Examples with Blob edu.stanford.muse.datacache.Blob used on opensource projects

Example 1 with Blob

use of edu.stanford.muse.datacache.Blob in project epadd by ePADD.

the class IndexUtils method getBlobsForAttachmentTypes.

/**
 * returns set of all blobs whose type match ANY of the given extensions.
 * ("none" is a valid type) and matches attachments that don't have an
 * extension.
 */
public static Set<Blob> getBlobsForAttachmentTypes(Collection<? extends Document> docs, String[] attachmentTypes) {
    Set<Blob> result = new LinkedHashSet<>();
    // convert to a set for fast lookup
    Set<String> attachmentTypesSet = new LinkedHashSet<>();
    Collections.addAll(attachmentTypesSet, attachmentTypes);
    for (Document d : docs) {
        if (!(d instanceof EmailDocument))
            continue;
        EmailDocument ed = (EmailDocument) d;
        if (ed.attachments == null)
            continue;
        for (Blob b : ed.attachments) {
            String ext = Util.getExtension(b.filename);
            if (ext == null)
                ext = "none";
            ext = ext.toLowerCase();
            if (attachmentTypesSet.contains(ext)) {
                result.add(b);
            }
        }
    }
    return result;
}

Example 2 with Blob

use of edu.stanford.muse.datacache.Blob in project epadd by ePADD.

the class Indexer method indexAttachments.

/**
 * returns whether indexAttachments succeeded
 */
private synchronized boolean indexAttachments(EmailDocument e, BlobStore blobStore, Set<Blob> processedBlobSet, IndexStats stats) throws IOException {
    boolean result = true;
    // bail out if no attachments
    if (e.attachments == null)
        return true;
    final String DELIMITER = "\n";
    for (Blob b : e.attachments) {
        if (processedBlobSet != null && processedBlobSet.contains(b))
            // skip if already processed (blob may be shared by multiple docs)
            continue;
        /*int id_int = iwriter_blob.numDocs();
			String id = Integer.toString(++id_int);*/
        int id_int = blobStore.index(b);
        String id = Integer.toString(id_int);
        if (processedBlobSet != null)
            processedBlobSet.add(b);
        attachmentDocIdToBlob.put(id, b);
        // not to be confused with edu.stanford.muse.index.Document
        org.apache.lucene.document.Document doc = new org.apache.lucene.document.Document();
        Pair<String, String> content = b.getContent(blobStore);
        if (content == null) {
            // failed to process blob
            result = false;
            log.warn("Failed to fetch content from: " + b.filename + " content type: " + b.contentType + " size: " + b.getSize());
            // but try to continue the process
            continue;
        }
        // imp: for id, should use Field.Index.NOT_ANALYZED field should be http://vuknikolic.wordpress.com/2011/01/03/lucenes-field-options-store-and-index-aka-rtfm/
        // note: id for attachments index is just sequential numbers, 1, 2, 3. etc.
        // it is not the full unique id (<folder>-<num>) that the emails index has.
        // NOTE: docid, emaildocid and languages fields can be stored without position (hence ft as FieldType)
        // because user can not do a phrase query on these fields.
        doc.add(new Field("docId", id, ft));
        // Field type ft instead of StoredFiled so as to be able to search over this field
        doc.add(new Field("emailDocId", e.getUniqueId(), ft));
        String documentText = content.first + DELIMITER + content.second;
        // we'll store all languages detected in the doc as a field in the index
        Set<String> languages = Languages.getAllLanguages(documentText);
        String lang_str = Util.join(languages, LANGUAGE_FIELD_DELIMITER);
        doc.add(new Field("languages", lang_str, ft));
        if (edu.stanford.muse.Config.OPENNLP_NER) {
            Set<String> names = setNameFieldsOpenNLP(documentText, doc);
            // just some connector for storing the field
            String s = Util.join(names, NAMES_FIELD_DELIMITER);
            doc.add(new Field("names", s, full_ft));
            if (stats != null)
                stats.nIndexedNames_blob += names.size();
        }
        // log.info ("blob metadata = " + content.first);
        // meta data does not contain the fileName
        doc.add(new Field("meta", content.first, full_ft));
        doc.add(new Field("fileName", b.filename, full_ft));
        doc.add(new Field("body", content.second, full_ft));
        iwriter_blob.addDocument(doc);
        // log.info("Indexed attachment #" + id + " : text = '" + documentText + "' names = '" + s + "'");
        if (stats != null) {
            stats.indexedTextLength_blob += documentText.length();
        }
    }
    return result;
}

