orcs-oss/orcs-code
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(api): expose cache metrics in REPL + normalize across providers (#813)

Browse Source 
* feat(api): expose cache metrics in REPL + /cache-stats command

* fix(api): normalize Kimi/DeepSeek/Gemini cache fields through shim layer

* test(api): cover /cache-stats rendering + fix CacheMetrics docstring drift

* fix(api): always reset cache turn counter + include date in /cache-stats rows

* refactor(api): unify shim usage builder + add cost-tracker wiring test

* fix(api): classify private-IP/self-hosted OpenAI endpoints as N/A instead of cold

* fix(api): require colon guard on IPv6 ULA prefix to avoid public-host over-match

* perf(api): ring buffer for cache history + hit rate clamp + .localhost TLD

* fix(api): null guards on formatters + document Codex Responses API shape

* fix(api): defensive start-of-turn reset + config gate fallback + env var docs

* fix(api): trust forwarded cache data on self-hosted URLs (data-driven)

* refactor(api): delegate streaming Responses usage to shared makeUsage helper
This commit is contained in:
viudes
2026-04-25 01:38:25 -03:00
committed by GitHub
parent 9070220292
commit 9e23c2bec4
20 changed files with 2749 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
src/services/api/openaiShim.ts
View File

@@ -46,6 +46,7 @@ import {
type AnthropicUsage,
type ShimCreateParams,
} from './codexShim.js'
import { buildAnthropicUsageFromRawUsage } from './cacheMetrics.js'
import { compressToolHistory } from './compressToolHistory.js'
import { fetchWithProxyRetry } from './fetchWithProxyRetry.js'
import {
@@ -845,16 +846,12 @@ function convertChunkUsage(
usage: OpenAIStreamChunk['usage'] | undefined,
): Partial<AnthropicUsage> | undefined {
if (!usage) return undefined
const cached = usage.prompt_tokens_details?.cached_tokens ?? 0
return {
// Subtract cached tokens: OpenAI includes them in prompt_tokens,
// but Anthropic convention treats input_tokens as non-cached only.
input_tokens: (usage.prompt_tokens ?? 0) - cached,
output_tokens: usage.completion_tokens ?? 0,
cache_creation_input_tokens: 0,
cache_read_input_tokens: cached,
}
// Delegates to the shared helper so this path, codexShim.makeUsage,
// the non-streaming response below, and the integration tests all
// produce byte-identical output for the same raw input.
return buildAnthropicUsageFromRawUsage(
usage as unknown as Record<string, unknown>,
)
}
const JSON_REPAIR_SUFFIXES = [
@@ -2154,12 +2151,9 @@ class OpenAIShimMessages {
model: data.model ?? model,
stop_reason: stopReason,
stop_sequence: null,
usage: {
input_tokens: data.usage?.prompt_tokens ?? 0,
output_tokens: data.usage?.completion_tokens ?? 0,
cache_creation_input_tokens: 0,
cache_read_input_tokens: data.usage?.prompt_tokens_details?.cached_tokens ?? 0,
},
usage: buildAnthropicUsageFromRawUsage(
data.usage as unknown as Record<string, unknown> | undefined,
),
}
}
}