Search in sources :

Example 1 with MatchGroup

use of annis.service.objects.MatchGroup in project ANNIS by korpling.

the class AnnisRunner method doFind.

public void doFind(String annisQuery) {
    List<Match> matches = queryDao.find(analyzeQuery(annisQuery, "find"));
    MatchGroup group = new MatchGroup(matches);
    out.println(group.toString());
}
Also used : MatchGroup(annis.service.objects.MatchGroup) Match(annis.service.objects.Match) AnnotatedMatch(annis.dao.objects.AnnotatedMatch)

Example 2 with MatchGroup

use of annis.service.objects.MatchGroup in project ANNIS by korpling.

the class GeneralTextExporter method convertText.

@Override
public Exception convertText(String queryAnnisQL, int contextLeft, int contextRight, Set<String> corpora, List<String> keys, String argsAsString, boolean alignmc, WebResource annisResource, Writer out, EventBus eventBus, Map<String, CorpusConfig> corpusConfigs) {
    try {
        if (keys == null || keys.isEmpty()) {
            // auto set
            keys = new LinkedList<>();
            keys.add("tok");
            List<AnnisAttribute> attributes = new LinkedList<>();
            for (String corpus : corpora) {
                attributes.addAll(annisResource.path("corpora").path(urlPathEscape.escape(corpus)).path("annotations").queryParam("fetchvalues", "false").queryParam("onlymostfrequentvalues", "false").get(new AnnisAttributeListType()));
            }
            for (AnnisAttribute a : attributes) {
                if (a.getName() != null) {
                    String[] namespaceAndName = a.getName().split(":", 2);
                    if (namespaceAndName.length > 1) {
                        keys.add(namespaceAndName[1]);
                    } else {
                        keys.add(namespaceAndName[0]);
                    }
                }
            }
        }
        Map<String, String> args = new HashMap<>();
        for (String s : argsAsString.split("&|;")) {
            String[] splitted = s.split("=", 2);
            String key = splitted[0];
            String val = "";
            if (splitted.length > 1) {
                val = splitted[1];
            }
            args.put(key, val);
        }
        int stepSize = 10;
        // 1. Get all the matches as Salt ID
        InputStream matchStream = annisResource.path("search/find/").queryParam("q", Helper.encodeJersey(queryAnnisQL)).queryParam("corpora", StringUtils.join(corpora, ",")).accept(MediaType.TEXT_PLAIN_TYPE).get(InputStream.class);
        try (BufferedReader inReader = new BufferedReader(new InputStreamReader(matchStream, "UTF-8"))) {
            WebResource subgraphRes = annisResource.path("search/subgraph");
            MatchGroup currentMatches = new MatchGroup();
            String currentLine;
            int offset = 0;
            // 2. iterate over all matches and get the sub-graph for a group of matches
            while (!Thread.currentThread().isInterrupted() && (currentLine = inReader.readLine()) != null) {
                Match match = Match.parseFromString(currentLine);
                currentMatches.getMatches().add(match);
                if (currentMatches.getMatches().size() >= stepSize) {
                    WebResource res = subgraphRes.queryParam("left", "" + contextLeft).queryParam("right", "" + contextRight);
                    if (args.containsKey("segmentation")) {
                        res = res.queryParam("segmentation", args.get("segmentation"));
                    }
                    SubgraphFilter filter = getSubgraphFilter();
                    if (filter != null) {
                        res = res.queryParam("filter", filter.name());
                    }
                    Stopwatch stopwatch = Stopwatch.createUnstarted();
                    stopwatch.start();
                    SaltProject p = res.post(SaltProject.class, currentMatches);
                    stopwatch.stop();
                    // export was fast enough
                    if (stopwatch.elapsed(TimeUnit.MILLISECONDS) < 500 && stepSize < 50) {
                        stepSize += 10;
                    }
                    convertText(LegacyGraphConverter.convertToResultSet(p), keys, args, out, offset - currentMatches.getMatches().size());
                    currentMatches.getMatches().clear();
                    if (eventBus != null) {
                        eventBus.post(offset + 1);
                    }
                }
                offset++;
            }
            if (Thread.interrupted()) {
                return new InterruptedException("Exporter job was interrupted");
            }
            // query the left over matches
            if (!currentMatches.getMatches().isEmpty()) {
                WebResource res = subgraphRes.queryParam("left", "" + contextLeft).queryParam("right", "" + contextRight);
                if (args.containsKey("segmentation")) {
                    res = res.queryParam("segmentation", args.get("segmentation"));
                }
                SubgraphFilter filter = getSubgraphFilter();
                if (filter != null) {
                    res = res.queryParam("filter", filter.name());
                }
                SaltProject p = res.post(SaltProject.class, currentMatches);
                convertText(LegacyGraphConverter.convertToResultSet(p), keys, args, out, offset - currentMatches.getMatches().size() - 1);
            }
            offset = 0;
        }
        out.append("\n");
        out.append("\n");
        out.append("finished");
        return null;
    } catch (AnnisQLSemanticsException | AnnisQLSyntaxException | AnnisCorpusAccessException | UniformInterfaceException | IOException ex) {
        return ex;
    }
}
Also used : HashMap(java.util.HashMap) AnnisAttribute(annis.service.objects.AnnisAttribute) Stopwatch(com.google.common.base.Stopwatch) WebResource(com.sun.jersey.api.client.WebResource) Match(annis.service.objects.Match) AnnisQLSyntaxException(annis.exceptions.AnnisQLSyntaxException) InputStreamReader(java.io.InputStreamReader) InputStream(java.io.InputStream) AnnisQLSemanticsException(annis.exceptions.AnnisQLSemanticsException) SaltProject(org.corpus_tools.salt.common.SaltProject) IOException(java.io.IOException) SubgraphFilter(annis.service.objects.SubgraphFilter) LinkedList(java.util.LinkedList) UniformInterfaceException(com.sun.jersey.api.client.UniformInterfaceException) AnnisCorpusAccessException(annis.exceptions.AnnisCorpusAccessException) MatchGroup(annis.service.objects.MatchGroup) BufferedReader(java.io.BufferedReader)

