|
| 1 | +package parser |
| 2 | + |
| 3 | +import ( |
| 4 | + "bytes" |
| 5 | + "errors" |
| 6 | + "fmt" |
| 7 | + "math" |
| 8 | + "regexp" |
| 9 | + "strconv" |
| 10 | + "strings" |
| 11 | + |
| 12 | + "expenses/internal/models" |
| 13 | + "expenses/pkg/logger" |
| 14 | + "expenses/pkg/utils" |
| 15 | + |
| 16 | + "github.com/xuri/excelize/v2" |
| 17 | +) |
| 18 | + |
| 19 | +// AxisCreditParser parses Axis credit card statements exported as XLSX |
| 20 | +type AxisCreditParser struct{} |
| 21 | + |
| 22 | +func (p *AxisCreditParser) Parse(fileBytes []byte, metadata string, fileName string, password string) ([]models.CreateTransactionInput, error) { |
| 23 | + f, err := excelize.OpenReader(bytes.NewReader(fileBytes)) |
| 24 | + if err != nil { |
| 25 | + return nil, fmt.Errorf("failed to open xlsx: %w", err) |
| 26 | + } |
| 27 | + defer func() { |
| 28 | + _ = f.Close() |
| 29 | + }() |
| 30 | + |
| 31 | + for _, sheet := range f.GetSheetList() { |
| 32 | + rows, err := f.GetRows(sheet) |
| 33 | + if err != nil { |
| 34 | + continue |
| 35 | + } |
| 36 | + |
| 37 | + headerIndex := -1 |
| 38 | + var dateIdx, descIdx, amountIdx, signIdx int |
| 39 | + for i, row := range rows { |
| 40 | + if len(row) == 0 { |
| 41 | + continue |
| 42 | + } |
| 43 | + joined := strings.ToLower(strings.Join(row, " ")) |
| 44 | + if strings.Contains(joined, "date") && strings.Contains(joined, "amount") && (strings.Contains(joined, "debit") || strings.Contains(joined, "credit") || strings.Contains(joined, "debit/credit")) { |
| 45 | + headerIndex = i |
| 46 | + // determine column indexes |
| 47 | + dateIdx, descIdx, amountIdx, signIdx = -1, -1, -1, -1 |
| 48 | + for idx, cell := range row { |
| 49 | + lc := strings.ToLower(strings.TrimSpace(cell)) |
| 50 | + if dateIdx == -1 && strings.Contains(lc, "date") { |
| 51 | + dateIdx = idx |
| 52 | + } |
| 53 | + if descIdx == -1 && (strings.Contains(lc, "transaction") || strings.Contains(lc, "transaction details") || strings.Contains(lc, "details")) { |
| 54 | + descIdx = idx |
| 55 | + } |
| 56 | + if amountIdx == -1 && strings.Contains(lc, "amount") { |
| 57 | + amountIdx = idx |
| 58 | + } |
| 59 | + if signIdx == -1 && (strings.Contains(lc, "debit") || strings.Contains(lc, "credit")) { |
| 60 | + signIdx = idx |
| 61 | + } |
| 62 | + } |
| 63 | + break |
| 64 | + } |
| 65 | + } |
| 66 | + |
| 67 | + if headerIndex == -1 { |
| 68 | + // try next sheet |
| 69 | + continue |
| 70 | + } |
| 71 | + |
| 72 | + var transactions []models.CreateTransactionInput |
| 73 | + for i := headerIndex + 1; i < len(rows); i++ { |
| 74 | + row := rows[i] |
| 75 | + // Skip empty rows |
| 76 | + if len(row) == 0 { |
| 77 | + continue |
| 78 | + } |
| 79 | + |
| 80 | + // Normalize cell slice to avoid index issues |
| 81 | + for len(row) <= descIdx && len(row) < 5 { |
| 82 | + row = append(row, "") |
| 83 | + } |
| 84 | + |
| 85 | + txn, err := p.parseTransactionRow(row, dateIdx, descIdx, amountIdx, signIdx) |
| 86 | + if err != nil { |
| 87 | + logger.Warnf("Failed to parse row %d in sheet %s: %v", i+1, sheet, err) |
| 88 | + continue |
