Search in sources :

Example 1 with AnnisAttribute

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

the class AnnisRunner method doAnnotations.

public void doAnnotations(String doListValues) {
    boolean listValues = "values".equals(doListValues);
    List<AnnisAttribute> annotations = queryDao.listAnnotations(getCorpusList(), listValues, true);
    try {
        ObjectMapper om = new ObjectMapper();
        AnnotationIntrospector ai = new JaxbAnnotationIntrospector();
        DeserializationConfig config = om.getDeserializationConfig().withAnnotationIntrospector(ai);
        om.setDeserializationConfig(config);
        om.configure(SerializationConfig.Feature.INDENT_OUTPUT, true);
        System.out.println(om.writeValueAsString(annotations));
    } catch (IOException ex) {
        log.error("problems with writing result", ex);
    }
}
Also used : AnnisAttribute(annis.service.objects.AnnisAttribute) JaxbAnnotationIntrospector(org.codehaus.jackson.xc.JaxbAnnotationIntrospector) AnnotationIntrospector(org.codehaus.jackson.map.AnnotationIntrospector) DeserializationConfig(org.codehaus.jackson.map.DeserializationConfig) IOException(java.io.IOException) ObjectMapper(org.codehaus.jackson.map.ObjectMapper) JaxbAnnotationIntrospector(org.codehaus.jackson.xc.JaxbAnnotationIntrospector)

Example 2 with AnnisAttribute

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

the class AnnotationExistenceValidator method transform.

@Override
public QueryData transform(QueryData data) {
    List<Long> corpusList = data.getCorpusList();
    if (queryDao != null && (corpusList != null) && !corpusList.isEmpty()) {
        // get first corpus name
        // List<AnnisCorpus> mycorpora = queryDao.listCorpora();
        // String firstcorpusname =  mycorpora.get(0).getName();
        Set<String> result = new TreeSet<>();
        /*get a list of all annotations in a similar way that TigerQueryBuilder gets it through 
       QueryServiceImpl in queryDao.listAnnotations()*/
        List<AnnisAttribute> atts = queryDao.listAnnotations(corpusList, false, true);
        // among them, get only node annotations
        for (AnnisAttribute a : atts) {
            if (a.getType() == AnnisAttribute.Type.node) {
                List<String> splitted = Splitter.on(":").limit(2).omitEmptyStrings().trimResults().splitToList(a.getName());
                result.add(splitted.get(splitted.size() - 1));
            // result is a set of strings of available annotations
            }
        }
        List<AqlParseError> errors = new LinkedList<>();
        for (List<QueryNode> alternative : data.getAlternatives()) {
            for (QueryNode n : alternative) {
                Set<QueryAnnotation> m = n.getNodeAnnotations();
                // always get the first one
                if (!m.isEmpty()) {
                    // name is the node name string, ready to check if name is in the list of
                    // available names
                    String name = m.iterator().next().getName();
                    if (!result.contains(name)) {
                        errors.add(new AqlParseError(n, "\"" + name + "\"" + " is not a valid annotation name in selected corpora "));
                    }
                }
            }
        }
        if (!errors.isEmpty()) {
            throw new AnnisQLSemanticsException("Invalid annotation names detected.", errors);
        }
    }
    return data;
}
Also used : QueryAnnotation(annis.model.QueryAnnotation) AnnisAttribute(annis.service.objects.AnnisAttribute) AqlParseError(annis.model.AqlParseError) AnnisQLSemanticsException(annis.exceptions.AnnisQLSemanticsException) LinkedList(java.util.LinkedList) TreeSet(java.util.TreeSet) QueryNode(annis.model.QueryNode)

Example 3 with AnnisAttribute

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

the class ListAnnotationsSqlHelper method extractData.

@Override
public ArrayList<AnnisAttribute> extractData(ResultSet resultSet) throws SQLException, DataAccessException {
    Map<String, AnnisAttribute> attributesByName = new HashMap<>();
    while (resultSet.next()) {
        String namespace = resultSet.getString("namespace");
        String name = resultSet.getString("name");
        String qName = AnnisNode.qName(namespace, name);
        String edgeNamespace = resultSet.getString("edge_namespace");
        String edgeName = resultSet.getString("edge_name");
        String qEdgeName = AnnisNode.qName(edgeNamespace, edgeName);
        String key = qName;
        if (qEdgeName != null) {
            key += "_" + qEdgeName;
        }
        if (!attributesByName.containsKey(key)) {
            attributesByName.put(key, new AnnisAttribute());
        }
        AnnisAttribute attribute = attributesByName.get(key);
        attribute.setName(qName);
        attribute.setEdgeName(qEdgeName);
        AnnisAttribute.Type t = AnnisAttribute.Type.unknown;
        try {
            t = AnnisAttribute.Type.valueOf(resultSet.getString("type"));
        } catch (Exception ex) {
            log.warn("annotation type is unknown {}", ex);
        }
        attribute.setType(t);
        AnnisAttribute.SubType st = AnnisAttribute.SubType.unknown;
        String subTypeValue = resultSet.getString("subtype");
        if (subTypeValue != null) {
            try {
                st = AnnisAttribute.SubType.valueOf(subTypeValue);
            } catch (Exception ex) {
                log.warn("annotation sub type is unknown {}", ex);
            }
        }
        attribute.setSubtype(st);
        String value = resultSet.getString("value");
        if (value != null) {
            attribute.addValue(value);
        }
    }
    return new ArrayList<>(attributesByName.values());
}
Also used : HashMap(java.util.HashMap) AnnisAttribute(annis.service.objects.AnnisAttribute) ArrayList(java.util.ArrayList) DataAccessException(org.springframework.dao.DataAccessException) SQLException(java.sql.SQLException)

Example 4 with AnnisAttribute

use of annis.service.objects.AnnisAttribute 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 5 with AnnisAttribute

use of annis.service.objects.AnnisAttribute 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)

Aggregations

AnnisAttribute (annis.service.objects.AnnisAttribute)15 LinkedList (java.util.LinkedList)10 UniformInterfaceException (com.sun.jersey.api.client.UniformInterfaceException)9 WebResource (com.sun.jersey.api.client.WebResource)9 ClientHandlerException (com.sun.jersey.api.client.ClientHandlerException)7 TreeSet (java.util.TreeSet)7 GenericType (com.sun.jersey.api.client.GenericType)5 AnnisQLSemanticsException (annis.exceptions.AnnisQLSemanticsException)3 IOException (java.io.IOException)3 ArrayList (java.util.ArrayList)3 HashMap (java.util.HashMap)3 List (java.util.List)3 AnnisCorpusAccessException (annis.exceptions.AnnisCorpusAccessException)2 AnnisQLSyntaxException (annis.exceptions.AnnisQLSyntaxException)2 QueryNode (annis.model.QueryNode)2 Match (annis.service.objects.Match)2 MatchGroup (annis.service.objects.MatchGroup)2 SubgraphFilter (annis.service.objects.SubgraphFilter)2 Stopwatch (com.google.common.base.Stopwatch)2 BufferedReader (java.io.BufferedReader)2