( + client: &reqwest::Client, + source_name: &str, + source_url: &str, +) -> Result { + for attempt in 1..=MODEL_PRICE_SYNC_ATTEMPTS { + match fetch_model_price_catalog_once(client, source_name, source_url).await { + Ok(content) => return Ok(content), + Err((true, _)) if attempt < MODEL_PRICE_SYNC_ATTEMPTS => { + tokio::time::sleep(Duration::from_millis(400 * attempt as u64)).await; + } + Err((retryable, error)) => { + return Err(if retryable { + format!("{error}（已尝试 {attempt} 次）") + } else { + error + }); + } + } + } + unreachable!("至少尝试读取一次价格目录") +} + #[tauri::command] -pub(crate) async fn sync_usage_model_prices( +pub(crate) async fn preview_usage_model_prices( query: UsageQuery, + source: String, gui_config_state: tauri::State<'_, GuiConfigState>, -) -> Result { +) -> Result { + let (source_name, source_url) = match source.as_str() { + "models-dev" => ("Models.dev", MODEL_PRICE_SYNC_URL), + "litellm" => ("LiteLLM", LITELLM_PRICE_SYNC_URL), + _ => return Err("不支持的价格来源".to_string()), + }; let config = gui_config_state.snapshot()?; let client_builder = reqwest::Client::builder() - .connect_timeout(Duration::from_secs(10)) - .timeout(Duration::from_secs(30)); + .user_agent("EasyCLIProxyAPI/0.2.97") + .connect_timeout(Duration::from_secs(5)) + .timeout(Duration::from_secs(60)); let proxy_url = config.proxy_url.trim(); let client = apply_configured_proxy(client_builder, proxy_url) - .ok() - .and_then(|builder| builder.build().ok()); - let remote_content = match client { - Some(client) => match client.get(MODEL_PRICE_SYNC_URL).send().await { - Ok(response) if response.status().is_success() => response.text().await.ok(), - _ => None, - }, - None => None, - }; + .map_err(|error| format!("配置价格同步代理失败: {error}"))? + .build() + .map_err(|error| format!("创建价格同步客户端失败: {error}"))?; + let content = fetch_model_price_catalog(&client, source_name, source_url).await?; let now = Local::now().timestamp_millis(); - let (remote_prices, used_builtin) = match remote_content - .as_deref() - .and_then(|content| parse_model_price_catalog(content, "github", now).ok()) - { - Some(prices) => (prices, false), - None => (bundled_model_prices()?, true), + let remote_prices = if source == "models-dev" { + parse_models_dev_prices(&content, now)? + } else { + parse_litellm_prices(&content, now)? }; run_usage_task(move || { - let mut connection = open_usage_database()?; - let current_prices = load_model_prices(&connection)?; - let manual_models = current_prices - .values() - .filter(|price| price.source == "manual") - .map(|price| price.model.to_ascii_lowercase()) - .collect::>(); - let mut result = ModelPriceSyncResult { - used_builtin, - ..ModelPriceSyncResult::default() - }; - if !used_builtin { - let transaction = connection - .transaction() - .map_err(|error| format!("开始更新模型价格失败: {error}"))?; - transaction - .execute("DELETE FROM model_prices WHERE source = 'github'", []) - .map_err(|error| format!("清理旧模型价格失败: {error}"))?; - for price in remote_prices.values() { - if manual_models.contains(&price.model.to_ascii_lowercase()) { - result.skipped += 1; - continue; - } - upsert_model_price(&transaction, price)?; - result.imported += 1; - } - transaction - .commit() - .map_err(|error| format!("提交模型价格更新失败: {error}"))?; - } else { - connection - .execute("DELETE FROM model_prices WHERE source = 'github'", []) - .map_err(|error| format!