mirror of
https://github.com/Ranchero-Software/NetNewsWire
synced 2025-08-12 06:26:36 +00:00
Move local modules into a folder named Modules.
This commit is contained in:
76
Modules/Parser/Sources/Parser/Feeds/FeedParser.swift
Normal file
76
Modules/Parser/Sources/Parser/Feeds/FeedParser.swift
Normal file
@@ -0,0 +1,76 @@
|
||||
//
|
||||
// FeedParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import ParserObjC
|
||||
|
||||
// FeedParser handles RSS, Atom, JSON Feed, and RSS-in-JSON.
|
||||
// You don’t need to know the type of feed.
|
||||
|
||||
public struct FeedParser {
|
||||
|
||||
public static func canParse(_ parserData: ParserData) -> Bool {
|
||||
|
||||
let type = feedType(parserData)
|
||||
|
||||
switch type {
|
||||
case .jsonFeed, .rssInJSON, .rss, .atom:
|
||||
return true
|
||||
default:
|
||||
return false
|
||||
}
|
||||
}
|
||||
|
||||
public static func parse(_ parserData: ParserData) async throws -> ParsedFeed? {
|
||||
|
||||
let type = feedType(parserData)
|
||||
|
||||
switch type {
|
||||
|
||||
case .jsonFeed:
|
||||
return try JSONFeedParser.parse(parserData)
|
||||
|
||||
case .rssInJSON:
|
||||
return try RSSInJSONParser.parse(parserData)
|
||||
|
||||
case .rss:
|
||||
return RSSParser.parse(parserData)
|
||||
|
||||
case .atom:
|
||||
return AtomParser.parse(parserData)
|
||||
|
||||
case .unknown, .notAFeed:
|
||||
return nil
|
||||
}
|
||||
}
|
||||
|
||||
/// For unit tests measuring performance.
|
||||
public static func parseSync(_ parserData: ParserData) throws -> ParsedFeed? {
|
||||
|
||||
let type = feedType(parserData)
|
||||
|
||||
switch type {
|
||||
|
||||
case .jsonFeed:
|
||||
return try JSONFeedParser.parse(parserData)
|
||||
|
||||
case .rssInJSON:
|
||||
return try RSSInJSONParser.parse(parserData)
|
||||
|
||||
case .rss:
|
||||
return RSSParser.parse(parserData)
|
||||
|
||||
case .atom:
|
||||
return AtomParser.parse(parserData)
|
||||
|
||||
case .unknown, .notAFeed:
|
||||
return nil
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
29
Modules/Parser/Sources/Parser/Feeds/FeedParserError.swift
Normal file
29
Modules/Parser/Sources/Parser/Feeds/FeedParserError.swift
Normal file
@@ -0,0 +1,29 @@
|
||||
//
|
||||
// FeedParserError.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/24/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct FeedParserError: Error, Sendable {
|
||||
|
||||
public enum FeedParserErrorType: Sendable {
|
||||
|
||||
case rssChannelNotFound
|
||||
case rssItemsNotFound
|
||||
case jsonFeedVersionNotFound
|
||||
case jsonFeedItemsNotFound
|
||||
case jsonFeedTitleNotFound
|
||||
case invalidJSON
|
||||
}
|
||||
|
||||
public let errorType: FeedParserErrorType
|
||||
|
||||
public init(_ errorType: FeedParserErrorType) {
|
||||
|
||||
self.errorType = errorType
|
||||
}
|
||||
}
|
||||
64
Modules/Parser/Sources/Parser/Feeds/FeedType.swift
Normal file
64
Modules/Parser/Sources/Parser/Feeds/FeedType.swift
Normal file
@@ -0,0 +1,64 @@
|
||||
//
|
||||
// FeedType.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
#if SWIFT_PACKAGE
|
||||
import ParserObjC
|
||||
#endif
|
||||
|
||||
public enum FeedType: Sendable {
|
||||
case rss
|
||||
case atom
|
||||
case jsonFeed
|
||||
case rssInJSON
|
||||
case unknown
|
||||
case notAFeed
|
||||
}
|
||||
|
||||
|
||||
private let minNumberOfBytesRequired = 128
|
||||
|
||||
public func feedType(_ parserData: ParserData, isPartialData: Bool = false) -> FeedType {
|
||||
|
||||
// Can call with partial data — while still downloading, for instance.
|
||||
// If there’s not enough data, return .unknown. Ask again when there’s more data.
|
||||
// If it’s definitely not a feed, return .notAFeed.
|
||||
//
|
||||
// This is fast enough to call on the main thread.
|
||||
|
||||
if parserData.data.count < minNumberOfBytesRequired {
|
||||
return .unknown
|
||||
}
|
||||
|
||||
let nsdata = parserData.data as NSData
|
||||
|
||||
if nsdata.isProbablyJSONFeed() {
|
||||
return .jsonFeed
|
||||
}
|
||||
if nsdata.isProbablyRSSInJSON() {
|
||||
return .rssInJSON
|
||||
}
|
||||
if nsdata.isProbablyRSS() {
|
||||
return .rss
|
||||
}
|
||||
if nsdata.isProbablyAtom() {
|
||||
return .atom
|
||||
}
|
||||
|
||||
if isPartialData && nsdata.isProbablyJSON() {
|
||||
// Might not be able to detect a JSON Feed without all data.
