From db6ea5d529ea437f193afb5b5c1892f63e409600 Mon Sep 17 00:00:00 2001
From: androidlover5842 <androidlover5842@gmail.com>
Date: Sat, 31 Jan 2026 12:46:48 +0530
Subject: [PATCH] Normalize extracted address fields

---
 .../component/DocumentExtractionService.kt    | 59 +++++++++++++++++++
 1 file changed, 59 insertions(+)

diff --git a/src/main/kotlin/com/android/trisolarisserver/component/DocumentExtractionService.kt b/src/main/kotlin/com/android/trisolarisserver/component/DocumentExtractionService.kt
index 9799294..29c7983 100644
--- a/src/main/kotlin/com/android/trisolarisserver/component/DocumentExtractionService.kt
+++ b/src/main/kotlin/com/android/trisolarisserver/component/DocumentExtractionService.kt
@@ -244,6 +244,7 @@ class DocumentExtractionService(
         normalizePinCode(results)
         normalizeIdNumber(results)
         markAadhaarIfValid(results)
+        normalizeAddress(results)
         applyBookingCityUpdates(document, results)
         results["docType"] = computeDocType(results, handled)
         applyGuestUpdates(document, propertyId, results)
@@ -282,6 +283,13 @@ class DocumentExtractionService(
         }
     }
 
+    private fun normalizeAddress(results: MutableMap<String, String>) {
+        val key = DocumentPrompts.ADDRESS.first
+        val raw = cleanedValue(results[key]) ?: return
+        val normalized = cleanAddress(raw) ?: return
+        results[key] = normalized
+    }
+
     private fun markAadhaarIfValid(results: MutableMap<String, String>) {
         val idKey = DocumentPrompts.ID_NUMBER.first
         val digits = normalizeDigits(cleanedValue(results[idKey]))
@@ -519,3 +527,54 @@ private fun isValidPin(value: String?): Boolean {
     if (value.isNullOrBlank()) return false
     return pinCodeRegex.matches(value)
 }
+
+private fun cleanAddress(raw: String): String? {
+    val relationRegex = Regex("^\\s*(S/O|D/O|W/O|C/O|H/O|F/O)\\b", RegexOption.IGNORE_CASE)
+    val prefixRegexes = listOf(
+        Regex("^\\s*ADDRESS\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*HOUSE\\s*/\\s*BLDG\\.?\\s*/\\s*APT\\.?\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*HOUSE-?BLDG\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*HOUSE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*H\\s*NO\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*HOUSE\\s*NO\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*STREET/ROAD/LANE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*STREET\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*ROAD\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*LANE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*AREA\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*SECTOR\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*COLONY\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*VILLAGE/TOWN/CITY\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*VILLAGE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*TOWN\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*CITY\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*P\\.?\\s*O\\.?\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*POST\\s*OFFICE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*P\\.?\\s*DIST\\.?\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*DISTRICT\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*DIST\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*STATE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*PIN\\s*CODE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*PINCODE\\b[:\\-\\s]*", RegexOption.IGNORE_CASE),
+        Regex("^\\s*PIN\\b[:\\-\\s]*", RegexOption.IGNORE_CASE)
+    )
+    val dropPhrases = setOf(
+        "address", "addr", "area", "area was", "street/road/lane", "village/town/city", "colony"
+    )
+    val parts = raw.replace("\n", ",").split(",")
+    val cleanedParts = mutableListOf<String>()
+    for (part in parts) {
+        var value = part.trim()
+        if (value.isBlank()) continue
+        if (relationRegex.containsMatchIn(value)) continue
+        for (regex in prefixRegexes) {
+            value = regex.replace(value, "").trim()
+        }
+        value = value.replace(Regex("\\s+"), " ").trim()
+        if (value.isBlank()) continue
+        if (value.length < 3) continue
+        if (dropPhrases.contains(value.lowercase())) continue
+        cleanedParts.add(value)
+    }
+    return if (cleanedParts.isEmpty()) null else cleanedParts.joinToString(", ")
+}