Example 3 with MatchGroup

use of annis.service.objects.MatchGroup in project ANNIS by korpling.

the class SaltBasedExporter method convertText.

@Override
public Exception convertText(String queryAnnisQL, int contextLeft, int contextRight, Set<String> corpora, List<String> keys, String argsAsString, boolean alignmc, WebResource annisResource, Writer out, EventBus eventBus, Map<String, CorpusConfig> corpusConfigs) {
    CacheManager cacheManager = CacheManager.create();
    try {
        Cache cache = cacheManager.getCache("saltProjectsCache");
        if (keys == null || keys.isEmpty()) {
            // auto set
            keys = new LinkedList<>();
            keys.add("tok");
            List<AnnisAttribute> attributes = new LinkedList<>();
            for (String corpus : corpora) {
                attributes.addAll(annisResource.path("corpora").path(urlPathEscape.escape(corpus)).path("annotations").queryParam("fetchvalues", "false").queryParam("onlymostfrequentvalues", "false").get(new AnnisAttributeListType()));
            }
            for (AnnisAttribute a : attributes) {
                if (a.getName() != null) {
                    String[] namespaceAndName = a.getName().split(":", 2);
                    if (namespaceAndName.length > 1) {
                        keys.add(namespaceAndName[1]);
                    } else {
                        keys.add(namespaceAndName[0]);
                    }
                }
            }
        }
        Map<String, String> args = new HashMap<>();
        for (String s : argsAsString.split("&|;")) {
            String[] splitted = s.split("=", 2);
            String key = splitted[0];
            String val = "";
            if (splitted.length > 1) {
                val = splitted[1];
            }
            args.put(key, val);
        }
        int stepSize = 10;
        int pCounter = 1;
        Map<Integer, Integer> offsets = new HashMap<Integer, Integer>();
        // 1. Get all the matches as Salt ID
        InputStream matchStream = annisResource.path("search/find/").queryParam("q", Helper.encodeJersey(queryAnnisQL)).queryParam("corpora", StringUtils.join(corpora, ",")).accept(MediaType.TEXT_PLAIN_TYPE).get(InputStream.class);
        // get node count for the query
        WebResource resource = Helper.getAnnisWebResource();
        List<QueryNode> nodes = resource.path("query/parse/nodes").queryParam("q", Helper.encodeJersey(queryAnnisQL)).get(new GenericType<List<QueryNode>>() {
        });
        Integer nodeCount = nodes.size();
        try (BufferedReader inReader = new BufferedReader(new InputStreamReader(matchStream, "UTF-8"))) {
            WebResource subgraphRes = annisResource.path("search/subgraph");
            MatchGroup currentMatches = new MatchGroup();
            String currentLine;
            int offset = 1;
            // 2. iterate over all matches and get the sub-graph for a group of matches
            while (!Thread.currentThread().isInterrupted() && (currentLine = inReader.readLine()) != null) {
                Match match = Match.parseFromString(currentLine);
                currentMatches.getMatches().add(match);
                if (currentMatches.getMatches().size() >= stepSize) {
                    WebResource res = subgraphRes.queryParam("left", "" + contextLeft).queryParam("right", "" + contextRight);
                    if (args.containsKey("segmentation")) {
                        res = res.queryParam("segmentation", args.get("segmentation"));
                    }
                    SubgraphFilter filter = getSubgraphFilter();
                    if (filter != null) {
                        res = res.queryParam("filter", filter.name());
                    }
                    Stopwatch stopwatch = Stopwatch.createStarted();
                    SaltProject p = res.post(SaltProject.class, currentMatches);
                    stopwatch.stop();
