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 { | |
100a8395 | 23 | SLASHDOT, PIPEDOT, LWN, LEMONDE, |
73785268 NR |
24 | } |
25 | ||
20217360 NR |
26 | /** |
27 | * Used to process an element into lines. | |
28 | * | |
29 | * @author niki | |
30 | */ | |
31 | public interface ElementProcessor { | |
32 | /** | |
33 | * Detect if this node is a quote and should be trated as such. | |
34 | * | |
35 | * @param node | |
36 | * the node to check | |
37 | * @return TRUE if it is | |
38 | */ | |
27008a87 NR |
39 | public boolean detectQuote(Node node); |
40 | ||
20217360 NR |
41 | /** |
42 | * Process text content (will be called on each text element, allowing | |
43 | * you to modify it if needed). | |
44 | * | |
45 | * @param text | |
46 | * the text to process | |
47 | * @return | |
48 | */ | |
27008a87 NR |
49 | public String processText(String text); |
50 | ||
20217360 NR |
51 | /** |
52 | * Ignore this node. | |
53 | * | |
54 | * @param node | |
55 | * the node to ignore | |
56 | * @return TRUE if it has to be ignored | |
57 | */ | |
27008a87 | 58 | public boolean ignoreNode(Node node); |
100a8395 NR |
59 | |
60 | /** | |
20217360 NR |
61 | * Manually process this node (and return the manual processing value) |
62 | * if so desired. | |
63 | * <p> | |
64 | * If the node is manually processed, it and its children will not be | |
65 | * automatically processed. | |
100a8395 NR |
66 | * |
67 | * @param node | |
68 | * the node to optionally process | |
69 | * | |
20217360 NR |
70 | * @return NULL if not processed (will thus be automatically processed |
71 | * as usual), a {@link String} (may be empty) if we process it | |
72 | * manually -- the given {@link String} will be used instead of | |
73 | * the usual automatic processing if not NULL | |
100a8395 NR |
74 | */ |
75 | public String manualProcessing(Node node); | |
27008a87 NR |
76 | } |
77 | ||
20217360 NR |
78 | /** |
79 | * A default {@link ElementProcessor} (will not detect or process anything | |
80 | * manually). | |
81 | * | |
82 | * @author niki | |
83 | */ | |
84 | public class BasicElementProcessor implements ElementProcessor { | |
85 | @Override | |
86 | public boolean detectQuote(Node node) { | |
87 | return false; | |
88 | } | |
89 | ||
90 | @Override | |
91 | public String processText(String text) { | |
92 | return text; | |
93 | } | |
94 | ||
95 | @Override | |
96 | public boolean ignoreNode(Node node) { | |
97 | return false; | |
98 | } | |
99 | ||
100 | @Override | |
101 | public String manualProcessing(Node node) { | |
102 | return null; | |
103 | } | |
104 | } | |
105 | ||
73785268 NR |
106 | static private String preselector; |
107 | ||
108 | private Type type; | |
109 | ||
100a8395 NR |
110 | /** |
111 | * List all the recent items, but only assure the ID and internal URL to | |
112 | * fetch it later on (until it has been fetched, the rest of the | |
113 | * {@link Story} is not confirmed). | |
114 | * | |
115 | * @return the list of new stories | |
116 | * | |
117 | * @throws IOException | |
118 | * in case of I/O | |
119 | */ | |
73785268 NR |
120 | abstract public List<Story> list() throws IOException; |
121 | ||
5c056aad NR |
122 | /** |
123 | * Fetch the full article content as well as all the comments associated to | |
124 | * this {@link Story}, if any (can be empty, but not NULL). | |
125 | * | |
126 | * @param story | |
127 | * the story to fetch the comments of | |
128 | * | |
129 | * @throws IOException | |
130 | * in case of I/O error | |
131 | */ | |
132 | abstract public void fetch(Story story) throws IOException; | |
73785268 NR |
133 | |
134 | abstract public String getDescription(); | |
