Dot stripping

iktdev-no · Nov 19, 2024 · 01df7ef · 01df7ef
1 parent 9ecdcd8
commit 01df7ef
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 37 deletions.
diff --git a/...coordinator/src/test/kotlin/no/iktdev/mediaprocessing/coordinator/reader/KafkaTestBase.kt b/...coordinator/src/test/kotlin/no/iktdev/mediaprocessing/coordinator/reader/KafkaTestBase.kt
diff --git a/.../common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/FileNameParser.kt b/.../common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/FileNameParser.kt
@@ -1,6 +1,5 @@
 package no.iktdev.mediaprocessing.shared.common.parsing
 
-import org.apache.kafka.common.protocol.types.Field.Str
 
 class FileNameParser(val fileName: String) {
     var cleanedFileName: String
@@ -122,7 +121,7 @@ class FileNameParser(val fileName: String) {
     fun removeDot(input: String): String {
         //var text = Regex("(?<=\\s)\\.|\\.(?=\\s)").replace(input, "")
         //return Regex("\\.(?<!(Dr|Mr|Ms|Mrs|Lt|Capt|Prof|St|Ave)\\.)\\b").replace(text, " ")
-        return Regex("(?<!\\b(?:Dr|Mr|Ms|Mrs|Lt|Capt|Prof|St|Ave))\\.+(?=\\s|\\w)").replace(input, " ")
+        return Regex("(?<!\\b(?:Dr|Mr|Ms|Mrs|Lt|Capt|Prof|St|Ave))\\.").replace(input, " ")
     }
 
     fun removeInBetweenCharacters(text: String): String {

diff --git a/shared/common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/NameHelper.kt b/shared/common/src/main/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/NameHelper.kt
@@ -8,14 +8,14 @@ object NameHelper {
         val normalized = Normalizer.normalize(text, Normalizer.Form.NFC)
         val result = normalized.replace("\\p{M}".toRegex(), "")
         val cleaned = "[^A-Za-z0-9 -]".toRegex().replace(result, "")
-        return StringUtils.stripAccents(cleaned)
+        return StringUtils.stripAccents(cleaned).trim()
     }
 
     fun cleanup(input: String): String {
         var cleaned = Regex("(?<=\\w)[_.](?=\\w)").replace(input, " ")
         cleaned = Regexes.illegalCharacters.replace(cleaned, " - ")
         cleaned = Regexes.trimWhiteSpaces.replace(cleaned, " ")
-        return NameHelper.normalize(cleaned)
+        return NameHelper.normalize(cleaned).trim()
     }
 }
 

diff --git a/...mon/src/test/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/FileNameParserTest.kt b/...mon/src/test/kotlin/no/iktdev/mediaprocessing/shared/common/parsing/FileNameParserTest.kt
@@ -36,6 +36,15 @@ class FileNameParserTest {
 
     }
 
+    @Test
+    fun assertDotRemoval() {
+        val input = "Like.a.Potato.Chef.S01E01.Departure.\\u0026.Skills.1080p.Potato"
+        val parser = FileNameParser(input)
+        val result = parser.guessDesiredTitle()
+        assertThat(result).isEqualTo("Like a Potato Chef")
+        assertThat(parser.guessDesiredFileName()).isEqualTo("Like a Potato Chef S01E01 Departure \\u0026 Skills")
+    }
+
     @Test
     fun movieName() {
         val inName = "Wicket.Wicker.Potato.4.2023.UHD.BluRay.2160p"
@@ -89,5 +98,4 @@ class FileNameParserTest {
         val result = FileNameParser(input).guessDesiredTitle()
         assertThat(result).isEmpty()
     }
-
 }