autoDetectColumnType intelligently detects if a column contains numeric, date, or string data Returns colTypeDate for dates, colTypeFloat for numbers, colTypeStr for strings
(colIndex int)
| 476 | // autoDetectColumnType intelligently detects if a column contains numeric, date, or string data |
| 477 | // Returns colTypeDate for dates, colTypeFloat for numbers, colTypeStr for strings |
| 478 | func (b *Buffer) autoDetectColumnType(colIndex int) int { |
| 479 | b.mu.RLock() |
| 480 | defer b.mu.RUnlock() |
| 481 | |
| 482 | if colIndex < 0 || colIndex >= b.colLen { |
| 483 | return colTypeStr |
| 484 | } |
| 485 | |
| 486 | // Sample size for type detection |
| 487 | startRow := b.rowFreeze |
| 488 | endRow := b.rowLen |
| 489 | |
| 490 | // For large datasets, sample smartly (first N rows + some middle + last N) |
| 491 | sampleSize := 100 |
| 492 | sampleRows := []int{} |
| 493 | |
| 494 | if endRow-startRow > sampleSize { |
| 495 | // Sample first 50 rows |
| 496 | for i := startRow; i < startRow+50 && i < endRow; i++ { |
| 497 | sampleRows = append(sampleRows, i) |
| 498 | } |
| 499 | // Sample middle 25 rows |
| 500 | midPoint := (startRow + endRow) / 2 |
| 501 | for i := midPoint; i < midPoint+25 && i < endRow; i++ { |
| 502 | sampleRows = append(sampleRows, i) |
| 503 | } |
| 504 | // Sample last 25 rows |
| 505 | for i := endRow - 25; i < endRow; i++ { |
| 506 | if i > startRow { |
| 507 | sampleRows = append(sampleRows, i) |
| 508 | } |
| 509 | } |
| 510 | } else { |
| 511 | // For small datasets, check all rows |
| 512 | for i := startRow; i < endRow; i++ { |
| 513 | sampleRows = append(sampleRows, i) |
| 514 | } |
| 515 | } |
| 516 | |
| 517 | // Analyze samples |
| 518 | dateCount := 0 |
| 519 | numericCount := 0 |
| 520 | totalCount := 0 |
| 521 | |
| 522 | for _, rowIdx := range sampleRows { |
| 523 | if rowIdx >= b.rowLen || colIndex >= len(b.cont[rowIdx]) { |
| 524 | continue |
| 525 | } |
| 526 | |
| 527 | value := strings.TrimSpace(b.cont[rowIdx][colIndex]) |
| 528 | |
| 529 | // Skip empty/null cells |
| 530 | if value == "" || value == "NA" || value == "N/A" || value == "NaN" || value == "null" { |
| 531 | continue |
| 532 | } |
| 533 | |
| 534 | totalCount++ |
| 535 |