Commit | Line | Data |
---|---|---|
73785268 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.net.URLConnection; | |
27008a87 | 7 | import java.util.ArrayList; |
73785268 NR |
8 | import java.util.List; |
9 | import java.util.zip.GZIPInputStream; | |
10 | ||
27008a87 NR |
11 | import org.jsoup.helper.StringUtil; |
12 | import org.jsoup.nodes.Element; | |
13 | import org.jsoup.nodes.Node; | |
14 | import org.jsoup.nodes.TextNode; | |
15 | import org.jsoup.select.Elements; | |
16 | import org.jsoup.select.NodeTraversor; | |
17 | import org.jsoup.select.NodeVisitor; | |
18 | ||
73785268 NR |
19 | import be.nikiroo.gofetch.data.Story; |
20 | ||
21 | public abstract class BasicSupport { | |
22 | public enum Type { | |
eaaeae39 | 23 | SLASHDOT, PIPEDOT, LWN, |
73785268 NR |
24 | } |
25 | ||
27008a87 NR |
26 | public interface QuoteProcessor { |
27 | public boolean detectQuote(Node node); | |
28 | ||
29 | public String processText(String text); | |
30 | ||
31 | public boolean ignoreNode(Node node); | |
32 | } | |
33 | ||
73785268 NR |
34 | static private String preselector; |
35 | ||
36 | private Type type; | |
37 | ||
38 | abstract public List<Story> list() throws IOException; | |
39 | ||
5c056aad NR |
40 | /** |
41 | * Fetch the full article content as well as all the comments associated to | |
42 | * this {@link Story}, if any (can be empty, but not NULL). | |
43 | * | |
44 | * @param story | |
45 | * the story to fetch the comments of | |
46 | * | |
47 | * @throws IOException | |
48 | * in case of I/O error | |
49 | */ | |
50 | abstract public void fetch(Story story) throws IOException; | |
73785268 NR |
51 | |
52 | abstract public String getDescription(); | |
2d95a873 | 53 | |
73785268 NR |
54 | public String getSelector() { |
55 | return getSelector(type); | |
56 | } | |
57 | ||
58 | public Type getType() { | |
59 | return type; | |
60 | } | |
61 | ||
62 | protected void setType(Type type) { | |
63 | this.type = type; | |
64 | } | |
65 | ||
66 | /** | |
67 | * @param preselector | |
68 | * the preselector to set | |
69 | */ | |
70 | static public void setPreselector(String preselector) { | |
71 | BasicSupport.preselector = preselector; | |
72 | } | |
73 | ||
74 | static public BasicSupport getSupport(Type type) { | |
75 | BasicSupport support = null; | |
76 | ||
77 | if (type != null) { | |
78 | switch (type) { | |
79 | case SLASHDOT: | |
80 | support = new Slashdot(); | |
81 | break; | |
2d95a873 NR |
82 | case PIPEDOT: |
83 | support = new Pipedot(); | |
84 | break; | |
eaaeae39 NR |
85 | case LWN: |
86 | support = new LWN(); | |
87 | break; | |
73785268 NR |
88 | } |
89 | ||
90 | if (support != null) { | |
91 | support.setType(type); | |
92 | } | |
93 | } | |
94 | ||
95 | return support; | |
96 | } | |
97 | ||
98 | static public String getSelector(Type type) { | |
99 | return preselector + "/" + type + "/"; | |
100 | } | |
101 | ||
102 | // TODO: check Downloader.java? | |
103 | static protected InputStream open(URL url) throws IOException { | |
104 | URLConnection conn = url.openConnection(); | |
105 | conn.connect(); | |
106 | InputStream in = conn.getInputStream(); | |
107 | if ("gzip".equals(conn.getContentEncoding())) { | |
108 | in = new GZIPInputStream(in); | |
109 | } | |
110 | ||
111 | return in; | |
112 | } | |
27008a87 NR |
113 | |
114 | /** | |
115 | * Get the first {@link Element} of the given class, or an empty span | |
116 | * {@link Element} if none found. | |
117 | * | |
118 | * @param element | |
119 | * the element to look in | |
120 | * @param className | |
121 | * the class to look for | |
122 | * | |
123 | * @return the value or an empty span {@link Element} | |
124 | */ | |
125 | static protected Element firstOrEmpty(Element element, String className) { | |
126 | Elements subElements = element.getElementsByClass(className); | |
