| 302 | } |
| 303 | |
| 304 | private void putOutlink(WebPage page, Outlink outlink, String toUrl) { |
| 305 | try { |
| 306 | toUrl = normalizers.normalize(toUrl, URLNormalizers.SCOPE_OUTLINK); |
| 307 | toUrl = filters.filter(toUrl); |
| 308 | } catch (MalformedURLException e2) { |
| 309 | return; |
| 310 | } catch (URLFilterException e) { |
| 311 | return; |
| 312 | } |
| 313 | if (toUrl == null) { |
| 314 | return; |
| 315 | } |
| 316 | Utf8 utf8ToUrl = new Utf8(toUrl); |
| 317 | if (page.getOutlinks().get(utf8ToUrl) != null) { |
| 318 | // skip duplicate outlinks |
| 319 | return; |
| 320 | } |
| 321 | page.getOutlinks().put(utf8ToUrl, new Utf8(outlink.getAnchor())); |
| 322 | } |
| 323 | |
| 324 | /** |
| 325 | * Parses given web page and stores parsed content within page. Puts a |