Example 3 with Blob

use of edu.stanford.muse.datacache.Blob in project epadd by ePADD.

the class SearchResult method selectBlobs.

/**
 * this map is used only by attachments page right now, not advanced search.
 * TODO: make adv. search page also use it
 */
public static SearchResult selectBlobs(SearchResult inputSet) {
    Collection<Document> docs = inputSet.archive.getAllDocs();
    String neededFilesize = JSPHelper.getParam(inputSet.queryParams, "attachmentFilesize");
    String[] extensions = JSPHelper.getParams(inputSet.queryParams, "attachmentExtension").toArray(new String[0]);
    // should also have lower-case strings, no "." included
    Set<String> extensionsToMatch = new LinkedHashSet<>();
    if (!Util.nullOrEmpty(extensions)) {
        extensionsToMatch = new LinkedHashSet<>();
        for (String s : extensions) extensionsToMatch.add(s.trim().toLowerCase());
    }
    // or given extensions with extensions due to attachment type
    // this will have more semicolon separated extensions
    String[] types = JSPHelper.getParams(inputSet.queryParams, "attachmentType").toArray(new String[0]);
    if (!Util.nullOrEmpty(types)) {
        for (String t : types) {
            String exts = Config.attachmentTypeToExtensions.get(t);
            if (exts == null)
                exts = t;
            // continue;
            // Front end should uniformly pass attachment types as extensions like mp3;mov;ogg etc. Earlier it was passing vide, audio, doc etc.
            // In order to accommodate both cases we first check if there is ampping from the extension type to actual extensions using .get(t)
            // if no such mapping is present then we assume that the input extension types are of the form mp3;mov;ogg and work on that.
            String[] components = exts.split(";");
            Collections.addAll(extensionsToMatch, components);
        }
    }
    // a variable to select if the extensions needed contain others.
    boolean isOtherSelected = extensionsToMatch.contains("others");
    // get the options that were displayed for attachment types. This will be used to select attachment extensions if the option 'other'
    // was selected by the user in the drop down box of export.jsp.
    List<String> attachmentTypeOptions = Config.attachmentTypeToExtensions.values().stream().map(x -> Util.tokenize(x, ";")).flatMap(Collection::stream).collect(Collectors.toList());
    SearchResult outputSet = filterDocsByDate(inputSet);
    // Collection<EmailDocument> eDocs = (Collection) filterDocsByDate (params, new HashSet<>((Collection) docs));
    Map<Document, Pair<BodyHLInfo, AttachmentHLInfo>> outputDocs = new HashMap<>();
    for (Document k : outputSet.matchedDocs.keySet()) {
        EmailDocument ed = (EmailDocument) k;
        Set<Blob> matchedBlobs = new HashSet<>();
        for (Blob b : ed.attachments) {
            if (!Util.filesizeCheck(neededFilesize, b.getSize()))
                continue;
            if (!(Util.nullOrEmpty(extensionsToMatch))) {
                Pair<String, String> pair = Util.splitIntoFileBaseAndExtension(b.getName());
                String ext = pair.getSecond();
                if (ext == null)
                    continue;
                ext = ext.toLowerCase();
                // Proceed to add this attachment only if either
                // 1. other is selected and this extension is not present in the list attachmentOptionType, or
                // 2. this extension is present in the variable neededExtensions [Q. What if there is a file with extension .others?]
                boolean firstcondition = isOtherSelected && !attachmentTypeOptions.contains(ext);
                boolean secondcondition = extensionsToMatch.contains(ext);
                if (!firstcondition && !secondcondition)
                    continue;
            }
            // ok, we've survived all filters, add b
            matchedBlobs.add(b);
        }
        // of this document
        if (matchedBlobs.size() != 0) {
            BodyHLInfo bhlinfo = inputSet.matchedDocs.get(k).first;
            AttachmentHLInfo attachmentHLInfo = inputSet.matchedDocs.get(k).second;
            attachmentHLInfo.addMultipleInfo(matchedBlobs);
            outputDocs.put(k, new Pair(bhlinfo, attachmentHLInfo));
        }
    }
    // Collections.reverse (allAttachments); // reverse, so most recent attachment is first
    return new SearchResult(outputDocs, inputSet.archive, inputSet.queryParams, inputSet.commonHLInfo, inputSet.regexToHighlight);
}

