|
| 1 | +package com.rarchives.ripme.ripper.rippers; |
| 2 | + |
| 3 | +import java.io.IOException; |
| 4 | +import java.net.MalformedURLException; |
| 5 | +import java.net.URL; |
| 6 | +import java.util.ArrayList; |
| 7 | +import java.util.List; |
| 8 | +import java.util.regex.Matcher; |
| 9 | +import java.util.regex.Pattern; |
| 10 | + |
| 11 | +import org.jsoup.nodes.Document; |
| 12 | +import org.apache.logging.log4j.LogManager; |
| 13 | +import org.apache.logging.log4j.Logger; |
| 14 | +import org.jsoup.nodes.Element; |
| 15 | +import org.jsoup.select.Elements; |
| 16 | + |
| 17 | +import com.rarchives.ripme.ripper.AbstractHTMLRipper; |
| 18 | + |
| 19 | +public class FapwizRipper extends AbstractHTMLRipper { |
| 20 | + |
| 21 | + private static final Logger logger = LogManager.getLogger(FapwizRipper.class); |
| 22 | + |
| 23 | + private static final Pattern CATEGORY_PATTERN = |
| 24 | + Pattern.compile("https?://fapwiz.com/category/([a-zA-Z1-9_-]+)/?$"); |
| 25 | + |
| 26 | + private static final Pattern USER_PATTERN = |
| 27 | + Pattern.compile("https?://fapwiz.com/([a-zA-Z1-9_-]+)/?$"); |
| 28 | + |
| 29 | + private static final Pattern POST_PATTERN = |
| 30 | + Pattern.compile("https?://fapwiz.com/([a-zA-Z1-9_-]+)/([a-zA-Z1-9_-]+)/?$"); |
| 31 | + |
| 32 | + public FapwizRipper(URL url) throws IOException { |
| 33 | + super(url); |
| 34 | + } |
| 35 | + |
| 36 | + @Override |
| 37 | + public String getHost() { |
| 38 | + return "fapwiz"; |
| 39 | + } |
| 40 | + |
| 41 | + @Override |
| 42 | + public String getDomain() { |
| 43 | + return "fapwiz.com"; |
| 44 | + } |
| 45 | + |
| 46 | + @Override |
| 47 | + public String getGID(URL url) throws MalformedURLException { |
| 48 | + Matcher m; |
| 49 | + |
| 50 | + m = CATEGORY_PATTERN.matcher(url.toExternalForm()); |
| 51 | + if (m.matches()) { |
| 52 | + return "category_" + m.group(1); |
| 53 | + } |
| 54 | + |
| 55 | + m = USER_PATTERN.matcher(url.toExternalForm()); |
| 56 | + if (m.matches()) { |
| 57 | + return "user_" + m.group(1); |
| 58 | + } |
| 59 | + |
| 60 | + m = POST_PATTERN.matcher(url.toExternalForm()); |
| 61 | + if (m.matches()) { |
| 62 | + return m.group(1) + "_" + m.group(2); |
| 63 | + } |
| 64 | + |
| 65 | + throw new MalformedURLException("Expected fapwiz URL format: " + |
| 66 | + "fapwiz.com/NAME - got " + url + " instead"); |
| 67 | + } |
| 68 | + |
| 69 | + @Override |
| 70 | + public List<String> getURLsFromPage(Document doc) { |
| 71 | + List<String> result = new ArrayList<>(); |
| 72 | + |
| 73 | + Matcher m; |
| 74 | + |
| 75 | + m = CATEGORY_PATTERN.matcher(url.toExternalForm()); |
| 76 | + if (m.matches()) { |
| 77 | + // structure of user page - does it work for category pages too? |
| 78 | + doc.select(".post-items-holder img").forEach(e -> { |
| 79 | + String imgSrc = e.attr("src"); |
| 80 | + // Replace -thumbnail.jpg with .mp4 |
| 81 | + String videoSrc = imgSrc.replace("-thumbnail.jpg", ".mp4"); |
| 82 | + result.add(videoSrc); |
| 83 | + }); |
| 84 | + } |
| 85 | + |
| 86 | + m = USER_PATTERN.matcher(url.toExternalForm()); |
| 87 | + if (m.matches()) { |
| 88 | + // structure of user page |
| 89 | + doc.select(".post-items-holder img").forEach(e -> { |
| 90 | + String imgSrc = e.attr("src"); |
| 91 | + // Replace -thumbnail.jpg with .mp4 |
| 92 | + String videoSrc = imgSrc.replace("-thumbnail.jpg", ".mp4"); |
| 93 | + result.add(videoSrc); |
| 94 | + }); |
| 95 | + } |
| 96 | + |
| 97 | + m = POST_PATTERN.matcher(url.toExternalForm()); |
| 98 | + if (m.matches()) { |
| 99 | + doc.select("video source").forEach(video -> { |
| 100 | + result.add(video.attr("src")); |
| 101 | + }); |
| 102 | + } |
| 103 | + |
| 104 | + return result; |
| 105 | + } |
| 106 | + |
| 107 | + @Override |
| 108 | + public void downloadURL(URL url, int index) { |
| 109 | + sleep(2000); |
| 110 | + addURLToDownload(url, getPrefix(index)); |
| 111 | + } |
| 112 | +} |
0 commit comments