Files
cc-switch/src-tauri/src/services/usage_stats.rs
T
Dex MillerandGitHub 5376ea042b Feat/usage improvements (#508)
* i18n: update cache terminology across all languages

- Change 'Cache Read' to 'Cache Hit' in all languages
- Change 'Cache Write' to 'Cache Creation' in all languages
- Update zh: 缓存读取 → 缓存命中, 缓存写入 → 缓存创建
- Update en: Cache Read → Cache Hit, Cache Write → Cache Creation
- Update ja: キャッシュ読取 → キャッシュヒット, キャッシュ書込 → キャッシュ作成

Affected keys: cacheReadTokens, cacheCreationTokens, cacheReadCost,
cacheWriteCost, cacheRead, cacheWrite

* feat(usage): add cache metrics to trend chart

- Add cache creation tokens visualization (orange line)
- Add cache hit tokens visualization (purple line)
- Add gradient definitions for new cache metrics
- Include cache data in hourly aggregation
- Display cache metrics alongside input/output tokens

This provides better visibility into cache usage patterns over time.

* fix(usage): fix timezone handling in datetime picker

- Add timestampToLocalDatetime() to convert Unix timestamp to local datetime
- Add localDatetimeToTimestamp() with validation for incomplete input
- Fix issue where typing hours/minutes would jump to previous day
- Validate datetime format completeness before conversion
- Use local timezone instead of UTC for datetime-local input

This resolves the issue where users couldn't fine-tune time selection
and the input would jump unexpectedly when editing hours or minutes.

* feat(usage): add auto-refresh for usage statistics

- Add 30-second auto-refresh interval for all usage queries
- Disable background refresh to save resources
- Apply to: summary, trends, provider stats, model stats, request logs
- Queries automatically update when tab is active
- Pause refresh when user switches to another tab

This keeps usage data fresh without manual refresh.

* fix(proxy): improve usage logging and cache token parsing

- Log requests even when usage parsing fails (with default values)
- Add detailed debug logging for usage metrics
- Support cache_read_input_tokens field in Codex responses
- Fallback to input_tokens_details.cached_tokens if needed
- Add test case for cached_tokens in input_tokens_details
- Ensure all requests are tracked in database for analytics

This fixes missing request logs when API responses lack usage data
and improves cache token detection across different response formats.

* style(rust): use inline format args in format! macros

- Replace format!("...", var) with format!("...{var}")
- Update universal provider ID formatting
- Update error message formatting
- Update config.toml generation in Codex provider

Fixes clippy::uninlined_format_args warnings.

* feat(proxy): enhance provider router logging

- Add debug logs for failover queue provider count
- Log circuit breaker state for each provider check
- Add logs for missing current provider scenarios
- Log when no current provider is configured
- Use inline format args for better readability

This improves debugging of provider selection and failover behavior.

* feat(database): update model pricing data

- Update Claude models to full version format (e.g. claude-opus-4-5-20251101)
- Add GPT-5.2 series model pricing (10 models)
- Add GPT-5.1 series model pricing (10 models)
- Add GPT-5 series model pricing (12 models)
- Add Gemini 3 series model pricing (2 models)
- Update Gemini 2.5 series model ID format (use dot separator)
- Unify display names by removing thinking level suffixes

* fix(usage): correct Gemini output token calculation

Fix Gemini API output token parsing to use totalTokenCount - promptTokenCount
instead of candidatesTokenCount alone. This ensures thoughtsTokenCount is
included in output statistics.

- Update from_gemini_response to calculate output from total - input
- Update from_gemini_stream_chunks with same logic for consistency
- Fix from_codex_stream_events to use adjusted token calculation
- Add test case for responses with thoughtsTokenCount
- Update existing tests to match new calculation logic

* fix(usage): correct cache token billing and add Codex format auto-detection

- Avoid double-billing cache tokens by subtracting from input before calculation
- Add smart Codex parser that auto-detects OpenAI vs Codex API format
- Extract model name from Codex responses for accurate tracking

* fix(proxy): improve takeover detection with live config check

- Add live config takeover detection for hot-switch decision
- Rebuild takeover when backup is missing or placeholder remains
- Make detect_takeover_in_live_config_for_app public
- Fix is_takeover_active to use actual takeover status

* refactor(usage): simplify model pricing lookup by removing suffix fallback

Replace complex suffix-stripping fallback with direct prefix/suffix cleanup.
Model IDs are now cleaned by removing vendor prefix (before /) and colon
suffix (after :), then matched exactly against pricing table.

