|
1
6 package net.sourceforge.cvsgrab.web;
7
8 import java.util.ArrayList  ;
9 import java.util.Iterator ;
10 import java.util.List ;
11 import java.util.Properties ;
12
13 import net.sourceforge.cvsgrab.CVSGrab;
14 import net.sourceforge.cvsgrab.CvsWebInterface;
15 import net.sourceforge.cvsgrab.InvalidVersionException;
16 import net.sourceforge.cvsgrab.MarkerNotFoundException;
17 import net.sourceforge.cvsgrab.RemoteFile;
18 import net.sourceforge.cvsgrab.WebBrowser;
19
20 import org.apache.commons.httpclient.URIException;
21 import org.apache.commons.httpclient.util.URIUtil;
22 import org.apache.commons.jxpath.JXPathContext;
23 import org.apache.commons.jxpath.Pointer;
24 import org.w3c.dom.Document ;
25
26
33 public abstract class ViewCvsInterface extends CvsWebInterface {
34
35 private String _type;
36 private String _filesXpath = "//TR[TD/A/IMG/@alt = '(file)']";
37 private String _fileNameXpath = "TD[1]/A/@name";
38 private String _fileVersionXpath = "TD[A/IMG/@alt != '(graph)'][2]/A/B";
39 private String _directoriesXpath = "//TR[TD/A/IMG/@alt = '(dir)'][TD/A/@name != 'Attic']";
40 private String _directoryXpath = "TD[1]/A/@name";
41 private String _checkoutPath = "*checkout*/";
42 private String _webInterfaceType = "viewcvs";
43 private String _tagParam = "only_with_tag";
44 private String _cvsrootParam = "cvsroot";
45
46
50 public ViewCvsInterface(CVSGrab grabber) {
51 super(grabber);
52 }
53
54
57 public void init() throws Exception {
58 _type = getId();
59 }
60
61
67 public void detect(Document htmlPage) throws MarkerNotFoundException, InvalidVersionException {
68
69 JXPathContext context = JXPathContext.newContext(htmlPage);
70 Iterator viewCvsTexts = context.iterate("//META[@name = 'generator']/@content[starts-with(.,'ViewCVS')] | //A[@href]/text()[starts-with(.,'ViewCVS')]");
71 _type = null;
72 String viewCvsVersion = null;
73 while (viewCvsTexts.hasNext()) {
74 viewCvsVersion = (String ) viewCvsTexts.next();
75 if (viewCvsVersion.startsWith(getVersionMarker())) {
76 _type = viewCvsVersion;
77 break;
78 }
79 }
80 if (_type == null) {
81 throw new MarkerNotFoundException("Expected marker " + getVersionMarker() + ", found " + viewCvsVersion);
82 }
83 }
84
85
88 public String getId() {
89 String className = getClass().getName();
90 className = className.substring(className.lastIndexOf('.') + 1);
91 className = className.substring(0, className.indexOf("Interface"));
92 return className;
93 }
94
95
98 public String getType() {
99 return _type;
100 }
101
102
105 public String getBaseUrl() {
106 String url = WebBrowser.forceFinalSlash(getGrabber().getRootUrl());
107 url += getGrabber().getPackagePath();
108 if (getProjectRoot() != null) {
109 url = WebBrowser.addQueryParam(url, _cvsrootParam, getProjectRoot());
110 }
111 url = WebBrowser.addQueryParam(url, getGrabber().getQueryParams());
112 return url;
113 }
114
115
120 public String getDirectoryUrl(String rootUrl, String directoryName) {
121 try {
122 String tag = getVersionTag();
123 String url = WebBrowser.forceFinalSlash(rootUrl);
124 url += WebBrowser.forceFinalSlash(quote(directoryName));
125 if (getProjectRoot() != null) {
126 url = WebBrowser.addQueryParam(url, _cvsrootParam, getProjectRoot());
127 }
128 url = WebBrowser.addQueryParam(url, _tagParam, tag);
129 url = WebBrowser.addQueryParam(url, getQueryParams());
