| 285 | } |
| 286 | |
| 287 | private MapFieldValueFilter<String, WebPage> getBatchIdFilter(String batchId) { |
| 288 | if (batchId.equals(REPARSE.toString()) |
| 289 | || batchId.equals(Nutch.ALL_CRAWL_ID.toString())) { |
| 290 | return null; |
| 291 | } |
| 292 | MapFieldValueFilter<String, WebPage> filter = new MapFieldValueFilter<String, WebPage>(); |
| 293 | filter.setFieldName(WebPage.Field.MARKERS.toString()); |
| 294 | filter.setFilterOp(FilterOp.EQUALS); |
| 295 | filter.setFilterIfMissing(true); |
| 296 | filter.setMapKey(Mark.FETCH_MARK.getName()); |
| 297 | filter.getOperands().add(new Utf8(batchId)); |
| 298 | return filter; |
| 299 | } |
| 300 | |
| 301 | public int parse(String batchId, boolean shouldResume, boolean force) |
| 302 | throws Exception { |