|
||||
// Dr. Drang’s JSON Feed (see althis.json and allthis-partial.json in tests)
|
||||
// has, at this writing, the JSON version element at the end of the feed,
|
||||
// which is totally legal — but it means not being able to detect
|
||||
// that it’s a JSON Feed without all the data.
|
||||
// So this returns .unknown instead of .notAFeed.
|
||||
return .unknown
|
||||
}
|
||||
|
||||
return .notAFeed
|
||||
}
|
||||
250
Modules/Parser/Sources/Parser/Feeds/JSON/JSONFeedParser.swift
Normal file
250
Modules/Parser/Sources/Parser/Feeds/JSON/JSONFeedParser.swift
Normal file
@@ -0,0 +1,250 @@
|
||||
//
|
||||
// JSONFeedParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/25/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
#if SWIFT_PACKAGE
|
||||
import ParserObjC
|
||||
#endif
|
||||
|
||||
// See https://jsonfeed.org/version/1.1
|
||||
|
||||
public struct JSONFeedParser {
|
||||
|
||||
struct Key {
|
||||
static let version = "version"
|
||||
static let items = "items"
|
||||
static let title = "title"
|
||||
static let homePageURL = "home_page_url"
|
||||
static let feedURL = "feed_url"
|
||||
static let feedDescription = "description"
|
||||
static let nextURL = "next_url"
|
||||
static let icon = "icon"
|
||||
static let favicon = "favicon"
|
||||
static let expired = "expired"
|
||||
static let author = "author"
|
||||
static let authors = "authors"
|
||||
static let name = "name"
|
||||
static let url = "url"
|
||||
static let avatar = "avatar"
|
||||
static let hubs = "hubs"
|
||||
static let type = "type"
|
||||
static let contentHTML = "content_html"
|
||||
static let contentText = "content_text"
|
||||
static let externalURL = "external_url"
|
||||
static let summary = "summary"
|
||||
static let image = "image"
|
||||
static let bannerImage = "banner_image"
|
||||
static let datePublished = "date_published"
|
||||
static let dateModified = "date_modified"
|
||||
static let tags = "tags"
|
||||
static let uniqueID = "id"
|
||||
static let attachments = "attachments"
|
||||
static let mimeType = "mime_type"
|
||||
static let sizeInBytes = "size_in_bytes"
|
||||
static let durationInSeconds = "duration_in_seconds"
|
||||
static let language = "language"
|
||||
}
|
||||
|
||||
static let jsonFeedVersionMarker = "://jsonfeed.org/version/" // Allow for the mistake of not getting the scheme exactly correct.
|
||||
|
||||
public static func parse(_ parserData: ParserData) throws -> ParsedFeed? {
|
||||
|
||||
guard let d = JSONUtilities.dictionary(with: parserData.data) else {
|
||||
throw FeedParserError(.invalidJSON)
|
||||
}
|
||||
|
||||
guard let version = d[Key.version] as? String, let _ = version.range(of: JSONFeedParser.jsonFeedVersionMarker) else {
|
||||
throw FeedParserError(.jsonFeedVersionNotFound)
|
||||
}
|
||||
guard let itemsArray = d[Key.items] as? JSONArray else {
|
||||
throw FeedParserError(.jsonFeedItemsNotFound)
|
||||
}
|
||||
guard let title = d[Key.title] as? String else {
|
||||
throw FeedParserError(.jsonFeedTitleNotFound)
|
||||
}
|
||||
|
||||
let authors = parseAuthors(d)
|
||||
let homePageURL = d[Key.homePageURL] as? String
|
||||
let feedURL = d[Key.feedURL] as? String ?? parserData.url
|
||||
let feedDescription = d[Key.feedDescription] as? String
|
||||
let nextURL = d[Key.nextURL] as? String
|
||||
let iconURL = d[Key.icon] as? String
|
||||
let faviconURL = d[Key.favicon] as? String
|
||||
let expired = d[Key.expired] as? Bool ?? false
|
||||
let hubs = parseHubs(d)
|
||||
let language = d[Key.language] as? String
|
||||
|
||||
let items = parseItems(itemsArray, parserData.url)
|
||||
|
||||
return ParsedFeed(type: .jsonFeed, title: title, homePageURL: homePageURL, feedURL: feedURL, language: language, feedDescription: feedDescription, nextURL: nextURL, iconURL: iconURL, faviconURL: faviconURL, authors: authors, expired: expired, hubs: hubs, items: items)
|
||||
}
|
||||
}
|
||||
|
||||
private extension JSONFeedParser {
|
||||
|
||||
static func parseAuthors(_ dictionary: JSONDictionary) -> Set<ParsedAuthor>? {
|
||||
|
||||
if let authorsArray = dictionary[Key.authors] as? JSONArray {
|
||||
var authors = Set<ParsedAuthor>()
|
||||
for author in authorsArray {
|
||||
if let parsedAuthor = parseAuthor(author) {
|
||||
authors.insert(parsedAuthor)
|
||||
}
|
||||
}
|
||||
return authors
|
||||
}
|
||||
|
||||
guard let authorDictionary = dictionary[Key.author] as? JSONDictionary,
