ADFA-3938 | Fix OCR property scaling issues

cv-image-to-xml/src/main/java/org/appdevforall/codeonthego/computervision/domain/parser/ValueCleanersImpl.kt

@@ -45,19 +45,45 @@ internal object DimensionCleaner : ValueCleaner {
     private val matchKeywords = setOf("match", "parent")
     private val wrapKeywords = setOf("wrap", "content", "wrapcan")
     private val DIMENSION_CONSTANTS = listOf("wrap_content", "match_parent")
+    private val explicitDimensionRegex = Regex("^(-?\\d+)(dp|sp|px|dip)$")
 
     override fun clean(rawValue: String): String {
-        val normalized = rawValue.lowercase().replace(" ", "_")
+        val trimmedValue = rawValue.trim()
+        val normalized = trimmedValue.lowercase().replace(" ", "_")
+
         if (matchKeywords.any { it in normalized }) return "match_parent"
         if (wrapKeywords.any { it in normalized }) return "wrap_content"
 
         val fuzzyResult = FuzzySearch.extractOne(normalized, DIMENSION_CONSTANTS)
         if (fuzzyResult.score >= 60) return fuzzyResult.string
 
         val fixedUnit = normalized.replace(Regex("0p$|op$|olp$"), "dp")
+        explicitDimensionRegex.matchEntire(fixedUnit)?.let { match ->
+            val normalizedNumber = normalizeOcrDimensionNumber(match.groupValues[1])
+            return normalizedNumber + match.groupValues[2]
+        }
+
         val numericPart = NumberCleaner.clean(fixedUnit.replace("_", ""))
+        val normalizedNumericPart = normalizeOcrDimensionNumber(numericPart)
+
+        return if (numericPart != fixedUnit) "${normalizedNumericPart}dp" else trimmedValue
+    }
+
+    private fun normalizeOcrDimensionNumber(numericPart: String): String {
+        if (!numericPart.matches(Regex("-?\\d+"))) return numericPart
+
+        val isNegative = numericPart.startsWith("-")
+        val numericValue = numericPart.toLongOrNull() ?: return numericPart
+        val canonical = numericValue.toString()
+        val unsignedCanonical = canonical.removePrefix("-")
+
+        // OCR sometimes reads the trailing "dp" as a single zero, turning 150dp into 1500.
+        if (unsignedCanonical.endsWith('0') && unsignedCanonical.toLong() >= 1000L) {
+            val normalizedValue = numericValue / 10L
+            return normalizedValue.toString()
+        }
 
-        return if (numericPart != fixedUnit) "${numericPart}dp" else rawValue
+        return if (isNegative && numericValue == 0L) "0" else canonical
     }
 }
 

cv-image-to-xml/src/test/java/org/appdevforall/codeonthego/computervision/domain/FuzzyAttributeParserTest.kt

@@ -132,6 +132,30 @@ class FuzzyAttributeParserTest {
         assertEquals("wrap_content", result["android:layout_height"])
     }
 
+    @Test
+    fun `OCR dimension with trailing zero before dp is normalized`() {
+        val annotation = "height: 1500dp | width: 100dp"
+        val result = FuzzyAttributeParser.parse(annotation, "EditText")
+
+        assertEquals("150dp", result["android:layout_height"])
+    }
+
+    @Test
+    fun `zero padded dimension values are normalized`() {
+        val annotation = "height: 0010dp | width: 100dp"
+        val result = FuzzyAttributeParser.parse(annotation, "EditText")
+
+        assertEquals("10dp", result["android:layout_height"])
+    }
+
+    @Test
+    fun `all zero padded dimension values normalize to zero`() {
+        val annotation = "height: 0000dp | width: 100dp"
+        val result = FuzzyAttributeParser.parse(annotation, "EditText")
+
+        assertEquals("0dp", result["android:layout_height"])
+    }
+
     @Test
     fun `empty chunks between pipes are skipped`() {
         val annotation = "width: 100dp | | height: 80dp"