claude-code-proxy/proxy/internal/handler/utils.go

217 lines
5.6 KiB
Go
Raw Normal View History

2025-06-29 19:27:00 -04:00
package handler
import (
"crypto/sha256"
"fmt"
"net/http"
Local fork: hardening + ops improvements (timeout knob, demotion, /livez, drain) This commit captures both the prior accumulated work-in-progress (framework migration web/→svelte/, postgres storage, conversation viewer, dashboard auth, OpenAPI spec, integration tests) AND today's operational improvements layered on top. History wasn't checkpointed incrementally; happy to split it via interactive rebase if a reviewer wants smaller commits. Today's changes (in addition to the older WIP): 1. Configurable upstream response-header timeout - ANTHROPIC_RESPONSE_HEADER_TIMEOUT env (default 300s) - Replaces hardcoded 300s in provider/anthropic.go that was firing on opus + 1M-context + extended thinking non-streaming requests - Files: internal/config/config.go, internal/provider/anthropic.go 2. Structured forward-error diagnostic logging - When a forward to Anthropic fails, log a single key=value line with request_id, model, stream, body_bytes, has_thinking, anthropic_beta, query, elapsed, ctx_err — alongside the existing human-readable error line for back-compat - Files: internal/handler/handlers.go (logForwardFailure) 3. Full SSE protocol passthrough + Flusher fix - handler/handlers.go: forward all SSE lines verbatim (event:, id:, retry:, : comments, blank-line terminators), not only data:. Previous code produced malformed SSE for strict parsers. - middleware/logging.go: explicit Flush() method on responseWriter. Embedding http.ResponseWriter (interface) does not auto-promote Flush(), so every w.(http.Flusher) check in the streaming handler was returning ok=false and SSE writes buffered in net/http until the body closed. 4. Non-streaming → streaming demotion (feature-flagged) - ANTHROPIC_DEMOTE_NONSTREAMING env (default false) - When enabled and the routed provider is anthropic, force stream=true upstream for clients that asked for stream=false. Receive SSE, accumulate via accumulateSSEToMessage (handles text, tool_use with partial_json reassembly, thinking, signature, citations_delta, usage merge), and synthesize a single non-streaming JSON response. - Eliminates the ResponseHeaderTimeout class of failure entirely. - Body rewrite uses json.Decoder + UseNumber() to preserve integer precision in unknown nested fields (tool inputs from prior turns). - Files: internal/config/config.go, internal/handler/handlers.go, cmd/proxy/main.go, cmd/proxy/main_test.go 5. Live operational state: /livez gauge + graceful drain - New internal/runtime package: atomic in-flight counter + draining flag - New middleware/inflight.go: increments runtime gauge, applied to /v1/* subrouter so Messages, ChatCompletions, and ProxyPassthrough are all counted - /v1/* moved to a gorilla/mux subrouter so the InFlight middleware applies surgically; /health, /livez, /openapi.* remain on parent router (unauthenticated, uncounted) - Health handler returns 503 draining when runtime.IsDraining() is true, so Traefik stops routing to a slot before drain begins - New /livez handler returns {status, in_flight, draining, timestamp} - SIGTERM handler in main.go: SetDraining(true), poll for in_flight==0 with 32-min ceiling and 1s tick (logs every 10s), then srv.Shutdown - Auth bypass list extended with /livez - Files: internal/runtime/runtime.go (new), internal/middleware/inflight.go (new), internal/middleware/auth.go, internal/handler/handlers.go (Health, Livez, runtime import), cmd/proxy/main.go (subrouter, drain loop) 6. OpenAPI spec updates - Document Health 503 response and new DrainingResponse schema - Add /livez path with LivezResponse schema - Files: internal/handler/openapi.go Verified: go build ./... clean, go test ./... all pass, go vet clean. Three rounds of codex peer review across changes 1-5; all feedback addressed (citations_delta, json.Number precision, drain-loop logging via lastLog timestamp, PathPrefix tightened to "/v1/").
