test(gateway): cover chat pii redaction formats

This commit is contained in:
Kayphoon
2026-05-26 22:29:45 +08:00
parent a2f91b4108
commit 7530bec7de
18 changed files with 2049 additions and 364 deletions
@@ -17,6 +17,7 @@ mod passthrough;
mod plan_builders;
mod pool_scheduler;
pub(crate) mod pool_scores;
mod redaction;
mod report_context;
mod route;
mod runtime_miss;
@@ -70,7 +70,7 @@ pub(crate) async fn maybe_build_sync_local_same_format_provider_decision_payload
maybe_build_local_same_format_provider_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -137,7 +137,7 @@ pub(crate) async fn maybe_build_stream_local_same_format_provider_decision_paylo
maybe_build_local_same_format_provider_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -22,7 +22,7 @@ use crate::ai_serving::transport::{
};
use crate::{
append_execution_contract_fields_to_value, append_local_failover_policy_to_value,
AiExecutionDecision, AppState,
AiExecutionDecision, AppState, GatewayError,
};
use aether_scheduler_core::SchedulerMinimalCandidateSelectionCandidate;
@@ -40,7 +40,7 @@ pub(crate) async fn maybe_build_local_same_format_provider_decision_payload_for_
input: &LocalSameFormatProviderDecisionInput,
attempt: LocalSameFormatProviderCandidateAttempt,
spec: LocalSameFormatProviderSpec,
) -> Option<AiExecutionDecision> {
) -> Result<Option<AiExecutionDecision>, GatewayError> {
let spec_metadata = local_same_format_provider_spec_metadata(spec);
let LocalSameFormatProviderCandidateAttempt {
eligible,
@@ -51,10 +51,18 @@ pub(crate) async fn maybe_build_local_same_format_provider_decision_payload_for_
let candidate = &eligible.candidate;
let (execution_strategy, conversion_mode) =
ai_local_execution_contract_for_formats(spec_metadata.api_format, spec_metadata.api_format);
let resolved = resolve_local_same_format_provider_candidate_payload_parts(
let Some(resolved) = resolve_local_same_format_provider_candidate_payload_parts(
state, parts, trace_id, body_json, input, &attempt, spec,
)
.await?;
.await?
else {
return Ok(None);
};
let original_request_body_json = if resolved.request_redacted {
Some(&resolved.provider_request_body)
} else {
Some(body_json)
};
let prompt_cache_key = resolved
.provider_request_body
@@ -116,7 +124,7 @@ pub(crate) async fn maybe_build_local_same_format_provider_decision_payload_for_
request_path: Some(parts.uri.path()),
request_query_string: parts.uri.query(),
request_origin: Some(crate::ai_serving::request_origin_from_parts(parts)),
original_request_body_json: Some(body_json),
original_request_body_json,
original_request_body_base64: None,
client_session_affinity: input.client_session_affinity.as_ref(),
scheduler_affinity_epoch: eligible.orchestration.scheduler_affinity_epoch,
@@ -149,9 +157,10 @@ pub(crate) async fn maybe_build_local_same_format_provider_decision_payload_for_
upstream_url,
provider_request_headers,
provider_request_body,
request_redacted: _,
} = resolved;
Some(build_ai_execution_decision_response(
Ok(Some(build_ai_execution_decision_response(
AiExecutionDecisionResponseParts {
decision_is_stream: spec_metadata.require_streaming,
decision_kind: spec_metadata.decision_kind.to_string(),
@@ -185,7 +194,7 @@ pub(crate) async fn maybe_build_local_same_format_provider_decision_payload_for_
report_context: Some(report_context),
auth_context: input.auth_context.clone(),
},
))
)))
}
pub(super) async fn mark_skipped_local_same_format_provider_candidate(
@@ -6,6 +6,9 @@ use serde_json::Value;
use crate::ai_serving::planner::common::{
enforce_provider_body_stream_policy, request_requires_body_stream_field,
};
use crate::ai_serving::planner::redaction::{
request_identity_response_encoding_when_redacted, resolve_provider_chat_pii_redaction,
};
use crate::ai_serving::transport::antigravity::{
build_antigravity_safe_v1internal_request, build_antigravity_static_identity_headers,
classify_local_antigravity_request_support, AntigravityEnvelopeRequestType,
@@ -15,7 +18,7 @@ use crate::ai_serving::transport::{
build_same_format_provider_headers, SameFormatProviderHeadersInput,
};
use crate::ai_serving::{CandidateFailureDiagnostic, GatewayProviderTransportSnapshot};
use crate::AppState;
use crate::{AppState, GatewayError};
mod policy;
mod prepare;
@@ -96,6 +99,7 @@ pub(crate) struct LocalSameFormatProviderCandidatePayloadParts {
pub(super) upstream_url: String,
pub(super) provider_request_headers: BTreeMap<String, String>,
pub(super) provider_request_body: Value,
pub(super) request_redacted: bool,
}
pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
@@ -106,9 +110,9 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
input: &LocalSameFormatProviderDecisionInput,
attempt: &LocalSameFormatProviderCandidateAttempt,
spec: LocalSameFormatProviderSpec,
) -> Option<LocalSameFormatProviderCandidatePayloadParts> {
) -> Result<Option<LocalSameFormatProviderCandidatePayloadParts>, GatewayError> {
let candidate = &attempt.eligible.candidate;
let prepared = prepare_local_same_format_provider_candidate(
