|
| 1 | +package anonymizer |
| 2 | + |
| 3 | +import ( |
| 4 | + "bufio" |
| 5 | + "encoding/csv" |
| 6 | + "errors" |
| 7 | + "fmt" |
| 8 | + "io/fs" |
| 9 | + "os" |
| 10 | + "path/filepath" |
| 11 | + "strings" |
| 12 | + |
| 13 | + "golang.org/x/exp/rand" |
| 14 | +) |
| 15 | + |
| 16 | +type Anonimizer struct { |
| 17 | + csvData []map[string]string |
| 18 | + anonData map[string][]string |
| 19 | + fieldNames []string |
| 20 | + randFunc func(int) int |
| 21 | + Output []string |
| 22 | +} |
| 23 | + |
| 24 | +func New() *Anonimizer { |
| 25 | + return &Anonimizer{anonData: make(map[string][]string), randFunc: rand.Intn} |
| 26 | +} |
| 27 | + |
| 28 | +func (an *Anonimizer) ReadCSVLogs(filename string) error { |
| 29 | + fmt.Println("Reading CSV file:", filename) |
| 30 | + file, err := os.Open(filename) |
| 31 | + if err != nil { |
| 32 | + return err |
| 33 | + } |
| 34 | + defer file.Close() |
| 35 | + |
| 36 | + csvReader := csv.NewReader(file) |
| 37 | + rows, err := csvReader.ReadAll() |
| 38 | + if err != nil { |
| 39 | + return err |
| 40 | + } |
| 41 | + |
| 42 | + // First element of the csvReader contains field names |
| 43 | + an.fieldNames = rows[0] |
| 44 | + |
| 45 | + for _, row := range rows[1:] { |
| 46 | + m := make(map[string]string) |
| 47 | + for i, val := range row { |
| 48 | + m[an.fieldNames[i]] = string(val) |
| 49 | + } |
| 50 | + an.csvData = append(an.csvData, m) |
| 51 | + } |
| 52 | + fmt.Println("Field names in the CSV file:", an.fieldNames) |
| 53 | + fmt.Println("Number of rows found in the CSV file:", len(an.csvData)) |
| 54 | + |
| 55 | + return nil |
| 56 | +} |
| 57 | + |
| 58 | +func (an *Anonimizer) LoadAnonData(anonDataDir string) error { |
| 59 | + for i := range an.fieldNames { |
| 60 | + if an.fieldNames[i] == "raw" { |
| 61 | + continue |
| 62 | + } |
| 63 | + |
| 64 | + filepath := filepath.Join(anonDataDir, an.fieldNames[i]) |
| 65 | + _, err := os.Stat(filepath) |
| 66 | + if err != nil { |
| 67 | + if errors.Is(err, fs.ErrNotExist) { |
| 68 | + fmt.Printf("Anonymizing data not found for field %s. Skipping.\n", an.fieldNames[i]) |
| 69 | + continue |
| 70 | + } |
| 71 | + return err |
| 72 | + } |
| 73 | + |
| 74 | + data, err := an.loadAnonymizingData(filepath) |
| 75 | + if err != nil { |
| 76 | + return fmt.Errorf("loading anonymizing data file %s: %v", filepath, err) |
| 77 | + } |
| 78 | + |
| 79 | + an.anonData[an.fieldNames[i]] = data |
| 80 | + fmt.Printf("Loaded anonymizing data for field: %s; values loaded: %d\n", an.fieldNames[i], len(data)) |
| 81 | + } |
| 82 | + |
| 83 | + return nil |
| 84 | +} |
| 85 | + |
| 86 | +func (an *Anonimizer) loadAnonymizingData(filepath string) ([]string, error) { |
| 87 | + anonDataFile, err := os.OpenFile(filepath, os.O_RDONLY, os.ModePerm) |
| 88 | + if err != nil { |
| 89 | + return nil, err |
| 90 | + } |
| 91 | + defer anonDataFile.Close() |
| 92 | + |
| 93 | + var anonData []string |
| 94 | + anonDataFileScanner := bufio.NewScanner(anonDataFile) |
| 95 | + for anonDataFileScanner.Scan() { |
| 96 | + anonData = append(anonData, anonDataFileScanner.Text()) |
| 97 | + } |
| 98 | + |
| 99 | + return anonData, nil |
| 100 | +} |
| 101 | + |
| 102 | +func (an *Anonimizer) Run() { |
| 103 | + fmt.Println("Anonymizing data") |
| 104 | + |
| 105 | + for _, logLine := range an.csvData { |
| 106 | + for field, value := range logLine { |
| 107 | + if field == "raw" { |
| 108 | + continue |
| 109 | + } |
| 110 | + |
| 111 | + if value == "" { |
| 112 | + continue |
| 113 | + } |
| 114 | + |
| 115 | + if anonValues, exists := an.anonData[field]; exists { |
| 116 | + newAnonValue := anonValues[an.randFunc(len(anonValues))] |
| 117 | + fmt.Printf("Replacing the values for field %s. From %s to %s.\n", field, value, newAnonValue) |
| 118 | + |
| 119 | + logLine["raw"] = strings.Replace(logLine["raw"], value, newAnonValue, -1) |
| 120 | + } |
| 121 | + } |
| 122 | + |
| 123 | + an.Output = append(an.Output, fmt.Sprint(logLine["raw"])) |
| 124 | + } |
| 125 | +} |
| 126 | + |
| 127 | +func (an *Anonimizer) WriteAnonymizedOutput(filename string) error { |
| 128 | + fmt.Printf("Writing anonymized data to output file: %s", filename) |
| 129 | + file, err := os.Create(filename) |
| 130 | + if err != nil { |
| 131 | + return err |
| 132 | + } |
| 133 | + defer file.Close() |
| 134 | + |
| 135 | + for _, line := range an.Output { |
| 136 | + _, err := file.WriteString(line + "\n") |
| 137 | + if err != nil { |
| 138 | + return fmt.Errorf("writing anonymized data to output file %s: %v", filename, err) |
| 139 | + } |
| 140 | + } |
| 141 | + fmt.Println("Writing done") |
| 142 | + |
| 143 | + return nil |
| 144 | +} |
| 145 | + |
| 146 | +func (an *Anonimizer) SetRandFunc(randFunc func(int) int) { |
| 147 | + an.randFunc = randFunc |
| 148 | +} |
0 commit comments