2026-05-02 15:15:58 -06:00
"strconv"
2025-06-29 19:27:00 -04:00
"strings"
"github.com/seifghazi/claude-code-monitor/internal/model"
)
Local fork: hardening + ops improvements (timeout knob, demotion, /livez, drain) This commit captures both the prior accumulated work-in-progress (framework migration web/→svelte/, postgres storage, conversation viewer, dashboard auth, OpenAPI spec, integration tests) AND today's operational improvements layered on top. History wasn't checkpointed incrementally; happy to split it via interactive rebase if a reviewer wants smaller commits. Today's changes (in addition to the older WIP): 1. Configurable upstream response-header timeout - ANTHROPIC_RESPONSE_HEADER_TIMEOUT env (default 300s) - Replaces hardcoded 300s in provider/anthropic.go that was firing on opus + 1M-context + extended thinking non-streaming requests - Files: internal/config/config.go, internal/provider/anthropic.go 2. Structured forward-error diagnostic logging - When a forward to Anthropic fails, log a single key=value line with request_id, model, stream, body_bytes, has_thinking, anthropic_beta, query, elapsed, ctx_err — alongside the existing human-readable error line for back-compat - Files: internal/handler/handlers.go (logForwardFailure) 3. Full SSE protocol passthrough + Flusher fix - handler/handlers.go: forward all SSE lines verbatim (event:, id:, retry:, : comments, blank-line terminators), not only data:. Previous code produced malformed SSE for strict parsers. - middleware/logging.go: explicit Flush() method on responseWriter. Embedding http.ResponseWriter (interface) does not auto-promote Flush(), so every w.(http.Flusher) check in the streaming handler was returning ok=false and SSE writes buffered in net/http until the body closed. 4. Non-streaming → streaming demotion (feature-flagged) - ANTHROPIC_DEMOTE_NONSTREAMING env (default false) - When enabled and the routed provider is anthropic, force stream=true upstream for clients that asked for stream=false. Receive SSE, accumulate via accumulateSSEToMessage (handles text, tool_use with partial_json reassembly, thinking, signature, citations_delta, usage merge), and synthesize a single non-streaming JSON response. - Eliminates the ResponseHeaderTimeout class of failure entirely. - Body rewrite uses json.Decoder + UseNumber() to preserve integer precision in unknown nested fields (tool inputs from prior turns). - Files: internal/config/config.go, internal/handler/handlers.go, cmd/proxy/main.go, cmd/proxy/main_test.go 5. Live operational state: /livez gauge + graceful drain - New internal/runtime package: atomic in-flight counter + draining flag - New middleware/inflight.go: increments runtime gauge, applied to /v1/* subrouter so Messages, ChatCompletions, and ProxyPassthrough are all counted - /v1/* moved to a gorilla/mux subrouter so the InFlight middleware applies surgically; /health, /livez, /openapi.* remain on parent router (unauthenticated, uncounted) - Health handler returns 503 draining when runtime.IsDraining() is true, so Traefik stops routing to a slot before drain begins - New /livez handler returns {status, in_flight, draining, timestamp} - SIGTERM handler in main.go: SetDraining(true), poll for in_flight==0 with 32-min ceiling and 1s tick (logs every 10s), then srv.Shutdown - Auth bypass list extended with /livez - Files: internal/runtime/runtime.go (new), internal/middleware/inflight.go (new), internal/middleware/auth.go, internal/handler/handlers.go (Health, Livez, runtime import), cmd/proxy/main.go (subrouter, drain loop) 6. OpenAPI spec updates - Document Health 503 response and new DrainingResponse schema - Add /livez path with LivezResponse schema - Files: internal/handler/openapi.go Verified: go build ./... clean, go test ./... all pass, go vet clean. Three rounds of codex peer review across changes 1-5; all feedback addressed (citations_delta, json.Number precision, drain-loop logging via lastLog timestamp, PathPrefix tightened to "/v1/").