let Some(prepared) = prepare_local_same_format_provider_candidate(
state,
trace_id,
input,
@@ -117,7 +121,10 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
&attempt.candidate_id,
spec,
)
.await?;
.await
else {
return Ok(None);
};
let enable_model_directives =
crate::system_features::reasoning_model_directive_enabled_for_api_format_and_model(
state,
@@ -125,6 +132,16 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
Some(&input.requested_model),
)
.await;
let redaction = resolve_provider_chat_pii_redaction(
state,
parts,
body_json,
&input.auth_context,
spec.api_format,
&attempt.candidate_id,
)
.await?;
let body_json = redaction.body_json.as_ref();
let Some(mut base_provider_request_body) =
super::super::request::build_same_format_provider_request_body(
@@ -161,7 +178,7 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
if let Some(mapping) =
crate::system_features::reasoning_model_directive_mapping_for_api_format_and_model(
@@ -207,7 +224,7 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
"transport_unsupported",
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -239,7 +256,7 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -270,14 +287,14 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
let extra_headers = antigravity_auth
.as_ref()
.map(build_antigravity_static_identity_headers)
.unwrap_or_default();
let Some(provider_request_headers) =
let Some(mut provider_request_headers) =
build_same_format_provider_headers(SameFormatProviderHeadersInput {
headers: &parts.headers,
provider_request_body: &provider_request_body,
@@ -310,10 +327,14 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
request_identity_response_encoding_when_redacted(
&mut provider_request_headers,
redaction.redacted,
);
Some(LocalSameFormatProviderCandidatePayloadParts {
Ok(Some(LocalSameFormatProviderCandidatePayloadParts {
transport: prepared.transport,
is_antigravity: prepared.is_antigravity,
is_kiro: prepared.is_kiro,
@@ -326,5 +347,6 @@ pub(crate) async fn resolve_local_same_format_provider_candidate_payload_parts(
upstream_url,
provider_request_headers,
provider_request_body,
})
request_redacted: redaction.redacted,
}))
}
@@ -249,7 +249,7 @@ impl LocalSameFormatProviderSyncAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -287,7 +287,7 @@ impl LocalSameFormatProviderStreamAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -365,7 +365,7 @@ pub(crate) async fn build_local_sync_plan_and_reports(
let Some(payload) = maybe_build_local_same_format_provider_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};
@@ -450,7 +450,7 @@ pub(crate) async fn build_local_stream_plan_and_reports(
let Some(payload) = maybe_build_local_same_format_provider_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};
@@ -0,0 +1,192 @@
use std::borrow::Cow;
use std::time::{SystemTime, UNIX_EPOCH};
use serde_json::Value;
use tracing::warn;
use crate::ai_serving::ExecutionRuntimeAuthContext;
use crate::privacy::{
build_redaction_session_config, read_chat_pii_redaction_runtime_config,
try_mask_chat_pii_request_json_with_cache_options, ChatPiiRedactionRequestFormat,
MaskChatRequestOptions, RedactionMaskError, RedactionSessionSlot, RedisRedactionMappingCache,
};
use crate::{AppState, GatewayError};
pub(crate) struct ProviderRequestRedaction<'a> {
pub(crate) body_json: Cow<'a, Value>,
pub(crate) redacted: bool,
}
impl<'a> ProviderRequestRedaction<'a> {
fn disabled(body_json: &'a Value) -> Self {
Self {
body_json: Cow::Borrowed(body_json),
redacted: false,
}
}
}
#[derive(Clone, Copy, Debug, Default)]
struct ChatPiiRedactionFeatureSettings {
enabled: Option<bool>,
inject_model_instruction: Option<bool>,
}
impl ChatPiiRedactionFeatureSettings {
fn merge_from_value(&mut self, value: Option<&Value>) {
let Some(settings) = value
.and_then(Value::as_object)
.and_then(|features| features.get("chat_pii_redaction"))
.and_then(Value::as_object)
else {
return;
};
if let Some(enabled) = settings.get("enabled").and_then(Value::as_bool) {
self.enabled = Some(enabled);
}
if let Some(inject_model_instruction) = settings
.get("inject_model_instruction")
.and_then(Value::as_bool)
{
self.inject_model_instruction = Some(inject_model_instruction);
}
}
fn effective_enabled(self) -> bool {
self.enabled.unwrap_or(false)
}
fn effective_inject_model_instruction(self) -> bool {
self.inject_model_instruction.unwrap_or(true)
}
}
pub(crate) fn request_identity_response_encoding_when_redacted(
headers: &mut std::collections::BTreeMap<String, String>,
redacted: bool,
) {
if redacted {
headers.insert("accept-encoding".to_string(), "identity".to_string());
}
}
pub(crate) async fn resolve_provider_chat_pii_redaction<'a>(
state: &AppState,
parts: &http::request::Parts,
body_json: &'a Value,
auth_context: &ExecutionRuntimeAuthContext,
client_api_format: &str,
candidate_id: &str,
) -> Result<ProviderRequestRedaction<'a>, GatewayError> {
let Some(format) = ChatPiiRedactionRequestFormat::from_api_format(client_api_format) else {