2d95a873 | 135 | |
73785268 NR |
136 | public String getSelector() { |
137 | return getSelector(type); | |
138 | } | |
139 | ||
140 | public Type getType() { | |
141 | return type; | |
142 | } | |
143 | ||
144 | protected void setType(Type type) { | |
145 | this.type = type; | |
146 | } | |
147 | ||
148 | /** | |
149 | * @param preselector | |
150 | * the preselector to set | |
151 | */ | |
152 | static public void setPreselector(String preselector) { | |
153 | BasicSupport.preselector = preselector; | |
154 | } | |
155 | ||
20217360 NR |
156 | /** |
157 | * Return a {@link BasicSupport} that is compatible with the given | |
158 | * {@link Type} if it exists (or NULL if not). | |
159 | * | |
160 | * @param type | |
161 | * the type | |
162 | * | |
163 | * @return a compatible {@link BasicSupport} if it exists (or NULL if not) | |
164 | */ | |
73785268 NR |
165 | static public BasicSupport getSupport(Type type) { |
166 | BasicSupport support = null; | |
167 | ||
168 | if (type != null) { | |
169 | switch (type) { | |
170 | case SLASHDOT: | |
171 | support = new Slashdot(); | |
172 | break; | |
2d95a873 NR |
173 | case PIPEDOT: |
174 | support = new Pipedot(); | |
175 | break; | |
eaaeae39 NR |
176 | case LWN: |
177 | support = new LWN(); | |
178 | break; | |
100a8395 NR |
179 | case LEMONDE: |
180 | support = new LeMonde(); | |
181 | break; | |
73785268 NR |
182 | } |
183 | ||
184 | if (support != null) { | |
185 | support.setType(type); | |
186 | } | |
187 | } | |
188 | ||
189 | return support; | |
190 | } | |
191 | ||
192 | static public String getSelector(Type type) { | |
193 | return preselector + "/" + type + "/"; | |
194 | } | |
195 | ||
196 | // TODO: check Downloader.java? | |
197 | static protected InputStream open(URL url) throws IOException { | |
198 | URLConnection conn = url.openConnection(); | |
199 | conn.connect(); | |
200 | InputStream in = conn.getInputStream(); | |
201 | if ("gzip".equals(conn.getContentEncoding())) { | |
202 | in = new GZIPInputStream(in); | |
203 | } | |
204 | ||
205 | return in; | |
206 | } | |
27008a87 NR |
207 | |
208 | /** | |
209 | * Get the first {@link Element} of the given class, or an empty span | |
210 | * {@link Element} if none found. | |
211 | * | |
212 | * @param element | |
213 | * the element to look in | |
214 | * @param className | |
215 | * the class to look for | |
216 | * | |
217 | * @return the value or an empty span {@link Element} | |
218 | */ | |
219 | static protected Element firstOrEmpty(Element element, String className) { | |
220 | Elements subElements = element.getElementsByClass(className); | |
221 | if (subElements.size() > 0) { | |
222 | return subElements.get(0); | |
223 | } | |
224 | ||
225 | return new Element("span"); | |
226 | } | |
227 | ||
228 | /** | |
229 | * Get the first {@link Element} of the given tag, or an empty span | |
230 | * {@link Element} if none found. | |
231 | * | |
232 | * @param element | |
233 | * the element to look in | |
234 | * @param tagName | |
235 | * the tag to look for | |
236 | * | |
237 | * @return the value or an empty span {@link Element} | |
238 | */ | |
239 | static protected Element firstOrEmptyTag(Element element, String tagName) { | |
240 | Elements subElements = element.getElementsByTag(tagName); | |
241 | if (subElements.size() > 0) { | |
242 | return subElements.get(0); | |
243 | } | |
244 | ||
245 | return new Element("span"); | |
246 | } | |
247 | ||
20217360 NR |
248 | /** |
249 | * Process the given element into text (each line is a text paragraph and | |
250 | * can be prepended with ">" signs to indicate a quote or sub-quote or | |
251 | * sub-sub-quote...). | |
252 | * | |