2026-05-02 15:15:58 -06:00
var hopByHopHeaders = map[string]bool{
"connection": true,
"keep-alive": true,
"proxy-authenticate": true,
"proxy-authorization": true,
"te": true,
"trailers": true,
"transfer-encoding": true,
"upgrade": true,
"content-encoding": true, // We handle decompression ourselves
"content-length": true, // May change after decompression
}
Local fork: hardening + ops improvements (timeout knob, demotion, /livez, drain) This commit captures both the prior accumulated work-in-progress (framework migration web/→svelte/, postgres storage, conversation viewer, dashboard auth, OpenAPI spec, integration tests) AND today's operational improvements layered on top. History wasn't checkpointed incrementally; happy to split it via interactive rebase if a reviewer wants smaller commits. Today's changes (in addition to the older WIP): 1. Configurable upstream response-header timeout - ANTHROPIC_RESPONSE_HEADER_TIMEOUT env (default 300s) - Replaces hardcoded 300s in provider/anthropic.go that was firing on opus + 1M-context + extended thinking non-streaming requests - Files: internal/config/config.go, internal/provider/anthropic.go 2. Structured forward-error diagnostic logging - When a forward to Anthropic fails, log a single key=value line with request_id, model, stream, body_bytes, has_thinking, anthropic_beta, query, elapsed, ctx_err — alongside the existing human-readable error line for back-compat - Files: internal/handler/handlers.go (logForwardFailure) 3. Full SSE protocol passthrough + Flusher fix - handler/handlers.go: forward all SSE lines verbatim (event:, id:, retry:, : comments, blank-line terminators), not only data:. Previous code produced malformed SSE for strict parsers. - middleware/logging.go: explicit Flush() method on responseWriter. Embedding http.ResponseWriter (interface) does not auto-promote Flush(), so every w.(http.Flusher) check in the streaming handler was returning ok=false and SSE writes buffered in net/http until the body closed. 4. Non-streaming → streaming demotion (feature-flagged) - ANTHROPIC_DEMOTE_NONSTREAMING env (default false) - When enabled and the routed provider is anthropic, force stream=true upstream for clients that asked for stream=false. Receive SSE, accumulate via accumulateSSEToMessage (handles text, tool_use with partial_json reassembly, thinking, signature, citations_delta, usage merge), and synthesize a single non-streaming JSON response. - Eliminates the ResponseHeaderTimeout class of failure entirely. - Body rewrite uses json.Decoder + UseNumber() to preserve integer precision in unknown nested fields (tool inputs from prior turns). - Files: internal/config/config.go, internal/handler/handlers.go, cmd/proxy/main.go, cmd/proxy/main_test.go 5. Live operational state: /livez gauge + graceful drain - New internal/runtime package: atomic in-flight counter + draining flag - New middleware/inflight.go: increments runtime gauge, applied to /v1/* subrouter so Messages, ChatCompletions, and ProxyPassthrough are all counted - /v1/* moved to a gorilla/mux subrouter so the InFlight middleware applies surgically; /health, /livez, /openapi.* remain on parent router (unauthenticated, uncounted) - Health handler returns 503 draining when runtime.IsDraining() is true, so Traefik stops routing to a slot before drain begins - New /livez handler returns {status, in_flight, draining, timestamp} - SIGTERM handler in main.go: SetDraining(true), poll for in_flight==0 with 32-min ceiling and 1s tick (logs every 10s), then srv.Shutdown - Auth bypass list extended with /livez - Files: internal/runtime/runtime.go (new), internal/middleware/inflight.go (new), internal/middleware/auth.go, internal/handler/handlers.go (Health, Livez, runtime import), cmd/proxy/main.go (subrouter, drain loop) 6. OpenAPI spec updates - Document Health 503 response and new DrainingResponse schema - Add /livez path with LivezResponse schema - Files: internal/handler/openapi.go Verified: go build ./... clean, go test ./... all pass, go vet clean. Three rounds of codex peer review across changes 1-5; all feedback addressed (citations_delta, json.Number precision, drain-loop logging via lastLog timestamp, PathPrefix tightened to "/v1/").