return Ok(ProviderRequestRedaction::disabled(body_json));
};
let Some(slot) = parts.extensions.get::<RedactionSessionSlot>() else {
return Ok(ProviderRequestRedaction::disabled(body_json));
};
let runtime_config = read_chat_pii_redaction_runtime_config(state)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read chat pii redaction runtime config"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
if !runtime_config.enabled {
return Ok(ProviderRequestRedaction::disabled(body_json));
}
let feature_settings = resolve_chat_pii_redaction_feature_settings(state, auth_context).await?;
if !feature_settings.effective_enabled() {
return Ok(ProviderRequestRedaction::disabled(body_json));
}
let Some(hmac_key) = state.encryption_key().map(str::as_bytes).map(Vec::from) else {
warn!("gateway chat pii redaction is enabled but encryption key is unavailable");
return Err(GatewayError::Internal(
"chat pii redaction setup failed".to_string(),
));
};
let body_bytes = serde_json::to_vec(body_json).map_err(|err| {
warn!(
error = ?err,
"gateway failed to serialize provider chat pii redaction body"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let now_unix_secs = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap_or_default()
.as_secs();
let cache = RedisRedactionMappingCache::new(state.runtime_state.as_ref());
let masked = try_mask_chat_pii_request_json_with_cache_options(
&body_bytes,
format,
build_redaction_session_config(hmac_key, &runtime_config, now_unix_secs),
MaskChatRequestOptions::runtime(feature_settings.effective_inject_model_instruction()),
Some(&cache),
)
.await
.map_err(redaction_mask_error_to_gateway_error)?;
if !masked.redacted {
return Ok(ProviderRequestRedaction {
body_json: Cow::Borrowed(body_json),
redacted: false,
});
}
let masked_body_json = serde_json::from_slice::<Value>(&masked.body).map_err(|err| {
warn!(
error = ?err,
"gateway failed to decode redacted provider chat pii body"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
slot.put_for_candidate(candidate_id, masked.session);
Ok(ProviderRequestRedaction {
body_json: Cow::Owned(masked_body_json),
redacted: true,
})
}
async fn resolve_chat_pii_redaction_feature_settings(
state: &AppState,
auth_context: &ExecutionRuntimeAuthContext,
) -> Result<ChatPiiRedactionFeatureSettings, GatewayError> {
let user_settings = state
.read_user_feature_settings(&auth_context.user_id)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read user chat pii redaction feature settings"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let key_settings = state
.read_auth_api_key_feature_settings(
&auth_context.user_id,
&auth_context.api_key_id,
auth_context.api_key_is_standalone,
)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read api key chat pii redaction feature settings"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let mut settings = ChatPiiRedactionFeatureSettings::default();
settings.merge_from_value(user_settings.as_ref());
settings.merge_from_value(key_settings.as_ref());
Ok(settings)
}
fn redaction_mask_error_to_gateway_error(error: RedactionMaskError) -> GatewayError {
match error {
RedactionMaskError::Limit(limit) => GatewayError::Client {
status: limit.client_status(),
message: limit.safe_message().to_string(),
},
}
}
@@ -233,7 +233,7 @@ impl LocalStandardSyncAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -270,7 +270,7 @@ impl LocalStandardStreamAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -331,7 +331,7 @@ pub(crate) async fn maybe_build_sync_via_standard_family_payload(
if let Some(payload) = maybe_build_local_standard_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -380,7 +380,7 @@ pub(crate) async fn maybe_build_stream_via_standard_family_payload(
if let Some(payload) = maybe_build_local_standard_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -438,7 +438,7 @@ pub(crate) async fn build_local_sync_plan_and_reports(
let Some(payload) = maybe_build_local_standard_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};
@@ -512,7 +512,7 @@ pub(crate) async fn build_local_stream_plan_and_reports(
let Some(payload) = maybe_build_local_standard_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};
@@ -24,7 +24,7 @@ use crate::ai_serving::{
};
use crate::{
append_execution_contract_fields_to_value, append_local_failover_policy_to_value,
AiExecutionDecision, AppState,
AiExecutionDecision, AppState, GatewayError,
};
use super::request::resolve_local_standard_candidate_payload_parts;
@@ -38,7 +38,7 @@ pub(super) async fn maybe_build_local_standard_decision_payload_for_candidate(
input: &LocalStandardDecisionInput,
attempt: LocalStandardCandidateAttempt,
spec: LocalStandardSpec,
) -> Option<AiExecutionDecision> {
) -> Result<Option<AiExecutionDecision>, GatewayError> {
let spec_metadata = local_standard_spec_metadata(spec);
if api_format_alias_matches(
&attempt.eligible.provider_api_format,
@@ -70,10 +70,18 @@ pub(super) async fn maybe_build_local_standard_decision_payload_for_candidate(
..