* feat(database): add Chinese AI model pricing data

Add pricing for domestic AI models (CNY/1M tokens):
- Doubao-Seed-Code (ByteDance)
- DeepSeek V3/V3.1/V3.2
- Kimi K2/K2-Thinking/K2-Turbo (Moonshot)
- MiniMax M2/M2.1/M2.1-Lightning
- GLM-4.6/4.7 (Zhipu)
- Mimo V2 Flash (Xiaomi)

Also fix test case to use correct model ID and remove invalid currency column.

* refactor(proxy): improve header forwarding with blacklist approach

Change from whitelist to blacklist mode for request header forwarding.
Only skip headers that will be overridden (auth, host, content-length).
This preserves client's original headers and improves compatibility.

* fix(proxy): bypass timeout and retry configs when failover is disabled

When auto_failover_enabled is false, timeout and retry configurations
should not affect normal request flow. This change ensures:

- create_forwarder: passes 0 for all timeout/retry params when failover
  is disabled, effectively bypassing these checks
- streaming_timeout_config: returns 0 for both first_byte_timeout and
  idle_timeout when failover is disabled

This prevents unnecessary timeout errors and retry attempts when users
have explicitly disabled the failover feature.

* fix(proxy): handle zero value input in failover config fields

* refactor(proxy): remove retry logic and add enabled check for failover

* refactor(proxy): distinguish circuit-open from no-provider errors

* Align usage stats to sliding windows

* feat(proxy): add body and header filtering for upstream requests

* feat(proxy): enable transparent passthrough for headers

- Passthrough anthropic-beta header as-is from client
- Passthrough anthropic-version header from client
- Passthrough client IP headers (x-forwarded-for, x-real-ip) by default
- Filter private params (underscore-prefixed fields) from request body
- No database changes required

* feat(proxy): extract session ID from client requests for logging

- Add SessionIdExtractor to parse session ID from Claude/Codex requests
- Support extraction from metadata.user_id, headers, previous_response_id
- Pass session_id through RequestContext to usage logger
- Enable request correlation by session in proxy_request_logs
2025-12-31 22:57:00 +08:00

