1 | package org.greenstone.gsdl3.gs3build.util;
|
---|
2 |
|
---|
3 | import java.applet.*;
|
---|
4 | import java.net.*;
|
---|
5 |
|
---|
6 | public class URLString
|
---|
7 | { private String url;
|
---|
8 |
|
---|
9 | private String urlbase(String url)
|
---|
10 | { int index;
|
---|
11 | String tailString, bodyString;
|
---|
12 | int body;
|
---|
13 |
|
---|
14 | index = url.indexOf("//");
|
---|
15 | tailString = url.substring(index+2, url.length());
|
---|
16 |
|
---|
17 | body = tailString.indexOf('/');
|
---|
18 | bodyString = tailString.substring(0, body+1);
|
---|
19 |
|
---|
20 | tailString = url.substring(0, index+2) + bodyString;
|
---|
21 | return tailString;
|
---|
22 | }
|
---|
23 |
|
---|
24 | public URLString(String s)
|
---|
25 | { this.url = s;
|
---|
26 | }
|
---|
27 |
|
---|
28 | public URLString(String s, String target)
|
---|
29 | { // reconcile
|
---|
30 | int sref;
|
---|
31 | int cref;
|
---|
32 | int c;
|
---|
33 |
|
---|
34 | // trim fragment identifiers
|
---|
35 | target = noRef(target);
|
---|
36 | if (target == null)
|
---|
37 | { this.url = s;
|
---|
38 | return;
|
---|
39 | }
|
---|
40 |
|
---|
41 | // test for absolute
|
---|
42 | cref = target.indexOf(':');
|
---|
43 | if (cref >= 0)
|
---|
44 | { for (c = 0; c < cref; c ++)
|
---|
45 | { if (target.charAt(c) != '+' &&
|
---|
46 | target.charAt(c) != '-' &&
|
---|
47 | target.charAt(c) != '.' &&
|
---|
48 | Character.isLetterOrDigit(target.charAt(c)) == false)
|
---|
49 | { break;
|
---|
50 | }
|
---|
51 | }
|
---|
52 | }
|
---|
53 | else
|
---|
54 | { c = 0;
|
---|
55 | }
|
---|
56 |
|
---|
57 | if (c == cref)
|
---|
58 | { // absolute reference
|
---|
59 | String tmp;
|
---|
60 |
|
---|
61 | tmp = target.substring(0, c);
|
---|
62 | tmp = tmp.toLowerCase();
|
---|
63 | if (target.indexOf("//") == cref + 1)
|
---|
64 | { if (target.lastIndexOf('/') == cref + 2)
|
---|
65 | { target = target + "/";
|
---|
66 | }
|
---|
67 | }
|
---|
68 | /* if (tmp.equals("http") == false)
|
---|
69 | { this.url = null;
|
---|
70 | return;
|
---|
71 | }*/
|
---|
72 | this.url = target;
|
---|
73 | }
|
---|
74 | else if (target.indexOf("//") == 0)
|
---|
75 | { // absolute - implicit scheme (ie http).; also check for no trailing '/'
|
---|
76 | if (target.lastIndexOf('/') == 1)
|
---|
77 | { target = target + "/";
|
---|
78 | }
|
---|
79 | target = "http:" + target;
|
---|
80 | this.url = target;
|
---|
81 | }
|
---|
82 | else
|
---|
83 | { // relative - trim silly '/' at end
|
---|
84 | if (target.charAt(0) == '/')
|
---|
85 | { // if we get a leading '/' then add it to the base of the url
|
---|
86 | s = urlbase(s);
|
---|
87 | target = target.substring(1);
|
---|
88 | this.url = s + target;
|
---|
89 | // System.out.println(this.url+"<"+s+"<"+target);
|
---|
90 | return;
|
---|
91 | }
|
---|
92 |
|
---|
93 | while(target.length() >= 2 && target.charAt(0) == '.' && target.charAt(1) == '.')