|
||||
let parsedAuthor = parseAuthor(authorDictionary) else {
|
||||
return nil
|
||||
}
|
||||
|
||||
return Set([parsedAuthor])
|
||||
}
|
||||
|
||||
static func parseAuthor(_ dictionary: JSONDictionary) -> ParsedAuthor? {
|
||||
let name = dictionary[Key.name] as? String
|
||||
let url = dictionary[Key.url] as? String
|
||||
let avatar = dictionary[Key.avatar] as? String
|
||||
if name == nil && url == nil && avatar == nil {
|
||||
return nil
|
||||
}
|
||||
return ParsedAuthor(name: name, url: url, avatarURL: avatar, emailAddress: nil)
|
||||
}
|
||||
|
||||
static func parseHubs(_ dictionary: JSONDictionary) -> Set<ParsedHub>? {
|
||||
|
||||
guard let hubsArray = dictionary[Key.hubs] as? JSONArray else {
|
||||
return nil
|
||||
}
|
||||
|
||||
let hubs = hubsArray.compactMap { (hubDictionary) -> ParsedHub? in
|
||||
guard let hubURL = hubDictionary[Key.url] as? String, let hubType = hubDictionary[Key.type] as? String else {
|
||||
return nil
|
||||
}
|
||||
return ParsedHub(type: hubType, url: hubURL)
|
||||
}
|
||||
return hubs.isEmpty ? nil : Set(hubs)
|
||||
}
|
||||
|
||||
static func parseItems(_ itemsArray: JSONArray, _ feedURL: String) -> Set<ParsedItem> {
|
||||
|
||||
return Set(itemsArray.compactMap { (oneItemDictionary) -> ParsedItem? in
|
||||
return parseItem(oneItemDictionary, feedURL)
|
||||
})
|
||||
}
|
||||
|
||||
static func parseItem(_ itemDictionary: JSONDictionary, _ feedURL: String) -> ParsedItem? {
|
||||
|
||||
guard let uniqueID = parseUniqueID(itemDictionary) else {
|
||||
return nil
|
||||
}
|
||||
|
||||
let contentHTML = itemDictionary[Key.contentHTML] as? String
|
||||
let contentText = itemDictionary[Key.contentText] as? String
|
||||
if contentHTML == nil && contentText == nil {
|
||||
return nil
|
||||
}
|
||||
|
||||
let url = itemDictionary[Key.url] as? String
|
||||
let externalURL = itemDictionary[Key.externalURL] as? String
|
||||
let title = parseTitle(itemDictionary, feedURL)
|
||||
let language = itemDictionary[Key.language] as? String
|
||||
let summary = itemDictionary[Key.summary] as? String
|
||||
let imageURL = itemDictionary[Key.image] as? String
|
||||
let bannerImageURL = itemDictionary[Key.bannerImage] as? String
|
||||
|
||||
let datePublished = parseDate(itemDictionary[Key.datePublished] as? String)
|
||||
let dateModified = parseDate(itemDictionary[Key.dateModified] as? String)
|
||||
|
||||
let authors = parseAuthors(itemDictionary)
|
||||
var tags: Set<String>? = nil
|
||||
if let tagsArray = itemDictionary[Key.tags] as? [String] {
|
||||
tags = Set(tagsArray)
|
||||
}
|
||||
let attachments = parseAttachments(itemDictionary)
|
||||
|
||||
return ParsedItem(syncServiceID: nil, uniqueID: uniqueID, feedURL: feedURL, url: url, externalURL: externalURL, title: title, language: language, contentHTML: contentHTML, contentText: contentText, summary: summary, imageURL: imageURL, bannerImageURL: bannerImageURL, datePublished: datePublished, dateModified: dateModified, authors: authors, tags: tags, attachments: attachments)
|
||||
}
|
||||
|
||||
static func parseTitle(_ itemDictionary: JSONDictionary, _ feedURL: String) -> String? {
|
||||
|
||||
guard let title = itemDictionary[Key.title] as? String else {
|
||||
return nil
|
||||
}
|
||||
|
||||
if isSpecialCaseTitleWithEntitiesFeed(feedURL) {
|
||||
return (title as NSString).rsparser_stringByDecodingHTMLEntities()
|
||||
}
|
||||
|
||||
return title
|
||||
}
|
||||
|
||||
static func isSpecialCaseTitleWithEntitiesFeed(_ feedURL: String) -> Bool {
|
||||
|
||||
// As of 16 Feb. 2018, Kottke’s and Heer’s feeds includes HTML entities in the title elements.
|
||||
// If we find more feeds like this, we’ll add them here. If these feeds get fixed, we’ll remove them.
|
||||
|
||||
let lowerFeedURL = feedURL.lowercased()
|
||||
let matchStrings = ["kottke.org", "pxlnv.com", "macstories.net", "macobserver.com"]
|
||||
for matchString in matchStrings {
|
||||
if lowerFeedURL.contains(matchString) {
|
||||
return true
|
||||
}
|
||||
}
|
||||
|
||||
return false
|
||||
}
|
||||
|
||||
static func parseUniqueID(_ itemDictionary: JSONDictionary) -> String? {
|
||||
|
||||
if let uniqueID = itemDictionary[Key.uniqueID] as? String {
|
||||
return uniqueID // Spec says it must be a string
|
||||
}
|
||||
// Version 1 spec also says that if it’s a number, even though that’s incorrect, it should be coerced to a string.