962 lines
34 KiB
Rust
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
//! 使用统计服务
//!
//! 提供使用量数据的聚合查询功能
use crate::database::{lock_conn, Database};
use crate::error::AppError;
use chrono::{Local, TimeZone};
use rusqlite::{params, Connection, OptionalExtension};
use serde::{Deserialize, Serialize};
use serde_json::Value;
use std::collections::HashMap;
use std::str::FromStr;
/// 使用量汇总
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct UsageSummary {
pub total_requests: u64,
pub total_cost: String,
pub total_input_tokens: u64,
pub total_output_tokens: u64,
pub total_cache_creation_tokens: u64,
pub total_cache_read_tokens: u64,
pub success_rate: f32,
}
/// 每日统计
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct DailyStats {
pub date: String,
pub request_count: u64,
pub total_cost: String,
pub total_tokens: u64,
pub total_input_tokens: u64,
pub total_output_tokens: u64,
pub total_cache_creation_tokens: u64,
pub total_cache_read_tokens: u64,
}
/// Provider 统计
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct ProviderStats {
pub provider_id: String,
pub provider_name: String,
pub request_count: u64,
pub total_tokens: u64,
pub total_cost: String,
pub success_rate: f32,
pub avg_latency_ms: u64,
}
/// 模型统计
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct ModelStats {
pub model: String,
pub request_count: u64,
pub total_tokens: u64,
pub total_cost: String,
pub avg_cost_per_request: String,
}
/// 请求日志过滤器
#[derive(Debug, Clone, Default, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct LogFilters {
pub app_type: Option<String>,
pub provider_name: Option<String>,
pub model: Option<String>,
pub status_code: Option<u16>,
pub start_date: Option<i64>,
pub end_date: Option<i64>,
}
/// 分页请求日志响应
#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct PaginatedLogs {
pub data: Vec<RequestLogDetail>,
pub total: u32,
pub page: u32,
pub page_size: u32,
}
/// 请求日志详情
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct RequestLogDetail {
pub request_id: String,
pub provider_id: String,
#[serde(skip_serializing_if = "Option::is_none")]
pub provider_name: Option<String>,
pub app_type: String,
pub model: String,
pub input_tokens: u32,
pub output_tokens: u32,
pub cache_read_tokens: u32,
pub cache_creation_tokens: u32,
pub input_cost_usd: String,
pub output_cost_usd: String,
pub cache_read_cost_usd: String,
pub cache_creation_cost_usd: String,
pub total_cost_usd: String,
pub is_streaming: bool,
pub latency_ms: u64,
pub first_token_ms: Option<u64>,
pub duration_ms: Option<u64>,
pub status_code: u16,
pub error_message: Option<String>,
pub created_at: i64,
}
impl Database {
/// 获取使用量汇总
pub fn get_usage_summary(
&self,
start_date: Option<i64>,
end_date: Option<i64>,
) -> Result<UsageSummary, AppError> {
let conn = lock_conn!(self.conn);
let (where_clause, params_vec) = if start_date.is_some() || end_date.is_some() {
let mut conditions = Vec::new();
let mut params = Vec::new();
if let Some(start) = start_date {
conditions.push("created_at >= ?");
params.push(start);
}
if let Some(end) = end_date {
conditions.push("created_at <= ?");
params.push(end);
}
(format!("WHERE {}", conditions.join(" AND ")), params)
} else {
(String::new(), Vec::new())
};
let sql = format!(
"SELECT
COUNT(*) as total_requests,
COALESCE(SUM(CAST(total_cost_usd AS REAL)), 0) as total_cost,
COALESCE(SUM(input_tokens), 0) as total_input_tokens,
COALESCE(SUM(output_tokens), 0) as total_output_tokens,
COALESCE(SUM(cache_creation_tokens), 0) as total_cache_creation_tokens,
COALESCE(SUM(cache_read_tokens), 0) as total_cache_read_tokens,
COALESCE(SUM(CASE WHEN status_code >= 200 AND status_code < 300 THEN 1 ELSE 0 END), 0) as success_count
FROM proxy_request_logs
{where_clause}"
);
let result = conn.query_row(&sql, rusqlite::params_from_iter(params_vec), |row| {
let total_requests: i64 = row.get(0)?;
let total_cost: f64 = row.get(1)?;
let total_input_tokens: i64 = row.get(2)?;
let total_output_tokens: i64 = row.get(3)?;
let total_cache_creation_tokens: i64 = row.get(4)?;
let total_cache_read_tokens: i64 = row.get(5)?;
let success_count: i64 = row.get(6)?;
let success_rate = if total_requests > 0 {