|
---|
94 | { // move up a level on the left
|
---|
95 |
|
---|
96 | sref = s.lastIndexOf('/');
|
---|
97 | s = s.substring(0, sref);
|
---|
98 |
|
---|
99 | target = target.substring(3);
|
---|
100 | }
|
---|
101 | // strip trailing leaf from s
|
---|
102 | sref = s.lastIndexOf('/');
|
---|
103 | s = s.substring(0, sref+1);
|
---|
104 | // append s+target
|
---|
105 | this.url = s + target;
|
---|
106 | }
|
---|
107 |
|
---|
108 | // abandon the url if it is a call to a script
|
---|
109 | if (this.url.indexOf('?') >= 0)
|
---|
110 | { this.url = null;
|
---|
111 | }
|
---|
112 | }
|
---|
113 |
|
---|
114 | public static boolean dirParent(String parent, String child)
|
---|
115 | { while (child.length() >= parent.length())
|
---|
116 | { child = child.substring(0, child.lastIndexOf('/'));
|
---|
117 |
|
---|
118 | if (parent.equals(child))
|
---|
119 | { return true;
|
---|
120 | }
|
---|
121 | }
|
---|
122 | return false;
|
---|
123 | }
|
---|
124 |
|
---|
125 | public static int dirPos(String from, String to)
|
---|
126 | { String fromdir;
|
---|
127 | String todir;
|
---|
128 |
|
---|
129 | fromdir = from.substring(0, from.lastIndexOf('/'));
|
---|
130 | todir = to.substring(0, to.lastIndexOf('/'));
|
---|
131 |
|
---|
132 | if (fromdir.equals(todir))
|
---|
133 | { return 0;
|
---|
134 | }
|
---|
135 | if (fromdir.length() > todir.length())
|
---|
136 | { if (dirParent(todir, fromdir) == true)
|
---|
137 | { return -1;
|
---|
138 | }
|
---|
139 | }
|
---|
140 | else
|
---|
141 | { if (dirParent(fromdir, todir) == true)
|
---|
142 | { return 1;
|
---|
143 | }
|
---|
144 | }
|
---|
145 | return -2;
|
---|
146 | }
|
---|
147 |
|
---|
148 | public static int commonLevels(String a, String b)
|
---|
149 | { String path_a, path_b;
|
---|
150 | int offset_a, offset_b;
|
---|
151 | String dir_a, dir_b;
|
---|
152 | int common;
|
---|
153 |
|
---|
154 | path_a = path(a);
|
---|
155 | path_b = path(b);
|
---|
156 | common = 0;
|
---|
157 |
|
---|
158 | if (path_a == null || path_b == null)
|
---|
159 | { return 0;
|
---|
160 | }
|
---|
161 |
|
---|
162 | do
|
---|
163 | { offset_a = path_a.indexOf('/');
|
---|
164 | offset_b = path_b.indexOf('/');
|
---|
165 |
|
---|
166 | if (offset_a > 0)
|
---|
167 | { dir_a = path_a.substring(0, offset_a);
|
---|
168 | if (path_a.length() > offset_a + 1)
|
---|
169 | { path_a = path_a.substring(offset_a+1);
|
---|
170 | }
|
---|
171 | else
|
---|
172 | { path_a = null;
|
---|
173 | }
|
---|
174 | }
|
---|
175 | else
|
---|
176 | { dir_a = path_a;
|
---|
177 | path_a = null;
|
---|
178 | }
|
---|
179 |
|
---|
180 | if (offset_b > 0)
|
---|
181 | { dir_b = path_b.substring(0, offset_b);
|
---|
182 | if (path_b.length() > offset_b + 1)
|
---|
183 | { path_b = path_b.substring(offset_b+1);
|
---|
184 | }
|
---|
185 | else
|
---|
186 | { path_b = null;
|
---|
187 | }
|
---|
188 | }
|
---|
189 | else
|
---|
190 | { dir_b = path_b;
|
---|
191 | path_b = null;
|
---|
192 | }
|
---|
193 |
|
---|
194 | if (dir_a.length() != dir_b.length() ||
|
---|
195 | dir_a.equals(dir_b) == false)
|
---|
