Updated

apps/coordinator/src/main/kotlin/no/iktdev/mediaprocessing/coordinator/tasks/event/MetadataAndBaseInfoToCoverTask.kt

@@ -3,6 +3,7 @@ package no.iktdev.mediaprocessing.coordinator.tasks.event
 import mu.KotlinLogging
 import no.iktdev.mediaprocessing.coordinator.Coordinator
 import no.iktdev.mediaprocessing.coordinator.TaskCreator
+import no.iktdev.mediaprocessing.shared.common.parsing.NameHelper
 import no.iktdev.mediaprocessing.shared.common.persistance.PersistentMessage
 import no.iktdev.mediaprocessing.shared.kafka.core.KafkaEvents
 import no.iktdev.mediaprocessing.shared.kafka.dto.MessageDataWrapper
@@ -51,7 +52,7 @@ class MetadataAndBaseInfoToCoverTask(@Autowired override var coordinator: Coordi
             CoverInfoPerformed(
                 status = Status.COMPLETED,
                 url = coverUrl,
-                outFileBaseName = coverTitle,
+                outFileBaseName = NameHelper.normalize(coverTitle),
                 outDir = fileOut.outDirectory,
                 derivedFromEventId = event.eventId
             )

apps/coordinator/src/main/kotlin/no/iktdev/mediaprocessing/coordinator/tasks/event/MetadataAndBaseInfoToFileOut.kt

@@ -9,6 +9,7 @@ import no.iktdev.mediaprocessing.shared.common.SharedConfig
 import no.iktdev.mediaprocessing.shared.common.datasource.toEpochSeconds
 import no.iktdev.mediaprocessing.shared.common.lastOrSuccessOf
 import no.iktdev.mediaprocessing.shared.common.parsing.FileNameDeterminate
+import no.iktdev.mediaprocessing.shared.common.parsing.NameHelper
 import no.iktdev.mediaprocessing.shared.common.parsing.Regexes
 import no.iktdev.mediaprocessing.shared.common.persistance.PersistentMessage
 import no.iktdev.mediaprocessing.shared.kafka.core.KafkaEnv
@@ -134,7 +135,7 @@ class MetadataAndBaseInfoToFileOut(@Autowired override var coordinator: Coordina
             }
         }
 
-        fun getOutputDirectory() = SharedConfig.outgoingContent.using(getTitle())
+        fun getOutputDirectory() = SharedConfig.outgoingContent.using(NameHelper.normalize(getTitle()))
 
 
 

shared/common/build.gradle.kts

@@ -39,6 +39,8 @@ dependencies {
     implementation("org.jetbrains.exposed:exposed-jdbc:$exposedVersion")
     implementation("org.jetbrains.exposed:exposed-java-time:$exposedVersion")
     implementation ("mysql:mysql-connector-java:8.0.29")
+    implementation("org.apache.commons:commons-lang3:3.12.0")
+
 
     implementation(project(mapOf("path" to ":shared:kafka")))
     implementation(project(mapOf("path" to ":shared:contract")))

shared/common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/FileNameDeterminate.kt

@@ -76,7 +76,7 @@ class FileNameDeterminate(val title: String, val sanitizedName: String, val ctyp
         var cleaned = Regex("(?<=\\w)[_.](?=\\w)").replace(input, " ")
         cleaned = Regexes.illegalCharacters.replace(cleaned, " - ")
         cleaned = Regexes.trimWhiteSpaces.replace(cleaned, " ")
-        return cleaned
+        return NameHelper.normalize(cleaned)
     }
 
     open internal class Base(val title: String, val sanitizedName: String) {
@@ -130,8 +130,7 @@ class FileNameDeterminate(val title: String, val sanitizedName: String, val ctyp
          *      Cool - S1 13
          */
         fun findSeasonAndEpisode(inputText: String): Pair<String?, String?> {
-            val regex = Regex("""(?i)\b(?:S|Season)\s*(\d+).*?(?:E|Episode)?\s*(\d+)\b""")
+            val matchResult = Regexes.SeasonEpisodeBlock.find(inputText)
-            val matchResult = regex.find(inputText)
             val season = matchResult?.groups?.get(1)?.value
             val episode = matchResult?.groups?.get(2)?.value
             return season to episode
@@ -149,10 +148,15 @@ class FileNameDeterminate(val title: String, val sanitizedName: String, val ctyp
         }
 
         fun findEpisodeTitle(): String? {
+            var startPosition: Int = 0
+            startPosition = Regexes.SeasonEpisodeBlock.find(sanitizedName)?.value?.let { block ->
+                sanitizedName.indexOf(block) + block.length
+            } ?: 0
+
             val seCombo = findSeasonAndEpisode(sanitizedName)
             val episodeNumber = findEpisodeNumber()
 
-            val startPosition = if (seCombo.second != null) sanitizedName.indexOf(seCombo.second!!)+ seCombo.second!!.length
+            startPosition = if (startPosition != 0) startPosition else if (seCombo.second != null) sanitizedName.indexOf(seCombo.second!!)+ seCombo.second!!.length
             else if (episodeNumber != null) sanitizedName.indexOf(episodeNumber) + episodeNumber.length else 0
             val availableText = sanitizedName.substring(startPosition)
 

shared/common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/NameHelper.kt

@@ -0,0 +1,12 @@
+package no.iktdev.mediaprocessing.shared.common.parsing
+
+import org.apache.commons.lang3.StringUtils
+import java.text.Normalizer
+
+object NameHelper {
+    fun normalize(text: String): String {
+        val normalized = Normalizer.normalize(text, Normalizer.Form.NFC)
+        val result = normalized.replace("\\p{M}".toRegex(), "")
+        return StringUtils.stripAccents(result)
+    }
+}

shared/common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/Regexes.kt

@@ -3,6 +3,6 @@ package no.iktdev.mediaprocessing.shared.common.parsing
 object Regexes {
     val illegalCharacters = Regex("""[\\\\\\/\\:\\*\\?\\\"\\<\\>|!`()\[\]]""")
     val trimWhiteSpaces = Regex("\\s{2,}")
-    val seasonAndEpisode = Regex("""(?i)\b(?:S|Season)\s*(\d+).*?(?:E|Episode)?\s*(\d+)\b""", RegexOption.IGNORE_CASE)
+    val SeasonEpisodeBlock = Regex("""(?i)\b(?:S|Season)\s*(\d+).*?(?:E|Episode)?\s*(\d+)\b""", RegexOption.IGNORE_CASE)
     val season = Regex("""(?i)\b(?:S|Season)\s*-?\s*(\d+)""", RegexOption.IGNORE_CASE)
 }