(success_count as f32 / total_requests as f32) * 100.0
} else {
0.0
};
Ok(UsageSummary {
total_requests: total_requests as u64,
total_cost: format!("{total_cost:.6}"),
total_input_tokens: total_input_tokens as u64,
total_output_tokens: total_output_tokens as u64,
total_cache_creation_tokens: total_cache_creation_tokens as u64,
total_cache_read_tokens: total_cache_read_tokens as u64,
success_rate,
})
})?;
Ok(result)
}
/// 获取每日趋势(滑动窗口,<=24h 按小时,>24h 按天,窗口与汇总一致)
pub fn get_daily_trends(
&self,
start_date: Option<i64>,
end_date: Option<i64>,
) -> Result<Vec<DailyStats>, AppError> {
let conn = lock_conn!(self.conn);
let end_ts = end_date.unwrap_or_else(|| Local::now().timestamp());
let mut start_ts = start_date.unwrap_or_else(|| end_ts - 24 * 60 * 60);
if start_ts >= end_ts {
start_ts = end_ts - 24 * 60 * 60;
}
let duration = end_ts - start_ts;
let bucket_seconds: i64 = if duration <= 24 * 60 * 60 {
60 * 60
} else {
24 * 60 * 60
};
let mut bucket_count: i64 = if duration <= 0 {
1
} else {
((duration as f64) / bucket_seconds as f64).ceil() as i64
};
// 固定 24 小时窗口为 24 个小时桶,避免浮点误差
if bucket_seconds == 60 * 60 {
bucket_count = 24;
}
if bucket_count < 1 {
bucket_count = 1;
}
let sql = "
SELECT
CAST((created_at - ?1) / ?3 AS INTEGER) as bucket_idx,
COUNT(*) as request_count,
COALESCE(SUM(CAST(total_cost_usd AS REAL)), 0) as total_cost,
COALESCE(SUM(input_tokens + output_tokens), 0) as total_tokens,
COALESCE(SUM(input_tokens), 0) as total_input_tokens,
COALESCE(SUM(output_tokens), 0) as total_output_tokens,
COALESCE(SUM(cache_creation_tokens), 0) as total_cache_creation_tokens,
COALESCE(SUM(cache_read_tokens), 0) as total_cache_read_tokens
FROM proxy_request_logs
WHERE created_at >= ?1 AND created_at <= ?2
GROUP BY bucket_idx
ORDER BY bucket_idx ASC";
let mut stmt = conn.prepare(sql)?;
let rows = stmt.query_map(params![start_ts, end_ts, bucket_seconds], |row| {
Ok((
row.get::<_, i64>(0)?,
DailyStats {
date: String::new(),
request_count: row.get::<_, i64>(1)? as u64,
total_cost: format!("{:.6}", row.get::<_, f64>(2)?),
total_tokens: row.get::<_, i64>(3)? as u64,
total_input_tokens: row.get::<_, i64>(4)? as u64,
total_output_tokens: row.get::<_, i64>(5)? as u64,
total_cache_creation_tokens: row.get::<_, i64>(6)? as u64,
total_cache_read_tokens: row.get::<_, i64>(7)? as u64,
},
))
})?;
let mut map: HashMap<i64, DailyStats> = HashMap::new();
for row in rows {
let (mut bucket_idx, stat) = row?;
if bucket_idx < 0 {
continue;
}
if bucket_idx >= bucket_count {
bucket_idx = bucket_count - 1;
}
map.insert(bucket_idx, stat);
}
let mut stats = Vec::with_capacity(bucket_count as usize);
for i in 0..bucket_count {
let bucket_start_ts = start_ts + i * bucket_seconds;
let bucket_start = Local
.timestamp_opt(bucket_start_ts, 0)
.single()
.unwrap_or_else(Local::now);
let date = bucket_start.format("%Y-%m-%dT%H:%M:%S").to_string();
if let Some(mut stat) = map.remove(&i) {
stat.date = date;
stats.push(stat);
} else {
stats.push(DailyStats {
date,
request_count: 0,
total_cost: "0.000000".to_string(),
total_tokens: 0,
total_input_tokens: 0,
total_output_tokens: 0,
total_cache_creation_tokens: 0,
total_cache_read_tokens: 0,
});
}
}
Ok(stats)
}
/// 获取 Provider 统计
pub fn get_provider_stats(&self) -> Result<Vec<ProviderStats>, AppError> {
let conn = lock_conn!(self.conn);
let sql = "SELECT
l.provider_id,
p.name as provider_name,
COUNT(*) as request_count,
COALESCE(SUM(l.input_tokens + l.output_tokens), 0) as total_tokens,
COALESCE(SUM(CAST(l.total_cost_usd AS REAL)), 0) as total_cost,
COALESCE(SUM(CASE WHEN l.status_code >= 200 AND l.status_code < 300 THEN 1 ELSE 0 END), 0) as success_count,
COALESCE(AVG(l.latency_ms), 0) as avg_latency
FROM proxy_request_logs l
LEFT JOIN providers p ON l.provider_id = p.id AND l.app_type = p.app_type
GROUP BY l.provider_id, l.app_type
ORDER BY total_cost DESC";
let mut stmt = conn.prepare(sql)?;
let rows = stmt.query_map([], |row| {
let request_count: i64 = row.get(2)?;
let success_count: i64 = row.get(5)?;
let success_rate = if request_count > 0 {
(success_count as f32 / request_count as f32) * 100.0
} else {
0.0
};
Ok(ProviderStats {