("恢复软件内置模型价格失败: {error}"))?; - } + let connection = open_usage_database()?; + let models = collect_price_preview_models(&connection, &query)?; + Ok(build_model_price_preview( + &models, + &remote_prices, + source_name, + source_url, + )) + }) + .await +} - let filter = build_usage_filter(&query); - let models = load_usage_cost_groups(&connection, &filter)? - .into_iter() - .map(|group| group.model) - .collect::>(); - let effective_prices = load_model_prices(&connection)?; - for model in models { - if resolve_model_price(&model, "", &effective_prices).is_none() { - result.unmatched.push(model); - } +#[tauri::command] +pub(crate) async fn apply_usage_model_prices(prices: Vec) -> Result { + let mut seen = std::collections::HashSet::new(); + for price in &prices { + validate_model_price(price)?; + if !matches!(price.source.as_str(), "models.dev" | "litellm-sync") + || price.source_model_id.trim().is_empty() + { + return Err("无效的同步价格来源".to_string()); } - Ok(result) + if !seen.insert(price.model.to_ascii_lowercase()) { + return Err(format!("重复的模型价格: {}", price.model)); + } + } + run_usage_task(move || { + let mut connection = open_usage_database()?; + let transaction = connection + .transaction() + .map_err(|error| format!("开始保存模型价格失败: {error}"))?; + for mut price in prices.iter().cloned() { + price.model = price.model.trim().to_string(); + price.updated_at_ms = Local::now().timestamp_millis(); + upsert_model_price(&transaction, &price)?; + } + transaction + .commit() + .map_err(|error| format!("提交模型价格失败: {error}"))?; + Ok(ModelPriceSyncResult { imported: prices.len() }) }) .await } @@ -3703,7 +3977,7 @@ fn load_usage_analysis( let sql = format!( r#" SELECT - COALESCE(NULLIF(TRIM(model), ''), 'unknown'), + {USAGE_DISPLAY_MODEL_SQL}, COALESCE(NULLIF(TRIM(provider), ''), '未知 Provider'), COALESCE(NULLIF(TRIM(source), ''), '未知来源'), COALESCE(NULLIF(TRIM(api_key_hash), ''), '未记录密钥'), @@ -4291,6 +4565,7 @@ mod tests { }, [" Model-A ", "model-a", "", "unknown", "Model-B"][index % 5], ); + record.alias = ["", " model-alias ", ""][index % 3].to_string(); record.provider = ["openai", "claude", " "][index % 3].to_string(); record.source = ["source-a", " ", "source-b", "sk-secret-source"][index % 4].to_string(); @@ -4349,7 +4624,10 @@ mod tests { let actual = load_usage_analysis(&connection, &query, &config).unwrap(); assert_eq!( canonical(actual.models), - canonical(load_simple_categories(&connection, &query, "model", "unknown").unwrap()) + canonical( + load_simple_categories(&connection, &query, USAGE_DISPLAY_MODEL_SQL, "unknown") + .unwrap() + ) ); assert_eq!( canonical(actual.providers), @@ -4369,6 +4647,56 @@ mod tests { } } + #[test] + fn model_categories_and_filter_keep_aliases_separate() { + let mut connection = Connection::open_in_memory().unwrap(); + initialize_usage_schema(&connection).unwrap(); + let mut records = Vec::new(); + for (index, alias) in ["", "model-fast", "model-high", "model-fast"] + .into_iter() + .enumerate() + { + let mut record = sample_record( + &format!("model-alias-{index}"), + "2026-09-25T00:00:00Z", + "base-model", + ); + record.alias = alias.to_string(); + records.push(record); + } + insert_usage_records(&mut connection, &records).unwrap(); + let config = GuiConfigFile::default(); + let analysis = load_usage_analysis(&connection, &UsageQuery::default(), &config).unwrap(); + let counts = analysis + .models + .iter() + .map(|category| (category.key.as_str(), category.requests)) + .collect::>(); + assert_eq!(counts.len(), 3); + assert_eq!(counts["base-model"], 1); + assert_eq!(counts["model-fast"], 2); + assert_eq!(counts["model-high"], 1); + let overview = load_usage_overview(&connection, &UsageQuery::default()).unwrap(); + let trend_models = overview + .timeline + .iter() + .flat_map(|point| point.models.iter().map(|model| model.key.as_str())) + .collect::>(); + assert_eq!