Atte149/book-story
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

🛠️ XML parser for FB2

Browse source 
* New XmlTextParser to parse FB2
This commit is contained in:
Acclorite 2025-03-05 18:24:22 +02:00
parent 78b3b04df3
commit 697efb118a
2 changed files with 55 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

6
app/src/main/java/ua/acclorite/book_story/data/parser/TextParserImpl.kt
View file

@ -13,6 +13,7 @@ import ua.acclorite.book_story.data.parser.epub.EpubTextParser
import ua.acclorite.book_story.data.parser.html.HtmlTextParser
import ua.acclorite.book_story.data.parser.pdf.PdfTextParser
import ua.acclorite.book_story.data.parser.txt.TxtTextParser
import ua.acclorite.book_story.data.parser.xml.XmlTextParser
import ua.acclorite.book_story.domain.file.CachedFile
import ua.acclorite.book_story.domain.reader.ReaderText
import javax.inject.Inject
@ -26,7 +27,8 @@ class TextParserImpl @Inject constructor(
// Document parser (HTML+Markdown)
private val epubTextParser: EpubTextParser,
private val htmlTextParser: HtmlTextParser
private val htmlTextParser: HtmlTextParser,
private val xmlTextParser: XmlTextParser
) : TextParser {
override suspend fun parse(cachedFile: CachedFile): List<ReaderText> {
@ -51,7 +53,7 @@ class TextParserImpl @Inject constructor(
}
".fb2" -> {
htmlTextParser.parse(cachedFile)
xmlTextParser.parse(cachedFile)
}
".html" -> {

51
app/src/main/java/ua/acclorite/book_story/data/parser/xml/XmlTextParser.kt Normal file
View file

@ -0,0 +1,51 @@
/*
* Book's Story — free and open-source Material You eBook reader.
* Copyright (C) 2024-2025 Acclorite
* SPDX-License-Identifier: GPL-3.0-only
*/
package ua.acclorite.book_story.data.parser.xml
import android.util.Log
import kotlinx.coroutines.yield
import org.jsoup.Jsoup
import org.jsoup.parser.Parser
import ua.acclorite.book_story.data.parser.DocumentParser
import ua.acclorite.book_story.data.parser.TextParser
import ua.acclorite.book_story.domain.file.CachedFile
import ua.acclorite.book_story.domain.reader.ReaderText
import javax.inject.Inject
private const val XML_TAG = "XML Parser"
class XmlTextParser @Inject constructor(
private val documentParser: DocumentParser
) : TextParser {
override suspend fun parse(cachedFile: CachedFile): List<ReaderText> {
Log.i(XML_TAG, "Started XML parsing: ${cachedFile.name}.")
return try {
val readerText = cachedFile.openInputStream()?.use { stream ->
documentParser.parseDocument(Jsoup.parse(stream, null, "", Parser.xmlParser()))
}
yield()
if (
readerText.isNullOrEmpty() ||
readerText.filterIsInstance<ReaderText.Text>().isEmpty() ||
readerText.filterIsInstance<ReaderText.Chapter>().isEmpty()
) {
Log.e(XML_TAG, "Could not extract text from XML.")
return emptyList()
}
Log.i(XML_TAG, "Successfully finished XML parsing.")
readerText
} catch (e: Exception) {
e.printStackTrace()
emptyList()
}
}
}