130 return url;
131 } catch (URIException ex) {
132 ex.printStackTrace();
133 throw new RuntimeException ("Cannot create URI");
134 }
135 }
136
137
140 public RemoteFile[] getFiles(Document htmlPage) {
141 JXPathContext context = JXPathContext.newContext(htmlPage);
142 List files = new ArrayList ();
143 Iterator i = context.iteratePointers(getFilesXpath());
144 while (i.hasNext()) {
145 Pointer pointer = (Pointer) i.next();
146 JXPathContext nodeContext = context.getRelativeContext(pointer);
147 String fileName = (String ) nodeContext.getValue(getFileNameXpath());
148 String version = (String ) nodeContext.getValue(getFileVersionXpath());
149 RemoteFile file = new RemoteFile(fileName, version);
150 adjustFile(file, nodeContext);
151 files.add(file);
152 }
153 return (RemoteFile[]) files.toArray(new RemoteFile[files.size()]);
154 }
155
156
159 public String [] getDirectories(Document htmlPage) {
160 JXPathContext context = JXPathContext.newContext(htmlPage);
161 context.registerNamespace("HTML", "http://www.w3.org/1999/xhtml");
162 context.registerNamespace("", "http://www.w3.org/1999/xhtml");
163 List directories = new ArrayList ();
164 Iterator i = context.iteratePointers(getDirectoriesXpath());
165 while (i.hasNext()) {
166 Pointer pointer = (Pointer) i.next();
167 JXPathContext nodeContext = context.getRelativeContext(pointer);
168 try {
169 String dir = (String ) nodeContext.getValue(getDirectoryXpath());
170 directories.add(dir);
171 } catch (RuntimeException e) {
172 CVSGrab.getLog().error("Cannot localise directory name in document location " + nodeContext.getPointer("."), e);
173 }
174 }
175 return (String []) directories.toArray(new String [directories.size()]);
176 }
177
178 public String getDownloadUrl(RemoteFile file) {
179 try {
180 String url = WebBrowser.forceFinalSlash(file.getDirectory().getRemoteRepository().getRootUrl());
182 String dir = file.getDirectory().getDirectoryPath();
183 url += getCheckoutPath();
184 url += WebBrowser.forceFinalSlash(quote(dir));
185 if (file.isInAttic()) {
186 url += "Attic/";
187 }
188 url += quote(file.getName());
189 if (getProjectRoot() != null) {
190 url = WebBrowser.addQueryParam(url, _cvsrootParam, getProjectRoot());
191 }
192 url = WebBrowser.addQueryParam(url, "rev", file.getVersion());
193 url = WebBrowser.addQueryParam(url, getQueryParams());
194 return url;
195 } catch (URIException ex) {
196 ex.printStackTrace();
197 throw new RuntimeException ("Cannot create URI");
198 }
199 }
200
201 public Properties guessWebProperties(String url) {
202 Properties properties = new Properties ();
203 int keywordPosition = url.toLowerCase().indexOf(_webInterfaceType);
205 if (keywordPosition > 0) {
206 int rootUrlPosition = url.indexOf('/', keywordPosition) + 1;
207 int cgiFolderPos = url.indexOf("cgi/", rootUrlPosition);
208 if (cgiFolderPos > 0) {
209 rootUrlPosition = cgiFolderPos + 4;
210 }
211 int nextSlashPos = url.indexOf('/', rootUrlPosition) + 1;
212 int magicScriptPos = url.indexOf(".cgi", rootUrlPosition);
213 if (magicScriptPos < 0 ) {
214 magicScriptPos = url.indexOf(".py", rootUrlPosition);
215 }
216 if (magicScriptPos > 0 && magicScriptPos < nextSlashPos) {
217 rootUrlPosition = nextSlashPos;
218 }
219 String guessedRootUrl = url.substring(0, rootUrlPosition);
220 String guessedPackagePath = url.substring(rootUrlPosition);
