Example usage for org.apache.lucene.search IndexSearcher IndexSearcher

List of usage examples for org.apache.lucene.search IndexSearcher IndexSearcher

  1. HOME
  2. Java
  3. org.apache.lucene
  4. org.apache.lucene.search.*
  5. IndexSearcher
  6. IndexSearcher

Introduction

In this page you can find the example usage for org.apache.lucene.search IndexSearcher IndexSearcher.

Prototype

public IndexSearcher(IndexReaderContext context)

Source Link

Document

Creates a searcher searching the provided top-level IndexReaderContext .

Usage

From source file:com.aliasi.lingmed.test.functional.medline.TestDateSearch.java

License:Lingpipe license

public static void main(String[] args) throws Exception {
    File luceneDir = new File(args[0]);
    Searcher medlineLocalSearcher = new IndexSearcher(FSDirectory.getDirectory(luceneDir));
    MedlineSearcher medlineSearcher = new MedlineSearcherImpl(new MedlineCodec(), medlineLocalSearcher);

    System.out.println("test1a: citations between 1985-1985");
    try {/*from  w  ww .  j av a  2s  .  co  m*/
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("1985", "1985");
        System.out.println("citations in 1985: " + hits.size());
        for (MedlineCitation hit : hits) {
            PubDate pubDate = hit.article().journal().journalIssue().pubDate();
            if (pubDate.isStructured()) {
                System.out.println("citation: " + hit.pmid() + "\tpubdate " + pubDate.toString());
            } else {
                System.out.println("citation: " + hit.pmid() + "\tUNSTRUCTURED " + pubDate.medlineDate());
            }
        }
        System.out.println("citations in 1985: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1a failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1b: citations between 2000-2005");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("2000", "2005");
        System.out.println("citations in 2000-2005: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1b failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1c: citations between 2004-2005");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("2004", "2005");
        System.out.println("citations in 2004-2005: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1c failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1d: citations between 1964-1965");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("1964", "1965");
        System.out.println("citations in 1964-1965: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1d failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1e: citations between 0000-9999");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("0000", "9999");
        System.out.println("citations in 0000-9999: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1e failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1f: citations between 0000-1000");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("0000", "1000");
        System.out.println("citations in 0000-1000: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1f failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test1g: citations between 3000-5000");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("3000", "5000");
        System.out.println("citations in 3000-5000: " + hits.size());
    } catch (DaoException de) {
        System.out.println("test1g failed: " + de.getMessage());
        de.printStackTrace();
    }

    System.out.println("test2: citations between 1999-1998");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("1999", "1998");
    } catch (DaoException de) {
        System.out.println("test2 passed (i.e. failed): " + de.getMessage());
    }

    System.out.println("test3: citations between foo-bar");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("foo", "bar");
    } catch (DaoException de) {
        System.out.println("test3 passed (i.e. failed): " + de.getMessage());
    }

    System.out.println("test4: citations between empty range");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("", "");
    } catch (DaoException de) {
        System.out.println("test4 passed (i.e. failed): " + de.getMessage());
    }

    System.out.println("test5: citations between empty range");
    try {
        SearchResults<MedlineCitation> hits = medlineSearcher.getCitationsInYearRange("2000", "");
    } catch (DaoException de) {
        System.out.println("test4 passed (i.e. failed): " + de.getMessage());
    }

    System.exit(0);
}

From source file:com.aliasi.lingmed.test.functional.medline.VerifyDates.java

License:Lingpipe license

public static void main(String[] args) throws Exception {
    File luceneDir = new File(args[0]);
    Searcher medlineLocalSearcher = new IndexSearcher(FSDirectory.getDirectory(luceneDir));
    MedlineSearcher medlineSearcher = new MedlineSearcherImpl(new MedlineCodec(), medlineLocalSearcher);

    int ct = 0;/* w  w  w.  j ava  2  s .co  m*/
    for (MedlineCitation citation : medlineSearcher) {
        if (citation.article() == null || citation.article().journal() == null
                || citation.article().journal().journalIssue() == null
                || citation.article().journal().journalIssue().pubDate() == null) {
            continue;
        }
        PubDate pubDate = citation.article().journal().journalIssue().pubDate();
        if (pubDate.isStructured()) {
            System.out.println("citation: " + citation.pmid() + "\tpubdate " + pubDate.toString());
        } else {
            System.out.println("citation: " + citation.pmid() + "\tUNSTRUCTURED " + pubDate.medlineDate());
        }