} = &attempt;
let candidate = &eligible.candidate;
let resolved = resolve_local_standard_candidate_payload_parts(
let Some(resolved) = resolve_local_standard_candidate_payload_parts(
state, parts, trace_id, body_json, input, &attempt, spec,
)
.await?;
.await?
else {
return Ok(None);
};
let original_request_body_json = if resolved.request_redacted {
Some(&resolved.provider_request_body)
} else {
Some(body_json)
};
let proxy = state
.resolve_transport_proxy_snapshot_with_tunnel_affinity(&resolved.transport)
.await;
@@ -124,7 +132,7 @@ pub(super) async fn maybe_build_local_standard_decision_payload_for_candidate(
request_path: Some(parts.uri.path()),
request_query_string: parts.uri.query(),
request_origin: Some(crate::ai_serving::request_origin_from_parts(parts)),
original_request_body_json: Some(body_json),
original_request_body_json,
original_request_body_base64: None,
client_session_affinity: input.client_session_affinity.as_ref(),
scheduler_affinity_epoch: eligible.orchestration.scheduler_affinity_epoch,
@@ -157,9 +165,10 @@ pub(super) async fn maybe_build_local_standard_decision_payload_for_candidate(
upstream_is_stream,
envelope_name: _,
transport,
request_redacted: _,
} = resolved;
Some(build_ai_execution_decision_response(
Ok(Some(build_ai_execution_decision_response(
AiExecutionDecisionResponseParts {
decision_is_stream: spec_metadata.require_streaming,
decision_kind: spec_metadata.decision_kind.to_string(),
@@ -193,7 +202,7 @@ pub(super) async fn maybe_build_local_standard_decision_payload_for_candidate(
report_context: Some(report_context),
auth_context: input.auth_context.clone(),
},
))
)))
}
pub(super) async fn mark_skipped_local_standard_candidate(
@@ -512,6 +521,7 @@ mod tests {
claude_stream_spec(),
)
.await
.expect("same-format candidate should not error")
.expect("same-format candidate should build a standard-family payload");
assert_eq!(payload.endpoint_id.as_deref(), Some("endpoint-claude"));
@@ -547,6 +557,7 @@ mod tests {
claude_stream_spec(),
)
.await
.expect("cross-format candidate should not error")
.expect("cross-format candidate should still build after the same-format candidate");
assert_eq!(
@@ -11,6 +11,9 @@ use crate::ai_serving::planner::common::{
endpoint_config_forces_body_stream_field, enforce_provider_body_stream_policy,
request_requires_body_stream_field, resolve_upstream_is_stream_for_provider,
};
use crate::ai_serving::planner::redaction::{
request_identity_response_encoding_when_redacted, resolve_provider_chat_pii_redaction,
};
use crate::ai_serving::planner::spec_metadata::local_standard_spec_metadata;
use crate::ai_serving::planner::standard::{
apply_codex_openai_responses_special_headers, request_body_build_failure_extra_data,
@@ -30,7 +33,7 @@ use crate::ai_serving::{
build_openai_image_request_body_from_gemini_image_request, gemini_request_is_image_generation,
CandidateFailureDiagnostic, GatewayProviderTransportSnapshot, LocalResolvedOAuthRequestAuth,
};
use crate::AppState;
use crate::{AppState, GatewayError};
use super::payload::{
mark_skipped_local_standard_candidate, mark_skipped_local_standard_candidate_with_extra_data,
@@ -49,6 +52,7 @@ pub(crate) struct LocalStandardCandidatePayloadParts {
pub(super) upstream_is_stream: bool,
pub(super) envelope_name: Option<&'static str>,
pub(super) transport: Arc<GatewayProviderTransportSnapshot>,
pub(super) request_redacted: bool,
}
pub(crate) async fn resolve_local_standard_candidate_payload_parts(
@@ -59,7 +63,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
input: &LocalStandardDecisionInput,
attempt: &LocalStandardCandidateAttempt,
spec: LocalStandardSpec,
) -> Option<LocalStandardCandidatePayloadParts> {
) -> Result<Option<LocalStandardCandidatePayloadParts>, GatewayError> {
let spec_metadata = local_standard_spec_metadata(spec);
let planner_state = crate::ai_serving::PlannerAppState::new(state);
let candidate = &attempt.eligible.candidate;
@@ -69,16 +73,18 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
&& provider_api_format == "openai:image"
&& gemini_request_is_image_generation(body_json)
{
return resolve_local_gemini_image_to_openai_image_candidate_payload_parts(
state, parts, trace_id, body_json, input, attempt,
)
.await;
return Ok(
resolve_local_gemini_image_to_openai_image_candidate_payload_parts(
state, parts, trace_id, body_json, input, attempt,
)
.await,
);
}
let is_kiro_claude_cli = is_kiro_claude_messages_transport(transport, provider_api_format);
let Some(conversion_kind) =
crate::ai_serving::request_conversion_kind(spec_metadata.api_format, provider_api_format)
else {
return None;
return Ok(None);
};
if let Some(skip_reason) = crate::ai_serving::request_conversion_transport_unsupported_reason(
@@ -95,7 +101,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
let oauth_context = OauthPreparationContext {
@@ -123,7 +129,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
"transport_auth_unavailable",
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -148,7 +154,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -173,7 +179,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
}
};
@@ -194,6 +200,16 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
Some(&input.requested_model),
)
.await;
let redaction = resolve_provider_chat_pii_redaction(
state,
parts,
body_json,
&input.auth_context,
spec_metadata.api_format,
&attempt.candidate_id,
)
.await?;
let body_json = redaction.body_json.as_ref();
let mut provider_request_body =