provider_id: row.get(0)?,
provider_name: row
.get::<_, Option<String>>(1)?
.unwrap_or_else(|| "Unknown".to_string()),
request_count: request_count as u64,
total_tokens: row.get::<_, i64>(3)? as u64,
total_cost: format!("{:.6}", row.get::<_, f64>(4)?),
success_rate,
avg_latency_ms: row.get::<_, f64>(6)? as u64,
})
})?;
let mut stats = Vec::new();
for row in rows {
stats.push(row?);
}
Ok(stats)
}
/// 获取模型统计
pub fn get_model_stats(&self) -> Result<Vec<ModelStats>, AppError> {
let conn = lock_conn!(self.conn);
let sql = "SELECT
model,
COUNT(*) as request_count,
COALESCE(SUM(input_tokens + output_tokens), 0) as total_tokens,
COALESCE(SUM(CAST(total_cost_usd AS REAL)), 0) as total_cost
FROM proxy_request_logs
GROUP BY model
ORDER BY total_cost DESC";
let mut stmt = conn.prepare(sql)?;
let rows = stmt.query_map([], |row| {
let request_count: i64 = row.get(1)?;
let total_cost: f64 = row.get(3)?;
let avg_cost = if request_count > 0 {
total_cost / request_count as f64
} else {
0.0
};
Ok(ModelStats {
model: row.get(0)?,
request_count: request_count as u64,
total_tokens: row.get::<_, i64>(2)? as u64,
total_cost: format!("{total_cost:.6}"),
avg_cost_per_request: format!("{avg_cost:.6}"),
})
})?;
let mut stats = Vec::new();
for row in rows {
stats.push(row?);
}
Ok(stats)
}
/// 获取请求日志列表(分页)
pub fn get_request_logs(
&self,
filters: &LogFilters,
page: u32,
page_size: u32,
) -> Result<PaginatedLogs, AppError> {
let conn = lock_conn!(self.conn);
let mut conditions = Vec::new();
let mut params: Vec<Box<dyn rusqlite::ToSql>> = Vec::new();
if let Some(ref app_type) = filters.app_type {
conditions.push("l.app_type = ?");
params.push(Box::new(app_type.clone()));
}
if let Some(ref provider_name) = filters.provider_name {
conditions.push("p.name LIKE ?");
params.push(Box::new(format!("%{provider_name}%")));
}
if let Some(ref model) = filters.model {
conditions.push("l.model LIKE ?");
params.push(Box::new(format!("%{model}%")));
}
if let Some(status) = filters.status_code {
conditions.push("l.status_code = ?");
params.push(Box::new(status as i64));
}
if let Some(start) = filters.start_date {
conditions.push("l.created_at >= ?");
params.push(Box::new(start));
}
if let Some(end) = filters.end_date {
conditions.push("l.created_at <= ?");
params.push(Box::new(end));
}
let where_clause = if conditions.is_empty() {
String::new()
} else {
format!("WHERE {}", conditions.join(" AND "))
};
// 获取总数
let count_sql = format!(
"SELECT COUNT(*) FROM proxy_request_logs l
LEFT JOIN providers p ON l.provider_id = p.id AND l.app_type = p.app_type
{where_clause}"
);
let count_params: Vec<&dyn rusqlite::ToSql> = params.iter().map(|p| p.as_ref()).collect();
let total: u32 = conn.query_row(&count_sql, count_params.as_slice(), |row| {
row.get::<_, i64>(0).map(|v| v as u32)
})?;
// 获取数据
let offset = page * page_size;
params.push(Box::new(page_size as i64));
params.push(Box::new(offset as i64));
let sql = format!(
"SELECT l.request_id, l.provider_id, p.name as provider_name, l.app_type, l.model,
l.input_tokens, l.output_tokens, l.cache_read_tokens, l.cache_creation_tokens,
l.input_cost_usd, l.output_cost_usd, l.cache_read_cost_usd, l.cache_creation_cost_usd, l.total_cost_usd,
l.is_streaming, l.latency_ms, l.first_token_ms, l.duration_ms,
l.status_code, l.error_message, l.created_at
FROM proxy_request_logs l
LEFT JOIN providers p ON l.provider_id = p.id AND l.app_type = p.app_type
{where_clause}
ORDER BY l.created_at DESC
LIMIT ? OFFSET ?"
);
let mut stmt = conn.prepare(&sql)?;
let params_refs: Vec<&dyn rusqlite::ToSql> = params.iter().map(|p| p.as_ref()).collect();
let rows = stmt.query_map(params_refs.as_slice(), |row| {
Ok(RequestLogDetail {
request_id: row.get(0)?,
provider_id: row.get(1)?,
provider_name: row.get(2)?,
app_type: row.get(3)?,
model: row.get(4)?,
input_tokens: row.get::<_, i64>(5)? as u32,
output_tokens: row.get::<_, i64>(6)? as u32,
cache_read_tokens: row.get::<_, i64>(7)? as u32,
cache_creation_tokens: row.get::<_, i64>(8)? as u32,
input_cost_usd: row.get(9)?,
output_cost_usd: row.get(10)?,
cache_read_cost_usd: row.get(11)?,
cache_creation_cost_usd: row.get(12)?,