| 89 | + } |
| 90 | + if txn != nil { |
| 91 | + transactions = append(transactions, *txn) |
| 92 | + } |
| 93 | + } |
| 94 | + |
| 95 | + if len(transactions) > 0 { |
| 96 | + return transactions, nil |
| 97 | + } |
| 98 | + } |
| 99 | + |
| 100 | + return nil, errors.New("transaction header row not found in Axis credit statement") |
| 101 | +} |
| 102 | + |
| 103 | +func (p *AxisCreditParser) parseTransactionRow(row []string, dateIdx int, descIdx int, amountIdx int, signIdx int) (*models.CreateTransactionInput, error) { |
| 104 | + // Validate indexes |
| 105 | + if dateIdx < 0 || descIdx < 0 { |
| 106 | + return nil, nil |
| 107 | + } |
| 108 | + |
| 109 | + get := func(idx int) string { |
| 110 | + if idx >= 0 && idx < len(row) { |
| 111 | + return strings.TrimSpace(row[idx]) |
| 112 | + } |
| 113 | + return "" |
| 114 | + } |
| 115 | + |
| 116 | + dateStr := get(dateIdx) |
| 117 | + if dateStr == "" || strings.EqualFold(dateStr, "date") { |
| 118 | + return nil, nil |
| 119 | + } |
| 120 | + |
| 121 | + // Normalize two-digit year like 14 Nov '25 -> 14 Nov 2025 |
| 122 | + re := regexp.MustCompile(`'(?P<yy>\d{2})`) |
| 123 | + dateStr = re.ReplaceAllString(dateStr, "20$1") |
| 124 | + |
| 125 | + txnDate, err := utils.ParseDate(dateStr) |
| 126 | + if err != nil { |
| 127 | + // Not a valid transaction row |
| 128 | + return nil, nil |
| 129 | + } |
| 130 | + |
| 131 | + desc := get(descIdx) |
| 132 | + // If description is empty, try next available column |
| 133 | + if desc == "" { |
| 134 | + if len(row) > descIdx+1 { |
| 135 | + desc = strings.TrimSpace(row[descIdx+1]) |
| 136 | + } |
| 137 | + } |
| 138 | + |
| 139 | + amountStr := get(amountIdx) |
| 140 | + var inferredSign string |
| 141 | + var amount float64 |
| 142 | + // If amount not found in configured column, try to find numeric cell after description |
| 143 | + if amountStr == "" { |
| 144 | + for j := descIdx + 1; j < len(row); j++ { |
| 145 | + cell := strings.TrimSpace(row[j]) |
| 146 | + if cell == "" { |
| 147 | + continue |
| 148 | + } |
| 149 | + if v, sgn, err := p.parseAmount(cell); err == nil { |
| 150 | + amountStr = cell |
| 151 | + inferredSign = sgn |
| 152 | + amount = v |
| 153 | + break |
| 154 | + } |
| 155 | + } |
| 156 | + } |
| 157 | + |
| 158 | + if amountStr == "" { |
| 159 | + return nil, fmt.Errorf("amount not found for row: %v", row) |
| 160 | + } |
| 161 | + |
| 162 | + // Try parse amount from amountStr if not already parsed |
| 163 | + if amount == 0 { |
| 164 | + v, sgn, err := p.parseAmount(amountStr) |
| 165 | + if err != nil { |
| 166 | + return nil, fmt.Errorf("failed to parse amount '%s': %w", amountStr, err) |
| 167 | + } |
| 168 | + amount = v |
| 169 | + if inferredSign == "" { |
| 170 | + inferredSign = sgn |
| 171 | + } |
| 172 | + } |
| 173 | + |
| 174 | + // sign handling: prefer explicit sign column; fall back to inferred sign |
| 175 | + sign := strings.ToLower(get(signIdx)) |
| 176 | + if sign != "" { |
| 177 | + if strings.Contains(sign, "debit") || strings.Contains(sign, "dr") { |
| 178 | + amount = math.Abs(amount) |