    }
    System.exit(0);
}

From source file:com.aperigeek.dropvault.web.service.IndexService.java

License:Open Source License

private IndexSearcher getIndexSearcher(String username, String password) throws IOException {
    IndexSearcher searcher = new IndexSearcher(getDirectory(username, password));
    return searcher;
}

From source file:com.appeligo.alerts.KeywordAlertThread.java

License:Apache License

private IndexSearcher getIndexSearcher() throws IOException {
    if ((currentSearcher != null)
            && ((System.currentTimeMillis() - currentSearcherCreated) > shortestTimeBetweenQueriesMs)) {
        currentSearcher.close();// w ww .j a v a2s  .co  m
        currentSearcher = null;
    }
    if (currentSearcher == null) {
        if (log.isDebugEnabled())
            log.debug("KeywordAlertThread time to get a new searcher");
        currentSearcherCreated = System.currentTimeMillis();
        currentSearcher = new IndexSearcher(liveIndexDir);
    }
    return currentSearcher;
}

From source file:com.appeligo.amazon.AmazonSearcher.java

License:Apache License

protected void loadProductSearcher() {
    Configuration config = ConfigUtils.getAmazonConfig();
    String indexLocation = config.getString("programIndex");
    try {//from  w  w  w .  ja v a 2s.  c om
        productSearcher = new IndexSearcher(indexLocation);
    } catch (IOException e) {
        if (log.isErrorEnabled()) {
            log.error("Cannot load product index: " + indexLocation, e);
        }
    }
}

From source file:com.appeligo.amazon.ProgramIndexer.java

License:Apache License

public void run() {
    if (!indexLocation.exists()) {
        //create the index directory
        indexLocation.mkdirs();/*from   w w w. j av a  2  s  . com*/
    }

    try {
        savedSearches.clear();
        writer = new IndexWriter(indexLocation, analyzer);
        searcher = new IndexSearcher(indexLocation.getPath());

        deleteExpiredPrograms();
        writer.flush();

        //reopen searcher so the deleted documents are not present
        close(searcher);
        searcher = new IndexSearcher(indexLocation.getPath());

        addNewPrograms();

    } catch (Exception e) {
        log.error("Error occurred during indexing.", e);

    } finally {
        savedSearches.clear();
        close(searcher);
        searcher = null;

        close(writer);
        writer = null;

        close(conn);
        conn = null;
    }
}

From source file:com.appeligo.captions.DeleteOldProgramsThread.java

License:Apache License

@Override
public void run() {

    while (true) {
        LuceneIndexer liveIndex = LuceneIndexer.getInstance(liveIndexLocation);
        Calendar wayback = Calendar.getInstance();
        wayback.add(Calendar.MONTH, -3);
        Calendar tenminutesago = Calendar.getInstance();
        tenminutesago.add(Calendar.MINUTE, -10);
        log.info("Deleting old programs from live index, between " + wayback.getTime() + " and "
                + tenminutesago.getTime());
        String dateField = "lineup-" + liveLineup + "-endTime";
        ConstantScoreRangeQuery dateQuery = new ConstantScoreRangeQuery(dateField,
                DateTools.dateToString(wayback.getTime(), DateTools.Resolution.MINUTE),
                DateTools.dateToString(tenminutesago.getTime(), DateTools.Resolution.MINUTE), true, true);
        IndexSearcher searcher = null;//  w  w w  . j a v a2 s .  c  o  m
        try {
            searcher = new IndexSearcher(liveIndexLocation);
            Hits hits = searcher.search(dateQuery);
            Set<Term> terms = new HashSet<Term>();
            if (hits.length() > 0) {
                for (int index = 0; index < hits.length(); index++) {
                    Document doc = hits.doc(index);
                    Term term = new Term(dateField, doc.get(dateField));
                    terms.add(term);
                }
            }
            liveIndex.deleteDocuments(terms.toArray(new Term[terms.size()]));
        } catch (IOException e) {
            log.error("Error deleting old programs from live index", e);
        } finally {
            if (searcher != null) {
                try {
                    searcher.close();
                } catch (IOException e) {
                    log.error("Error closing searcher when deleting old programs from live index", e);