total_cost_usd: row.get(13)?,
is_streaming: row.get::<_, i64>(14)? != 0,
latency_ms: row.get::<_, i64>(15)? as u64,
first_token_ms: row.get::<_, Option<i64>>(16)?.map(|v| v as u64),
duration_ms: row.get::<_, Option<i64>>(17)?.map(|v| v as u64),
status_code: row.get::<_, i64>(18)? as u16,
error_message: row.get(19)?,
created_at: row.get(20)?,
})
})?;
let mut logs = Vec::new();
let mut provider_cache = HashMap::new();
let mut pricing_cache = HashMap::new();
for row in rows {
let mut log = row?;
Self::maybe_backfill_log_costs(
&conn,
&mut log,
&mut provider_cache,
&mut pricing_cache,
)?;
logs.push(log);
}
Ok(PaginatedLogs {
data: logs,
total,
page,
page_size,
})
}
/// 获取单个请求详情
pub fn get_request_detail(
&self,
request_id: &str,
) -> Result<Option<RequestLogDetail>, AppError> {
let conn = lock_conn!(self.conn);
let result = conn.query_row(
"SELECT l.request_id, l.provider_id, p.name as provider_name, l.app_type, l.model,
input_tokens, output_tokens, cache_read_tokens, cache_creation_tokens,
input_cost_usd, output_cost_usd, cache_read_cost_usd, cache_creation_cost_usd, total_cost_usd,
is_streaming, latency_ms, first_token_ms, duration_ms,
status_code, error_message, created_at
FROM proxy_request_logs l
LEFT JOIN providers p ON l.provider_id = p.id AND l.app_type = p.app_type
WHERE l.request_id = ?",
[request_id],
|row| {
Ok(RequestLogDetail {
request_id: row.get(0)?,
provider_id: row.get(1)?,
provider_name: row.get(2)?,
app_type: row.get(3)?,
model: row.get(4)?,
input_tokens: row.get::<_, i64>(5)? as u32,
output_tokens: row.get::<_, i64>(6)? as u32,
cache_read_tokens: row.get::<_, i64>(7)? as u32,
cache_creation_tokens: row.get::<_, i64>(8)? as u32,
input_cost_usd: row.get(9)?,
output_cost_usd: row.get(10)?,
cache_read_cost_usd: row.get(11)?,
cache_creation_cost_usd: row.get(12)?,
total_cost_usd: row.get(13)?,
is_streaming: row.get::<_, i64>(14)? != 0,
latency_ms: row.get::<_, i64>(15)? as u64,
first_token_ms: row.get::<_, Option<i64>>(16)?.map(|v| v as u64),
duration_ms: row.get::<_, Option<i64>>(17)?.map(|v| v as u64),
status_code: row.get::<_, i64>(18)? as u16,
error_message: row.get(19)?,
created_at: row.get(20)?,
})
},
);
match result {
Ok(mut detail) => {
let mut provider_cache = HashMap::new();
let mut pricing_cache = HashMap::new();
Self::maybe_backfill_log_costs(
&conn,
&mut detail,
&mut provider_cache,
&mut pricing_cache,
)?;
Ok(Some(detail))
}
Err(rusqlite::Error::QueryReturnedNoRows) => Ok(None),
Err(e) => Err(AppError::Database(e.to_string())),
}
}
/// 检查 Provider 使用限额
pub fn check_provider_limits(
&self,
provider_id: &str,
app_type: &str,
) -> Result<ProviderLimitStatus, AppError> {
let conn = lock_conn!(self.conn);
// 获取 provider 的限额设置
let (limit_daily, limit_monthly) = conn
.query_row(
"SELECT meta FROM providers WHERE id = ? AND app_type = ?",
params![provider_id, app_type],
|row| {
let meta_str: String = row.get(0)?;
Ok(meta_str)
},
)
.ok()
.and_then(|meta_str| serde_json::from_str::<serde_json::Value>(&meta_str).ok())
.map(|meta| {
let daily = meta
.get("limitDailyUsd")
.and_then(|v| v.as_str())
.and_then(|s| s.parse::<f64>().ok());
let monthly = meta
.get("limitMonthlyUsd")
.and_then(|v| v.as_str())
.and_then(|s| s.parse::<f64>().ok());
(daily, monthly)
})
.unwrap_or((None, None));
// 计算今日使用量
let daily_usage: f64 = conn
.query_row(
"SELECT COALESCE(SUM(CAST(total_cost_usd AS REAL)), 0)
FROM proxy_request_logs
WHERE provider_id = ? AND app_type = ?
AND date(datetime(created_at, 'unixepoch', 'localtime')) = date('now', 'localtime')",
params![provider_id, app_type],
|row| row.get(0),
)
.unwrap_or(0.0);
// 计算本月使用量
let monthly_usage: f64 = conn
.query_row(
"SELECT COALESCE(SUM(CAST(total_cost_usd AS REAL)), 0)
FROM proxy_request_logs
WHERE provider_id = ? AND app_type = ?
AND strftime('%Y-%m', datetime(created_at, 'unixepoch', 'localtime')) = strftime('%Y-%m', 'now', 'localtime')",
params![provider_id, app_type],
|row| row.get(0),
)
.unwrap_or(0.0);
let daily_exceeded = limit_daily
.map(|limit| daily_usage >= limit)
.unwrap_or(false);
let monthly_exceeded = limit_monthly
.map(|limit| monthly_usage >= limit)
.unwrap_or(false);
Ok(ProviderLimitStatus {