(trend_models, ["base-model", "model-fast", "model-high"].into()); + + for (model, expected) in [("base-model", 1), ("model-fast", 2), ("model-high", 1)] { + let query = UsageQuery { + model: Some(model.to_string()), + ..UsageQuery::default() + }; + let filtered = load_usage_analysis(&connection, &query, &config).unwrap(); + assert_eq!(filtered.models.len(), 1); + assert_eq!(filtered.models[0].key, model); + assert_eq!(filtered.models[0].requests, expected); + assert_eq!(load_usage_events(&connection, &query, &config).unwrap().total, expected as usize); + } + } + #[test] fn collector_total_records_apply_insertions_and_retention_deletions() { let state = UsageCollectorState::default(); @@ -6123,6 +6451,56 @@ mod tests { assert!(official_model_price("unpriced-model").is_none()); } + #[test] + fn estimates_gpt6_fast_mode_for_short_and_long_context() { + let sol = official_model_price("gpt-6-sol").unwrap(); + let short_tokens = CostTokens { + input: 1_000_000, + output: 1_000_000, + ..CostTokens::default() + }; + assert!( + (cost_for_price("gpt-6-sol", "fast", &short_tokens, &sol) - 24.0).abs() < 0.000001 + ); + + let long_tokens = CostTokens { + input: 300_000, + output: 200_000, + cache_read: 100_000, + long_input: 300_000, + long_output: 200_000, + long_cache_read: 100_000, + ..CostTokens::default() + }; + assert!( + (cost_for_price("gpt-6-sol", "fast", &long_tokens, &sol) - 7.68).abs() < 0.000001 + ); + } + + #[test] + fn grok_47_uses_200k_threshold_and_doubles_long_context_output_price() { + let mut connection = Connection::open_in_memory().unwrap(); + initialize_usage_schema(&connection).unwrap(); + let mut short = sample_record("grok-short", "2026-09-25T00:00:00Z", "grok-4.7"); + short.tokens.input_tokens = 200_000; + short.tokens.output_tokens = 100_000; + short.tokens.cache_read_tokens = 0; + short.tokens.total_tokens = 300_000; + let mut long = sample_record("grok-long", "2026-09-25T00:00:00Z", "xai/grok-4.7"); + long.tokens.input_tokens = 250_000; + long.tokens.output_tokens = 100_000; + long.tokens.cache_read_tokens = 0; + long.tokens.total_tokens = 350_000; + insert_usage_records(&mut connection, &[short, long]).unwrap(); + + let query = UsageQuery::default(); + let groups = load_usage_cost_groups(&connection, &build_usage_filter(&query)).unwrap(); + assert_eq!(groups.iter().find(|group| group.model == "grok-4.7").unwrap().tokens.long_input, 0); + assert_eq!(groups.iter().find(|group| group.model == "xai/grok-4.7").unwrap().tokens.long_input, 250_000); + let pricing = load_usage_pricing(&connection, &query).unwrap(); + assert!((pricing.total_cost - 3.2).abs() < 0.000001); + } + #[test] fn bundled_model_prices_are_available_offline_and_override_legacy_cloud_cache() { let root = test_root("bundled-pricing"); @@ -6157,6 +6535,166 @@ mod tests { fs::remove_dir_all(root).unwrap(); } + #[test] + fn models_dev_sync_uses_first_party_prices_and_preserves_explicit_zero_cache() { + let catalog = serde_json::json!({ + "openai": { "models": { + "gpt-6-sol": { "cost": { "input": 2.25, "output": 10.0, "cache_read": 0.0 } }, + "missing-output": { "cost": { "input": 1.0 } } + } }, + "anthropic": { "models": {} }, + "google": { "models": {} }, + "deepseek": { "models": {} }, + "xai": { "models": {} }, + "openrouter": { "models": { + "gpt-6-sol": { "cost": { "input": 99.0, "output": 99.0 } } + } } + }); + let prices = parse_models_dev_prices(&catalog.to_string(), 123).unwrap(); + assert_eq!(prices.len(), 2); + let price = prices.iter().find(|price| price.source_model_id == "openai/gpt-6-sol").unwrap(); + assert_eq!(price.model, "gpt-6-sol"); + assert_eq!(price.prompt, 2.25); + assert_eq!(price.cache_read, 0.0); + assert!(price.cache_read_configured); + assert_eq!(price.source, "models.dev"); + assert_eq!