                }
            }
        }

        Calendar cal = Calendar.getInstance();
        int minute = cal.get(Calendar.MINUTE);
        if (minute < 15) {
            cal.set(Calendar.MINUTE, 15);
        } else if (minute >= 45) {
            cal.set(Calendar.MINUTE, 15);
            cal.add(Calendar.HOUR, 1);
        } else {
            cal.set(Calendar.MINUTE, 45);
        }
        log.info("queued up that delete, now we're waiting until " + cal.getTime());
        Utils.sleepUntil(cal.getTimeInMillis());
    }
}

From source file:com.appeligo.lucene.LuceneIndexer.java

License:Apache License

private Document getProgramDocument(String programId) {
    IndexSearcher searcher = null;//from  www  .ja v a 2  s .  c o m
    Document programDoc = null;
    try {
        searcher = new IndexSearcher(indexLocation);
        TermQuery termQuery = new TermQuery(new Term("programID", programId));
        Hits hits = searcher.search(termQuery);
        if (hits.length() > 0) {
            programDoc = hits.doc(0);
        }
    } catch (IOException e) {
        log.error(e);
    } finally {
        try {
            searcher.close();
        } catch (IOException e) {
            log.error(e);
        }
    }
    return programDoc;
}

From source file:com.appeligo.search.actions.SearchResults.java

License:Apache License

public IndexSearcher newIndexSearcher() throws IOException {
    log.debug("getting index " + indexLocation);
    return new IndexSearcher(indexLocation);
}

From source file:com.appeligo.search.actions.SearchResults.java

License:Apache License

public List<SearchResult> getSearchResults(int startIndex) {

    initializeStatics();/*from   www  . j a v  a2s .  c o  m*/

    hasMoreResults = false;
    try {
        IndexSearcher searcher = null;

        try {
            searcher = newIndexSearcher();
            IndexReader reader = searcher.getIndexReader();

            Query luceneQuery = generateLuceneQuery(searcher);
            luceneQuery = luceneQuery.rewrite(reader);
            Hits hits = searcher.search(luceneQuery);

            usingSuggestedQuery = false;
            suggestedQuery = null;
            if ((didYouMeanParser != null)
                    && ((hits.length() < minimumHits) || (calcScore(searcher, getQuery()) < minimumScore))) {
                if (log.isDebugEnabled()) {
                    log.debug("Need to suggest because either num hits " + hits.length() + " < " + minimumHits
                            + "\n or top hit score " + (hits.length() > 0 ? hits.score(0) : "[NO HITS]") + " < "
                            + minimumScore);
                }
                IndexSearcher compositeSearcher = new IndexSearcher(compositeIndexLocation);
                try {
                    log.debug("calling suggest() with query=" + getQuery() + " and composite index from "
                            + compositeIndexLocation);
                    //Query didYouMean = didYouMeanParser.suggest(getQuery(), compositeSearcher.getIndexReader());
                    Query suggestedQueries[] = didYouMeanParser.getSuggestions(getQuery(),
                            compositeSearcher.getIndexReader());
                    TreeSet<Suggestion> suggestions = new TreeSet<Suggestion>();