match crate::ai_serving::planner::standard::build_standard_request_body_with_model_directives_and_request_headers(
body_json,
@@ -229,7 +245,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
};
enforce_provider_body_stream_policy(
@@ -261,7 +277,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
}
if let Some(kiro_auth) = kiro_auth.as_ref() {
return build_kiro_cross_format_payload_parts(
return Ok(build_kiro_cross_format_payload_parts(
state,
parts,
trace_id,
@@ -276,8 +292,9 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
provider_request_body,
upstream_is_stream,
kiro_auth,
redaction.redacted,
)
.await;
.await);
}
let upstream_url = match crate::ai_serving::planner::standard::build_standard_upstream_url(
@@ -304,7 +321,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
};
let Some(resolved_headers) =
@@ -337,7 +354,7 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
let mut provider_request_headers = resolved_headers.headers;
apply_codex_openai_responses_special_headers(
@@ -349,8 +366,12 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
Some(trace_id),
transport.key.decrypted_auth_config.as_deref(),
);
request_identity_response_encoding_when_redacted(
&mut provider_request_headers,
redaction.redacted,
);
Some(LocalStandardCandidatePayloadParts {
Ok(Some(LocalStandardCandidatePayloadParts {
auth_header: resolved_headers.auth_header,
auth_value: resolved_headers.auth_value,
mapped_model: prepared_candidate.mapped_model,
@@ -361,7 +382,8 @@ pub(crate) async fn resolve_local_standard_candidate_payload_parts(
upstream_is_stream,
envelope_name: None,
transport: Arc::clone(transport),
})
request_redacted: redaction.redacted,
}))
}
async fn resolve_local_gemini_image_to_openai_image_candidate_payload_parts(
@@ -496,6 +518,7 @@ async fn resolve_local_gemini_image_to_openai_image_candidate_payload_parts(
upstream_is_stream,
envelope_name: None,
transport: Arc::clone(transport),
request_redacted: false,
})
}
@@ -515,6 +538,7 @@ async fn build_kiro_cross_format_payload_parts(
claude_request_body: Value,
upstream_is_stream: bool,
kiro_auth: &KiroRequestAuth,
request_redacted: bool,
) -> Option<LocalStandardCandidatePayloadParts> {
let candidate = &attempt.eligible.candidate;
let provider_request_body = match build_kiro_provider_request_body(
@@ -572,7 +596,7 @@ async fn build_kiro_cross_format_payload_parts(
return None;
}
};
let provider_request_headers = match build_kiro_provider_headers(KiroProviderHeadersInput {
let mut provider_request_headers = match build_kiro_provider_headers(KiroProviderHeadersInput {
headers: &parts.headers,
provider_request_body: &provider_request_body,
original_request_body: original_body_json,
@@ -602,6 +626,10 @@ async fn build_kiro_cross_format_payload_parts(
return None;
}
};
request_identity_response_encoding_when_redacted(
&mut provider_request_headers,
request_redacted,
);
Some(LocalStandardCandidatePayloadParts {
auth_header,
@@ -614,5 +642,6 @@ async fn build_kiro_cross_format_payload_parts(
upstream_is_stream,
envelope_name: Some(KIRO_ENVELOPE_NAME),
transport: Arc::clone(transport),
request_redacted,
})
}
@@ -1,7 +1,5 @@
use std::borrow::Cow;
use std::collections::BTreeMap;
use std::sync::Arc;
use std::time::{SystemTime, UNIX_EPOCH};
use serde_json::Value;
@@ -14,6 +12,9 @@ use crate::ai_serving::planner::common::{
endpoint_config_forces_body_stream_field, enforce_provider_body_stream_policy,
request_requires_body_stream_field, OPENAI_CHAT_STREAM_PLAN_KIND,
};
use crate::ai_serving::planner::redaction::{
request_identity_response_encoding_when_redacted, resolve_provider_chat_pii_redaction,
};
use crate::ai_serving::planner::standard::{
apply_codex_openai_responses_special_headers, build_cross_format_openai_chat_request_body,
build_cross_format_openai_chat_upstream_url, build_local_openai_chat_request_body,
@@ -36,13 +37,7 @@ use crate::ai_serving::{
LocalResolvedOAuthRequestAuth,
};
use crate::ai_serving::{ConversionMode, ExecutionStrategy};
use crate::privacy::{
build_redaction_session_config, read_chat_pii_redaction_runtime_config,
try_mask_chat_request_json_with_cache_options, MaskChatRequestOptions, RedactionMaskError,
RedactionSessionSlot, RedisRedactionMappingCache,
};
use crate::{AppState, GatewayError};
use tracing::warn;
use super::support::{
mark_skipped_local_openai_chat_candidate,
@@ -66,99 +61,6 @@ pub(crate) struct LocalOpenAiChatCandidatePayloadParts {
pub(super) request_redacted: bool,
}
fn request_identity_response_encoding_when_redacted(
headers: &mut BTreeMap<String, String>,
redacted: bool,
) {
if redacted {
headers.insert("accept-encoding".to_string(), "identity".to_string());
}
}
struct ProviderChatRequestRedaction<'a> {
body_json: Cow<'a, Value>,
redacted: bool,
}
impl<'a> ProviderChatRequestRedaction<'a> {
fn disabled(body_json: &'a Value, _parts: &http::request::Parts) -> Self {
Self {
body_json: Cow::Borrowed(body_json),
redacted: false,
}
}
}
#[derive(Clone, Copy, Debug, Default)]
struct ChatPiiRedactionFeatureSettings {
enabled: Option<bool>,
inject_model_instruction: Option<bool>,
}
impl ChatPiiRedactionFeatureSettings {
fn merge_from_value(&mut self, value: Option<&Value>) {
let Some(settings) = value
.and_then(Value::as_object)
.and_then(|features| features.get("chat_pii_redaction"))
.and_then(Value::as_object)
else {
return;
};
if let Some(enabled) = settings.get("enabled").and_then(Value::as_bool) {
self.enabled = Some(enabled);
}
if let Some(inject_model_instruction) = settings
