Examples with JavaRegexParseException datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException used on opensource projects

Example 1 with JavaRegexParseException

use of datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException in project datawave by NationalSecurityAgency.

the class DiscoveryLogic method makeRanges.

/**
 * Makes two collections of ranges: one for the forward index (value0) and one for the reverse index (value1).
 *
 * If a literal has a field name, then the Range for that term will include the column family. If there are multiple fields, then multiple ranges are
 * created.
 *
 * @param config
 * @return
 * @throws TableNotFoundException
 * @throws ExecutionException
 */
@SuppressWarnings("unchecked")
public static Pair<Set<Range>, Set<Range>> makeRanges(DiscoveryQueryConfiguration config, Set<Text> familiesToSeek, MetadataHelper metadataHelper) throws TableNotFoundException, ExecutionException {
    Set<Range> forwardRanges = new HashSet<>();
    for (Entry<String, String> literalAndField : config.getLiterals().entries()) {
        String literal = literalAndField.getKey(), field = literalAndField.getValue();
        // if we're _ANYFIELD_, then use null when making the literal range
        field = Constants.ANY_FIELD.equals(field) ? null : field;
        if (field != null) {
            familiesToSeek.add(new Text(field));
        }
        forwardRanges.add(ShardIndexQueryTableStaticMethods.getLiteralRange(field, literal));
    }
    for (Entry<String, LiteralRange<String>> rangeEntry : config.getRanges().entries()) {
        LiteralRange<String> range = rangeEntry.getValue();
        String field = rangeEntry.getKey();
        // if we're _ANYFIELD_, then use null when making the literal range
        field = Constants.ANY_FIELD.equals(field) ? null : field;
        if (field != null) {
            familiesToSeek.add(new Text(field));
        }
        try {
            forwardRanges.add(ShardIndexQueryTableStaticMethods.getBoundedRangeRange(range));
        } catch (IllegalRangeArgumentException e) {
            log.error("Error using range [" + range + "]", e);
            continue;
        }
    }
    Set<Range> reverseRanges = new HashSet<>();
    for (Entry<String, String> patternAndField : config.getPatterns().entries()) {
        String pattern = patternAndField.getKey(), field = patternAndField.getValue();
        // if we're _ANYFIELD_, then use null when making the literal range
        field = Constants.ANY_FIELD.equals(field) ? null : field;
        ShardIndexQueryTableStaticMethods.RefactoredRangeDescription description;
        try {
            if (field != null) {
                familiesToSeek.add(new Text(field));
            }
            description = ShardIndexQueryTableStaticMethods.getRegexRange(field, pattern, false, metadataHelper, config);
        } catch (JavaRegexParseException e) {
            log.error("Error parsing pattern [" + pattern + "]", e);
            continue;
        }
        if (description.isForReverseIndex) {
            reverseRanges.add(description.range);
        } else {
            forwardRanges.add(description.range);
        }
    }
    return Pair.with(forwardRanges, reverseRanges);
}

Example 2 with JavaRegexParseException

use of datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException in project datawave by NationalSecurityAgency.

the class IteratorBuildingVisitor method buildLiteralRange.

public static LiteralRange<?> buildLiteralRange(ASTERNode node) {
    JavaRegexAnalyzer analyzer;
    try {
        analyzer = new JavaRegexAnalyzer(String.valueOf(JexlASTHelper.getLiteralValue(node)));
        LiteralRange<String> range = new LiteralRange<>(JexlASTHelper.getIdentifier(node), NodeOperand.AND);
        if (!analyzer.isLeadingLiteral()) {
            // if the range is a leading wildcard we have to seek over the whole range since it's forward indexed only
            range.updateLower(Constants.NULL_BYTE_STRING, true, node);
            range.updateUpper(Constants.MAX_UNICODE_STRING, true, node);
        } else {
            range.updateLower(analyzer.getLeadingLiteral(), true, node);
            if (analyzer.hasWildCard()) {
                range.updateUpper(analyzer.getLeadingLiteral() + Constants.MAX_UNICODE_STRING, true, node);
            } else {
                range.updateUpper(analyzer.getLeadingLiteral(), true, node);
            }
        }
        return range;
    } catch (JavaRegexParseException | NoSuchElementException e) {
        throw new DatawaveFatalQueryException(e);
    }
}

Example 3 with JavaRegexParseException

use of datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException in project datawave by NationalSecurityAgency.

the class IteratorBuildingVisitor method buildLiteralRange.

LiteralRange<?> buildLiteralRange(ASTNRNode node) {
    JavaRegexAnalyzer analyzer;
    try {
        analyzer = new JavaRegexAnalyzer(String.valueOf(JexlASTHelper.getLiteralValue(node)));
        LiteralRange<String> range = new LiteralRange<>(JexlASTHelper.getIdentifier(node), NodeOperand.AND);
        range.updateLower(analyzer.getLeadingOrTrailingLiteral(), true, node);
        range.updateUpper(analyzer.getLeadingOrTrailingLiteral() + Constants.MAX_UNICODE_STRING, true, node);
        return range;
    } catch (JavaRegexParseException | NoSuchElementException e) {
        throw new DatawaveFatalQueryException(e);
    }
}

Example 4 with JavaRegexParseException

use of datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException in project datawave by NationalSecurityAgency.

the class FullTableScan method getCost.