                    // export was fast enough
                    if (stopwatch.elapsed(TimeUnit.MILLISECONDS) < 500 && stepSize < 50) {
                        stepSize += 10;
                    }
                    convertSaltProject(p, keys, args, alignmc, offset - currentMatches.getMatches().size(), corpusConfigs, out, nodeCount);
                    offsets.put(pCounter, offset - currentMatches.getMatches().size());
                    cache.put(new Element(pCounter++, p));
                    currentMatches.getMatches().clear();
                    if (eventBus != null) {
                        eventBus.post(offset + 1);
                    }
                }
                offset++;
            }
            if (Thread.interrupted()) {
                return new InterruptedException("Exporter job was interrupted");
            }
            // query the left over matches
            if (!currentMatches.getMatches().isEmpty()) {
                WebResource res = subgraphRes.queryParam("left", "" + contextLeft).queryParam("right", "" + contextRight);
                if (args.containsKey("segmentation")) {
                    res = res.queryParam("segmentation", args.get("segmentation"));
                }
                SubgraphFilter filter = getSubgraphFilter();
                if (filter != null) {
                    res = res.queryParam("filter", filter.name());
                }
                SaltProject p = res.post(SaltProject.class, currentMatches);
                convertSaltProject(p, keys, args, alignmc, offset - currentMatches.getMatches().size() - 1, corpusConfigs, out, nodeCount);
                offsets.put(pCounter, offset - currentMatches.getMatches().size() - 1);
                cache.put(new Element(pCounter++, p));
            }
            offset = 1;
        }
        // build the list of ordered match numbers (ordering by occurrence in text)
        getOrderedMatchNumbers();
        @SuppressWarnings("unchecked") List<Integer> cacheKeys = cache.getKeys();
        List<Integer> listOfKeys = new ArrayList<Integer>();
        for (Integer key : cacheKeys) {
            listOfKeys.add(key);
        }
        Collections.sort(listOfKeys);
        for (Integer key : listOfKeys) {
            SaltProject p = (SaltProject) cache.get(key).getObjectValue();
            convertSaltProject(p, keys, args, alignmc, offsets.get(key), corpusConfigs, out, null);
        }
        out.append(System.lineSeparator());
        return null;
    } catch (AnnisQLSemanticsException | AnnisQLSyntaxException | AnnisCorpusAccessException | UniformInterfaceException | IOException | CacheException | IllegalStateException | ClassCastException ex) {
        return ex;
    } finally {
        cacheManager.removalAll();
        cacheManager.shutdown();
    }
}
Also used : HashMap(java.util.HashMap) CacheException(net.sf.ehcache.CacheException) AnnisAttribute(annis.service.objects.AnnisAttribute) Element(net.sf.ehcache.Element) Stopwatch(com.google.common.base.Stopwatch) ArrayList(java.util.ArrayList) WebResource(com.sun.jersey.api.client.WebResource) Match(annis.service.objects.Match) AnnisQLSyntaxException(annis.exceptions.AnnisQLSyntaxException) CacheManager(net.sf.ehcache.CacheManager) ArrayList(java.util.ArrayList) LinkedList(java.util.LinkedList) List(java.util.List) InputStreamReader(java.io.InputStreamReader) InputStream(java.io.InputStream) AnnisQLSemanticsException(annis.exceptions.AnnisQLSemanticsException) SaltProject(org.corpus_tools.salt.common.SaltProject) IOException(java.io.IOException) SubgraphFilter(annis.service.objects.SubgraphFilter) LinkedList(java.util.LinkedList) UniformInterfaceException(com.sun.jersey.api.client.UniformInterfaceException) AnnisCorpusAccessException(annis.exceptions.AnnisCorpusAccessException) QueryNode(annis.model.QueryNode) MatchGroup(annis.service.objects.MatchGroup) BufferedReader(java.io.BufferedReader) Cache(net.sf.ehcache.Cache)