2026-05-02 15:15:58 -06:00
// ApplyHeaderRules applies block/set/replace rules to an http.Header in-place.
func ApplyHeaderRules(headers http.Header, rules []model.HeaderRule) {
for _, rule := range rules {
if !rule.Enabled {
continue
}
key := http.CanonicalHeaderKey(rule.Header)
switch rule.Action {
case "block":
headers.Del(key)
case "set":
headers.Set(key, rule.Value)
case "replace":
if rule.Find == "" {
continue
}
for i, v := range headers.Values(key) {
if strings.Contains(v, rule.Find) {
replaced := strings.ReplaceAll(v, rule.Find, rule.Value)
if i == 0 {
headers.Set(key, replaced)
} else {
headers.Add(key, replaced)
}
}
}
}
}
}
Local fork: hardening + ops improvements (timeout knob, demotion, /livez, drain) This commit captures both the prior accumulated work-in-progress (framework migration web/→svelte/, postgres storage, conversation viewer, dashboard auth, OpenAPI spec, integration tests) AND today's operational improvements layered on top. History wasn't checkpointed incrementally; happy to split it via interactive rebase if a reviewer wants smaller commits. Today's changes (in addition to the older WIP): 1. Configurable upstream response-header timeout - ANTHROPIC_RESPONSE_HEADER_TIMEOUT env (default 300s) - Replaces hardcoded 300s in provider/anthropic.go that was firing on opus + 1M-context + extended thinking non-streaming requests - Files: internal/config/config.go, internal/provider/anthropic.go 2. Structured forward-error diagnostic logging - When a forward to Anthropic fails, log a single key=value line with request_id, model, stream, body_bytes, has_thinking, anthropic_beta, query, elapsed, ctx_err — alongside the existing human-readable error line for back-compat - Files: internal/handler/handlers.go (logForwardFailure) 3. Full SSE protocol passthrough + Flusher fix - handler/handlers.go: forward all SSE lines verbatim (event:, id:, retry:, : comments, blank-line terminators), not only data:. Previous code produced malformed SSE for strict parsers. - middleware/logging.go: explicit Flush() method on responseWriter. Embedding http.ResponseWriter (interface) does not auto-promote Flush(), so every w.(http.Flusher) check in the streaming handler was returning ok=false and SSE writes buffered in net/http until the body closed. 4. Non-streaming → streaming demotion (feature-flagged) - ANTHROPIC_DEMOTE_NONSTREAMING env (default false) - When enabled and the routed provider is anthropic, force stream=true upstream for clients that asked for stream=false. Receive SSE, accumulate via accumulateSSEToMessage (handles text, tool_use with partial_json reassembly, thinking, signature, citations_delta, usage merge), and synthesize a single non-streaming JSON response. - Eliminates the ResponseHeaderTimeout class of failure entirely. - Body rewrite uses json.Decoder + UseNumber() to preserve integer precision in unknown nested fields (tool inputs from prior turns). - Files: internal/config/config.go, internal/handler/handlers.go, cmd/proxy/main.go, cmd/proxy/main_test.go 5. Live operational state: /livez gauge + graceful drain - New internal/runtime package: atomic in-flight counter + draining flag - New middleware/inflight.go: increments runtime gauge, applied to /v1/* subrouter so Messages, ChatCompletions, and ProxyPassthrough are all counted - /v1/* moved to a gorilla/mux subrouter so the InFlight middleware applies surgically; /health, /livez, /openapi.* remain on parent router (unauthenticated, uncounted) - Health handler returns 503 draining when runtime.IsDraining() is true, so Traefik stops routing to a slot before drain begins - New /livez handler returns {status, in_flight, draining, timestamp} - SIGTERM handler in main.go: SetDraining(true), poll for in_flight==0 with 32-min ceiling and 1s tick (logs every 10s), then srv.Shutdown - Auth bypass list extended with /livez - Files: internal/runtime/runtime.go (new), internal/middleware/inflight.go (new), internal/middleware/auth.go, internal/handler/handlers.go (Health, Livez, runtime import), cmd/proxy/main.go (subrouter, drain loop) 6. OpenAPI spec updates - Document Health 503 response and new DrainingResponse schema - Add /livez path with LivezResponse schema - Files: internal/handler/openapi.go Verified: go build ./... clean, go test ./... all pass, go vet clean. Three rounds of codex peer review across changes 1-5; all feedback addressed (citations_delta, json.Number precision, drain-loop logging via lastLog timestamp, PathPrefix tightened to "/v1/").