|
||||
if let uniqueID = itemDictionary[Key.uniqueID] as? Int {
|
||||
return "\(uniqueID)"
|
||||
}
|
||||
if let uniqueID = itemDictionary[Key.uniqueID] as? Double {
|
||||
return "\(uniqueID)"
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
static func parseDate(_ dateString: String?) -> Date? {
|
||||
|
||||
guard let dateString = dateString, !dateString.isEmpty else {
|
||||
return nil
|
||||
}
|
||||
return RSDateWithString(dateString)
|
||||
}
|
||||
|
||||
static func parseAttachments(_ itemDictionary: JSONDictionary) -> Set<ParsedAttachment>? {
|
||||
|
||||
guard let attachmentsArray = itemDictionary[Key.attachments] as? JSONArray else {
|
||||
return nil
|
||||
}
|
||||
return Set(attachmentsArray.compactMap { parseAttachment($0) })
|
||||
}
|
||||
|
||||
static func parseAttachment(_ attachmentObject: JSONDictionary) -> ParsedAttachment? {
|
||||
|
||||
guard let url = attachmentObject[Key.url] as? String else {
|
||||
return nil
|
||||
}
|
||||
guard let mimeType = attachmentObject[Key.mimeType] as? String else {
|
||||
return nil
|
||||
}
|
||||
|
||||
let title = attachmentObject[Key.title] as? String
|
||||
let sizeInBytes = attachmentObject[Key.sizeInBytes] as? Int
|
||||
let durationInSeconds = attachmentObject[Key.durationInSeconds] as? Int
|
||||
|
||||
return ParsedAttachment(url: url, mimeType: mimeType, title: title, sizeInBytes: sizeInBytes, durationInSeconds: durationInSeconds)
|
||||
}
|
||||
}
|
||||
184
Modules/Parser/Sources/Parser/Feeds/JSON/RSSInJSONParser.swift
Normal file
184
Modules/Parser/Sources/Parser/Feeds/JSON/RSSInJSONParser.swift
Normal file
@@ -0,0 +1,184 @@
|
||||
//
|
||||
// RSSInJSONParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/24/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
#if SWIFT_PACKAGE
|
||||
import ParserObjC
|
||||
#endif
|
||||
|
||||
// See https://github.com/scripting/Scripting-News/blob/master/rss-in-json/README.md
|
||||
// Also: http://cyber.harvard.edu/rss/rss.html
|
||||
|
||||
public struct RSSInJSONParser {
|
||||
|
||||
public static func parse(_ parserData: ParserData) throws -> ParsedFeed? {
|
||||
|
||||
do {
|
||||
guard let parsedObject = try JSONSerialization.jsonObject(with: parserData.data) as? JSONDictionary else {
|
||||
throw FeedParserError(.invalidJSON)
|
||||
}
|
||||
guard let rssObject = parsedObject["rss"] as? JSONDictionary else {
|
||||
throw FeedParserError(.rssChannelNotFound)
|
||||
}
|
||||
guard let channelObject = rssObject["channel"] as? JSONDictionary else {
|
||||
throw FeedParserError(.rssChannelNotFound)
|
||||
}
|
||||
|
||||
// I’d bet money that in practice the items array won’t always appear correctly inside the channel object.
|
||||
// I’d also bet that sometimes it gets called "items" instead of "item".
|
||||
var itemsObject = channelObject["item"] as? JSONArray
|
||||
if itemsObject == nil {
|
||||
itemsObject = parsedObject["item"] as? JSONArray
|
||||
}
|
||||
if itemsObject == nil {
|
||||
itemsObject = channelObject["items"] as? JSONArray
|
||||
}
|
||||
if itemsObject == nil {
|
||||
itemsObject = parsedObject["items"] as? JSONArray
|
||||
}
|
||||
if itemsObject == nil {
|
||||
throw FeedParserError(.rssItemsNotFound)
|
||||
}
|
||||
|
||||
let title = channelObject["title"] as? String
|
||||
let homePageURL = channelObject["link"] as? String
|
||||
let feedURL = parserData.url
|
||||
let feedDescription = channelObject["description"] as? String
|
||||
let feedLanguage = channelObject["language"] as? String
|
||||
|
||||
let items = parseItems(itemsObject!, parserData.url)
|
||||
|
||||
return ParsedFeed(type: .rssInJSON, title: title, homePageURL: homePageURL, feedURL: feedURL, language: feedLanguage, feedDescription: feedDescription, nextURL: nil, iconURL: nil, faviconURL: nil, authors: nil, expired: false, hubs: nil, items: items)
|
||||
|
||||
}
|
||||
catch { throw error }
|
||||
}
|
||||
}
|
||||
|
||||
private extension RSSInJSONParser {
|
||||
|
||||
static func parseItems(_ itemsObject: JSONArray, _ feedURL: String) -> Set<ParsedItem> {
|
||||
|
||||
return Set(itemsObject.compactMap{ (oneItemDictionary) -> ParsedItem? in
|
||||
|
||||
return parsedItemWithDictionary(oneItemDictionary, feedURL)
|
||||
})
|
||||
}
|
||||
|
||||
static func parsedItemWithDictionary(_ itemDictionary: JSONDictionary, _ feedURL: String) -> ParsedItem? {
|
||||
|
||||
let externalURL = itemDictionary["link"] as? String
|
||||
let title = itemDictionary["title"] as? String
|
||||
|
||||
var contentHTML = itemDictionary["description"] as? String
|
||||
var contentText: String? = nil
|
||||
if contentHTML != nil && !(contentHTML!.contains("<")) {
|
||||
contentText = contentHTML
|
||||
contentHTML = nil
|
||||
}
|
||||
if contentHTML == nil && contentText == nil && title == nil {
|
||||
return nil
|
||||
}
|
||||
|
||||
var datePublished: Date? = nil
|
||||
if let datePublishedString = itemDictionary["pubDate"] as? String {
|
||||
datePublished = RSDateWithString(datePublishedString)
|
||||
}
|
||||
|
||||
let authors = parseAuthors(itemDictionary)
|
||||
let tags = parseTags(itemDictionary)
|
||||
let attachments = parseAttachments(itemDictionary)
|
||||
|
||||
var uniqueID: String? = itemDictionary["guid"] as? String
|
||||
if uniqueID == nil {
|
||||
|
||||
// Calculate a uniqueID based on a combination of non-empty elements. Then hash the result.