(price.source_model_id, "openai/gpt-6-sol"); + + let effective = enriched_model_price("gpt-6-sol", price); + assert_eq!(effective.cache_read, 0.0); + assert_eq!(effective.cache_creation, 0.0); + } + + #[test] + fn price_preview_matches_official_models_without_guessing_dynamic_aliases() { + let catalog = serde_json::json!({ + "openai": { "models": { "gpt-6-sol": { "cost": { "input": 2.0, "output": 10.0 } } } }, + "google": { "models": { "gpt-6-sol": { "cost": { "input": 90.0, "output": 90.0 } } } } + }); + let remote = parse_models_dev_prices(&catalog.to_string(), 1).unwrap(); + let models = vec![ + ("gpt-6-sol".to_string(), String::new()), + ("gpt-daybreak-blue-latest".to_string(), String::new()), + ("my-alias".to_string(), "gpt-6-sol".to_string()), + ]; + let preview = build_model_price_preview(&models, &remote, "Models.dev", MODEL_PRICE_SYNC_URL); + assert_eq!(preview.matches.len(), 2); + assert_eq!(preview.matches[0].prompt, 2.0); + assert_eq!(preview.matches[0].source_model_id, "openai/gpt-6-sol"); + assert_eq!(preview.matches[1].model, "my-alias"); + assert_eq!(preview.unmatched, ["gpt-daybreak-blue-latest"]); + } + + #[test] + fn price_preview_uses_usage_alias_for_models_with_existing_prices() { + let root = test_root("price-preview-existing-alias"); + let connection = open_test_database(&root); + upsert_model_price( + &connection, + &ModelPrice { + model: "my-alias".to_string(), + prompt: 1.0, + completion: 2.0, + source: "manual".to_string(), + ..ModelPrice::default() + }, + ) + .unwrap(); + connection + .execute( + "INSERT INTO usage_events (event_key, timestamp, timestamp_ms, local_hour, model, alias, created_at) VALUES ('preview-alias', '2026-09-24T00:00:00Z', 1, '2026-09-24T00', 'my-alias', 'gpt-6-sol', '2026-09-24T00:00:00Z')", + [], + ) + .unwrap(); + + let models = collect_price_preview_models(&connection, &UsageQuery::default()).unwrap(); + let remote = parse_models_dev_prices( + &serde_json::json!({ + "openai": { "models": { "gpt-6-sol": { "cost": { "input": 2.0, "output": 10.0 } } } } + }) + .to_string(), + 1, + ) + .unwrap(); + let preview = + build_model_price_preview(&models, &remote, "Models.dev", MODEL_PRICE_SYNC_URL); + let matched = preview + .matches + .iter() + .find(|price| price.model == "my-alias") + .unwrap(); + assert_eq!(matched.source_model_id, "openai/gpt-6-sol"); + assert!(!preview.unmatched.contains(&"my-alias".to_string())); + + drop(connection); + fs::remove_dir_all(root).unwrap(); + } + + #[test] + fn litellm_preview_reads_only_base_text_token_prices() { + let catalog = serde_json::json!({ + "openai/test": { + "mode": "chat", "input_cost_per_token": 0.000002, + "output_cost_per_token": 0.000010, + "cache_read_input_token_cost": 0.0, + "input_cost_per_token_above_200k_tokens": 0.000004 + }, + "audio/test": { + "mode": "chat", "supported_output_modalities": ["audio"], + "input_cost_per_token": 0.000002, "output_cost_per_token": 0.000010 + } + }); + let prices = parse_litellm_prices(&catalog.to_string(), 1).unwrap(); + assert_eq!(prices.len(), 1); + assert_eq!(prices[0].prompt, 2.0); + assert_eq!(prices[0].completion, 10.0); + assert!(prices[0].cache_read_configured); + assert_eq!(prices[0].cache_read, 0.0); + } + + #[tokio::test] + async fn price_catalog_retries_after_truncated_body_and_decodes_gzip() { + use tokio::{ + io::{AsyncReadExt, AsyncWriteExt}, + net::TcpListener, + }; + + let listener = TcpListener::bind("127.0.0.1:0").await.unwrap(); + let url = format!