Commit | Line | Data |
---|---|---|
eaaeae39 NR |
1 | package be.nikiroo.gofetch.support; |
2 | ||
3 | import java.io.IOException; | |
4 | import java.io.InputStream; | |
5 | import java.net.URL; | |
6 | import java.util.ArrayList; | |
7 | import java.util.List; | |
8 | ||
9 | import org.jsoup.helper.DataUtil; | |
10 | import org.jsoup.nodes.Document; | |
11 | import org.jsoup.nodes.Element; | |
12 | import org.jsoup.select.Elements; | |
13 | ||
14 | import be.nikiroo.gofetch.data.Comment; | |
15 | import be.nikiroo.gofetch.data.Story; | |
16 | ||
17 | /** | |
18 | * Support <a href='https://lwn.net/'>https://lwn.net/</a>. | |
19 | * | |
20 | * @author niki | |
21 | */ | |
22 | public class LWN extends BasicSupport { | |
23 | @Override | |
24 | public String getDescription() { | |
25 | return "LWN: Linux Weekly Newsletter"; | |
26 | } | |
27 | ||
28 | @Override | |
29 | public List<Story> list() throws IOException { | |
30 | // TODO: comments + do not get comment for [$] stories | |
31 | // + update body on getComment (global change, also LinuxToday) | |
32 | ||
33 | List<Story> list = new ArrayList<Story>(); | |
34 | ||
35 | URL url = new URL("https://lwn.net/"); | |
36 | InputStream in = open(url); | |
37 | Document doc = DataUtil.load(in, "UTF-8", url.toString()); | |
38 | Elements stories = doc.getElementsByClass("pure-u-1"); | |
39 | for (Element story : stories) { | |
40 | Elements titles = story.getElementsByClass("Headline"); | |
41 | Elements listings = story.getElementsByClass("BlurbListing"); | |
42 | if (titles.size() == 0) { | |
43 | continue; | |
44 | } | |
45 | if (listings.size() == 0) { | |
46 | continue; | |
47 | } | |
48 | ||
49 | Element listing = listings.get(0); | |
50 | if (listing.children().size() < 2) { | |
51 | continue; | |
52 | } | |
53 | ||
54 | ||
55 | String title = titles.get(0).text(); | |
56 | String details = listing.children().get(0).text(); | |
57 | String body = listing.children().get(1).text(); | |
58 | ||
59 | String author = ""; | |
60 | int pos = details.indexOf(" by "); | |
61 | if (pos >= 0) { | |
62 | author = details.substring(pos + " by ".length()).trim(); | |
63 | } | |
64 | ||
65 | String date = ""; | |
66 | pos = details.indexOf(" Posted "); | |
67 | if (pos >= 0) { | |
68 | date = details.substring(pos + " Posted ".length()).trim(); | |
69 | } | |
70 | ||
71 | ||
72 | String id = ""; | |
73 | String intUrl = ""; | |
74 | String extUrl = ""; | |
75 | for (Element idElem : story.getElementsByTag("a")) { | |
76 | // Last link is the story link | |
77 | intUrl = idElem.absUrl("href"); | |
78 | pos = intUrl.indexOf("#Comments"); | |
79 | if (pos >= 0) { | |
80 | intUrl = intUrl.substring(0, pos -1); | |
81 | } | |
82 | id = intUrl.replaceAll("[^0-9]", ""); | |
83 | } | |
84 | ||
85 | list.add(new Story(getType(), id, title, details, intUrl, extUrl, body)); | |
86 | } | |
87 | ||
88 | return list; | |
89 | } | |
90 | ||
91 | @Override | |
92 | public List<Comment> getComments(Story story) throws IOException { | |
93 | List<Comment> comments = new ArrayList<Comment>(); | |
94 | ||
95 | /* | |
96 | URL url = new URL(story.getUrlInternal()); | |
97 | InputStream in = open(url); | |
98 | Document doc = DataUtil.load(in, "UTF-8", url.toString()); | |
99 | Elements listing = doc.getElementsByTag("main"); | |
100 | if (listing.size() > 0) { | |
101 | comments.addAll(getComments(listing.get(0))); | |
102 | } | |
103 | */ | |
104 | ||
105 | return comments; | |
106 | } | |
107 | ||
108 | private List<Comment> getComments(Element listing) { | |
109 | List<Comment> comments = new ArrayList<Comment>(); | |
110 | for (Element commentElement : listing.children()) { | |
111 | if (commentElement.hasClass("comment")) { | |
112 | Comment comment = getComment(commentElement); | |
113 | if (!comment.isEmpty()) { | |
114 | comments.add(comment); | |
115 | } | |
116 | } | |
117 | } | |
118 | return comments; | |
119 | } | |
120 | ||
121 | private Comment getComment(Element commentElement) { | |
122 | String title = firstOrEmptyTag(commentElement, "h3"); | |
123 | String author = firstOrEmpty(commentElement, "h4"); | |
124 | String content = firstOrEmpty(commentElement, "comment-body"); | |
125 | ||
126 | String date = ""; | |
127 | int pos = author.lastIndexOf(" on "); | |
128 | if (pos >= 0) { | |
129 | date = author.substring(pos + " on ".length()).trim(); | |
130 | author = author.substring(0, pos).trim(); | |
131 | } | |
132 | ||
133 | Comment comment = new Comment(commentElement.id(), author, title, date, | |
134 | content); | |
135 | ||
136 | Elements commentOutline = commentElement | |
137 | .getElementsByClass("comment-outline"); | |
138 | if (commentOutline.size() > 0) { | |
139 | comment.addAll(getComments(commentOutline.get(0))); | |
140 | } | |
141 | ||
142 | return comment; | |
143 | } | |
144 | ||
145 | /** | |
146 | * Get the first element of the given class, or an empty {@link String} if | |
147 | * none found. | |
148 | * | |
149 | * @param element | |
150 | * the element to look in | |
151 | * @param className | |
152 | * the class to look for | |
153 | * | |
154 | * @return the value or an empty {@link String} | |
155 | */ | |
156 | private String firstOrEmpty(Element element, String className) { | |
157 | Elements subElements = element.getElementsByClass(className); | |
158 | if (subElements.size() > 0) { | |
159 | return subElements.get(0).text(); | |
160 | } | |
161 | ||
162 | return ""; | |
163 | } | |
164 | ||
165 | /** | |
166 | * Get the first element of the given tag, or an empty {@link String} if | |
167 | * none found. | |
168 | * | |
169 | * @param element | |
170 | * the element to look in | |
171 | * @param tagName | |
172 | * the tag to look for | |
173 | * | |
174 | * @return the value or an empty {@link String} | |
175 | */ | |
176 | private String firstOrEmptyTag(Element element, String tagName) { | |
177 | Elements subElements = element.getElementsByTag(tagName); | |
178 | if (subElements.size() > 0) { | |
179 | return subElements.get(0).text(); | |
180 | } | |
181 | ||
182 | return ""; | |
183 | } | |
184 | } |