|
||||
// Items should have guids. When they don't, re-runs are very likely
|
||||
// because there's no other 100% reliable way to determine identity.
|
||||
// This calculated uniqueID is valid only for this particular feed. (Just like ids in JSON Feed.)
|
||||
|
||||
var s = ""
|
||||
if let datePublished = datePublished {
|
||||
s += "\(datePublished.timeIntervalSince1970)"
|
||||
}
|
||||
if let title = title {
|
||||
s += title
|
||||
}
|
||||
if let externalURL = externalURL {
|
||||
s += externalURL
|
||||
}
|
||||
if let authorEmailAddress = authors?.first?.emailAddress {
|
||||
s += authorEmailAddress
|
||||
}
|
||||
if let oneAttachmentURL = attachments?.first?.url {
|
||||
s += oneAttachmentURL
|
||||
}
|
||||
if s.isEmpty {
|
||||
// Sheesh. Tough case.
|
||||
if let _ = contentHTML {
|
||||
s = contentHTML!
|
||||
}
|
||||
if let _ = contentText {
|
||||
s = contentText!
|
||||
}
|
||||
}
|
||||
uniqueID = (s as NSString).rsparser_md5Hash()
|
||||
}
|
||||
|
||||
if let uniqueID = uniqueID {
|
||||
return ParsedItem(syncServiceID: nil, uniqueID: uniqueID, feedURL: feedURL, url: nil, externalURL: externalURL, title: title, language: nil, contentHTML: contentHTML, contentText: contentText, summary: nil, imageURL: nil, bannerImageURL: nil, datePublished: datePublished, dateModified: nil, authors: authors, tags: tags, attachments: attachments)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
static func parseAuthors(_ itemDictionary: JSONDictionary) -> Set<ParsedAuthor>? {
|
||||
|
||||
guard let authorEmailAddress = itemDictionary["author"] as? String else {
|
||||
return nil
|
||||
}
|
||||
let parsedAuthor = ParsedAuthor(name: nil, url: nil, avatarURL: nil, emailAddress: authorEmailAddress)
|
||||
return Set([parsedAuthor])
|
||||
}
|
||||
|
||||
static func parseTags(_ itemDictionary: JSONDictionary) -> Set<String>? {
|
||||
|
||||
if let categoryObject = itemDictionary["category"] as? JSONDictionary {
|
||||
if let oneTag = categoryObject["#value"] as? String {
|
||||
return Set([oneTag])
|
||||
}
|
||||
return nil
|
||||
}
|
||||
else if let categoryArray = itemDictionary["category"] as? JSONArray {
|
||||
return Set(categoryArray.compactMap{ $0["#value"] as? String })
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
static func parseAttachments(_ itemDictionary: JSONDictionary) -> Set<ParsedAttachment>? {
|
||||
|
||||
guard let enclosureObject = itemDictionary["enclosure"] as? JSONDictionary else {
|
||||
return nil
|
||||
}
|
||||
guard let attachmentURL = enclosureObject["url"] as? String else {
|
||||
return nil
|
||||
}
|
||||
|
||||
var attachmentSize = enclosureObject["length"] as? Int
|
||||
if attachmentSize == nil {
|
||||
if let attachmentSizeString = enclosureObject["length"] as? String {
|
||||
attachmentSize = (attachmentSizeString as NSString).integerValue
|
||||
}
|
||||
}
|
||||
|
||||
let type = enclosureObject["type"] as? String
|
||||
if let attachment = ParsedAttachment(url: attachmentURL, mimeType: type, title: nil, sizeInBytes: attachmentSize, durationInSeconds: nil) {
|
||||
return Set([attachment])
|
||||
}
|
||||
return nil
|
||||
}
|
||||
}
|
||||
36
Modules/Parser/Sources/Parser/Feeds/ParsedAttachment.swift
Normal file
36
Modules/Parser/Sources/Parser/Feeds/ParsedAttachment.swift
Normal file
@@ -0,0 +1,36 @@
|
||||
//
|
||||
// ParsedAttachment.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedAttachment: Hashable, Sendable {
|
||||
|
||||
public let url: String
|
||||
public let mimeType: String?
|
||||
public let title: String?
|
||||
public let sizeInBytes: Int?
|
||||
public let durationInSeconds: Int?