("http://{}/api.json", listener.local_addr().unwrap()); + let server = tokio::spawn(async move { + let mut encoder = flate2::write::GzEncoder::new(Vec::new(), flate2::Compression::default()); + std::io::Write::write_all(&mut encoder, b"{}").unwrap(); + let compressed = encoder.finish().unwrap(); + for attempt in 0..2 { + let (mut stream, _) = listener.accept().await.unwrap(); + let mut request = [0_u8; 2048]; + let size = stream.read(&mut request).await.unwrap(); + let request = String::from_utf8_lossy(&request[..size]).to_ascii_lowercase(); + assert!(request.contains("accept-encoding: gzip")); + if attempt == 0 { + stream.write_all(b"HTTP/1.1 200 OK\r\nContent-Length: 100\r\nConnection: close\r\n\r\n{}").await.unwrap(); + } else { + let header = format!( + "HTTP/1.1 200 OK\r\nContent-Length: {}\r\nContent-Encoding: gzip\r\nConnection: close\r\n\r\n", + compressed.len() + ); + stream.write_all(header.as_bytes()).await.unwrap(); + stream.write_all(&compressed).await.unwrap(); + } + } + }); + let client = reqwest::Client::builder().no_proxy().build().unwrap(); + let content = tokio::time::timeout( + Duration::from_secs(5), + fetch_model_price_catalog(&client, "test", &url), + ) + .await + .unwrap() + .unwrap(); + assert_eq!(content, "{}"); + server.await.unwrap(); + } + #[test] fn manual_model_prices_drive_pricing_and_overview_cost() { let root = test_root("manual-pricing"); @@ -6169,7 +6707,8 @@ mod tests { model: "custom-model".to_string(), prompt: 1.0, completion: 2.0, - cache: 0.1, + cache_read: 0.1, + cache_read_configured: true, prompt_configured: true, completion_configured: true, source: "manual".to_string(), @@ -6193,4 +6732,38 @@ mod tests { drop(connection); fs::remove_dir_all(root).unwrap(); } + + #[test] + fn missing_cache_prices_stay_zero_even_with_builtin_and_legacy_prices() { + let root = test_root("zero-cache-pricing"); + let connection = open_test_database(&root); + upsert_model_price( + &connection, + &ModelPrice { + model: "gpt-6-sol".to_string(), + prompt: 2.0, + completion: 10.0, + prompt_configured: true, + completion_configured: true, + source: "manual".to_string(), + ..ModelPrice::default() + }, + ) + .unwrap(); + connection + .execute("UPDATE model_prices SET cache_per_1m = 99 WHERE model = 'gpt-6-sol'", []) + .unwrap(); + let price = load_model_prices(&connection).unwrap().remove("gpt-6-sol").unwrap(); + assert_eq!(price.cache_read, 0.0); + assert_eq!(price.cache_creation, 0.0); + let tokens = CostTokens { + input: 200_000, + cache_read: 100_000, + cache_creation: 100_000, + ..CostTokens::default() + }; + assert_eq!(cost_for_price("gpt-6-sol", "", &tokens, &price), 0.0); + drop(connection); + fs::remove_dir_all(root).unwrap(); + } } diff --git a/src/i18n/ja.ts b/src/i18n/ja.ts index 3ae5c73..de99d98 100644 --- a/src/i18n/ja.ts +++ b/src/i18n/ja.ts @@ -518,20 +518,29 @@ export const jaOverrides = { 'usage.pricing.coverage': '{priced} / {total} 件を計算済み · 利用可能なモデル料金 {saved} 件', 'usage.pricing.search': 'モデルを検索', 'usage.pricing.add': '手動で追加', - 'usage.pricing.sync': '料金を同期', - 'usage.pricing.syncing': '同期中', - 'usage.pricing.syncResult': '料金を更新しました：{imported} 件を追加、手動料金 {skipped} 件を保持、{unmatched} 件は未一致', + 'usage.pricing.sync': '料金を更新', + 'usage.pricing.syncing': '読み込み中', + 'usage.pricing.syncResult': '{imported} 件のモデル料金を保存しました', + 'usage.pricing.syncSource': '料金ソース', + 'usage.pricing.syncPreview': '料金同期のプレビュー', + 'usage.pricing.syncNote': '選択した基本料金のみ保存します。時間帯料金、コンテキスト階層、動的エイリアスは確認してください。手動料金は最初は選択されません。', + 'usage.pricing.syncCounts': '一致 {matched} 件、未一致 {unmatched} 件', + 'usage.pricing.selectAll': 'すべて選択', + 'usage.pricing.selectNone': '選択解除', + 'usage.pricing.syncNoMatches': '一致するモデル料金がありません', + 'usage.pricing.syncUnmatched': '未一致の {count} 件を表示', + 'usage.pricing.syncInvalid': '0 以上の有効な料金を入力してください。