253 | * @param element | |
254 | * the element to process | |
255 | * @param elementProcessor | |
256 | * the element processor, must not be NULL | |
257 | * | |
258 | * @return text lines, each line is a paragraph | |
259 | */ | |
27008a87 | 260 | static protected List<String> toLines(Element element, |
20217360 | 261 | final ElementProcessor elementProcessor) { |
27008a87 NR |
262 | final List<String> lines = new ArrayList<String>(); |
263 | final StringBuilder currentLine = new StringBuilder(); | |
264 | final List<Integer> quoted = new ArrayList<Integer>(); | |
265 | final List<Node> ignoredNodes = new ArrayList<Node>(); | |
266 | ||
267 | if (element != null) { | |
268 | new NodeTraversor(new NodeVisitor() { | |
269 | @Override | |
270 | public void head(Node node, int depth) { | |
100a8395 | 271 | String manual = null; |
20217360 | 272 | boolean ignore = elementProcessor.ignoreNode(node) |
100a8395 NR |
273 | || ignoredNodes.contains(node.parentNode()); |
274 | if (!ignore) { | |
20217360 | 275 | manual = elementProcessor.manualProcessing(node); |
100a8395 NR |
276 | if (manual != null) { |
277 | currentLine.append(manual); | |
278 | ignore = true; | |
279 | } | |
280 | } | |
281 | ||
282 | if (ignore) { | |
27008a87 NR |
283 | ignoredNodes.add(node); |
284 | return; | |
285 | } | |
286 | ||
287 | String prep = ""; | |
288 | for (int i = 0; i < quoted.size(); i++) { | |
289 | prep += ">"; | |
290 | } | |
291 | prep += " "; | |
292 | ||
20217360 | 293 | boolean enterQuote = elementProcessor.detectQuote(node); |
27008a87 NR |
294 | boolean leaveQuote = quoted.contains(depth); |
295 | ||
296 | if (enterQuote) { | |
297 | quoted.add(depth); | |
298 | } | |
299 | ||
300 | if (leaveQuote) { | |
301 | quoted.remove(Integer.valueOf(depth)); | |
302 | } | |
303 | ||
304 | if (enterQuote || leaveQuote) { | |
305 | if (currentLine.length() > 0) { | |
306 | if (currentLine.charAt(currentLine.length() - 1) == '\n') { | |
307 | currentLine.setLength(currentLine.length() - 1); | |
308 | } | |
309 | for (String l : currentLine.toString().split("\n")) { | |
310 | lines.add(prep + l); | |
311 | } | |
312 | } | |
313 | currentLine.setLength(0); | |
314 | } | |
315 | ||
316 | if (node instanceof Element) { | |
317 | Element element = (Element) node; | |
318 | boolean block = element.isBlock() | |
319 | || element.tagName().equalsIgnoreCase("br"); | |
320 | if (block && currentLine.length() > 0) { | |
321 | currentLine.append("\n"); | |
322 | } | |
323 | } else if (node instanceof TextNode) { | |
324 | TextNode textNode = (TextNode) node; | |
325 | String line = StringUtil.normaliseWhitespace(textNode | |
326 | .getWholeText()); | |
327 | ||
20217360 | 328 | currentLine.append(elementProcessor.processText(line)); |
27008a87 NR |
329 | currentLine.append(" "); |
330 | } | |
331 | } | |
332 | ||
333 | @Override | |
334 | public void tail(Node node, int depth) { | |
335 | } | |
336 | }).traverse(element); | |
337 | } | |
338 | ||
339 | if (currentLine.length() > 0) { | |
340 | String prep = ""; | |
341 | for (int i = 0; i < quoted.size(); i++) { | |
342 | prep += ">"; | |
343 | } | |
344 | prep += " "; | |
345 | if (currentLine.length() > 0) { | |
346 | if (currentLine.charAt(currentLine.length() - 1) == '\n') { | |
347 | currentLine.setLength(currentLine.length() - 1); | |
348 | } | |
349 | for (String l : currentLine.toString().split("\n")) { | |
350 | lines.add(prep + l); | |
351 | } | |
352 | } | |
353 | } | |
354 | ||
355 | for (int i = 0; i < lines.size(); i++) { | |
356 | lines.set(i, lines.get(i).replace(" ", " ").trim()); | |
357 | } | |
358 | ||
359 | return lines; | |
360 | } | |
73785268 | 361 | } |