| 179 | + } else if strings.Contains(sign, "credit") || strings.Contains(sign, "cr") { |
| 180 | + amount = -math.Abs(amount) |
| 181 | + } |
| 182 | + } else { |
| 183 | + switch inferredSign { |
| 184 | + case "debit": |
| 185 | + amount = math.Abs(amount) |
| 186 | + case "credit": |
| 187 | + amount = -math.Abs(amount) |
| 188 | + } |
| 189 | + } |
| 190 | + |
| 191 | + name := strings.TrimSpace(desc) |
| 192 | + if len(name) > 40 { |
| 193 | + name = name[:37] + "..." |
| 194 | + } |
| 195 | + |
| 196 | + txn := &models.CreateTransactionInput{ |
| 197 | + CreateBaseTransactionInput: models.CreateBaseTransactionInput{ |
| 198 | + Name: name, |
| 199 | + Description: desc, |
| 200 | + Amount: &amount, |
| 201 | + Date: txnDate, |
| 202 | + }, |
| 203 | + CategoryIds: []int64{}, |
| 204 | + } |
| 205 | + return txn, nil |
| 206 | +} |
| 207 | + |
| 208 | +// parseAmount cleans the raw amount cell and returns numeric value and inferred sign ("debit" or "credit") if detectable. |
| 209 | +func (p *AxisCreditParser) parseAmount(raw string) (float64, string, error) { |
| 210 | + s := strings.TrimSpace(raw) |
| 211 | + if s == "" { |
| 212 | + return 0, "", fmt.Errorf("empty amount") |
| 213 | + } |
| 214 | + |
| 215 | + // Replace non-breaking spaces |
| 216 | + s = strings.ReplaceAll(s, "\u00A0", " ") |
| 217 | + lower := strings.ToLower(s) |
| 218 | + inferred := "" |
| 219 | + |
| 220 | + // Parentheses indicate negative amount (treat as credit/payment) |
| 221 | + if strings.Contains(s, "(") && strings.Contains(s, ")") { |
| 222 | + // don't make the numeric value negative here; infer sign as credit |
| 223 | + inferred = "credit" |
| 224 | + } |
| 225 | + |
| 226 | + // Trailing CR/DR indicators |
| 227 | + if strings.HasSuffix(strings.TrimSpace(lower), "cr") { |
| 228 | + inferred = "credit" |
| 229 | + s = strings.TrimSpace(s[:len(s)-2]) |
| 230 | + } else if strings.HasSuffix(strings.TrimSpace(lower), "dr") { |
| 231 | + inferred = "debit" |
| 232 | + s = strings.TrimSpace(s[:len(s)-2]) |
| 233 | + } |
| 234 | + // Remove currency symbols and words |
| 235 | + replacements := []string{"₹", "rs.", "rs", "inr"} |
| 236 | + for _, r := range replacements { |
| 237 | + s = strings.ReplaceAll(strings.ToLower(s), r, "") |
| 238 | + } |
| 239 | + |
| 240 | + // Remove commas, spaces and parentheses |
| 241 | + s = strings.ReplaceAll(s, ",", "") |
| 242 | + s = strings.ReplaceAll(s, " ", "") |
| 243 | + s = strings.ReplaceAll(s, "(", "") |
| 244 | + s = strings.ReplaceAll(s, ")", "") |
| 245 | + |
| 246 | + if s == "" { |
| 247 | + return 0, inferred, fmt.Errorf("empty amount after cleaning") |
| 248 | + } |
| 249 | + |
| 250 | + v, err := strconv.ParseFloat(s, 64) |
| 251 | + if err != nil { |
| 252 | + return 0, inferred, err |
| 253 | + } |
| 254 | + |
| 255 | + // If parentheses were present we already set inferred to credit; keep value positive |
| 256 | + return v, inferred, nil |
| 257 | +} |
| 258 | + |
| 259 | +func init() { |
| 260 | + RegisterParser(models.BankTypeAxisCredit, &AxisCreditParser{}) |
| 261 | +} |
0 commit comments