Improve chapter recognition (#8050)

This commit is contained in:
stevenyomi 2022-09-24 22:29:55 +08:00 committed by GitHub
parent b5dca2eb09
commit 4b4a138eee
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 32 additions and 46 deletions

View File

@ -4,35 +4,31 @@ package eu.kanade.tachiyomi.util.chapter
* -R> = regex conversion. * -R> = regex conversion.
*/ */
object ChapterRecognition { object ChapterRecognition {
private const val NUMBER_PATTERN = """([0-9]+)(\.[0-9]+)?(\.?[a-z]+)?"""
/** /**
* All cases with Ch.xx * All cases with Ch.xx
* Mokushiroku Alice Vol.1 Ch. 4: Misrepresentation -R> 4 * Mokushiroku Alice Vol.1 Ch. 4: Misrepresentation -R> 4
*/ */
private val basic = Regex("""(?<=ch\.) *([0-9]+)(\.[0-9]+)?(\.?[a-z]+)?""") private val basic = Regex("""(?<=ch\.) *$NUMBER_PATTERN""")
/** /**
* Regex used when only one number occurrence
* Example: Bleach 567: Down With Snowwhite -R> 567 * Example: Bleach 567: Down With Snowwhite -R> 567
*/ */
private val occurrence = Regex("""([0-9]+)(\.[0-9]+)?(\.?[a-z]+)?""") private val number = Regex(NUMBER_PATTERN)
/**
* Regex used when manga title removed
* Example: Solanin 028 Vol. 2 -> 028 Vol.2 -> 028Vol.2 -R> 028
*/
private val withoutManga = Regex("""^([0-9]+)(\.[0-9]+)?(\.?[a-z]+)?""")
/** /**
* Regex used to remove unwanted tags * Regex used to remove unwanted tags
* Example Prison School 12 v.1 vol004 version1243 volume64 -R> Prison School 12 * Example Prison School 12 v.1 vol004 version1243 volume64 -R> Prison School 12
*/ */
private val unwanted = Regex("""(?<![a-z])(v|ver|vol|version|volume|season|s).?[0-9]+""") private val unwanted = Regex("""\b(?:v|ver|vol|version|volume|season|s)[^a-z]?[0-9]+""")
/** /**
* Regex used to remove unwanted whitespace * Regex used to remove unwanted whitespace
* Example One Piece 12 special -R> One Piece 12special * Example One Piece 12 special -R> One Piece 12special
*/ */
private val unwantedWhiteSpace = Regex("""(\s)(extra|special|omake)""") private val unwantedWhiteSpace = Regex("""\s(?=extra|special|omake)""")
fun parseChapterNumber(mangaTitle: String, chapterName: String, chapterNumber: Float? = null): Float { fun parseChapterNumber(mangaTitle: String, chapterName: String, chapterNumber: Float? = null): Float {
// If chapter number is known return. // If chapter number is known return.
@ -43,40 +39,23 @@ object ChapterRecognition {
// Get chapter title with lower case // Get chapter title with lower case
var name = chapterName.lowercase() var name = chapterName.lowercase()
// Remove manga title from chapter title.
name = name.replace(mangaTitle.lowercase(), "").trim()
// Remove comma's or hyphens. // Remove comma's or hyphens.
name = name.replace(',', '.').replace('-', '.') name = name.replace(',', '.').replace('-', '.')
// Remove unwanted white spaces. // Remove unwanted white spaces.
unwantedWhiteSpace.findAll(name).let { name = unwantedWhiteSpace.replace(name, "")
it.forEach { occurrence -> name = name.replace(occurrence.value, occurrence.value.trim()) }
}
// Remove unwanted tags. // Remove unwanted tags.
unwanted.findAll(name).let { name = unwanted.replace(name, "")
it.forEach { occurrence -> name = name.replace(occurrence.value, "") }
}
// Check base case ch.xx // Check base case ch.xx
getChapterNumberFromMatch(basic.find(name))?.let { return it } basic.find(name)?.let { return getChapterNumberFromMatch(it) }
// Check one number occurrence.
val occurrences: MutableList<MatchResult> = arrayListOf()
occurrence.findAll(name).let {
it.forEach { occurrence -> occurrences.add(occurrence) }
}
if (occurrences.size == 1) {
getChapterNumberFromMatch(occurrences[0])?.let { return it }
}
// Remove manga title from chapter title.
val nameWithoutManga = name.replace(mangaTitle.lowercase(), "").trim()
// Check if first value is number after title remove.
getChapterNumberFromMatch(withoutManga.find(nameWithoutManga))?.let { return it }
// Take the first number encountered. // Take the first number encountered.
getChapterNumberFromMatch(occurrence.find(nameWithoutManga))?.let { return it } number.find(name)?.let { return getChapterNumberFromMatch(it) }
return chapterNumber ?: -1f return chapterNumber ?: -1f
} }
@ -86,8 +65,8 @@ object ChapterRecognition {
* @param match result of regex * @param match result of regex
* @return chapter number if found else null * @return chapter number if found else null
*/ */
private fun getChapterNumberFromMatch(match: MatchResult?): Float? { private fun getChapterNumberFromMatch(match: MatchResult): Float {
return match?.let { return match.let {
val initial = it.groups[1]?.value?.toFloat()!! val initial = it.groups[1]?.value?.toFloat()!!
val subChapterDecimal = it.groups[2]?.value val subChapterDecimal = it.groups[2]?.value
val subChapterAlpha = it.groups[3]?.value val subChapterAlpha = it.groups[3]?.value
@ -120,11 +99,9 @@ object ChapterRecognition {
return .97f return .97f
} }
return if (alpha[0] == '.') { val trimmedAlpha = alpha.trimStart('.')
// Take value after (.) if (trimmedAlpha.length == 1) {
parseAlphaPostFix(alpha[1]) return parseAlphaPostFix(trimmedAlpha[0])
} else {
parseAlphaPostFix(alpha[0])
} }
} }
@ -135,6 +112,8 @@ object ChapterRecognition {
* x.a -> x.1, x.b -> x.2, etc * x.a -> x.1, x.b -> x.2, etc
*/ */
private fun parseAlphaPostFix(alpha: Char): Float { private fun parseAlphaPostFix(alpha: Char): Float {
return ("0." + (alpha.code - 96).toString()).toFloat() val number = alpha.code - ('a'.code - 1)
if (number >= 10) return 0f
return number / 10f
} }
} }