196 | { return common;
|
---|
197 | }
|
---|
198 | common ++;
|
---|
199 | } while (path_a != null && path_b != null);
|
---|
200 |
|
---|
201 | return common;
|
---|
202 | }
|
---|
203 |
|
---|
204 | public static String dirLevelName(String url, int level)
|
---|
205 | { String lpath;
|
---|
206 | String dir;
|
---|
207 | int thislevel = 0;
|
---|
208 | int offset;
|
---|
209 |
|
---|
210 | lpath = path(url);
|
---|
211 | if (lpath == null)
|
---|
212 | { return null;
|
---|
213 | }
|
---|
214 | do
|
---|
215 | { offset = lpath.indexOf('/');
|
---|
216 |
|
---|
217 | if (offset >= 0)
|
---|
218 | { dir = lpath.substring(0, offset);
|
---|
219 | lpath = lpath.substring(offset+1);
|
---|
220 | }
|
---|
221 | else
|
---|
222 | { dir = lpath;
|
---|
223 | }
|
---|
224 |
|
---|
225 | thislevel ++;
|
---|
226 |
|
---|
227 | if (level == thislevel)
|
---|
228 | { return dir;
|
---|
229 | }
|
---|
230 | } while (offset >= 0);
|
---|
231 |
|
---|
232 | return null;
|
---|
233 | }
|
---|
234 |
|
---|
235 | public static int dirLevel(String url)
|
---|
236 | { String lpath;
|
---|
237 | int level;
|
---|
238 | int offset;
|
---|
239 |
|
---|
240 | lpath = path(url);
|
---|
241 | if (lpath == null)
|
---|
242 | { return 0;
|
---|
243 | }
|
---|
244 | level = 0;
|
---|
245 | do
|
---|
246 | { offset = lpath.indexOf('/');
|
---|
247 | if (offset >= 0)
|
---|
248 | { lpath = lpath.substring(offset+1);
|
---|
249 | }
|
---|
250 | level ++;
|
---|
251 | } while (offset >= 0);
|
---|
252 |
|
---|
253 | return level;
|
---|
254 | }
|
---|
255 |
|
---|
256 | public static String tidyURL(String urlString, boolean permitScript)
|
---|
257 | { int cref;
|
---|
258 | int c;
|
---|
259 | int sref;
|
---|
260 | String tmp, s;
|
---|
261 |
|
---|
262 | s = urlString;
|
---|
263 | cref = s.indexOf(':');
|
---|
264 | if (cref >= 0)
|
---|
265 | { for (c = 0; c < cref; c ++)
|
---|
266 | { if (s.charAt(c) != '+' &&
|
---|
267 | s.charAt(c) != '-' &&
|
---|
268 | s.charAt(c) != '.' &&
|
---|
269 | Character.isLetterOrDigit(s.charAt(c)) == false)
|
---|
270 | { break;
|
---|
271 | }
|
---|
272 | }
|
---|
273 | }
|
---|
274 | else
|
---|
275 | { c = 0;
|
---|
276 | }
|
---|
277 |
|
---|
278 | if (cref == c)
|
---|
279 | { tmp = s.substring(0, c).toLowerCase();
|
---|
280 | s = tmp + s.substring(c, s.length());
|
---|
281 | if (s.lastIndexOf('/') == cref + 2)
|
---|
282 | { s = s + "/";
|
---|
283 | }
|
---|
284 | }
|
---|
285 | else
|
---|
286 | { sref = s.indexOf("//");
|
---|
287 | if (sref == 0)
|
---|
288 | { s = "http:" + s;
|
---|
289 | }
|
---|
290 | else
|
---|
291 | { s = "http://" + s;
|
---|
292 | }
|
---|
293 | if (s.lastIndexOf('/') == 6)
|
---|
294 | { s = s + "/";
|
---|
295 | }
|
---|
296 | }
|
---|
297 | return s;
|
---|
298 | }
|
---|
299 |
|
---|
300 | public static String tidyURL(String urlString)
|
---|
301 | { return tidyURL(urlString, false);
|
---|
302 | }
|
---|
303 |
|
---|
304 | public static URL toURL(String urlString)
|
---|
305 | { URL reply;
|
---|
306 |
|
---|
307 | try
|
---|
308 | { reply = new URL(URLString.tidyURL(urlString));