入力・出力料金は必須です', + 'usage.pricing.applySelected': '選択した {count} 件を保存', 'usage.pricing.syncFallback': 'リモート料金を読み込めなかったため、内蔵料金を引き続き使用します', 'usage.pricing.model': 'モデル', 'usage.pricing.calls': 'リクエスト数', 'usage.pricing.cost': '推定コスト', 'usage.pricing.prompt': '入力料金 ($/1M)', 'usage.pricing.completion': '出力料金 ($/1M)', - 'usage.pricing.cache': '互換キャッシュ料金 ($/1M)', 'usage.pricing.cacheRead': 'キャッシュ読み取り料金 ($/1M)', 'usage.pricing.cacheCreation': 'キャッシュ作成料金 ($/1M)', 'usage.pricing.actions': '操作', - 'usage.pricing.optional': '任意。空欄の場合は代替料金を使用', + 'usage.pricing.optional': '任意。空欄の場合は 0 で計算', 'usage.pricing.modelRequired': 'モデル名を入力してください', 'usage.pricing.saving': '保存中', 'usage.pricing.saved': 'モデル料金を保存し、概要のコストを更新しました', diff --git a/src/i18n/locales/en.ts b/src/i18n/locales/en.ts index 653bea7..718fd7c 100644 --- a/src/i18n/locales/en.ts +++ b/src/i18n/locales/en.ts @@ -505,20 +505,29 @@ export const en: Record = { 'usage.pricing.coverage': '{priced} / {total} requests priced · {saved} model prices available', 'usage.pricing.search': 'Search model', 'usage.pricing.add': 'Add Manually', - 'usage.pricing.sync': 'Sync Prices', - 'usage.pricing.syncing': 'Syncing', - 'usage.pricing.syncResult': 'Prices updated: {imported} imported, {skipped} manual prices kept, {unmatched} unmatched', + 'usage.pricing.sync': 'Update Prices', + 'usage.pricing.syncing': 'Loading', + 'usage.pricing.syncResult': 'Saved prices for {imported} models', + 'usage.pricing.syncSource': 'Price source', + 'usage.pricing.syncPreview': 'Price sync preview', + 'usage.pricing.syncNote': 'Only selected base prices are saved. Check time-based prices, context tiers, and dynamic aliases yourself. Manual prices are initially unselected.', + 'usage.pricing.syncCounts': '{matched} matched, {unmatched} unmatched', + 'usage.pricing.selectAll': 'Select all', + 'usage.pricing.selectNone': 'Select none', + 'usage.pricing.syncNoMatches': 'No matching model prices', + 'usage.pricing.syncUnmatched': 'Show {count} unmatched models', + 'usage.pricing.syncInvalid': 'Enter valid nonnegative prices; input and output prices are required', + 'usage.pricing.applySelected': 'Save {count} selected', 'usage.pricing.syncFallback': 'Remote prices could not be read; built-in prices remain active', 'usage.pricing.model': 'Model', 'usage.pricing.calls': 'Requests', 'usage.pricing.cost': 'Estimated Cost', 'usage.pricing.prompt': 'Input ($/1M)', 'usage.pricing.completion': 'Output ($/1M)', - 'usage.pricing.cache': 'Legacy Cache ($/1M)', 'usage.pricing.cacheRead': 'Cache Read ($/1M)', 'usage.pricing.cacheCreation': 'Cache Creation ($/1M)', 'usage.pricing.actions': 'Actions', - 'usage.pricing.optional': 'Optional; blank uses fallback', + 'usage.pricing.optional': 'Optional; blank is billed at 0', 'usage.pricing.modelRequired': 'Enter a model name', 'usage.pricing.saving': 'Saving', 'usage.pricing.saved': 'Model price saved; overview cost updated', diff --git a/src/i18n/locales/zh-CN.ts b/src/i18n/locales/zh-CN.ts index 31332cf..79c3b51 100644 --- a/src/i18n/locales/zh-CN.ts +++ b/src/i18n/locales/zh-CN.ts @@ -504,20 +504,29 @@ export const zhCN = { 'usage.pricing.coverage': '已计价 {priced} / {total} 次请求 · 可用 {saved} 个模型价格', 'usage.pricing.search': '搜索模型', 'usage.pricing.add': '手动添加', - 'usage.pricing.sync': '同步价格', - 'usage.pricing.syncing': '同步中', - 'usage.pricing.syncResult': '价格更新完成：导入 {imported}，保留手动价格 {skipped}，未匹配 {unmatched}', + 'usage.pricing.sync': '更新价格', + 'usage.pricing.syncing': '读取中', + 'usage.pricing.syncResult': '已保存 {imported} 个模型价格', + 'usage.pricing.syncSource': '价格来源', + 'usage.pricing.syncPreview': '价格同步预览', + 'usage.pricing.syncNote': '仅保存勾选的基础价格。