.get("inject_model_instruction")
.and_then(Value::as_bool)
{
self.inject_model_instruction = Some(inject_model_instruction);
}
}
fn effective_enabled(self) -> bool {
self.enabled.unwrap_or(false)
}
fn effective_inject_model_instruction(self) -> bool {
self.inject_model_instruction.unwrap_or(true)
}
}
async fn resolve_chat_pii_redaction_feature_settings(
state: &AppState,
input: &LocalOpenAiChatDecisionInput,
) -> Result<ChatPiiRedactionFeatureSettings, GatewayError> {
let user_settings = state
.read_user_feature_settings(&input.auth_context.user_id)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read user chat pii redaction feature settings"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let key_settings = state
.read_auth_api_key_feature_settings(
&input.auth_context.user_id,
&input.auth_context.api_key_id,
input.auth_context.api_key_is_standalone,
)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read api key chat pii redaction feature settings"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let mut settings = ChatPiiRedactionFeatureSettings::default();
settings.merge_from_value(user_settings.as_ref());
settings.merge_from_value(key_settings.as_ref());
Ok(settings)
}
#[allow(clippy::too_many_arguments)]
pub(crate) async fn resolve_local_openai_chat_candidate_payload_parts(
state: &AppState,
@@ -186,9 +88,15 @@ pub(crate) async fn resolve_local_openai_chat_candidate_payload_parts(
Some(&input.requested_model),
)
.await;
let redaction =
resolve_provider_chat_request_redaction(state, parts, body_json, input, candidate_id)
.await?;
let redaction = resolve_provider_chat_pii_redaction(
state,
parts,
body_json,
&input.auth_context,
"openai:chat",
candidate_id,
)
.await?;
let body_json = redaction.body_json.as_ref();
if provider_api_format == "openai:chat" {
@@ -777,87 +685,3 @@ async fn build_kiro_openai_chat_cross_format_payload_parts(
request_redacted,
})
}
async fn resolve_provider_chat_request_redaction<'a>(
state: &AppState,
parts: &http::request::Parts,
body_json: &'a Value,
input: &LocalOpenAiChatDecisionInput,
candidate_id: &str,
) -> Result<ProviderChatRequestRedaction<'a>, GatewayError> {
if parts.uri.path() != "/v1/chat/completions" {
return Ok(ProviderChatRequestRedaction::disabled(body_json, parts));
}
let Some(slot) = parts.extensions.get::<RedactionSessionSlot>() else {
return Ok(ProviderChatRequestRedaction::disabled(body_json, parts));
};
let runtime_config = read_chat_pii_redaction_runtime_config(state)
.await
.map_err(|err| {
warn!(
error = ?err,
"gateway failed to read chat pii redaction runtime config"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
if !runtime_config.enabled {
return Ok(ProviderChatRequestRedaction::disabled(body_json, parts));
}
let feature_settings = resolve_chat_pii_redaction_feature_settings(state, input).await?;
if !feature_settings.effective_enabled() {
return Ok(ProviderChatRequestRedaction::disabled(body_json, parts));
}
let Some(hmac_key) = state.encryption_key().map(str::as_bytes).map(Vec::from) else {
warn!("gateway chat pii redaction is enabled but encryption key is unavailable");
return Err(GatewayError::Internal(
"chat pii redaction setup failed".to_string(),
));
};
let body_bytes = serde_json::to_vec(body_json).map_err(|err| {
warn!(
error = ?err,
"gateway failed to serialize provider chat pii redaction body"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
let now_unix_secs = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap_or_default()
.as_secs();
let cache = RedisRedactionMappingCache::new(state.runtime_state.as_ref());
let masked = try_mask_chat_request_json_with_cache_options(
&body_bytes,
build_redaction_session_config(hmac_key, &runtime_config, now_unix_secs),
MaskChatRequestOptions::runtime(feature_settings.effective_inject_model_instruction()),
Some(&cache),
)
.await
.map_err(redaction_mask_error_to_gateway_error)?;
if !masked.redacted {
return Ok(ProviderChatRequestRedaction {
body_json: Cow::Borrowed(body_json),
redacted: false,
});
}
let masked_body_json = serde_json::from_slice::<Value>(&masked.body).map_err(|err| {
warn!(
error = ?err,
"gateway failed to decode redacted provider chat pii body"
);
GatewayError::Internal("chat pii redaction setup failed".to_string())
})?;
slot.put_for_candidate(candidate_id, masked.session);
Ok(ProviderChatRequestRedaction {
body_json: Cow::Owned(masked_body_json),
redacted: true,
})
}
fn redaction_mask_error_to_gateway_error(error: RedactionMaskError) -> GatewayError {
match error {
RedactionMaskError::Limit(limit) => GatewayError::Client {
status: limit.client_status(),
message: limit.safe_message().to_string(),
},
}
}
@@ -15,7 +15,7 @@ use crate::ai_serving::transport::{
};
use crate::{
append_execution_contract_fields_to_value, append_local_failover_policy_to_value,
AiExecutionDecision, AppState,
AiExecutionDecision, AppState, GatewayError,
};
use super::request::resolve_local_openai_responses_candidate_payload_parts;
@@ -30,7 +30,7 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
input: &LocalOpenAiResponsesDecisionInput,
attempt: LocalOpenAiResponsesCandidateAttempt,
spec: LocalOpenAiResponsesSpec,
) -> Option<AiExecutionDecision> {
) -> Result<Option<AiExecutionDecision>, GatewayError> {
let spec_metadata = local_openai_responses_spec_metadata(spec);
let attempt_identity = attempt.attempt_identity();
let LocalOpenAiResponsesCandidateAttempt {
@@ -39,7 +39,7 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
candidate_id,
..