provider_id: provider_id.to_string(),
daily_usage: format!("{daily_usage:.6}"),
daily_limit: limit_daily.map(|l| format!("{l:.2}")),
daily_exceeded,
monthly_usage: format!("{monthly_usage:.6}"),
monthly_limit: limit_monthly.map(|l| format!("{l:.2}")),
monthly_exceeded,
})
}
}
/// Provider 限额状态
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct ProviderLimitStatus {
pub provider_id: String,
pub daily_usage: String,
pub daily_limit: Option<String>,
pub daily_exceeded: bool,
pub monthly_usage: String,
pub monthly_limit: Option<String>,
pub monthly_exceeded: bool,
}
#[derive(Clone)]
struct PricingInfo {
input: rust_decimal::Decimal,
output: rust_decimal::Decimal,
cache_read: rust_decimal::Decimal,
cache_creation: rust_decimal::Decimal,
}
impl Database {
fn maybe_backfill_log_costs(
conn: &Connection,
log: &mut RequestLogDetail,
provider_cache: &mut HashMap<(String, String), rust_decimal::Decimal>,
pricing_cache: &mut HashMap<String, PricingInfo>,
) -> Result<(), AppError> {
let total_cost = rust_decimal::Decimal::from_str(&log.total_cost_usd)
.unwrap_or(rust_decimal::Decimal::ZERO);
let has_cost = total_cost > rust_decimal::Decimal::ZERO;
let has_usage = log.input_tokens > 0
|| log.output_tokens > 0
|| log.cache_read_tokens > 0
|| log.cache_creation_tokens > 0;
if has_cost || !has_usage {
return Ok(());
}
let pricing = match Self::get_model_pricing_cached(conn, pricing_cache, &log.model)? {
Some(info) => info,
None => return Ok(()),
};
let multiplier = Self::get_cost_multiplier_cached(
conn,
provider_cache,
&log.provider_id,
&log.app_type,
)?;
let million = rust_decimal::Decimal::from(1_000_000u64);
let input_cost = rust_decimal::Decimal::from(log.input_tokens as u64) * pricing.input
/ million
* multiplier;
let output_cost = rust_decimal::Decimal::from(log.output_tokens as u64) * pricing.output
/ million
* multiplier;
let cache_read_cost = rust_decimal::Decimal::from(log.cache_read_tokens as u64)
* pricing.cache_read
/ million
* multiplier;
let cache_creation_cost = rust_decimal::Decimal::from(log.cache_creation_tokens as u64)
* pricing.cache_creation
/ million
* multiplier;
let total_cost = input_cost + output_cost + cache_read_cost + cache_creation_cost;
log.input_cost_usd = format!("{input_cost:.6}");
log.output_cost_usd = format!("{output_cost:.6}");
log.cache_read_cost_usd = format!("{cache_read_cost:.6}");
log.cache_creation_cost_usd = format!("{cache_creation_cost:.6}");
log.total_cost_usd = format!("{total_cost:.6}");
conn.execute(
"UPDATE proxy_request_logs
SET input_cost_usd = ?1,
output_cost_usd = ?2,
cache_read_cost_usd = ?3,
cache_creation_cost_usd = ?4,
total_cost_usd = ?5
WHERE request_id = ?6",
params![
log.input_cost_usd,
log.output_cost_usd,
log.cache_read_cost_usd,
log.cache_creation_cost_usd,
log.total_cost_usd,
log.request_id
],
)
.map_err(|e| AppError::Database(format!("更新请求成本失败: {e}")))?;
Ok(())
}
fn get_cost_multiplier_cached(
conn: &Connection,
cache: &mut HashMap<(String, String), rust_decimal::Decimal>,
provider_id: &str,
app_type: &str,
) -> Result<rust_decimal::Decimal, AppError> {
let key = (provider_id.to_string(), app_type.to_string());
if let Some(multiplier) = cache.get(&key) {
return Ok(*multiplier);
}
let meta_json: Option<String> = conn
.query_row(
"SELECT meta FROM providers WHERE id = ? AND app_type = ?",
params![provider_id, app_type],
|row| row.get(0),
)
.optional()
.map_err(|e| AppError::Database(format!("查询 provider meta 失败: {e}")))?;
let multiplier = meta_json
.and_then(|meta| serde_json::from_str::<Value>(&meta).ok())
.and_then(|value| value.get("costMultiplier").cloned())
.and_then(|val| {
val.as_str()
.and_then(|s| rust_decimal::Decimal::from_str(s).ok())
})
.unwrap_or(rust_decimal::Decimal::ONE);
cache.insert(key, multiplier);
Ok(multiplier)
}
fn get_model_pricing_cached(
conn: &Connection,
cache: &mut HashMap<String, PricingInfo>,
model: &str,
) -> Result<Option<PricingInfo>, AppError> {
if let Some(info) = cache.get(model) {
return Ok(Some(info.clone()));
}
let row = find_model_pricing_row(conn, model)?;
let Some((input, output, cache_read, cache_creation)) = row else {
return Ok(None);
};