                    if (suggestedQueries != null) {
                        for (int i = 0; i < suggestedQueries.length; i++) {
                            log.debug("trying suggested query: " + suggestedQueries[i].toString(defaultField));
                            String suggestedQueryString = suggestedQueries[i].toString(defaultField);
                            String constrainedQueryString = suggestedQueryString;
                            if (constrainedQueryString.indexOf('"') < 0
                                    && constrainedQueryString.indexOf('\'') < 0) {
                                constrainedQueryString = "\"" + constrainedQueryString + "\"~5"; // proximity/distance query (within 5 words of each other)
                            }
                            Query suggestedLuceneQuery = generateLuceneQuery(constrainedQueryString, searcher);
                            suggestedLuceneQuery = suggestedLuceneQuery.rewrite(reader);
                            Hits suggestedHits = searcher.search(suggestedLuceneQuery);

                            float score = calcScore(suggestedQueryString, suggestedHits);

                            log.debug("=========================================");
                            log.debug("SCORE = " + score);
                            log.debug("=========================================");

                            suggestions.add(
                                    new Suggestion(suggestedQueryString, suggestedLuceneQuery, suggestedHits,
                                            score, ((i == 0) ? didYouMeanParser.includesOriginal() : false)));
                            log.debug("hits=" + suggestedHits.length() + ", score=" + score);
                        }
                    }

                    Suggestion best = null;
                    if (suggestions.size() > 0) {
                        best = suggestions.last();
                    }

                    if (best != null && !best.isOriginal()) {
                        suggestedQuery = best.getQueryString();
                        if (suggestedQuery != null && suggestedQuery.indexOf('+') >= 0
                                && getQuery().indexOf('+') < 0) {
                            suggestedQuery = suggestedQuery.replaceAll("\\+", "");
                        }
                        if (hits.length() == 0) {
                            if (best.getHits().length() > 0) {
                                // Requery probably required because we added proximity before
                                String suggestedQueryString = best.getQueryString();
                                luceneQuery = generateLuceneQuery(suggestedQueryString, searcher);
                                luceneQuery = luceneQuery.rewrite(reader);
                                hits = searcher.search(luceneQuery);
                                //hits = best.getHits();
                                //luceneQuery = best.getLuceneQuery();
                                usingSuggestedQuery = true;
                            }
                        }
                        log.debug("DidYouMeanParser suggested " + suggestedQuery);
                    } else {
                        if (best != null && best.isOriginal()) {
                            log.debug("The suggestion was the original query after all");
                        }
                        log.debug("DidYouMeanParser did not suggest anything");
                    }
                } finally {
                    compositeSearcher.close();
                }
            }
            /*
            if (hits.length() == 0 && suggestedQuery != null) {
            // If we didn't find anything at all, go ahead and show them what the suggested query
            // will give them
            Query suggestedLuceneQuery = generateLuceneQuery(suggestedQuery, searcher);
            suggestedLuceneQuery = suggestedLuceneQuery.rewrite(reader);
               Hits suggestedHits = searcher.search(suggestedLuceneQuery);
               if (suggestedHits.length() > 0) {
             hits = suggestedHits;
             luceneQuery = suggestedLuceneQuery;
             usingSuggestedQuery = true;
               }
            }
               */
            totalHits = hits.length();
            //Get the genere matches:
            try {
                BitSetFacetHitCounter facetHitCounter = new BitSetFacetHitCounter();
                facetHitCounter.setSearcher(searcher);
                String baseQueryString = (isUsingSuggestedQuery() ? suggestedQuery : query);
                String quotedQueryString = baseQueryString;
                if (quotedQueryString.indexOf('"') == -1 && quotedQueryString.indexOf(' ') > -1) {
                    quotedQueryString = "\"" + quotedQueryString + "\"";
                }
                facetHitCounter.setBaseQuery(luceneQuery, baseQueryString);