时段价、上下文阶梯价和动态别名需自行核对；手动价格默认不勾选。', + 'usage.pricing.syncCounts': '匹配 {matched} 个，未匹配 {unmatched} 个', + 'usage.pricing.selectAll': '全选', + 'usage.pricing.selectNone': '取消全选', + 'usage.pricing.syncNoMatches': '没有可同步的模型价格', + 'usage.pricing.syncUnmatched': '查看 {count} 个未匹配模型', + 'usage.pricing.syncInvalid': '请输入有效的非负价格；输入价和输出价不能为空', + 'usage.pricing.applySelected': '保存所选 {count} 项', 'usage.pricing.syncFallback': '远程价格读取失败，已继续使用软件内置价格', 'usage.pricing.model': '模型', 'usage.pricing.calls': '请求数', 'usage.pricing.cost': '预估成本', 'usage.pricing.prompt': '输入价 ($/1M)', 'usage.pricing.completion': '输出价 ($/1M)', - 'usage.pricing.cache': '兼容缓存价 ($/1M)', 'usage.pricing.cacheRead': '缓存读取价 ($/1M)', 'usage.pricing.cacheCreation': '缓存创建价 ($/1M)', 'usage.pricing.actions': '操作', - 'usage.pricing.optional': '可选，留空使用回退价', + 'usage.pricing.optional': '可选，留空按 0 计费', 'usage.pricing.modelRequired': '请输入模型名称', 'usage.pricing.saving': '保存中', 'usage.pricing.saved': '模型价格已保存，总览成本已更新', diff --git a/src/pages/UsageRecordsPage.tsx b/src/pages/UsageRecordsPage.tsx index 65c0b7a..56f366d 100644 --- a/src/pages/UsageRecordsPage.tsx +++ b/src/pages/UsageRecordsPage.tsx @@ -135,7 +135,6 @@ type ModelPrice = { model: string; prompt: number; completion: number; - cache: number; cacheRead: number; cacheCreation: number; promptConfigured: boolean; @@ -183,9 +182,22 @@ type UsageStorageSettings = { type ModelPriceSyncResult = { imported: number; - skipped: number; +}; + +type ModelPriceSyncPreview = { + source: string; + sourceUrl: string; + matches: ModelPrice[]; unmatched: string[]; - usedBuiltin: boolean; +}; + +type SyncPriceDraft = { + selected: boolean; + price: ModelPrice; + prompt: string; + completion: string; + cacheRead: string; + cacheCreation: string; }; type UsageQuery = { @@ -2271,7 +2283,6 @@ type PriceDraft = { model: string; prompt: string; completion: string; - cache: string; cacheRead: string; cacheCreation: string; }; @@ -2280,7 +2291,6 @@ const emptyPriceDraft = (): PriceDraft => ({ model: '', prompt: '', completion: '', - cache: '', cacheRead: '', cacheCreation: '', }); @@ -2288,9 +2298,8 @@ const priceDraftFor = (model = '', price?: ModelPrice | null): PriceDraft => ({ model, prompt: price ? String(price.prompt) : '', completion: price ? String(price.completion) : '', - cache: price ? String(price.cache) : '', - cacheRead: price && (price.cacheReadConfigured || price.cacheRead > 0) ? String(price.cacheRead) : '', - cacheCreation: price && (price.cacheCreationConfigured || price.cacheCreation > 0) ? String(price.cacheCreation) : '', + cacheRead: price ? String(price.cacheRead) : '', + cacheCreation: price ? String(price.cacheCreation) : '', }); const parsePrice = (value: string) => { @@ -2315,6 +2324,15 @@ function PricingView({ const [draft, setDraft] = useState(null); const [saving, setSaving] = useState(false); const [syncing, setSyncing] = useState(false); + const [applyingSync, setApplyingSync] = useState(false); + const [syncSource, setSyncSource] = useState<'models-dev' | 'litellm'>(() => { + try { return localStorage.getItem('cpa-gui.pricing-sync-source.v1') === 'litellm' ? 'litellm' : 'models-dev'; } + catch { return 'models-dev'; } + }); + const [syncPreview, setSyncPreview] = useState(null); + const [syncDrafts, setSyncDrafts] = useState([]); + const [syncError, setSyncError] = useState(''); + const syncDialog = useRef(null); const { notice, revision, showNotice, clearNotice } = useAppNotice(); @@ -2336,7 +2354,6 @@ function PricingView({ model: draft.model.trim(), prompt: parsePrice(draft.prompt), completion: parsePrice(draft.completion), - cache: draft.cache.trim() ? parsePrice(draft.cache) : parsePrice(draft.prompt), cacheRead: parsePrice(draft.cacheRead), cacheCreation: parsePrice(draft.cacheCreation), promptConfigured: draft.prompt.trim() !