Example 4 with MatchGroup

use of annis.service.objects.MatchGroup in project ANNIS by korpling.

the class SingleResultFetchJob method call.

@Override
public SaltProject call() throws Exception {
    WebResource subgraphRes = Helper.getAnnisWebResource().path("query/search/subgraph");
    if (Thread.interrupted()) {
        return null;
    }
    List<Match> subList = new LinkedList<>();
    subList.add(match);
    SaltProject p = executeQuery(subgraphRes, new MatchGroup(subList), query.getLeftContext(), query.getRightContext(), query.getSegmentation(), SubgraphFilter.all);
    return p;
}
Also used : MatchGroup(annis.service.objects.MatchGroup) WebResource(com.sun.jersey.api.client.WebResource) SaltProject(org.corpus_tools.salt.common.SaltProject) LinkedList(java.util.LinkedList) Match(annis.service.objects.Match)

Example 5 with MatchGroup

use of annis.service.objects.MatchGroup in project ANNIS by korpling.

the class ResultFetchJob method run.

@Override
public void run() {
    WebResource subgraphRes = Helper.getAnnisWebResource().path("query/search/subgraph");
    // holds the ids of the matches.
    MatchGroup result;
    try {
        if (Thread.interrupted()) {
            return;
        }
        // set the the progress bar, for given the user some information about the loading process
        ui.accessSynchronously(new Runnable() {

            @Override
            public void run() {
                resultPanel.showMatchSearchInProgress(query);
            }
        });
        // get the matches
        result = futureMatches.get();
        // get the subgraph for each match, when the result is not empty
        if (result.getMatches().isEmpty()) {
            // check if thread was interrupted
            if (Thread.interrupted()) {
                return;
            }
            // nothing found, so inform the user about this.
            ui.access(new Runnable() {

                @Override
                public void run() {
                    resultPanel.showNoResult();
                }
            });
        } else {
            if (Thread.interrupted()) {
                return;
            }
            // since annis found something, inform the user that subgraphs are created
            ui.access(new Runnable() {

                @Override
                public void run() {
                    resultPanel.showSubgraphSearchInProgress(query, 0.0f);
                }
            });
            // prepare fetching subgraphs
            final BlockingQueue<SaltProject> queue = new ArrayBlockingQueue<>(result.getMatches().size());
            int current = 0;
            final ArrayList<Match> matchList = new ArrayList<>(result.getMatches());
            for (Match m : matchList) {
                if (Thread.interrupted()) {
                    return;
                }
                List<Match> subList = new LinkedList<>();
                subList.add(m);
                final SaltProject p = executeQuery(subgraphRes, new MatchGroup(subList), query.getLeftContext(), query.getRightContext(), query.getSegmentation(), SubgraphFilter.all);
                queue.put(p);
                log.debug("added match {} to queue", current + 1);
                if (current == 0) {
                    ui.access(new Runnable() {