|
---|
309 | }
|
---|
310 | catch (MalformedURLException ex)
|
---|
311 | { return null;
|
---|
312 | }
|
---|
313 | return reply;
|
---|
314 | }
|
---|
315 |
|
---|
316 | public void tidy()
|
---|
317 | { this.url = tidyURL(this.url);
|
---|
318 | }
|
---|
319 |
|
---|
320 | public boolean isNull()
|
---|
321 | { if (this.url == null)
|
---|
322 | { return true;
|
---|
323 | }
|
---|
324 | return false;
|
---|
325 | }
|
---|
326 |
|
---|
327 | public static boolean isForeign(String home, URL url)
|
---|
328 | { if (url.toString().length() < home.length())
|
---|
329 | { return true;
|
---|
330 | }
|
---|
331 |
|
---|
332 | if (home.equals(url.toString().substring(0, home.length())))
|
---|
333 | { return false;
|
---|
334 | }
|
---|
335 | return true;
|
---|
336 | }
|
---|
337 |
|
---|
338 | /**
|
---|
339 | * Remove any trailing "#localanchor" local reference from a URL string
|
---|
340 | *
|
---|
341 | * @param <code>String</code> the original url
|
---|
342 | * @return <code>String</code> the cleaned url, sans local references
|
---|
343 | */
|
---|
344 | public static String noRef(String target)
|
---|
345 | { if (target == null)
|
---|
346 | { return null;
|
---|
347 | }
|
---|
348 | if (target.indexOf('#') >= 0)
|
---|
349 | { if (target.indexOf('#') == 0)
|
---|
350 | { return null;
|
---|
351 | }
|
---|
352 | target = target.substring(0, target.indexOf('#'));
|
---|
353 | }
|
---|
354 | return target;
|
---|
355 | }
|
---|
356 |
|
---|
357 | /**
|
---|
358 | * Obtain the trailing local anchor reference from a URL,
|
---|
359 | * if it exists.
|
---|
360 | * e.g. for http://www.test.mdx.ac.uk/example/index.html#credits the
|
---|
361 | * response would be "credits"
|
---|
362 | *
|
---|
363 | * @param <code>String</code> the original url.
|
---|
364 | * @return <code>String</code> the local reference. <code>null</code> is
|
---|
365 | * returned if no local reference exists.
|
---|
366 | */
|
---|
367 | public static String subRef(String target)
|
---|
368 | { if (target.indexOf('#') >= 0)
|
---|
369 | { return target.substring(target.indexOf('#')+1);
|
---|
370 | }
|
---|
371 | return null;
|
---|
372 | }
|
---|
373 |
|
---|
374 | /**
|
---|
375 | * Clean a URL of the form GET parameters.
|
---|
376 | *
|
---|
377 | * @param <code>String</code> the original url, with GET message parameters
|
---|
378 | * @return <code>String</code> the cleaned url.
|
---|
379 | */
|
---|
380 | public static String noParameters(String url)
|
---|
381 | { int c;
|
---|
382 |
|
---|
383 | c = url.indexOf("?");
|
---|
384 | if (c < 0)
|
---|
385 | { return url;
|
---|
386 | }
|
---|
387 | return url.substring(0, c);
|
---|
388 | }
|
---|
389 |
|
---|
390 | /**
|
---|
391 | * Obtain the form parameters from a url
|
---|
392 | *
|
---|
393 | * @param <code>String</code> the original url, with GET message parameters
|
---|
394 | * @return <code>String</code> the form "GET" parameters. The leading '?' is
|
---|
395 | * preserved in the reply. If no parameters were
|
---|
396 | * given, the result is <code>null</code>.
