Add OCRConfidence

table.go

@@ -186,6 +186,26 @@ func (tr *TableRow) Cells() []Cell {
  return tr.cells
 }
 
+// OCRConfidence returns the OCR confidence for the table row.
+func (tr *TableRow) OCRConfidence() *OCRConfidence {
+ meanValues := make([]float64, 0, len(tr.cells))
+ maxValues := make([]float64, 0, len(tr.cells))
+ minValues := make([]float64, 0, len(tr.cells))
+
+ for _, cell := range tr.cells {
+ c := cell.OCRConfidence()
+ meanValues = append(meanValues, c.Mean())
+ maxValues = append(maxValues, c.Max())
+ minValues = append(minValues, c.Min())
+ }
+
+ return &OCRConfidence{
+ mean: internal.Mean(meanValues),
+ max: slices.Max(maxValues),
+ min: slices.Min(minValues),
+ }
+}
+
 type RowsOptions struct {
  IgnoreMergedCells bool
 }

table_cell.go

@@ -14,6 +14,7 @@ type Cell interface {
  Words() []*Word
  Text(optFns ...func(*TextLinearizationOptions)) string
  Confidence() float64
+ OCRConfidence() *OCRConfidence
  IsColumnHeader() bool
  IsTableTitle() bool
  IsTableFooter() bool
@@ -51,6 +52,26 @@ func (tmc *TableMergedCell) Text(_ ...func(*TextLinearizationOptions)) string {
  return strings.Join(texts, " ")
 }
 
+// OCRConfidence returns the OCR confidence for the merged cell.
+func (tmc *TableMergedCell) OCRConfidence() *OCRConfidence {
+ meanValues := make([]float64, 0, len(tmc.cells))
+ maxValues := make([]float64, 0, len(tmc.cells))
+ minValues := make([]float64, 0, len(tmc.cells))
+
+ for _, cell := range tmc.cells {
+ c := cell.OCRConfidence()
+ meanValues = append(meanValues, c.Mean())
+ maxValues = append(maxValues, c.Max())
+ minValues = append(minValues, c.Min())
+ }
+
+ return &OCRConfidence{
+ mean: internal.Mean(meanValues),
+ max: slices.Max(maxValues),
+ min: slices.Min(minValues),
+ }
+}
+
 // TableCell represents a cell in a table.
 type TableCell struct {
  cell
@@ -82,6 +103,28 @@ func (tc *TableCell) Text(optFns ...func(*TextLinearizationOptions)) string {
  return strings.Join(texts, " ")
 }
 
+// OCRConfidence returns the OCR confidence for the table cell.
+func (tc *TableCell) OCRConfidence() *OCRConfidence {
+ if tc.selectionElement != nil {
+ return &OCRConfidence{
+ mean: tc.SelectionElement().Confidence(),
+ min: tc.SelectionElement().Confidence(),
+ max: tc.SelectionElement().Confidence(),
+ }
+ }
+
+ scores := make([]float64, len(tc.words))
+ for i, w := range tc.words {
+ scores[i] = w.Confidence()
+ }
+
+ return &OCRConfidence{
+ mean: internal.Mean(scores),
+ max: slices.Max(scores),
+ min: slices.Min(scores),
+ }
+}
+
 // cell represents the base information shared among different types of table cells.
 type cell struct {
  base

table_cell_test.go

@@ -77,4 +77,19 @@ func TestTableCell(t *testing.T) {
  result := cell.Text()
  assert.Equal(t, "Hello World", result, "Text mismatch")
  })
+
+ t.Run("OCRConfindence", func(t *testing.T) {
+ // Create sample words
+ word1 := &Word{base: base{confidence: 0.9}}
+ word2 := &Word{base: base{confidence: 0.1}}
+
+ // Create a TableCell with sample words
+ cell := &TableCell{words: []*Word{word1, word2}}
+
+ // Call the OCRConfidence method and check if it returns the expected scores
+ result := cell.OCRConfidence()
+ assert.Equal(t, 0.5, result.Mean())
+ assert.Equal(t, 0.1, result.Min())
+ assert.Equal(t, 0.9, result.Max())
+ })
 }