} = attempt;
let resolved = resolve_local_openai_responses_candidate_payload_parts(
let Some(resolved) = resolve_local_openai_responses_candidate_payload_parts(
state,
parts,
trace_id,
@@ -50,8 +50,16 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
&candidate_id,
spec,
)
.await?;
.await?
else {
return Ok(None);
};
let candidate = &eligible.candidate;
let original_request_body_json = if resolved.request_redacted {
Some(&resolved.provider_request_body)
} else {
Some(body_json)
};
let prompt_cache_key = resolved
.provider_request_body
@@ -109,7 +117,7 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
request_path: Some(parts.uri.path()),
request_query_string: parts.uri.query(),
request_origin: Some(crate::ai_serving::request_origin_from_parts(parts)),
original_request_body_json: Some(body_json),
original_request_body_json,
original_request_body_base64: None,
client_session_affinity: input.client_session_affinity.as_ref(),
scheduler_affinity_epoch: eligible.orchestration.scheduler_affinity_epoch,
@@ -170,9 +178,10 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
envelope_name: _,
upstream_is_stream,
transport,
request_redacted: _,
} = resolved;
Some(build_ai_execution_decision_response(
Ok(Some(build_ai_execution_decision_response(
AiExecutionDecisionResponseParts {
decision_is_stream: spec_metadata.require_streaming,
decision_kind: spec_metadata.decision_kind.to_string(),
@@ -206,5 +215,5 @@ pub(crate) async fn maybe_build_local_openai_responses_decision_payload_for_cand
report_context: Some(report_context),
auth_context: input.auth_context.clone(),
},
))
)))
}
@@ -13,6 +13,9 @@ use crate::ai_serving::planner::common::{
endpoint_config_forces_body_stream_field, enforce_provider_body_stream_policy,
request_requires_body_stream_field, resolve_upstream_is_stream_for_provider,
};
use crate::ai_serving::planner::redaction::{
request_identity_response_encoding_when_redacted, resolve_provider_chat_pii_redaction,
};
use crate::ai_serving::planner::spec_metadata::local_openai_responses_spec_metadata;
use crate::ai_serving::planner::standard::{
apply_codex_openai_responses_special_headers, build_cross_format_openai_responses_request_body,
@@ -44,7 +47,7 @@ use crate::ai_serving::{
LocalResolvedOAuthRequestAuth, PlannerAppState,
};
use crate::ai_serving::{ConversionMode, ExecutionStrategy};
use crate::AppState;
use crate::{AppState, GatewayError};
use super::support::{
mark_skipped_local_openai_responses_candidate,
@@ -70,6 +73,7 @@ pub(crate) struct LocalOpenAiResponsesCandidatePayloadParts {
pub(super) envelope_name: Option<&'static str>,
pub(super) upstream_is_stream: bool,
pub(super) transport: Arc<GatewayProviderTransportSnapshot>,
pub(super) request_redacted: bool,
}
#[allow(clippy::too_many_arguments)]
@@ -83,7 +87,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
candidate_index: u32,
candidate_id: &str,
spec: LocalOpenAiResponsesSpec,
) -> Option<LocalOpenAiResponsesCandidatePayloadParts> {
) -> Result<Option<LocalOpenAiResponsesCandidatePayloadParts>, GatewayError> {
let spec_metadata = local_openai_responses_spec_metadata(spec);
let client_api_format = spec_metadata.api_format.trim().to_ascii_lowercase();
let planner_state = PlannerAppState::new(state);
@@ -119,7 +123,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
let oauth_context = OauthPreparationContext {
@@ -147,7 +151,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
"transport_auth_unavailable",
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -186,7 +190,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -211,7 +215,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
skip_reason,
)
.await;
return None;
return Ok(None);
}
}
};
@@ -225,6 +229,16 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
Some(&input.requested_model),
)
.await;
let redaction = resolve_provider_chat_pii_redaction(
state,
parts,
body_json,
&input.auth_context,
spec_metadata.api_format,
candidate_id,
)
.await?;
let body_json = redaction.body_json.as_ref();
let needs_bidirectional_conversion = !same_format && conversion_kind.is_some();
let upstream_is_stream = resolve_upstream_is_stream_for_provider(
@@ -287,7 +301,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
if let Some(mapping) =
crate::system_features::reasoning_model_directive_mapping_for_api_format_and_model(
@@ -328,7 +342,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
"transport_unsupported",