|
---|
397 | */
|
---|
398 | public static String formParameters(String url)
|
---|
399 | { int c;
|
---|
400 |
|
---|
401 | c = url.indexOf("?");
|
---|
402 | if (c < 0)
|
---|
403 | { return null;
|
---|
404 | }
|
---|
405 | return url.substring(c);
|
---|
406 | }
|
---|
407 |
|
---|
408 | /**
|
---|
409 | * Flatten a URL, making the '?' and '&' items into underscores
|
---|
410 | *
|
---|
411 | * @param <code>String</code> the original url, with GET message parameters
|
---|
412 | * @return <code>String</code> the "flattened" url
|
---|
413 | */
|
---|
414 | public static String flattenedUrl(String oldUrl)
|
---|
415 | { StringBuffer reply;
|
---|
416 | int c;
|
---|
417 | boolean modified = false;
|
---|
418 |
|
---|
419 | // replace ? with _
|
---|
420 | // replate & with _
|
---|
421 | reply = new StringBuffer(oldUrl);
|
---|
422 | for (c = 0; c < reply.length(); c ++)
|
---|
423 | { if (reply.charAt(c) == '?' ||
|
---|
424 | reply.charAt(c) == '&')
|
---|
425 | { reply.setCharAt(c, '_');
|
---|
426 | modified = true;
|
---|
427 | }
|
---|
428 | }
|
---|
429 |
|
---|
430 | if (modified == true)
|
---|
431 | { c = oldUrl.indexOf(".html");
|
---|
432 | if (c >= 0)
|
---|
433 | { while (c < oldUrl.length() - 5)
|
---|
434 | { reply.setCharAt(c, reply.charAt(c+5));
|
---|
435 | c ++;
|
---|
436 | }
|
---|
437 | reply.setCharAt(c ++, '.');
|
---|
438 | reply.setCharAt(c ++, 'h');
|
---|
439 | reply.setCharAt(c ++, 't');
|
---|
440 | reply.setCharAt(c ++, 'm');
|
---|
441 | reply.setCharAt(c ++, 'l');
|
---|
442 | }
|
---|
443 | }
|
---|
444 |
|
---|
445 | return reply.toString();
|
---|
446 | }
|
---|
447 |
|
---|
448 | /**
|
---|
449 | * Provide the complete pathname in the given url
|
---|
450 | * eg. "example/index.html" for http://www.test.mdx.ac.uk/example/index.html
|
---|
451 | *
|
---|
452 | * @param <code>String</code> url of page
|
---|
453 | * @return <code>String</code> leaf file name
|
---|
454 | */
|
---|
455 | public static String pathName(String url)
|
---|
456 | { int offset;
|
---|
457 | String remainder;
|
---|
458 |
|
---|
459 | if (url == null)
|
---|
460 | { return null;
|
---|
461 | }
|
---|
462 |
|
---|
463 | offset = url.indexOf("//");
|
---|
464 | if (offset < 0)
|
---|
465 | { return null;
|
---|
466 | }
|
---|
467 |
|
---|
468 | remainder = url.substring(offset + 2);
|
---|
469 | offset = remainder.indexOf('/');
|
---|
470 | if (offset < 0)
|
---|
471 | { return null;
|
---|
472 | }
|
---|
473 | return remainder.substring(offset + 1);
|
---|
474 | }
|
---|
475 |
|
---|
476 | /**
|
---|
477 | * Provide the leaf filename in the given url
|
---|
478 | * eg. "index.html" for http://www.test.mdx.ac.uk/example/index.html
|
---|
479 | *
|
---|
480 | * @param <code>String</code> url of page
|
---|
481 | * @return <code>String</code> leaf file name
|
---|
482 | */
|
---|
483 | public static String leafName(String url)
|
---|
484 | { int offset;
|
---|
485 | int dot;
|
---|
486 |
|
---|
487 | if (url == null)
|
---|
488 | { return null;
|
---|
489 | }
|
---|
490 | offset = url.lastIndexOf('/');
|
---|
491 | if (offset < 0)
|
---|
492 | { return null;
|
---|
493 | }
|
---|
494 | if (offset == url.length() - 1)
|
---|
495 | { return null;
|
---|
496 | }
|
---|
497 | dot = url.lastIndexOf('/');
|
---|
498 | if (dot < offset)
|
---|
499 | { return null;
|
---|
500 | }
|
---|
501 | return url.substring(offset+1);
|
---|
502 | }
|
---|
503 |
|
---|
504 | /**
|
---|
505 | * Provide the hostname in the given url
|
---|