Example 4 with Blob

use of edu.stanford.muse.datacache.Blob in project epadd by ePADD.

the class SearchResult method filterForAttachmentEntities.

/**
 ******************************ATTACHMENT SPECIFIC FILTERS************************************
 */
/**
 * returns only those docs with attachments matching params[attachmentEntity]
 * (this field is or-delimiter separated)
 * Todo: review usage of this and BlobStore.getKeywordsForBlob()
 */
private static SearchResult filterForAttachmentEntities(SearchResult inputSet) {
    String val = JSPHelper.getParam(inputSet.queryParams, "attachmentEntity");
    if (Util.nullOrEmpty(val))
        return inputSet;
    val = val.toLowerCase();
    Set<String> entities = Util.splitFieldForOr(val);
    BlobStore blobStore = inputSet.archive.blobStore;
    Map<Document, Pair<BodyHLInfo, AttachmentHLInfo>> outputDocs = new HashMap<>();
    inputSet.matchedDocs.keySet().stream().forEach((Document k) -> {
        EmailDocument ed = (EmailDocument) k;
        // Here.. check for all attachments of ed for match.
        Collection<Blob> blobs = ed.attachments;
        Set<Blob> matchedBlobs = new HashSet<>();
        for (Blob blob : blobs) {
            Collection<String> keywords = blobStore.getKeywordsForBlob(blob);
            if (keywords != null) {
                keywords.retainAll(entities);
                if (// it means this blob is of interest, add it to matchedBlobs.
                keywords.size() > 0)
                    matchedBlobs.add(blob);
            }
        }
        // of this document
        if (matchedBlobs.size() != 0) {
            BodyHLInfo bhlinfo = inputSet.matchedDocs.get(k).first;
            AttachmentHLInfo attachmentHLInfo = inputSet.matchedDocs.get(k).second;
            attachmentHLInfo.addMultipleInfo(matchedBlobs);
            outputDocs.put(k, new Pair(bhlinfo, attachmentHLInfo));
        }
    });
    return new SearchResult(outputDocs, inputSet.archive, inputSet.queryParams, inputSet.commonHLInfo, inputSet.regexToHighlight);
}

Example 5 with Blob

use of edu.stanford.muse.datacache.Blob in project epadd by ePADD.

the class SearchResult method searchForTerm.