2026-05-02 15:15:58 -06:00
// CopyAllResponseHeaders forwards all upstream response headers to the client,
// stripping only hop-by-hop headers that must not be forwarded by a proxy.
func CopyAllResponseHeaders(w http.ResponseWriter, resp *http.Response) {
for key, values := range resp.Header {
if hopByHopHeaders[strings.ToLower(key)] {
continue
}
for _, value := range values {
w.Header().Add(key, value)
}
}
}
Local fork: hardening + ops improvements (timeout knob, demotion, /livez, drain) This commit captures both the prior accumulated work-in-progress (framework migration web/→svelte/, postgres storage, conversation viewer, dashboard auth, OpenAPI spec, integration tests) AND today's operational improvements layered on top. History wasn't checkpointed incrementally; happy to split it via interactive rebase if a reviewer wants smaller commits. Today's changes (in addition to the older WIP): 1. Configurable upstream response-header timeout - ANTHROPIC_RESPONSE_HEADER_TIMEOUT env (default 300s) - Replaces hardcoded 300s in provider/anthropic.go that was firing on opus + 1M-context + extended thinking non-streaming requests - Files: internal/config/config.go, internal/provider/anthropic.go 2. Structured forward-error diagnostic logging - When a forward to Anthropic fails, log a single key=value line with request_id, model, stream, body_bytes, has_thinking, anthropic_beta, query, elapsed, ctx_err — alongside the existing human-readable error line for back-compat - Files: internal/handler/handlers.go (logForwardFailure) 3. Full SSE protocol passthrough + Flusher fix - handler/handlers.go: forward all SSE lines verbatim (event:, id:, retry:, : comments, blank-line terminators), not only data:. Previous code produced malformed SSE for strict parsers. - middleware/logging.go: explicit Flush() method on responseWriter. Embedding http.ResponseWriter (interface) does not auto-promote Flush(), so every w.(http.Flusher) check in the streaming handler was returning ok=false and SSE writes buffered in net/http until the body closed. 4. Non-streaming → streaming demotion (feature-flagged) - ANTHROPIC_DEMOTE_NONSTREAMING env (default false) - When enabled and the routed provider is anthropic, force stream=true upstream for clients that asked for stream=false. Receive SSE, accumulate via accumulateSSEToMessage (handles text, tool_use with partial_json reassembly, thinking, signature, citations_delta, usage merge), and synthesize a single non-streaming JSON response. - Eliminates the ResponseHeaderTimeout class of failure entirely. - Body rewrite uses json.Decoder + UseNumber() to preserve integer precision in unknown nested fields (tool inputs from prior turns). - Files: internal/config/config.go, internal/handler/handlers.go, cmd/proxy/main.go, cmd/proxy/main_test.go 5. Live operational state: /livez gauge + graceful drain - New internal/runtime package: atomic in-flight counter + draining flag - New middleware/inflight.go: increments runtime gauge, applied to /v1/* subrouter so Messages, ChatCompletions, and ProxyPassthrough are all counted - /v1/* moved to a gorilla/mux subrouter so the InFlight middleware applies surgically; /health, /livez, /openapi.* remain on parent router (unauthenticated, uncounted) - Health handler returns 503 draining when runtime.IsDraining() is true, so Traefik stops routing to a slot before drain begins - New /livez handler returns {status, in_flight, draining, timestamp} - SIGTERM handler in main.go: SetDraining(true), poll for in_flight==0 with 32-min ceiling and 1s tick (logs every 10s), then srv.Shutdown - Auth bypass list extended with /livez - Files: internal/runtime/runtime.go (new), internal/middleware/inflight.go (new), internal/middleware/auth.go, internal/handler/handlers.go (Health, Livez, runtime import), cmd/proxy/main.go (subrouter, drain loop) 6. OpenAPI spec updates - Document Health 503 response and new DrainingResponse schema - Add /livez path with LivezResponse schema - Files: internal/handler/openapi.go Verified: go build ./... clean, go test ./... all pass, go vet clean. Three rounds of codex peer review across changes 1-5; all feedback addressed (citations_delta, json.Number precision, drain-loop logging via lastLog timestamp, PathPrefix tightened to "/v1/").