506 | * eg. "www.test.mdx.ac.uk" for http://www.test.mdx.ac.uk/example/index.html
|
---|
507 | *
|
---|
508 | * @param <code>String</code> url of page
|
---|
509 | * @return <code>String</code> host name
|
---|
510 | */
|
---|
511 | public static String host(String url)
|
---|
512 | { String scheme, tail;
|
---|
513 | int spos;
|
---|
514 |
|
---|
515 | spos = url.indexOf("//");
|
---|
516 | if (spos >= 0)
|
---|
517 | { scheme = url.substring(0, spos + 2);
|
---|
518 | tail = url.substring(spos + 2, url.length());
|
---|
519 | // do sub-areas
|
---|
520 | spos = tail.indexOf('/');
|
---|
521 | if (spos >= 0)
|
---|
522 | { tail = tail.substring(0, spos+1);
|
---|
523 | }
|
---|
524 | scheme = scheme + tail;
|
---|
525 | return scheme;
|
---|
526 | }
|
---|
527 | return null;
|
---|
528 | }
|
---|
529 |
|
---|
530 | /**
|
---|
531 | * Return the logical path of the url on the remote server
|
---|
532 | * eg. "example" for http://www.test.mdx.ac.uk/example/index.html
|
---|
533 | *
|
---|
534 | * @param <code>String</code> url of page
|
---|
535 | * @return <code>String</code> logical path
|
---|
536 | */
|
---|
537 | public static String path(String url)
|
---|
538 | { String tail;
|
---|
539 | int offset;
|
---|
540 | String leaf;
|
---|
541 |
|
---|
542 | if (url == null)
|
---|
543 | { return null;
|
---|
544 | }
|
---|
545 |
|
---|
546 | offset = url.indexOf("//");
|
---|
547 | if (offset < 0)
|
---|
548 | { return null;
|
---|
549 | }
|
---|
550 |
|
---|
551 | tail = url.substring(offset+2);
|
---|
552 | offset = tail.indexOf('/');
|
---|
553 | if (offset < 0)
|
---|
554 | { return null;
|
---|
555 | }
|
---|
556 |
|
---|
557 | tail = tail.substring(offset+1);
|
---|
558 | if (tail == null || tail.length() == 0)
|
---|
559 | { return null;
|
---|
560 | }
|
---|
561 |
|
---|
562 | offset = tail.lastIndexOf('/');
|
---|
563 | if (offset < 0)
|
---|
564 | { return null;
|
---|
565 | }
|
---|
566 | else
|
---|
567 | { tail = tail.substring(0, offset);
|
---|
568 | }
|
---|
569 | return tail;
|
---|
570 | }
|
---|
571 |
|
---|
572 | /**
|
---|
573 | * Give the 'dot' extension of a url;
|
---|
574 | * eg. "html" for http://www.test.mdx.ac.uk/example/index.html
|
---|
575 | *
|
---|
576 | * @param <code>String</code> url
|
---|
577 | * @return <code>String</code> extension.
|
---|
578 | */
|
---|
579 | public static String extension(String url)
|
---|
580 | { int offset;
|
---|
581 | String leaf;
|
---|
582 |
|
---|
583 | if (url == null)
|
---|
584 | { return null;
|
---|
585 | }
|
---|
586 |
|
---|
587 | offset = url.lastIndexOf('/');
|
---|
588 | if (offset >= 0 && offset < url.length() - 1)
|
---|
589 | { leaf = url.substring(offset+1, url.length());
|
---|
590 | offset = leaf.lastIndexOf('.');
|
---|
591 |
|
---|
592 | // if we've got an extension then use it
|
---|
593 | if (offset >= 0)
|
---|
594 | { int poffset;
|
---|
595 |
|
---|
596 | // get parameter offset
|
---|
597 | poffset = leaf.indexOf('?');
|
---|
598 | if (poffset < 0)
|
---|
599 | { poffset = leaf.length();
|
---|
600 | }
|
---|
601 |
|
---|
602 | // give leaf value
|
---|
603 | return leaf.substring(offset, poffset);
|
---|
604 | }
|
---|
605 | else
|
---|
606 | { return null;
|
---|
607 | }
|
---|
608 | }
|
---|
609 | return null;
|
---|
610 | }
|
---|
611 |
|
---|
612 | public static String extension(URL url)
|
---|
613 | { return extension(url.toString());
|
---|
614 | }
|
---|
615 |
|
---|
616 | public String toString()
|
---|
617 | { return this.url;
|
---|
618 | }
|
---|
619 | }
|
---|