let pricing = PricingInfo {
input: rust_decimal::Decimal::from_str(&input)
.map_err(|e| AppError::Database(format!("解析输入价格失败: {e}")))?,
output: rust_decimal::Decimal::from_str(&output)
.map_err(|e| AppError::Database(format!("解析输出价格失败: {e}")))?,
cache_read: rust_decimal::Decimal::from_str(&cache_read)
.map_err(|e| AppError::Database(format!("解析缓存读取价格失败: {e}")))?,
cache_creation: rust_decimal::Decimal::from_str(&cache_creation)
.map_err(|e| AppError::Database(format!("解析缓存写入价格失败: {e}")))?,
};
cache.insert(model.to_string(), pricing.clone());
Ok(Some(pricing))
}
}
pub(crate) fn find_model_pricing_row(
conn: &Connection,
model_id: &str,
) -> Result<Option<(String, String, String, String)>, AppError> {
// 1) 去除供应商前缀(/ 之前)与冒号后缀(: 之后),例如 moonshotai/kimi-k2-0905:exa → kimi-k2-0905
let without_prefix = model_id
.rsplit_once('/')
.map(|(_, rest)| rest)
.unwrap_or(model_id);
let cleaned = without_prefix
.split(':')
.next()
.map(str::trim)
.unwrap_or(without_prefix);
// 2) 精确匹配清洗后的名称
let exact = conn
.query_row(
"SELECT input_cost_per_million, output_cost_per_million,
cache_read_cost_per_million, cache_creation_cost_per_million
FROM model_pricing
WHERE model_id = ?1",
[cleaned],
|row| {
Ok((
row.get::<_, String>(0)?,
row.get::<_, String>(1)?,
row.get::<_, String>(2)?,
row.get::<_, String>(3)?,
))
},
)
.optional()
.map_err(|e| AppError::Database(format!("查询模型定价失败: {e}")))?;
if exact.is_none() {
log::warn!("模型 {model_id}(清洗后: {cleaned})未找到定价信息,成本将记录为 0");
}
Ok(exact)
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_get_usage_summary() -> Result<(), AppError> {
let db = Database::memory()?;
// 插入测试数据
{
let conn = lock_conn!(db.conn);
conn.execute(
"INSERT INTO proxy_request_logs (
request_id, provider_id, app_type, model,
input_tokens, output_tokens, total_cost_usd,
latency_ms, status_code, created_at
) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
params!["req1", "p1", "claude", "claude-3", 100, 50, "0.01", 100, 200, 1000],
)?;
conn.execute(
"INSERT INTO proxy_request_logs (
request_id, provider_id, app_type, model,
input_tokens, output_tokens, total_cost_usd,
latency_ms, status_code, created_at
) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
params!["req2", "p1", "claude", "claude-3", 200, 100, "0.02", 150, 200, 2000],
)?;
}
let summary = db.get_usage_summary(None, None)?;
assert_eq!(summary.total_requests, 2);
assert_eq!(summary.success_rate, 100.0);
Ok(())
}
#[test]
fn test_get_model_stats() -> Result<(), AppError> {
let db = Database::memory()?;
// 插入测试数据
{
let conn = lock_conn!(db.conn);
conn.execute(
"INSERT INTO proxy_request_logs (
request_id, provider_id, app_type, model,
input_tokens, output_tokens, total_cost_usd,
latency_ms, status_code, created_at
) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
params![
"req1",
"p1",
"claude",
"claude-3-sonnet",
100,
50,
"0.01",
100,
200,
1000
],
)?;
}
let stats = db.get_model_stats()?;
assert_eq!(stats.len(), 1);
assert_eq!(stats[0].model, "claude-3-sonnet");
assert_eq!(stats[0].request_count, 1);
Ok(())
}
#[test]
fn test_model_pricing_matching() -> Result<(), AppError> {
let db = Database::memory()?;
let conn = lock_conn!(db.conn);
// 准备额外定价数据,覆盖前缀/后缀清洗场景
conn.execute(
"INSERT OR REPLACE INTO model_pricing (
model_id, display_name, input_cost_per_million, output_cost_per_million,
cache_read_cost_per_million, cache_creation_cost_per_million
) VALUES (?, ?, ?, ?, ?, ?)",
params![
"claude-haiku-4.5",
"Claude Haiku 4.5",
"1.0",
"2.0",
"0.0",
"0.0"
],
)?;
// 测试精确匹配(seed_model_pricing 已预置 claude-sonnet-4-5-20250929
let result = find_model_pricing_row(&conn, "claude-sonnet-4-5-20250929")?;
assert!(
result.is_some(),
"应该能精确匹配 claude-sonnet-4-5-20250929"
);
// 清洗:去除前缀和冒号后缀
let result = find_model_pricing_row(&conn, "anthropic/claude-haiku-4.5")?;
assert!(
result.is_some(),
"带前缀的模型 anthropic/claude-haiku-4.5 应能匹配到 claude-haiku-4.5"
);
let result = find_model_pricing_row(&conn, "moonshotai/kimi-k2-0905:exa")?;
assert!(
result.is_some(),
"带前缀+冒号后缀的模型应清洗后匹配到 kimi-k2-0905"
);
// 测试不存在的模型
let result = find_model_pricing_row(&conn, "unknown-model-123")?;
assert!(result.is_none(), "不应该匹配不存在的模型");
Ok(())
}
}