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -359,7 +373,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
}
} else {
@@ -386,6 +400,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
upstream_is_stream,
needs_bidirectional_conversion,
kiro_auth,
redaction.redacted,
)
.await;
}
@@ -421,7 +436,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
let extra_headers = antigravity_auth
.as_ref()
@@ -457,7 +472,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
),
)
.await;
return None;
return Ok(None);
};
let mut provider_request_headers = resolved_headers.headers;
apply_codex_openai_responses_special_headers(
@@ -469,6 +484,10 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
Some(trace_id),
transport.key.decrypted_auth_config.as_deref(),
);
request_identity_response_encoding_when_redacted(
&mut provider_request_headers,
redaction.redacted,
);
let (execution_strategy, conversion_mode) =
ai_local_execution_contract_for_formats(spec_metadata.api_format, provider_api_format);
@@ -497,7 +516,7 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
"gateway resolved local openai responses upstream url"
);
Some(LocalOpenAiResponsesCandidatePayloadParts {
Ok(Some(LocalOpenAiResponsesCandidatePayloadParts {
auth_header: resolved_headers.auth_header,
auth_value: resolved_headers.auth_value,
mapped_model,
@@ -516,7 +535,8 @@ pub(crate) async fn resolve_local_openai_responses_candidate_payload_parts(
},
upstream_is_stream,
transport: Arc::clone(transport),
})
request_redacted: redaction.redacted,
}))
}
fn api_format_alias_matches(left: &str, right: &str) -> bool {
@@ -543,7 +563,8 @@ async fn build_kiro_openai_responses_payload_parts(
upstream_is_stream: bool,
needs_bidirectional_conversion: bool,
kiro_auth: &KiroRequestAuth,
) -> Option<LocalOpenAiResponsesCandidatePayloadParts> {
request_redacted: bool,
) -> Result<Option<LocalOpenAiResponsesCandidatePayloadParts>, GatewayError> {
let candidate = &eligible.candidate;
let provider_request_body = match build_kiro_provider_request_body(
&claude_request_body,
@@ -569,7 +590,7 @@ async fn build_kiro_openai_responses_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
};
let upstream_url = match build_kiro_cross_format_upstream_url(
@@ -597,10 +618,10 @@ async fn build_kiro_openai_responses_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
};
let provider_request_headers = match build_kiro_provider_headers(KiroProviderHeadersInput {
let mut provider_request_headers = match build_kiro_provider_headers(KiroProviderHeadersInput {
headers: &parts.headers,
provider_request_body: &provider_request_body,
original_request_body: original_body_json,
@@ -627,7 +648,7 @@ async fn build_kiro_openai_responses_payload_parts(
),
)
.await;
return None;
return Ok(None);
}
};
let (execution_strategy, conversion_mode) =
@@ -652,7 +673,12 @@ async fn build_kiro_openai_responses_payload_parts(
"gateway resolved local openai responses kiro upstream url"
);
Some(LocalOpenAiResponsesCandidatePayloadParts {
request_identity_response_encoding_when_redacted(
&mut provider_request_headers,
request_redacted,
);
Ok(Some(LocalOpenAiResponsesCandidatePayloadParts {
auth_header,
auth_value,
mapped_model,
@@ -666,5 +692,6 @@ async fn build_kiro_openai_responses_payload_parts(
envelope_name: Some(KIRO_ENVELOPE_NAME),
upstream_is_stream,
transport: Arc::clone(transport),
})
request_redacted,
}))
}
@@ -117,7 +117,7 @@ pub(crate) async fn maybe_build_sync_local_openai_responses_decision_payload(
if let Some(payload) = maybe_build_local_openai_responses_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -155,7 +155,7 @@ pub(crate) async fn maybe_build_stream_local_openai_responses_decision_payload(
if let Some(payload) = maybe_build_local_openai_responses_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
{
return Ok(Some(payload));
}
@@ -219,7 +219,7 @@ impl LocalOpenAiResponsesSyncAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -257,7 +257,7 @@ impl LocalOpenAiResponsesStreamAttemptSource<'_> {
attempt,
self.spec,
)
.await
.await?
else {
return Ok(None);
};
@@ -325,7 +325,7 @@ pub(super) async fn build_local_sync_plan_and_reports(
let Some(payload) = maybe_build_local_openai_responses_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};
@@ -397,7 +397,7 @@ pub(super) async fn build_local_stream_plan_and_reports(
let Some(payload) = maybe_build_local_openai_responses_decision_payload_for_candidate(
state, parts, trace_id, body_json, &input, attempt, spec,
)
.await
.await?
else {
continue;
};