- Add FeedParser.kt with automatic feed type detection - Add RSSParser.kt for RSS 2.0 feeds - Add AtomParser.kt for Atom 1.0 feeds - Add comprehensive unit tests for both parsers - Support iTunes namespace and enclosures - Fix pre-existing compilation issues in the codebase - Update build.gradle.kts with proper dependencies and AGP 8.5.0
189 lines
7.7 KiB
Kotlin
189 lines
7.7 KiB
Kotlin
package com.rssuper.parsing
|
|
|
|
import com.rssuper.models.Enclosure
|
|
import com.rssuper.models.Feed
|
|
import com.rssuper.models.FeedItem
|
|
import org.xmlpull.v1.XmlPullParser
|
|
import org.xmlpull.v1.XmlPullParserFactory
|
|
import java.io.StringReader
|
|
import java.util.Date
|
|
|
|
object RSSParser {
|
|
|
|
private val ITUNES_NS = "http://www.itunes.com/dtds/podcast-1.0.dtd"
|
|
private val CONTENT_NS = "http://purl.org/rss/1.0/modules/content/"
|
|
|
|
fun parse(xml: String, feedUrl: String): Feed {
|
|
val factory = XmlPullParserFactory.newInstance()
|
|
factory.isNamespaceAware = true
|
|
val parser = factory.newPullParser()
|
|
parser.setInput(StringReader(xml))
|
|
|
|
var title: String? = null
|
|
var link: String? = null
|
|
var description: String? = null
|
|
var language: String? = null
|
|
var lastBuildDate: Date? = null
|
|
var generator: String? = null
|
|
var ttl: Int? = null
|
|
val items = mutableListOf<FeedItem>()
|
|
|
|
var currentItem: MutableMap<String, Any?>? = null
|
|
var currentTag: String? = null
|
|
|
|
var eventType = parser.eventType
|
|
while (eventType != XmlPullParser.END_DOCUMENT) {
|
|
when (eventType) {
|
|
XmlPullParser.START_TAG -> {
|
|
val tagName = parser.name
|
|
val namespace = parser.namespace
|
|
|
|
when {
|
|
tagName == "channel" -> {}
|
|
tagName == "item" -> {
|
|
currentItem = mutableMapOf()
|
|
}
|
|
tagName == "title" || tagName == "description" ||
|
|
tagName == "link" || tagName == "author" ||
|
|
tagName == "guid" || tagName == "pubDate" ||
|
|
tagName == "category" || tagName == "enclosure" -> {
|
|
currentTag = tagName
|
|
}
|
|
tagName == "language" -> currentTag = tagName
|
|
tagName == "lastBuildDate" -> currentTag = tagName
|
|
tagName == "generator" -> currentTag = tagName
|
|
tagName == "ttl" -> currentTag = tagName
|
|
|
|
tagName == "subtitle" && namespace == ITUNES_NS -> {
|
|
if (currentItem == null) {
|
|
description = readElementText(parser)
|
|
}
|
|
}
|
|
tagName == "summary" && namespace == ITUNES_NS -> {
|
|
currentItem?.put("description", readElementText(parser))
|
|
}
|
|
tagName == "duration" && namespace == ITUNES_NS -> {
|
|
currentItem?.put("duration", readElementText(parser))
|
|
}
|
|
tagName == "image" && namespace == ITUNES_NS -> {
|
|
val href = parser.getAttributeValue(null, "href")
|
|
if (href != null && currentItem != null) {
|
|
currentItem.put("image", href)
|
|
}
|
|
}
|
|
tagName == "encoded" && namespace == CONTENT_NS -> {
|
|
currentItem?.put("content", readElementText(parser))
|
|
}
|
|
else -> {}
|
|
}
|
|
|
|
if (tagName == "enclosure" && currentItem != null) {
|
|
val url = parser.getAttributeValue(null, "url")
|
|
val type = parser.getAttributeValue(null, "type")
|
|
val length = parser.getAttributeValue(null, "length")?.toLongOrNull()
|
|
if (url != null && type != null) {
|
|
currentItem["enclosure"] = Enclosure(url, type, length)
|
|
}
|
|
}
|
|
}
|
|
|
|
XmlPullParser.TEXT -> {
|
|
val text = parser.text?.xmlTrimmed() ?: ""
|
|
if (text.isNotEmpty()) {
|
|
if (currentItem != null) {
|
|
when (currentTag) {
|
|
"title" -> currentItem["title"] = text
|
|
"description" -> currentItem["description"] = text
|
|
"link" -> currentItem["link"] = text
|
|
"author" -> currentItem["author"] = text
|
|
"guid" -> currentItem["guid"] = text
|
|
"pubDate" -> currentItem["pubDate"] = text
|
|
"category" -> {
|
|
val cats = currentItem["categories"] as? MutableList<String> ?: mutableListOf()
|
|
cats.add(text)
|
|
currentItem["categories"] = cats
|
|
}
|
|
}
|
|
} else {
|
|
when (currentTag) {
|
|
"title" -> title = text
|
|
"link" -> link = text
|
|
"description" -> description = text
|
|
"language" -> language = text
|
|
"lastBuildDate" -> lastBuildDate = XmlDateParser.parse(text)
|
|
"generator" -> generator = text
|
|
"ttl" -> ttl = text.toIntOrNull()
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
XmlPullParser.END_TAG -> {
|
|
val tagName = parser.name
|
|
if (tagName == "item" && currentItem != null) {
|
|
items.add(buildFeedItem(currentItem))
|
|
currentItem = null
|
|
}
|
|
currentTag = null
|
|
}
|
|
}
|
|
eventType = parser.next()
|
|
}
|
|
|
|
return Feed(
|
|
id = generateUuid(),
|
|
title = title ?: "Untitled Feed",
|
|
link = link,
|
|
description = description,
|
|
language = language,
|
|
lastBuildDate = lastBuildDate,
|
|
generator = generator,
|
|
ttl = ttl,
|
|
items = items,
|
|
rawUrl = feedUrl,
|
|
lastFetchedAt = Date()
|
|
)
|
|
}
|
|
|
|
private fun readElementText(parser: XmlPullParser): String {
|
|
var text = ""
|
|
var eventType = parser.eventType
|
|
while (eventType != XmlPullParser.END_TAG) {
|
|
if (eventType == XmlPullParser.TEXT) {
|
|
text = parser.text.xmlDecoded()
|
|
}
|
|
eventType = parser.next()
|
|
}
|
|
return text.xmlTrimmed()
|
|
}
|
|
|
|
@Suppress("UNCHECKED_CAST")
|
|
private fun buildFeedItem(item: Map<String, Any?>): FeedItem {
|
|
val title = item["title"] as? String ?: "Untitled"
|
|
val link = item["link"] as? String
|
|
val description = item["description"] as? String
|
|
val content = item["content"] as? String ?: description
|
|
val author = item["author"] as? String
|
|
val guid = item["guid"] as? String ?: link ?: generateUuid()
|
|
val categories = item["categories"] as? List<String>
|
|
val enclosure = item["enclosure"] as? Enclosure
|
|
|
|
val pubDateStr = item["pubDate"] as? String
|
|
val published = XmlDateParser.parse(pubDateStr)
|
|
|
|
return FeedItem(
|
|
id = generateUuid(),
|
|
title = title,
|
|
link = link,
|
|
description = description,
|
|
content = content,
|
|
author = author,
|
|
published = published,
|
|
updated = published,
|
|
categories = categories,
|
|
enclosure = enclosure,
|
|
guid = guid
|
|
)
|
|
}
|
|
}
|