View File

@ -198,7 +198,6 @@ class ChapterRecognitionTest {
assertChapter(mangaTitle, "Fairy Tail 404.extravol002", 404.99f) assertChapter(mangaTitle, "Fairy Tail 404.extravol002", 404.99f)
assertChapter(mangaTitle, "Fairy Tail 404 extravol002", 404.99f) assertChapter(mangaTitle, "Fairy Tail 404 extravol002", 404.99f)
assertChapter(mangaTitle, "Fairy Tail 404.evol002", 404.5f)
} }
@Test @Test
@ -207,7 +206,6 @@ class ChapterRecognitionTest {
assertChapter(mangaTitle, "Fairy Tail 404.omakevol002", 404.98f) assertChapter(mangaTitle, "Fairy Tail 404.omakevol002", 404.98f)
assertChapter(mangaTitle, "Fairy Tail 404 omakevol002", 404.98f) assertChapter(mangaTitle, "Fairy Tail 404 omakevol002", 404.98f)
assertChapter(mangaTitle, "Fairy Tail 404.ovol002", 404.15f)
} }
@Test @Test
@ -216,7 +214,6 @@ class ChapterRecognitionTest {
assertChapter(mangaTitle, "Fairy Tail 404.specialvol002", 404.97f) assertChapter(mangaTitle, "Fairy Tail 404.specialvol002", 404.97f)
assertChapter(mangaTitle, "Fairy Tail 404 specialvol002", 404.97f) assertChapter(mangaTitle, "Fairy Tail 404 specialvol002", 404.97f)
assertChapter(mangaTitle, "Fairy Tail 404.svol002", 404.19f)
} }
@Test @Test
@ -253,6 +250,16 @@ class ChapterRecognitionTest {
assertChapter("One Outs", "One Outs 001", 1f) assertChapter("One Outs", "One Outs 001", 1f)
} }
@Test
fun `Chapters containing ordinals`() {
val mangaTitle = "The Sister of the Woods with a Thousand Young"
assertChapter(mangaTitle, "The 1st Night", 1f)
assertChapter(mangaTitle, "The 2nd Night", 2f)
assertChapter(mangaTitle, "The 3rd Night", 3f)
assertChapter(mangaTitle, "The 4th Night", 4f)
}
private fun assertChapter(mangaTitle: String, name: String, expected: Float) { private fun assertChapter(mangaTitle: String, name: String, expected: Float) {
val chapterNumber = ChapterRecognition.parseChapterNumber(mangaTitle, name) val chapterNumber = ChapterRecognition.parseChapterNumber(mangaTitle, name)
assertEquals(chapterNumber, expected) assertEquals(chapterNumber, expected)