|
||||
|
||||
public init?(url: String, mimeType: String?, title: String?, sizeInBytes: Int?, durationInSeconds: Int?) {
|
||||
if url.isEmpty {
|
||||
return nil
|
||||
}
|
||||
|
||||
self.url = url
|
||||
self.mimeType = mimeType
|
||||
self.title = title
|
||||
self.sizeInBytes = sizeInBytes
|
||||
self.durationInSeconds = durationInSeconds
|
||||
}
|
||||
|
||||
// MARK: - Hashable
|
||||
|
||||
public func hash(into hasher: inout Hasher) {
|
||||
hasher.combine(url)
|
||||
}
|
||||
}
|
||||
44
Modules/Parser/Sources/Parser/Feeds/ParsedAuthor.swift
Normal file
44
Modules/Parser/Sources/Parser/Feeds/ParsedAuthor.swift
Normal file
@@ -0,0 +1,44 @@
|
||||
//
|
||||
// ParsedAuthor.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedAuthor: Hashable, Codable, Sendable {
|
||||
|
||||
public let name: String?
|
||||
public let url: String?
|
||||
public let avatarURL: String?
|
||||
public let emailAddress: String?
|
||||
|
||||
public init(name: String?, url: String?, avatarURL: String?, emailAddress: String?) {
|
||||
self.name = name
|
||||
self.url = url
|
||||
self.avatarURL = avatarURL
|
||||
self.emailAddress = emailAddress
|
||||
}
|
||||
|
||||
// MARK: - Hashable
|
||||
|
||||
public func hash(into hasher: inout Hasher) {
|
||||
if let name {
|
||||
hasher.combine(name)
|
||||
}
|
||||
else if let url {
|
||||
hasher.combine(url)
|
||||
}
|
||||
else if let emailAddress {
|
||||
hasher.combine(emailAddress)
|
||||
}
|
||||
else if let avatarURL{
|
||||
hasher.combine(avatarURL)
|
||||
}
|
||||
else {
|
||||
hasher.combine("")
|
||||
}
|
||||
}
|
||||
}
|
||||
42
Modules/Parser/Sources/Parser/Feeds/ParsedFeed.swift
Normal file
42
Modules/Parser/Sources/Parser/Feeds/ParsedFeed.swift
Normal file
@@ -0,0 +1,42 @@
|
||||
//
|
||||
// ParsedFeed.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedFeed: Sendable {
|
||||
|
||||
public let type: FeedType
|
||||
public let title: String?
|
||||
public let homePageURL: String?
|
||||
public let feedURL: String?
|
||||
public let language: String?
|
||||
public let feedDescription: String?
|
||||
public let nextURL: String?
|
||||
public let iconURL: String?
|
||||
public let faviconURL: String?
|
||||
public let authors: Set<ParsedAuthor>?
|
||||
public let expired: Bool
|
||||
public let hubs: Set<ParsedHub>?
|
||||
public let items: Set<ParsedItem>
|
||||
|
||||
public init(type: FeedType, title: String?, homePageURL: String?, feedURL: String?, language: String?, feedDescription: String?, nextURL: String?, iconURL: String?, faviconURL: String?, authors: Set<ParsedAuthor>?, expired: Bool, hubs: Set<ParsedHub>?, items: Set<ParsedItem>) {
|
||||
self.type = type
|
||||
self.title = title
|
||||
self.homePageURL = homePageURL?.nilIfEmptyOrWhitespace
|
||||
self.feedURL = feedURL
|
||||
self.language = language
|
||||
self.feedDescription = feedDescription
|
||||
self.nextURL = nextURL
|
||||
self.iconURL = iconURL
|
||||
self.faviconURL = faviconURL
|
||||
self.authors = authors
|
||||
self.expired = expired
|
||||
self.hubs = hubs
|
||||
self.items = items
|
||||
}
|
||||
}
|
||||
15
Modules/Parser/Sources/Parser/Feeds/ParsedHub.swift
Normal file
15
Modules/Parser/Sources/Parser/Feeds/ParsedHub.swift
Normal file
@@ -0,0 +1,15 @@
|
||||
//
|
||||
// ParsedHub.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedHub: Hashable, Sendable {
|
||||
|
||||
public let type: String
|
||||
public let url: String
|
||||
}
|
||||
67
Modules/Parser/Sources/Parser/Feeds/ParsedItem.swift
Normal file
67
Modules/Parser/Sources/Parser/Feeds/ParsedItem.swift
Normal file
@@ -0,0 +1,67 @@
|
||||
//
|
||||
// ParsedItem.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/20/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedItem: Hashable, Sendable {
|
||||
|
||||
public let syncServiceID: String? //Nil when not syncing
|
||||
public let uniqueID: String //RSS guid, for instance; may be calculated
|
||||
public let feedURL: String
|
||||
public let url: String?
|
||||
public let externalURL: String?
|
||||
public let title: String?
|
||||
public let language: String?
|
||||
public let contentHTML: String?
|
||||
public let contentText: String?
|
||||
public let summary: String?
|
||||
public let imageURL: String?
|
||||
public let bannerImageURL: String?
|
||||
public let datePublished: Date?
|
||||
public let dateModified: Date?
|
||||
public let authors: Set<ParsedAuthor>?
|
||||
public let tags: Set<String>?
|
||||
public let attachments: Set<ParsedAttachment>?
