renun/NetNewsWire
1
0
Fork 0
mirror of https://github.com/Ranchero-Software/NetNewsWire synced 2025-08-12 06:26:36 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Create HTMLParserTests.

Browse Source
This commit is contained in:
Brent Simmons
2024-09-21 22:10:47 -07:00
parent 20b222f455
commit a7c4669bd6
9 changed files with 149 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
Modules/Parser/Sources/HTMLParser/HTMLLink.swift
View File

@@ -9,7 +9,14 @@ import Foundation
public final class HTMLLink {
let urlString: String? // Absolute URL string
let text: String?
let title: String? // Title attribute inside anchor tag
public var urlString: String? // Absolute URL string
public var text: String?
public var title: String? // Title attribute inside anchor tag
init(urlString: String? = nil, text: String? = nil, title: String? = nil) {
self.urlString = urlString
self.text = text
self.title = title
}
}

37
Modules/Parser/Sources/HTMLParser/HTMLLinkParser.swift
View File

@@ -15,7 +15,7 @@ public final class HTMLLinkParser {
private let parserData: ParserData
private let baseURL: URL?
public static func htmlLinks(parserData: ParserData) -> [HTMLLink] {
public static func htmlLinks(with parserData: ParserData) -> [HTMLLink] {
let parser = HTMLLinkParser(parserData)
parser.parse()
@@ -38,23 +38,23 @@ private extension HTMLLinkParser {
}
}
private extension HTMLLinkParser: SAXHTMLParserDelegate {
extension HTMLLinkParser: SAXHTMLParserDelegate {
var currentLink: HTMLLink? {
private var currentLink: HTMLLink? {
links.last
}
struct HTMLAttributeName {
let href = "href"
let title = "title"
private struct HTMLAttributeName {
static let href = "href"
static let title = "title"
}
func title(_ attributesDictionary: HTMLAttributesDictionary) -> String? {
private func title(with attributesDictionary: SAXHTMLParser.HTMLAttributesDictionary) -> String? {
attributesDictionary.object(object(forCaseInsensitiveKey: HTMLAttributeName.title))
attributesDictionary.object(forCaseInsensitiveKey: HTMLAttributeName.title)
}
func urlString(_ attributesDictionary: HTMLAttributesDictionary) -> String? {
private func urlString(with attributesDictionary: SAXHTMLParser.HTMLAttributesDictionary) -> String? {
guard let href = attributesDictionary.object(forCaseInsensitiveKey: HTMLAttributeName.href) else {
return nil
@@ -68,22 +68,22 @@ private extension HTMLLinkParser: SAXHTMLParserDelegate {
return absoluteURL.absoluteString
}
func handleLinkAttributes(_ attributesDictionary: HTMLAttributesDictionary) {
private func handleLinkAttributes(_ attributesDictionary: SAXHTMLParser.HTMLAttributesDictionary) {
guard let currentLink else {
assertionFailure("currentLink must not be nil")
return
}
link.urlString = urlString(attributesDictionary)
link.title = title(attributesDictionary)
currentLink.urlString = urlString(with: attributesDictionary)
currentLink.title = title(with: attributesDictionary)
}
struct HTMLName {
private struct HTMLName {
static let a = "a".utf8CString
}
func saxHTMLParser(_ saxHTMLParser: SAXHTMLParser, startElement name: XMLPointer, attributes: UnsafePointer<XMLPointer?>?) {
public func saxHTMLParser(_ saxHTMLParser: SAXHTMLParser, startElement name: XMLPointer, attributes: UnsafePointer<XMLPointer?>?) {
guard SAXEqualTags(name, HTMLName.a) else {
return
@@ -99,15 +99,20 @@ private extension HTMLLinkParser: SAXHTMLParserDelegate {
saxHTMLParser.beginStoringCharacters()
}
func saxHTMLParser(_ saxHTMLParser: SAXHTMLParser, endElement name: XMLPointer) {
public func saxHTMLParser(_ saxHTMLParser: SAXHTMLParser, endElement name: XMLPointer) {
guard SAXEqualTags(name, HTMLName.a) else {
return
}
guard let currentLink else {
assertionFailure("currentLink must not be nil.")
return
}
currentLink.text = saxHTMLParser.currentStringWithTrimmedWhitespace
}
func saxHTMLParser(_: SAXHTMLParser, charactersFound: XMLPointer, count: Int) {
public func saxHTMLParser(_: SAXHTMLParser, charactersFound: XMLPointer, count: Int) {
// Nothing needed.
}
}