2026-05-02 15:15:58 -06:00
// SanitizeResponseHeaders strips hop-by-hop proxy headers before applying the
// generic sensitive-header sanitization used for stored metadata.
func SanitizeResponseHeaders(headers http.Header) http.Header {
filtered := make(http.Header)
for key, values := range headers {
if hopByHopHeaders[strings.ToLower(key)] {
continue
}
copiedValues := append([]string(nil), values...)
filtered[key] = copiedValues
}
return SanitizeHeaders(filtered)
}
// ExtractRateLimitInfo parses rate limit headers from the upstream response
func ExtractRateLimitInfo(headers http.Header) *model.RateLimitInfo {
info := &model.RateLimitInfo{}
found := false
// Organization ID
if v := headers.Get("anthropic-organization-id"); v != "" {
info.OrganizationID = v
found = true
}
// Unified quota system (current Anthropic model)
if v := headers.Get("anthropic-ratelimit-unified-status"); v != "" {
info.UnifiedStatus = v
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-5h-utilization"); v != "" {
info.UnifiedUtilization5h, _ = strconv.ParseFloat(v, 64)
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-5h-reset"); v != "" {
info.UnifiedReset5h = v
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-7d-utilization"); v != "" {
info.UnifiedUtilization7d, _ = strconv.ParseFloat(v, 64)
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-7d-reset"); v != "" {
info.UnifiedReset7d = v
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-fallback-percentage"); v != "" {
info.UnifiedFallbackPercentage, _ = strconv.ParseFloat(v, 64)
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-overage-status"); v != "" {
info.UnifiedOverageStatus = v
found = true
}
if v := headers.Get("anthropic-ratelimit-unified-representative-claim"); v != "" {
info.UnifiedRepresentativeClaim = v
found = true
}
// Legacy per-resource rate limits
if v := headers.Get("anthropic-ratelimit-requests-limit"); v != "" {
info.RequestsLimit, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("anthropic-ratelimit-requests-remaining"); v != "" {
info.RequestsRemaining, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("anthropic-ratelimit-requests-reset"); v != "" {
info.RequestsReset = v
found = true
}
if v := headers.Get("anthropic-ratelimit-tokens-limit"); v != "" {
info.TokensLimit, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("anthropic-ratelimit-tokens-remaining"); v != "" {
info.TokensRemaining, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("anthropic-ratelimit-tokens-reset"); v != "" {
info.TokensReset = v
found = true
}
// Fall back to standard rate limit headers
if !found {
if v := headers.Get("x-ratelimit-limit"); v != "" {
info.RequestsLimit, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("x-ratelimit-remaining"); v != "" {
info.RequestsRemaining, _ = strconv.Atoi(v)
found = true
}
if v := headers.Get("x-ratelimit-reset"); v != "" {
info.RequestsReset = v
found = true
}
}
if !found {
return nil
}
return info
}
2025-06-29 19:27:00 -04:00
// SanitizeHeaders removes sensitive headers before logging/storage
func SanitizeHeaders(headers http.Header) http.Header {
sanitized := make(http.Header)
sensitiveHeaders := []string{
"x-api-key",
"api-key",
"authorization",
"anthropic-api-key",
"openai-api-key",
"bearer",
}
for key, values := range headers {
lowerKey := strings.ToLower(key)
isSensitive := false
for _, sensitive := range sensitiveHeaders {
if strings.Contains(lowerKey, sensitive) {
isSensitive = true
break
}
}
if isSensitive {
// Calculate SHA256 hash for each sensitive header value
hashedValues := make([]string, len(values))
for i, value := range values {
hash := sha256.Sum256([]byte(value))
hashedValues[i] = fmt.Sprintf("sha256:%x", hash)
}
sanitized[key] = hashedValues
2025-06-29 19:27:00 -04:00
} else {
sanitized[key] = values
}
}
return sanitized
}