/*
     * (non-Javadoc)
     * 
     * @see datawave.query.planner.pushdown.PushDown#getCost(org.apache.commons.jexl2.parser.JexlNode)
     */
@Override
public Cost getCost(JexlNode node) {
    String pattern = JexlASTHelper.getLiteralValue(node).toString();
    JavaRegexAnalyzer regex;
    try {
        regex = new JavaRegexAnalyzer(pattern);
        if (regex.isLeadingRegex() && regex.isTrailingRegex())
            return Cost.INFINITE;
    } catch (JavaRegexParseException e) {
        log.warn("Couldn't parse regex from ERNode: " + pattern);
    }
    return Cost.UNEVALUATED;
}

Example 5 with JavaRegexParseException

use of datawave.query.parser.JavaRegexAnalyzer.JavaRegexParseException in project datawave by NationalSecurityAgency.

the class RegexIndexLookup method submit.

@Override
public synchronized void submit() {
    if (indexLookupMap == null) {
        indexLookupMap = new IndexLookupMap(config.getMaxUnfieldedExpansionThreshold(), config.getMaxValueExpansionThreshold());
        indexLookupMap.setPatterns(patterns);
        Multimap<String, Range> forwardMap = ArrayListMultimap.create(), reverseMap = ArrayListMultimap.create();
        // Loop over all the patterns, classifying them as forward or reverse index satisfiable
        Iterator<Entry<Key, Value>> iter = Iterators.emptyIterator();
        ScannerSession bs;
        IteratorSetting fairnessIterator = null;
        if (config.getMaxIndexScanTimeMillis() > 0) {
            // The fairness iterator solves the problem whereby we have runaway iterators as a result of an evaluation that never finds anything
            fairnessIterator = new IteratorSetting(1, TimeoutIterator.class);
            long maxTime = (long) (config.getMaxIndexScanTimeMillis() * 1.25);
            fairnessIterator.addOption(TimeoutIterator.MAX_SESSION_TIME, Long.valueOf(maxTime).toString());
        }
        for (String pattern : patterns) {
            if (config.getDisallowedRegexPatterns().contains(pattern)) {
                PreConditionFailedQueryException qe = new PreConditionFailedQueryException(DatawaveErrorCode.IGNORE_PATTERN_FOR_INDEX_LOOKUP, MessageFormat.format("Pattern: {0}", pattern));
                log.debug(qe);
                throw new DoNotPerformOptimizedQueryException(qe);
            }
            ShardIndexQueryTableStaticMethods.RefactoredRangeDescription rangeDescription;
            try {
                rangeDescription = ShardIndexQueryTableStaticMethods.getRegexRange(null, pattern, config.getFullTableScanEnabled(), helper, config);
            } catch (IllegalArgumentException | JavaRegexParseException e) {
                log.debug("Ignoring pattern that was not capable of being looked up in the index: " + pattern, e);
                continue;
            } catch (TableNotFoundException e) {
                log.error(e);
                throw new DatawaveFatalQueryException(e);
            } catch (ExecutionException e) {
                throw new DatawaveFatalQueryException(e);
            }
            if (log.isTraceEnabled()) {
                log.trace("Adding pattern " + pattern);
                log.trace("Adding pattern " + rangeDescription);
            }
            if (rangeDescription.isForReverseIndex) {
                reverseMap.put(pattern, rangeDescription.range);
            } else {
                forwardMap.put(pattern, rangeDescription.range);
            }
        }
        if (!fields.isEmpty() && !forwardMap.isEmpty()) {
            for (String key : forwardMap.keySet()) {
                Collection<Range> ranges = forwardMap.get(key);
                try {
                    bs = ShardIndexQueryTableStaticMethods.configureLimitedDiscovery(config, scannerFactory, config.getIndexTableName(), ranges, Collections.emptySet(), Collections.singleton(key), false, true);
                    bs.setResourceClass(BatchResource.class);
                } catch (Exception e) {
                    throw new DatawaveFatalQueryException(e);
                }
                SessionOptions opts = bs.getOptions();
                if (null != fairnessIterator) {
                    opts.addScanIterator(fairnessIterator);
                    IteratorSetting cfg = new IteratorSetting(config.getBaseIteratorPriority() + 100, TimeoutExceptionIterator.class);
                    opts.addScanIterator(cfg);
                }
                for (String field : fields) {
                    opts.fetchColumnFamily(new Text(field));
                }
                forwardLookupData.getSessions().add(bs);
                iter = Iterators.concat(iter, bs);
            }
            forwardLookupData.setTimedScanFuture(execService.submit(createTimedCallable(iter, fields, forwardLookupData, indexLookupMap)));
        }
        if (!reverseFields.isEmpty() && !reverseMap.isEmpty()) {
            for (String key : reverseMap.keySet()) {
                Collection<Range> ranges = reverseMap.get(key);
                if (log.isTraceEnabled()) {
                    log.trace("adding " + ranges + " for reverse");
                }
                try {
                    bs = ShardIndexQueryTableStaticMethods.configureLimitedDiscovery(config, scannerFactory, config.getReverseIndexTableName(), ranges, Collections.emptySet(), Collections.singleton(key), true, true);
                    bs.setResourceClass(BatchResource.class);
                } catch (Exception e) {
                    throw new DatawaveFatalQueryException(e);
                }
                SessionOptions opts = bs.getOptions();
                if (null != fairnessIterator) {
                    opts.addScanIterator(fairnessIterator);
                    opts.addScanIterator(new IteratorSetting(config.getBaseIteratorPriority() + 100, TimeoutExceptionIterator.class));
                }
                for (String field : reverseFields) {
                    opts.fetchColumnFamily(new Text(field));
                }
                reverseLookupData.getSessions().add(bs);
                iter = Iterators.concat(iter, bs);
            }
            reverseLookupData.setTimedScanFuture(execService.submit(createTimedCallable(iter, reverseFields, reverseLookupData, indexLookupMap)));
        }
    }
}