package contentpolicy import ( "bytes" "context" "encoding/json" "errors" "fmt" "io" "log/slog" "net/http" "regexp" "slices" "strings" "sync/atomic" "time" "github.com/jackc/pgx/v5/pgxpool" ) var ErrBodyTooLarge = errors.New("request body is too large") type Rule struct { Name string `json:"name"` Pattern string `json:"pattern"` Replacement string `json:"replacement"` } type Policy struct { ID string `json:"id"` Name string `json:"name"` Description string `json:"description"` Action string `json:"action"` Priority int `json:"priority"` Paths []string `json:"paths"` Models []string `json:"models"` APIKeyIDs []string `json:"api_key_ids"` Rules []Rule `json:"rules"` Enabled bool `json:"enabled"` Revision int64 `json:"revision"` CreatedAt time.Time `json:"created_at"` UpdatedAt time.Time `json:"updated_at"` } type compiledRule struct { name, replacement string expression *regexp.Regexp } type compiledPolicy struct { Policy rules []compiledRule } type snapshot struct{ policies []compiledPolicy } type Engine struct { pool *pgxpool.Pool logger *slog.Logger refresh time.Duration current atomic.Pointer[snapshot] } type Match struct { PolicyID string `json:"policy_id"` PolicyName string `json:"policy_name"` Action string `json:"action"` Rules []string `json:"rules"` } type Result struct { Matches []Match Blocked, Redacted bool } func NewEngine(pool *pgxpool.Pool, refresh time.Duration, logger *slog.Logger) *Engine { if refresh <= 0 { refresh = 30 * time.Second } e := &Engine{pool: pool, refresh: refresh, logger: logger} e.current.Store(&snapshot{}) return e } func (e *Engine) Run(ctx context.Context) { _ = e.Reload(ctx) ticker := time.NewTicker(e.refresh) defer ticker.Stop() for { select { case <-ticker.C: if err := e.Reload(ctx); err != nil && e.logger != nil { e.logger.Warn("content policy refresh failed", "error", err) } case <-ctx.Done(): return } } } func (e *Engine) Reload(ctx context.Context) error { if e == nil || e.pool == nil { return errors.New("content policy store unavailable") } rows, err := e.pool.Query(ctx, `SELECT id::text,name,description,action,priority,paths,models,api_key_ids::text[],rules,enabled,revision,created_at,updated_at FROM gateway.content_policies WHERE enabled ORDER BY priority DESC,name,id`) if err != nil { return fmt.Errorf("load content policies: %w", err) } defer rows.Close() loaded := make([]compiledPolicy, 0) for rows.Next() { var policy Policy var raw []byte if err := rows.Scan(&policy.ID, &policy.Name, &policy.Description, &policy.Action, &policy.Priority, &policy.Paths, &policy.Models, &policy.APIKeyIDs, &raw, &policy.Enabled, &policy.Revision, &policy.CreatedAt, &policy.UpdatedAt); err != nil { return err } if err := json.Unmarshal(raw, &policy.Rules); err != nil { return fmt.Errorf("decode policy %s: %w", policy.ID, err) } compiled, err := compilePolicy(policy) if err != nil { return fmt.Errorf("compile policy %s: %w", policy.ID, err) } loaded = append(loaded, compiled) } if err := rows.Err(); err != nil { return err } e.current.Store(&snapshot{policies: loaded}) return nil } func compilePolicy(policy Policy) (compiledPolicy, error) { if policy.Action != "audit" && policy.Action != "block" && policy.Action != "redact" { return compiledPolicy{}, errors.New("unsupported action") } if len(policy.Rules) == 0 || len(policy.Rules) > 20 { return compiledPolicy{}, errors.New("rules must contain 1 to 20 entries") } result := compiledPolicy{Policy: policy, rules: make([]compiledRule, 0, len(policy.Rules))} for _, rule := range policy.Rules { if strings.TrimSpace(rule.Name) == "" || len(rule.Pattern) == 0 || len(rule.Pattern) > 512 { return compiledPolicy{}, errors.New("invalid rule name or pattern length") } expression, err := regexp.Compile(rule.Pattern) if err != nil { return compiledPolicy{}, err } replacement := rule.Replacement if replacement == "" { replacement = "[REDACTED]" } result.rules = append(result.rules, compiledRule{name: rule.Name, replacement: replacement, expression: expression}) } return result, nil } func Validate(policy Policy) error { _, err := compilePolicy(policy); return err } func (e *Engine) Apply(request *http.Request, maxBody int64, apiKeyID string) (Result, error) { if e == nil || request.Body == nil || request.Body == http.NoBody || request.Method == http.MethodGet { return Result{}, nil } body, err := io.ReadAll(io.LimitReader(request.Body, maxBody+1)) if err != nil { return Result{}, err } _ = request.Body.Close() if int64(len(body)) > maxBody { return Result{}, ErrBodyTooLarge } restoreBody(request, body) var document any decoder := json.NewDecoder(bytes.NewReader(body)) decoder.UseNumber() if decoder.Decode(&document) != nil { return Result{}, nil } model := findModel(document) result := Result{} for _, policy := range e.current.Load().policies { if !policyApplies(policy, request.URL.Path, model, apiKeyID) { continue } matched := make([]string, 0) changed := false walkText(&document, false, func(value string) string { for _, rule := range policy.rules { if rule.expression.MatchString(value) { if !slices.Contains(matched, rule.name) { matched = append(matched, rule.name) } if policy.Action == "redact" { next := rule.expression.ReplaceAllString(value, rule.replacement) changed = changed || next != value value = next } } } return value }) if len(matched) == 0 { continue } result.Matches = append(result.Matches, Match{PolicyID: policy.ID, PolicyName: policy.Name, Action: policy.Action, Rules: matched}) if policy.Action == "block" { result.Blocked = true break } result.Redacted = result.Redacted || changed } if result.Redacted && !result.Blocked { encoded, err := json.Marshal(document) if err != nil { return Result{}, err } restoreBody(request, encoded) } return result, nil } func policyApplies(policy compiledPolicy, path, model, apiKeyID string) bool { return matches(policy.Paths, path) && matches(policy.Models, model) && matches(policy.APIKeyIDs, apiKeyID) } func matches(values []string, candidate string) bool { if len(values) == 0 { return true } for _, value := range values { if value == "*" || value == candidate { return true } } return false } var textualFields = map[string]bool{"content": true, "text": true, "input": true, "prompt": true, "instructions": true} func walkText(value *any, selected bool, transform func(string) string) { switch current := (*value).(type) { case map[string]any: for key, child := range current { local := child walkText(&local, selected || textualFields[strings.ToLower(key)], transform) current[key] = local } case []any: for index, child := range current { local := child walkText(&local, selected, transform) current[index] = local } case string: if selected { *value = transform(current) } } } func findModel(document any) string { if object, ok := document.(map[string]any); ok { if model, ok := object["model"].(string); ok { return model } } return "" } func restoreBody(request *http.Request, body []byte) { request.Body = io.NopCloser(bytes.NewReader(body)) request.ContentLength = int64(len(body)) request.GetBody = func() (io.ReadCloser, error) { return io.NopCloser(bytes.NewReader(body)), nil } }