221 String versionTag = null;
222 String cvsroot = null;
223 String query = null;
224 int queryPos = guessedPackagePath.indexOf('?');
225 if (queryPos >= 0) {
226 query = guessedPackagePath.substring(queryPos + 1);
227 guessedPackagePath = guessedPackagePath.substring(0, queryPos);
228 Properties queryItems = WebBrowser.getQueryParams(query);
229 versionTag = (String ) queryItems.remove(_tagParam);
230 cvsroot = (String ) queryItems.remove(_cvsrootParam);
231 query = WebBrowser.toQueryParams(queryItems);
232 }
233 properties.put(CVSGrab.ROOT_URL_OPTION, guessedRootUrl);
234 properties.put(CVSGrab.PACKAGE_PATH_OPTION, guessedPackagePath);
235 if (versionTag != null && versionTag.trim().length() > 0) {
236 properties.put(CVSGrab.TAG_OPTION, versionTag);
237 }
238 if (cvsroot != null && cvsroot.trim().length() > 0) {
239 properties.put(CVSGrab.PROJECT_ROOT_OPTION, cvsroot);
240 }
241 if (query != null && query.trim().length() > 0) {
242 properties.put(CVSGrab.QUERY_PARAMS_OPTION, query);
243 }
244 }
245 return properties;
246 }
247
248 public String getFilesXpath() {
249 return _filesXpath;
250 }
251
252 public String getFileNameXpath() {
253 return _fileNameXpath;
254 }
255
256 public String getFileVersionXpath() {
257 return _fileVersionXpath;
258 }
259
260 public String getDirectoriesXpath() {
261 return _directoriesXpath;
262 }
263
264 public String getDirectoryXpath() {
265 return _directoryXpath;
266 }
267
268 protected String getCheckoutPath() {
269 return _checkoutPath;
270 }
271
272 protected void setCheckoutPath(String checkoutPath) {
273 _checkoutPath = checkoutPath;
274 }
275
276 public void setDirectoryXpath(String directoryXpath) {
277 _directoryXpath = directoryXpath;
278 }
279
280 public void setDirectoriesXpath(String directoriesXpath) {
281 _directoriesXpath = directoriesXpath;
282 }
283
284 public void setFileVersionXpath(String fileVersionXpath) {
285 _fileVersionXpath = fileVersionXpath;
286 }
287
288 public void setFileNameXpath(String fileNameXpath) {
289 _fileNameXpath = fileNameXpath;
290 }
291
292 public void setFilesXpath(String filesXpath) {
293 _filesXpath = filesXpath;
294 }
295
296 public String getTagParam() {
297 return _tagParam;
298 }
299
300 public void setTagParam(String param) {
301 _tagParam = param;
302 }
303
304 public String getWebInterfaceType() {
305 return _webInterfaceType;
306 }
307
308 protected void setWebInterfaceType(String webInterfaceType) {
309 this._webInterfaceType = webInterfaceType;
310 }
311
312 public String getCvsrootParam() {
313 return _cvsrootParam;
314 }
315
316 public void setCvsrootParam(String cvsrootParam) {
317 _cvsrootParam = cvsrootParam;
318 }
319
320
323 protected void setType(String type) {
324 _type = type;
325 }
326
327 protected abstract String getVersionMarker();
328
329 protected void adjustFile(RemoteFile file, JXPathContext nodeContext) {
330 String fileName = file.getName();
331 if (fileName.startsWith("Attic/")) {
332 file.setName(fileName.substring(6));
333 file.setInAttic(true);
334 }
335 }
336
337
343 protected String quote(String original) throws URIException {
344 return URIUtil.encodePath(original, "ISO-8859-1");
345 }
346
347 protected String getProjectRoot() {
348 return getGrabber().getProjectRoot();
349 }
350
351 }
352 |
Popular Tags |
Java API By Example, From Geeks To Geeks.