                        @Override
                        public void run() {
                            resultPanel.setQueryResultQueue(queue, query, matchList);
                        }
                    });
                }
                if (Thread.interrupted()) {
                    return;
                }
                current++;
            }
        }
    // end if no results
    } catch (InterruptedException ex) {
    // just return
    } catch (final ExecutionException root) {
        ui.accessSynchronously(new Runnable() {

            @Override
            public void run() {
                if (resultPanel != null && resultPanel.getPaging() != null) {
                    PagingComponent paging = resultPanel.getPaging();
                    Throwable cause = root.getCause();
                    if (cause instanceof UniformInterfaceException) {
                        UniformInterfaceException ex = (UniformInterfaceException) cause;
                        if (ex.getResponse().getStatus() == 400) {
                            List<AqlParseError> errors = ex.getResponse().getEntity(new GenericType<List<AqlParseError>>() {
                            });
                            String errMsg = Joiner.on(" | ").join(errors);
                            paging.setInfo("parsing error: " + errMsg);
                        } else if (ex.getResponse().getStatus() == 504) {
                            paging.setInfo("Timeout: query execution took too long");
                        } else if (ex.getResponse().getStatus() == 403) {
                            paging.setInfo("Not authorized to query this corpus.");
                        } else {
                            paging.setInfo("unknown error: " + ex);
                        }
                    } else {
                        log.error("Unexcepted ExecutionException cause", root);
                    }
                    resultPanel.showFinishedSubgraphSearch();
                }
            }
        });
    }
// end catch
}
Also used : ArrayList(java.util.ArrayList) AqlParseError(annis.model.AqlParseError) AsyncWebResource(com.sun.jersey.api.client.AsyncWebResource) WebResource(com.sun.jersey.api.client.WebResource) SaltProject(org.corpus_tools.salt.common.SaltProject) LinkedList(java.util.LinkedList) Match(annis.service.objects.Match) PagingComponent(annis.gui.paging.PagingComponent) UniformInterfaceException(com.sun.jersey.api.client.UniformInterfaceException) ArrayBlockingQueue(java.util.concurrent.ArrayBlockingQueue) MatchGroup(annis.service.objects.MatchGroup) ArrayList(java.util.ArrayList) LinkedList(java.util.LinkedList) List(java.util.List) ExecutionException(java.util.concurrent.ExecutionException)

Aggregations

MatchGroup (annis.service.objects.MatchGroup)12 Match (annis.service.objects.Match)9 SaltProject (org.corpus_tools.salt.common.SaltProject)7 LinkedList (java.util.LinkedList)5 WebResource (com.sun.jersey.api.client.WebResource)4 AnnotateQueryData (annis.sqlgen.extensions.AnnotateQueryData)3 UniformInterfaceException (com.sun.jersey.api.client.UniformInterfaceException)3 InputStreamReader (java.io.InputStreamReader)3 ArrayList (java.util.ArrayList)3 AnnisCorpusAccessException (annis.exceptions.AnnisCorpusAccessException)2 AnnisQLSemanticsException (annis.exceptions.AnnisQLSemanticsException)2 AnnisQLSyntaxException (annis.exceptions.AnnisQLSyntaxException)2 QueryData (annis.ql.parser.QueryData)2 AnnisAttribute (annis.service.objects.AnnisAttribute)2 SubgraphFilter (annis.service.objects.SubgraphFilter)2 LimitOffsetQueryData (annis.sqlgen.extensions.LimitOffsetQueryData)2 Stopwatch (com.google.common.base.Stopwatch)2 BufferedReader (java.io.BufferedReader)2 IOException (java.io.IOException)2 InputStream (java.io.InputStream)2