Package java.util.regex

Examples of java.util.regex.Pattern


  public void testNotCompiled()
    throws Exception
  {
    Regex node = new Regex(expression, expression, null);
    Pattern pattern = node.compile(new LiteralImpl("regex"), null);
    assertNotNull(pattern);
    assertEquals("regex", pattern.pattern());
  }
View Full Code Here


  private void assertRegexLike(String regex, String like)
    throws Exception
  {
    Regex node = new Regex(expression, like, true);
    Pattern pattern = node.compile(null, null);
    assertNotNull(pattern);
    assertEquals(regex, pattern.pattern());
  }
View Full Code Here

                    sb.crawlQueues.noticeURL.clear(NoticedURL.StackType.CORE);
                    try { sb.cleanProfiles(); } catch (final InterruptedException e) {/* ignore this */}
                } else if (option > INVALID) {
                    try {
                        // compiling the regular expression
                        final Pattern compiledPattern = Pattern.compile(pattern);
                       
                        if (option == PROFILE) {
                            // search and delete the crawl profile (_much_ faster, independant of queue size)
                            // XXX: what to do about the annoying LOST PROFILE messages in the log?
                            CrawlProfile entry;
                            for (final byte[] handle: sb.crawler.getActive()) {
                                entry = sb.crawler.getActive(handle);
                                final String name = entry.name();
                                if (name.equals(CrawlSwitchboard.CRAWL_PROFILE_PROXY) ||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_REMOTE) ||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_LOCAL_TEXT||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_GLOBAL_TEXT||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_LOCAL_MEDIA) ||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_GLOBAL_MEDIA) ||
                                        name.equals(CrawlSwitchboard.CRAWL_PROFILE_SURROGATE))
                                    continue;
                                if (compiledPattern.matcher(name).find()) sb.crawler.removeActive(entry.handle().getBytes());
                            }
                        } else {
                            // iterating through the list of URLs
                            final Iterator<Request> iter = sb.crawlQueues.noticeURL.iterator(NoticedURL.StackType.CORE);
                            Request entry;
                            final List<byte[]> removehashes = new ArrayList<byte[]>();
                            while (iter.hasNext()) {
                                if ((entry = iter.next()) == null) continue;
                                String value = null;
                               
                                location: switch (option) {
                                    case URL:       value = (entry.url() == null) ? null : entry.url().toString(); break location;
                                    case ANCHOR:    value = entry.name(); break location;
                                    case DEPTH:     value = Integer.toString(entry.depth()); break location;
                                    case INITIATOR:
                                        value = (entry.initiator() == null || entry.initiator().length == 0) ? "proxy" : ASCII.String(entry.initiator());
                                        break location;
                                    case MODIFIED:  value = daydate(entry.appdate()); break location;
                                    default: value = null; break location;
                                }
                               
                                if (value != null && compiledPattern.matcher(value).matches()) removehashes.add(entry.url().hash());
                            }
                            Log.logInfo("IndexCreateWWWLocalQueue", "created a remove list with " + removehashes.size() + " entries for pattern '" + pattern + "'");
                            for (final byte[] b: removehashes) {
                                sb.crawlQueues.noticeURL.removeByURLHash(b);
                            }
View Full Code Here

    if (is == null) {
      return;
    }

    Pattern p_comment = Pattern.compile("^#");
    Pattern p_parameter =
        Pattern.compile("(.*)\t(.*)\t(.*)\t(.*)\t(.*)\t(.*)");
    Matcher m;

    InputStreamReader isr = new InputStreamReader(is);
    BufferedReader br = new BufferedReader(isr);

    String line;
    while ((line = br.readLine()) != null) {
      if (p_comment.matcher(line).find()) {
        continue;
      }
      m = p_parameter.matcher(line);
      if (m.find() && (m.groupCount() == 6)) {
        GridParameter p = new GridParameter(Integer.parseInt(m.group(3)), m.group(4), m.group(6),
            m.group(5));
        getDiscipline(Integer.parseInt(m.group(1))).getCategory(
            Integer.parseInt(m.group(2))).setParameter(p);
View Full Code Here

            log.debug("Matched first message...");

            posList.add(new Long(0));
          }

          Pattern fromPattern = Pattern.compile(FROM__PATTERN);
          Matcher matcher;

          do {
            log.debug("scanning from " + String.valueOf(offset) + " to " + String.valueOf(offset+size));
            matcher = fromPattern.matcher(cb);
            while (matcher.find()) {
                // log.debug("Found match at [" + String.valueOf(offset+matcher.start()) + "]");

                // add one (1) to position to account for newline..
                posList.add(new Long(offset+matcher.start() + 1));
View Full Code Here

        final long    maxtime = Math.min((int) sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXTIME_DEFAULT, 3000), post.getLong("time", 3000)); // maximum number of wanted results
        final int     maxdist= post.getInt("maxdist", Integer.MAX_VALUE);
        final String  prefer = post.get("prefer", "");
        final String  contentdom = post.get("contentdom", "text");
        final String  filter = post.get("filter", ".*"); // a filter on the url
        final Pattern snippetPattern = Pattern.compile(post.get("snippet", ".*")); // a filter on the snippet
        String  sitehash = post.get("sitehash", ""); if (sitehash.length() == 0) sitehash = null;
        String  authorhash = post.get("authorhash", ""); if (authorhash.length() == 0) authorhash = null;
        String  language = post.get("language", "");
        if (language == null || language.length() == 0 || !ISO639.exists(language)) {
            // take language from the user agent
View Full Code Here

* @author stoughto
*/
public class ResourceList {

    public static Collection<String> getResources(String regex) {
        Pattern pattern;
        pattern = Pattern.compile(regex);
        return ResourceList.getResources(pattern);
    }
View Full Code Here

            // prepare search properties
            final boolean globalsearch = (global) && indexReceiveGranted;

            // do the search
            final HandleSet queryHashes = Word.words2hashesHandles(query[0]);
            final Pattern snippetPattern = QueryParams.stringSearchPattern(originalquerystring);

            // check filters
            try {
                Pattern.compile(urlmask);
            } catch (final PatternSyntaxException ex) {
View Full Code Here

    return exit;
  }

  private String[] parse(String command) {
    Pattern pattern = Pattern.compile("\"([^\"]*)\"|(\\S+)");
    Matcher matcher = pattern.matcher(command);
    List<String> tokens = new ArrayList<String>();
    while (matcher.find()) {
      if (matcher.group(1) != null) {
        tokens.add(matcher.group(1));
      }
View Full Code Here

    */
   
        final String counts = post.get("count", null);
        int maxcount = (counts == null || counts.equals("all")) ? Integer.MAX_VALUE : post.getInt("count", 10);
        String pattern = post.get("search", "");
        Pattern matcher = (pattern.isEmpty() || pattern.equals(".*")) ? null : Pattern.compile(".*" + pattern + ".*");
        prop.put("pattern", pattern);
       
        List<String> columns = new ArrayList<String>();
        for (final Map.Entry<String, String> entry: post.entrySet()) {
            if (entry.getKey().startsWith("col_")) {
View Full Code Here

TOP

Related Classes of java.util.regex.Pattern

Copyright © 2015 www.massapicom. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.