|
| 1 | +package pii |
| 2 | + |
| 3 | +import ( |
| 4 | + "encoding/json" |
| 5 | + "fmt" |
| 6 | + "net/http" |
| 7 | + "strings" |
| 8 | + "unicode" |
| 9 | +) |
| 10 | + |
| 11 | +type responsePIIConfig interface { |
| 12 | + PIIReverseInResponse() bool |
| 13 | +} |
| 14 | + |
| 15 | +type pseudonymizer struct { |
| 16 | + byValue map[string]string |
| 17 | + original map[string]string |
| 18 | + counts map[string]int |
| 19 | +} |
| 20 | + |
| 21 | +func newPseudonymizer() *pseudonymizer { |
| 22 | + return &pseudonymizer{ |
| 23 | + byValue: map[string]string{}, |
| 24 | + original: map[string]string{}, |
| 25 | + counts: map[string]int{}, |
| 26 | + } |
| 27 | +} |
| 28 | + |
| 29 | +func (p *pseudonymizer) replace(text string, spans []Span) string { |
| 30 | + var b strings.Builder |
| 31 | + last := 0 |
| 32 | + for _, span := range spans { |
| 33 | + if span.Action != ActionMask || span.Start < last || span.End > len(text) { |
| 34 | + continue |
| 35 | + } |
| 36 | + b.WriteString(text[last:span.Start]) |
| 37 | + value := text[span.Start:span.End] |
| 38 | + token, ok := p.byValue[value] |
| 39 | + if !ok { |
| 40 | + group := pseudonymGroup(span.Pattern) |
| 41 | + p.counts[group]++ |
| 42 | + token = fmt.Sprintf("%s_%03d", group, p.counts[group]) |
| 43 | + p.byValue[value] = token |
| 44 | + p.original[token] = value |
| 45 | + } |
| 46 | + b.WriteString(token) |
| 47 | + last = span.End |
| 48 | + } |
| 49 | + b.WriteString(text[last:]) |
| 50 | + return b.String() |
| 51 | +} |
| 52 | + |
| 53 | +func pseudonymGroup(pattern string) string { |
| 54 | + if i := strings.LastIndexByte(pattern, ':'); i >= 0 { |
| 55 | + pattern = pattern[i+1:] |
| 56 | + } |
| 57 | + var b strings.Builder |
| 58 | + for _, r := range strings.ToUpper(pattern) { |
| 59 | + if unicode.IsLetter(r) || unicode.IsDigit(r) { |
| 60 | + b.WriteRune(r) |
| 61 | + } else { |
| 62 | + b.WriteByte('_') |
| 63 | + } |
| 64 | + } |
| 65 | + if b.Len() == 0 { |
| 66 | + return "PII" |
| 67 | + } |
| 68 | + return b.String() |
| 69 | +} |
| 70 | + |
| 71 | +type restoringWriter struct { |
| 72 | + http.ResponseWriter |
| 73 | + pending string |
| 74 | + replacements map[string]string |
| 75 | +} |
| 76 | + |
| 77 | +func newRestoringWriter(w http.ResponseWriter, originals map[string]string) *restoringWriter { |
| 78 | + replacements := make(map[string]string, len(originals)) |
| 79 | + for token, original := range originals { |
| 80 | + encoded, _ := json.Marshal(original) |
| 81 | + replacements[token] = string(encoded[1 : len(encoded)-1]) |
| 82 | + } |
| 83 | + return &restoringWriter{ResponseWriter: w, replacements: replacements} |
| 84 | +} |
| 85 | + |
| 86 | +func (w *restoringWriter) Write(data []byte) (int, error) { |
| 87 | + w.pending += string(data) |
| 88 | + ready, pending := w.splitReady(w.replace(w.pending)) |
| 89 | + w.pending = pending |
| 90 | + if ready != "" { |
| 91 | + if _, err := w.ResponseWriter.Write([]byte(ready)); err != nil { |
| 92 | + return 0, err |
| 93 | + } |
| 94 | + } |
| 95 | + return len(data), nil |
| 96 | +} |
| 97 | + |
| 98 | +func (w *restoringWriter) Flush() { |
| 99 | + if f, ok := w.ResponseWriter.(http.Flusher); ok { |
| 100 | + f.Flush() |
| 101 | + } |
| 102 | +} |
| 103 | + |
| 104 | +func (w *restoringWriter) Finish() error { |
| 105 | + if w.pending == "" { |
| 106 | + return nil |
| 107 | + } |
| 108 | + _, err := w.ResponseWriter.Write([]byte(w.replace(w.pending))) |
| 109 | + w.pending = "" |
| 110 | + return err |
| 111 | +} |
| 112 | + |
| 113 | +func (w *restoringWriter) replace(s string) string { |
| 114 | + for token, original := range w.replacements { |
| 115 | + s = strings.ReplaceAll(s, token, original) |
| 116 | + } |
| 117 | + return s |
| 118 | +} |
| 119 | + |
| 120 | +func (w *restoringWriter) splitReady(s string) (string, string) { |
| 121 | + keep := 0 |
| 122 | + for token := range w.replacements { |
| 123 | + limit := min(len(token)-1, len(s)) |
| 124 | + for n := 1; n <= limit; n++ { |
| 125 | + if strings.HasSuffix(s, token[:n]) && n > keep { |
| 126 | + keep = n |
| 127 | + } |
| 128 | + } |
| 129 | + } |
| 130 | + return s[:len(s)-keep], s[len(s)-keep:] |
| 131 | +} |
0 commit comments