(String reversedUrl, WebPage page, Context context)
| 47 | private float scoreThreshold; |
| 48 | |
| 49 | @Override |
| 50 | public void map(String reversedUrl, WebPage page, Context context) |
| 51 | throws IOException, InterruptedException { |
| 52 | String url = TableUtil.unreverseUrl(reversedUrl); |
| 53 | |
| 54 | if (Mark.GENERATE_MARK.checkMark(page) != null) { |
| 55 | GeneratorJob.LOG.debug("Skipping {}; already generated", url); |
| 56 | return; |
| 57 | } |
| 58 | |
| 59 | // filter on distance |
| 60 | if (maxDistance > -1) { |
| 61 | CharSequence distanceUtf8 = page.getMarkers().get(DbUpdaterJob.DISTANCE); |
| 62 | if (distanceUtf8 != null) { |
| 63 | int distance = Integer.parseInt(distanceUtf8.toString()); |
| 64 | if (distance > maxDistance) { |
| 65 | return; |
| 66 | } |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | // If filtering is on don't generate URLs that don't pass URLFilters |
| 71 | try { |
| 72 | if (normalise) { |
| 73 | url = normalizers.normalize(url, |
| 74 | URLNormalizers.SCOPE_GENERATE_HOST_COUNT); |
| 75 | } |
| 76 | if (filter && filters.filter(url) == null) |
| 77 | return; |
| 78 | if ((sitemap && !URLFilters.isSitemap(page)) || !sitemap && URLFilters |
| 79 | .isSitemap(page)) |
| 80 | return; |
| 81 | } catch (URLFilterException | MalformedURLException e) { |
| 82 | GeneratorJob.LOG |
| 83 | .warn("Couldn't filter url: {} ({})", url, e); |
| 84 | return; |
| 85 | } |
| 86 | |
| 87 | // check fetch schedule |
| 88 | if (!schedule.shouldFetch(url, page, curTime)) { |
| 89 | if (GeneratorJob.LOG.isDebugEnabled()) { |
| 90 | GeneratorJob.LOG.debug("-shouldFetch rejected '" + url |
| 91 | + "', fetchTime=" + page.getFetchTime() + ", curTime=" + curTime); |
| 92 | } |
| 93 | return; |
| 94 | } |
| 95 | float score = page.getScore(); |
| 96 | try { |
| 97 | score = scoringFilters.generatorSortValue(url, page, score); |
| 98 | } catch (ScoringFilterException e) { |
| 99 | // ignore |
| 100 | } |
| 101 | |
| 102 | // consider only entries with a score superior to the threshold |
| 103 | if (scoreThreshold != Float.NaN && score < scoreThreshold) |
| 104 | return; |
| 105 | |
| 106 | entry.set(url, score); |
nothing calls this directly
no test coverage detected