127 | if (subElements.size() > 0) { | |
128 | return subElements.get(0); | |
129 | } | |
130 | ||
131 | return new Element("span"); | |
132 | } | |
133 | ||
134 | /** | |
135 | * Get the first {@link Element} of the given tag, or an empty span | |
136 | * {@link Element} if none found. | |
137 | * | |
138 | * @param element | |
139 | * the element to look in | |
140 | * @param tagName | |
141 | * the tag to look for | |
142 | * | |
143 | * @return the value or an empty span {@link Element} | |
144 | */ | |
145 | static protected Element firstOrEmptyTag(Element element, String tagName) { | |
146 | Elements subElements = element.getElementsByTag(tagName); | |
147 | if (subElements.size() > 0) { | |
148 | return subElements.get(0); | |
149 | } | |
150 | ||
151 | return new Element("span"); | |
152 | } | |
153 | ||
154 | static protected List<String> toLines(Element element, | |
155 | final QuoteProcessor quoteProcessor) { | |
156 | final List<String> lines = new ArrayList<String>(); | |
157 | final StringBuilder currentLine = new StringBuilder(); | |
158 | final List<Integer> quoted = new ArrayList<Integer>(); | |
159 | final List<Node> ignoredNodes = new ArrayList<Node>(); | |
160 | ||
161 | if (element != null) { | |
162 | new NodeTraversor(new NodeVisitor() { | |
163 | @Override | |
164 | public void head(Node node, int depth) { | |
165 | if (quoteProcessor.ignoreNode(node) | |
166 | || ignoredNodes.contains(node.parentNode())) { | |
167 | ignoredNodes.add(node); | |
168 | return; | |
169 | } | |
170 | ||
171 | String prep = ""; | |
172 | for (int i = 0; i < quoted.size(); i++) { | |
173 | prep += ">"; | |
174 | } | |
175 | prep += " "; | |
176 | ||
177 | boolean enterQuote = quoteProcessor.detectQuote(node); | |
178 | boolean leaveQuote = quoted.contains(depth); | |
179 | ||
180 | if (enterQuote) { | |
181 | quoted.add(depth); | |
182 | } | |
183 | ||
184 | if (leaveQuote) { | |
185 | quoted.remove(Integer.valueOf(depth)); | |
186 | } | |
187 | ||
188 | if (enterQuote || leaveQuote) { | |
189 | if (currentLine.length() > 0) { | |
190 | if (currentLine.charAt(currentLine.length() - 1) == '\n') { | |
191 | currentLine.setLength(currentLine.length() - 1); | |
192 | } | |
193 | for (String l : currentLine.toString().split("\n")) { | |
194 | lines.add(prep + l); | |
195 | } | |
196 | } | |
197 | currentLine.setLength(0); | |
198 | } | |
199 | ||
200 | if (node instanceof Element) { | |
201 | Element element = (Element) node; | |
202 | boolean block = element.isBlock() | |
203 | || element.tagName().equalsIgnoreCase("br"); | |
204 | if (block && currentLine.length() > 0) { | |
205 | currentLine.append("\n"); | |
206 | } | |
207 | } else if (node instanceof TextNode) { | |
208 | TextNode textNode = (TextNode) node; | |
209 | String line = StringUtil.normaliseWhitespace(textNode | |
210 | .getWholeText()); | |
211 | ||
212 | currentLine.append(quoteProcessor.processText(line)); | |
213 | currentLine.append(" "); | |
214 | } | |
215 | } | |
216 | ||
217 | @Override | |
218 | public void tail(Node node, int depth) { | |
219 | } | |
220 | }).traverse(element); | |
221 | } | |
222 | ||
223 | if (currentLine.length() > 0) { | |
224 | String prep = ""; | |
225 | for (int i = 0; i < quoted.size(); i++) { | |
226 | prep += ">"; | |
227 | } | |
228 | prep += " "; | |
229 | if (currentLine.length() > 0) { | |
230 | if (currentLine.charAt(currentLine.length() - 1) == '\n') { | |
231 | currentLine.setLength(currentLine.length() - 1); | |
232 | } | |
233 | for (String l : currentLine.toString().split("\n")) { | |
234 | lines.add(prep + l); | |
235 | } | |
236 | } | |
237 | } | |
238 | ||
239 | for (int i = 0; i < lines.size(); i++) { | |
240 | lines.set(i, lines.get(i).replace(" ", " ").trim()); | |
241 | } | |
242 | ||
243 | return lines; | |
244 | } | |
73785268 | 245 | } |