|
||||
|
||||
public init(syncServiceID: String?, uniqueID: String, feedURL: String, url: String?, externalURL: String?, title: String?,
|
||||
language: String?, contentHTML: String?, contentText: String?, summary: String?, imageURL: String?,
|
||||
bannerImageURL: String?,datePublished: Date?, dateModified: Date?, authors: Set<ParsedAuthor>?,
|
||||
tags: Set<String>?, attachments: Set<ParsedAttachment>?) {
|
||||
|
||||
self.syncServiceID = syncServiceID
|
||||
self.uniqueID = uniqueID
|
||||
self.feedURL = feedURL
|
||||
self.url = url
|
||||
self.externalURL = externalURL
|
||||
self.title = title
|
||||
self.language = language
|
||||
self.contentHTML = contentHTML
|
||||
self.contentText = contentText
|
||||
self.summary = summary
|
||||
self.imageURL = imageURL
|
||||
self.bannerImageURL = bannerImageURL
|
||||
self.datePublished = datePublished
|
||||
self.dateModified = dateModified
|
||||
self.authors = authors
|
||||
self.tags = tags
|
||||
self.attachments = attachments
|
||||
}
|
||||
|
||||
// MARK: - Hashable
|
||||
|
||||
public func hash(into hasher: inout Hasher) {
|
||||
if let syncServiceID = syncServiceID {
|
||||
hasher.combine(syncServiceID)
|
||||
}
|
||||
else {
|
||||
hasher.combine(uniqueID)
|
||||
hasher.combine(feedURL)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
32
Modules/Parser/Sources/Parser/Feeds/XML/AtomParser.swift
Normal file
32
Modules/Parser/Sources/Parser/Feeds/XML/AtomParser.swift
Normal file
@@ -0,0 +1,32 @@
|
||||
//
|
||||
// AtomParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/25/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
#if SWIFT_PACKAGE
|
||||
import ParserObjC
|
||||
#endif
|
||||
|
||||
// RSSParser wraps the Objective-C RSAtomParser.
|
||||
//
|
||||
// The Objective-C parser creates RSParsedFeed, RSParsedArticle, etc.
|
||||
// This wrapper then creates ParsedFeed, ParsedItem, etc. so that it creates
|
||||
// the same things that JSONFeedParser and RSSInJSONParser create.
|
||||
//
|
||||
// In general, you should see FeedParser.swift for all your feed-parsing needs.
|
||||
|
||||
public struct AtomParser {
|
||||
|
||||
public static func parse(_ parserData: ParserData) -> ParsedFeed? {
|
||||
|
||||
if let rsParsedFeed = RSAtomParser.parseFeed(with: parserData) {
|
||||
return RSParsedFeedTransformer.parsedFeed(rsParsedFeed)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,80 @@
|
||||
//
|
||||
// RSParsedFeedTransformer.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/25/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
#if SWIFT_PACKAGE
|
||||
import ParserObjC
|
||||
#endif
|
||||
|
||||
// RSRSSParser and RSAtomParser were written in Objective-C quite a while ago.
|
||||
// They create an RSParsedFeed object and related Objective-C objects.
|
||||
// These functions take an RSParsedFeed and return a Swift-y ParsedFeed,
|
||||
// which is part of providing a single API for feed parsing.
|
||||
|
||||
struct RSParsedFeedTransformer {
|
||||
|
||||
static func parsedFeed(_ rsParsedFeed: RSParsedFeed) -> ParsedFeed {
|
||||
|
||||
let items = parsedItems(rsParsedFeed.articles)
|
||||
return ParsedFeed(type: .rss, title: rsParsedFeed.title, homePageURL: rsParsedFeed.link, feedURL: rsParsedFeed.urlString, language: rsParsedFeed.language, feedDescription: nil, nextURL: nil, iconURL: nil, faviconURL: nil, authors: nil, expired: false, hubs: nil, items: items)
|
||||
}
|
||||
}
|
||||
|
||||
private extension RSParsedFeedTransformer {
|
||||
|
||||
static func parsedItems(_ parsedArticles: Set<RSParsedArticle>) -> Set<ParsedItem> {
|
||||
|
||||
// Create Set<ParsedItem> from Set<RSParsedArticle>
|
||||
|
||||
return Set(parsedArticles.map(parsedItem))
|
||||
}
|
||||
|
||||
static func parsedItem(_ parsedArticle: RSParsedArticle) -> ParsedItem {
|
||||
|
||||
let uniqueID = parsedArticle.articleID
|
||||
let url = parsedArticle.permalink
|
||||
let externalURL = parsedArticle.link
|
||||
let title = parsedArticle.title
|
||||
let language = parsedArticle.language
|
||||
let contentHTML = parsedArticle.body
|
||||
let datePublished = parsedArticle.datePublished
|
||||
let dateModified = parsedArticle.dateModified
|
||||
let authors = parsedAuthors(parsedArticle.authors)
|
||||
let attachments = parsedAttachments(parsedArticle.enclosures)
|
||||
|
||||
return ParsedItem(syncServiceID: nil, uniqueID: uniqueID, feedURL: parsedArticle.feedURL, url: url, externalURL: externalURL, title: title, language: language, contentHTML: contentHTML, contentText: nil, summary: nil, imageURL: nil, bannerImageURL: nil, datePublished: datePublished, dateModified: dateModified, authors: authors, tags: nil, attachments: attachments)