                List<HitCount> subQueries = new ArrayList<HitCount>();
                for (Map.Entry<String, Query> entry : genreQueries.entrySet()) {
                    subQueries.add(
                            new HitCount(entry.getKey(), entry.getValue(), entry.getValue().toString(), 0));
                }
                facetHitCounter.setSubQueries(subQueries);
                genreCounts = facetHitCounter.getFacetHitCounts(true);

                whatMatchedCounts = new ArrayList<HitCount>();
                whatMatchedCounts
                        .add(new HitCount("Title", getFieldQuery(baseQueryString, "programTitle", searcher),
                                "programTitle:" + quotedQueryString, 0));
                whatMatchedCounts.add(
                        new HitCount("Episode Title", getFieldQuery(baseQueryString, "episodeTitle", searcher),
                                "episodeTitle:" + quotedQueryString, 0));
                whatMatchedCounts.add(
                        new HitCount("Description", getFieldQuery(baseQueryString, "description", searcher),
                                "description:" + quotedQueryString, 0));
                whatMatchedCounts.add(new HitCount("Content", getFieldQuery(baseQueryString, "text", searcher),
                        "text:" + quotedQueryString, 0));
                whatMatchedCounts
                        .add(new HitCount("Credits", getFieldQuery(baseQueryString, "credits", searcher),
                                "credits:" + quotedQueryString, 0));
                facetHitCounter.setSubQueries(whatMatchedCounts);
                whatMatchedCounts = facetHitCounter.getFacetHitCounts(true);

                //Program Count  -- Not sure if there is a better way to do this.
                HashSet<String> programTitles = new HashSet<String>();
                programCounts = new ArrayList<HitCount>();
                for (int i = 0; i < hits.length() && programCounts.size() < 5; i++) {
                    String title = hits.doc(i).get("programTitle");
                    if (!programTitles.contains(title)) {
                        String queryTitle = title;
                        queryTitle = QueryParser.escape(title);
                        if (queryTitle.indexOf('"') > -1) {
                            queryTitle.replace("\"", "\\\"");
                        }
                        if (queryTitle.indexOf(' ') > -1) {
                            queryTitle = "\"" + queryTitle + "\"";
                        }

                        programCounts
                                .add(new HitCount(title, getFieldQuery(queryTitle, "programTitle", searcher),
                                        "programTitle:" + queryTitle, 0));
                        programTitles.add(title);
                    }
                }
                facetHitCounter.setSubQueries(programCounts);
                programCounts = facetHitCounter.getFacetHitCounts(false);
            } catch (Exception e) {
                e.printStackTrace();
            }

            results = new ArrayList<SearchResult>();
            programToSearchResult.clear();
            Query userQuery = getContentQuery(query, searcher);
            userQuery.rewrite(reader);
            Highlighter highlighter = new Highlighter(new TermFormatter(), new QueryScorer(userQuery, "text"));

            log.debug("#hits=" + hits.length());

            EPGProvider epgProvider = DefaultEpg.getInstance();

            boolean missingWebPaths = false; // We added this to the index midstream, so some do and some don't.
            // Next index rebuild, and they'll all have it.
            for (int i = 0; i < pageSize && i + startIndex < hits.length(); i++) {
                if (hits.doc(i + startIndex).get("webPath") == null) {
                    missingWebPaths = true;
                    break;
                }
            }
            Program[] programs = null;
            if (missingWebPaths) {
                List<String> programIds = new ArrayList<String>(pageSize);
                for (int i = 0; i < pageSize && i + startIndex < hits.length(); i++) {
                    programIds.add(hits.doc(i + startIndex).get("programID"));
                }
                programs = DefaultEpg.getInstance().getProgramList(programIds);
            }
            for (int i = 0; i < pageSize && i + startIndex < hits.length(); i++) {
                addDocument(hits.doc(i + startIndex), hits.score(i + startIndex), epgProvider, highlighter,
                        analyzer, null, null, (programs == null ? null : programs[i]));
            }
            if (results.size() + startIndex < hits.length()) {
                hasMoreResults = true;
            }
        } finally {
            if (searcher != null) {
                searcher.close();
            }
        }
    } catch (IOException e) {
        log.error("Error searching index", e);
    } catch (ParseException e) {
        log.error("Error searching index", e);
    }
    return results;
}