/**
 * returns SearchResult containing docs and attachments matching the given term.
 *
 * @param inputSet Input search result object on which this term filtering needs to be done
 * @param term     term to search for
 * @return searchresult obj
 */
public static SearchResult searchForTerm(SearchResult inputSet, String term) {
    // go in the order subject, body, attachment
    Set<Document> docsForTerm = new LinkedHashSet<>();
    SearchResult outputSet;
    if ("on".equals(JSPHelper.getParam(inputSet.queryParams, "termSubject"))) {
        Indexer.QueryOptions options = new Indexer.QueryOptions();
        options.setQueryType(Indexer.QueryType.SUBJECT);
        docsForTerm.addAll(inputSet.archive.docsForQuery(term, options));
    }
    if ("on".equals(JSPHelper.getParam(inputSet.queryParams, "termBody"))) {
        Indexer.QueryOptions options = new Indexer.QueryOptions();
        options.setQueryType(Indexer.QueryType.FULL);
        docsForTerm.addAll(inputSet.archive.docsForQuery(term, options));
    } else if ("on".equals(JSPHelper.getParam(inputSet.queryParams, "termOriginalBody"))) {
        // this is an else because we don't want to look at both body and body original
        Indexer.QueryOptions options = new Indexer.QueryOptions();
        options.setQueryType(Indexer.QueryType.ORIGINAL);
        docsForTerm.addAll(inputSet.archive.docsForQuery(term, options));
    }
    Map<Document, Pair<BodyHLInfo, AttachmentHLInfo>> attachmentSearchResult;
    if ("on".equals(JSPHelper.getParam(inputSet.queryParams, "termAttachments"))) {
        attachmentSearchResult = new HashMap<>();
        Set<Blob> blobsForTerm = inputSet.archive.blobsForQuery(term);
        // iterate over 'all attachments' of docs present in 'inputSet'
        inputSet.matchedDocs.keySet().stream().forEach(d -> {
            EmailDocument edoc = (EmailDocument) d;
            Set<Blob> commonAttachments = new HashSet<>(edoc.attachments);
            commonAttachments.retainAll(blobsForTerm);
            // 0         yes        term found in body but not in attachment. keep its info in bodyHLInfo only.
            if (commonAttachments.size() > 0) {
                if (docsForTerm.contains(edoc)) {
                    BodyHLInfo bhlinfo = inputSet.matchedDocs.get(d).first;
                    AttachmentHLInfo attachmentHLInfo = inputSet.matchedDocs.get(d).second;
                    // it means the body and the attachment matched the term. add this information in body highliter/attachment highlighter
                    bhlinfo.addTerm(term);
                    attachmentHLInfo.addMultipleInfo(commonAttachments);
                    attachmentSearchResult.put(d, new Pair(bhlinfo, attachmentHLInfo));
                } else {
                    // means only attachment matched the term. add this information in attachment highlighter
                    BodyHLInfo bhlinfo = inputSet.matchedDocs.get(d).first;
                    AttachmentHLInfo attachmentHLInfo = inputSet.matchedDocs.get(d).second;
                    attachmentHLInfo.addMultipleInfo(commonAttachments);
                    attachmentSearchResult.put(d, new Pair(bhlinfo, attachmentHLInfo));
                }
            } else if (commonAttachments.size() == 0 && docsForTerm.contains(d)) {
                // means the document had the term only in its body and not in the attachment.
                BodyHLInfo bhlinfo = inputSet.matchedDocs.get(d).first;
                AttachmentHLInfo attachmentHLInfo = inputSet.matchedDocs.get(d).second;
                bhlinfo.addTerm(term);
                attachmentSearchResult.put(d, new Pair(bhlinfo, attachmentHLInfo));
            }
        });
        outputSet = new SearchResult(attachmentSearchResult, inputSet.archive, inputSet.queryParams, inputSet.commonHLInfo, inputSet.regexToHighlight);
    } else {
        // just retain only those document in inputSet.matchedDocs which are present in docsForTerm set.
        inputSet.matchedDocs.keySet().retainAll(docsForTerm);
        outputSet = inputSet;
    }
    // blobsForTerm.retainAll(inputSet.matchInAttachment.second);
    /*
        //query for the docs where these blobs are present. Note that we do not need to search for these blobs in all docs
        //only those present in the input search object (matchInAttachment.first) are sufficient as by our invariant of
        //matchInAttachment, the set of documents where matchInAttachment.second are present is same as matchInAttachment.first.
        Set<Document> blobDocsForTerm = (Set<Document>) EmailUtils.getDocsForAttachments((Collection) inputSet.matchInAttachment.first, blobsForTerm);
        attachmentSearchResult = new Pair(blobDocsForTerm,blobsForTerm);
        */
    // Add term to common highlighting info (as it is without parsing) for highlighting.
    // The term will be in lucene syntax (OR,AND etc.)
    // lucene highlighter will take care of highlighting that.
    outputSet.commonHLInfo.addTerm(term);
    return outputSet;
}