== '', @@ -2370,20 +2387,23 @@ function PricingView({ } }; - const syncPrices = async () => { + const previewPrices = async () => { setSyncing(true); + setSyncError(''); clearNotice(); try { - const result = await invoke('sync_usage_model_prices', { query }); - showNotice({ - key: 'usage.pricing.syncResult', - variables: { - imported: result.imported, - skipped: result.skipped, - unmatched: result.unmatched.length, - }, - }); - await onChanged(); + const preview = await invoke('preview_usage_model_prices', { query, source: syncSource }); + const existing = new Map(pricing.rows.map((row) => [row.model.toLowerCase(), row.price])); + setSyncPreview(preview); + setSyncDrafts(preview.matches.map((price) => ({ + selected: existing.get(price.model.toLowerCase())?.source !== 'manual', + price, + prompt: String(price.prompt), + completion: String(price.completion), + cacheRead: String(price.cacheRead), + cacheCreation: String(price.cacheCreation), + }))); + syncDialog.current?.showModal(); } catch (syncError) { showNotice(String(syncError), 'error'); } finally { @@ -2391,6 +2411,49 @@ function PricingView({ } }; + const applySyncPrices = async () => { + const selected = syncDrafts.filter((item) => item.selected); + if (!selected.length) return; + const valid = (value: string) => value.trim() !== '' && Number.isFinite(Number(value)) && Number(value) >= 0; + if (selected.some((item) => !valid(item.prompt) || !valid(item.completion) + || (item.cacheRead.trim() !== '' && !valid(item.cacheRead)) + || (item.cacheCreation.trim() !== '' && !valid(item.cacheCreation)))) { + setSyncError(t('usage.pricing.syncInvalid')); + return; + } + const prices = selected.map((item): ModelPrice => ({ + ...item.price, + prompt: Number(item.prompt), + completion: Number(item.completion), + cacheRead: item.cacheRead.trim() === '' ? 0 : Number(item.cacheRead), + cacheCreation: item.cacheCreation.trim() === '' ? 0 : Number(item.cacheCreation), + cacheReadConfigured: item.cacheRead.trim() !== '', + cacheCreationConfigured: item.cacheCreation.trim() !== '', + })); + setApplyingSync(true); + setSyncError(''); + clearNotice(); + try { + const result = await invoke('apply_usage_model_prices', { prices }); + syncDialog.current?.close(); + showNotice({ key: 'usage.pricing.syncResult', variables: { imported: result.imported } }); + await onChanged(); + } catch (syncError) { + setSyncError(String(syncError)); + } finally { + setApplyingSync(false); + } + }; + + const updateSyncDraft = (index: number, update: Partial) => { + setSyncDrafts((current) => current.map((item, currentIndex) => currentIndex === index ? { ...item, ...update } : item)); + }; + + const changeSyncSource = (source: 'models-dev' | 'litellm') => { + setSyncSource(source); + try { localStorage.setItem('cpa-gui.pricing-sync-source.v1', source); } catch {} + }; + return (
{confirmationDialog} @@ -2415,8 +2478,11 @@ function PricingView({ - @@ -2424,6 +2490,53 @@ function PricingView({ + { if (applyingSync) event.preventDefault(); }} onClose={() => setSyncPreview(null)}> +
+
+

{t('usage.pricing.syncPreview')}

+ {syncPreview?.source} +
+ +
+

{t('usage.pricing.syncNote')}

+
+ {t('usage.pricing.syncCounts', { matched: syncDrafts.length, unmatched: syncPreview?.unmatched.length ?? 0 })} + + +
+
+ {syncDrafts.map((item, index) => ( +
+ + {(['prompt', 'completion', 'cacheRead', 'cacheCreation'] as const).map((field) => ( + + ))} +
+ ))} + {syncDrafts.length === 0 &&

{t('usage.pricing.syncNoMatches')}

} +
+ {(syncPreview?.unmatched.length ?? 0) > 0 && ( +
+ {t('usage.pricing.syncUnmatched', { count: syncPreview?.unmatched.length ?? 0 })} +

{syncPreview?.unmatched.join(', ')}

+
+ )} + {syncError &&

{syncError}

} +
+ + +
+ + {draft ? (
-