|
||||
}
|
||||
|
||||
static func parsedAuthors(_ authors: Set<RSParsedAuthor>?) -> Set<ParsedAuthor>? {
|
||||
|
||||
guard let authors = authors, !authors.isEmpty else {
|
||||
return nil
|
||||
}
|
||||
|
||||
let transformedAuthors = authors.compactMap { (author) -> ParsedAuthor? in
|
||||
return ParsedAuthor(name: author.name, url: author.url, avatarURL: nil, emailAddress: author.emailAddress)
|
||||
}
|
||||
|
||||
return transformedAuthors.isEmpty ? nil : Set(transformedAuthors)
|
||||
}
|
||||
|
||||
static func parsedAttachments(_ enclosures: Set<RSParsedEnclosure>?) -> Set<ParsedAttachment>? {
|
||||
|
||||
guard let enclosures = enclosures, !enclosures.isEmpty else {
|
||||
return nil
|
||||
}
|
||||
|
||||
let attachments = enclosures.compactMap { (enclosure) -> ParsedAttachment? in
|
||||
|
||||
let sizeInBytes = enclosure.length > 0 ? enclosure.length : nil
|
||||
return ParsedAttachment(url: enclosure.url, mimeType: enclosure.mimeType, title: nil, sizeInBytes: sizeInBytes, durationInSeconds: nil)
|
||||
}
|
||||
|
||||
return attachments.isEmpty ? nil : Set(attachments)
|
||||
}
|
||||
}
|
||||
29
Modules/Parser/Sources/Parser/Feeds/XML/RSSParser.swift
Normal file
29
Modules/Parser/Sources/Parser/Feeds/XML/RSSParser.swift
Normal file
@@ -0,0 +1,29 @@
|
||||
//
|
||||
// RSSParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/25/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import ParserObjC
|
||||
|
||||
// RSSParser wraps the Objective-C RSRSSParser.
|
||||
//
|
||||
// The Objective-C parser creates RSParsedFeed, RSParsedArticle, etc.
|
||||
// This wrapper then creates ParsedFeed, ParsedItem, etc. so that it creates
|
||||
// the same things that JSONFeedParser and RSSInJSONParser create.
|
||||
//
|
||||
// In general, you should see FeedParser.swift for all your feed-parsing needs.
|
||||
|
||||
public struct RSSParser {
|
||||
|
||||
public static func parse(_ parserData: ParserData) -> ParsedFeed? {
|
||||
|
||||
if let rsParsedFeed = RSRSSParser.parseFeed(with: parserData) {
|
||||
return RSParsedFeedTransformer.parsedFeed(rsParsedFeed)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
}
|
||||
12
Modules/Parser/Sources/Parser/JSON/JSONTypes.swift
Normal file
12
Modules/Parser/Sources/Parser/JSON/JSONTypes.swift
Normal file
@@ -0,0 +1,12 @@
|
||||
//
|
||||
// JSONDictionary.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 6/24/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public typealias JSONDictionary = [String: Any]
|
||||
public typealias JSONArray = [JSONDictionary]
|
||||
27
Modules/Parser/Sources/Parser/JSON/JSONUtilities.swift
Normal file
27
Modules/Parser/Sources/Parser/JSON/JSONUtilities.swift
Normal file
@@ -0,0 +1,27 @@
|
||||
//
|
||||
// JSONUtilities.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Brent Simmons on 12/10/17.
|
||||
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct JSONUtilities {
|
||||
|
||||
public static func object(with data: Data) -> Any? {
|
||||
|
||||
return try? JSONSerialization.jsonObject(with: data)
|
||||
}
|
||||
|
||||
public static func dictionary(with data: Data) -> JSONDictionary? {
|
||||
|
||||
return object(with: data) as? JSONDictionary
|
||||
}
|
||||
|
||||
public static func array(with data: Data) -> JSONArray? {
|
||||
|
||||
return object(with: data) as? JSONArray
|
||||
}
|
||||
}
|
||||
11
Modules/Parser/Sources/Parser/ParserData+Parser.swift
Normal file
11
Modules/Parser/Sources/Parser/ParserData+Parser.swift
Normal file
@@ -0,0 +1,11 @@
|
||||
//
|
||||
// File.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 4/7/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import ParserObjC
|
||||
|
||||
extension ParserData: @unchecked Sendable {}
|
||||
11
Modules/Parser/Sources/Parser/RSHTMLMetadata+Parser.swift
Normal file
11
Modules/Parser/Sources/Parser/RSHTMLMetadata+Parser.swift
Normal file
@@ -0,0 +1,11 @@
|
||||
//
|
||||
// File.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 4/7/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import ParserObjC
|
||||
|
||||
extension RSHTMLMetadataParser: @unchecked Sendable {}
|
||||
@@ -0,0 +1,17 @@
|
||||
//
|
||||
// String+RSParser.swift
|
||||
// RSParser
|
||||
//
|
||||
// Created by Nate Weaver on 2020-01-19.
|
||||
// Copyright © 2020 Ranchero Software, LLC. All rights reserved.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
extension String {
|
||||
|
||||
var nilIfEmptyOrWhitespace: String? {
|
||||
return self.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty ? nil : self
|
||||
}
|
||||
|
||||
}
|
||||
Reference in New Issue
Block a user