private Record toRecord(SolrInputDocument doc) {
Record record = new Record();
for (Entry<String, SolrInputField> entry : doc.entrySet()) {
record.getFields().putAll(entry.getKey(), entry.getValue().getValues());
}
return record;
}
private boolean hasAtLeastOneMimeType(Record record) {
if (!record.getFields().containsKey(Fields.ATTACHMENT_MIME_TYPE)) {
LOG.debug("Command failed because of missing MIME type for record: {}", record);
return false;
}
return true;
}
@Override
protected boolean doProcess(Record record) {
Timer.Context timerContext = elapsedTime.time();
try {
XContentBuilder documentBuilder = jsonBuilder().startObject();
Map<String, Collection<Object>> map = record.getFields().asMap();
for (Map.Entry<String, Collection<Object>> entry : map.entrySet()) {
String key = entry.getKey();
Iterator<Object> iterator = entry.getValue().iterator();
while (iterator.hasNext()) {
documentBuilder.field(key, iterator.next());
}
}
documentBuilder.endObject();
loader.addDocument(documentBuilder.bytes(), indexName, indexType, ttl);
} catch (Exception e) {
throw new MorphlineRuntimeException(e);
} finally {
timerContext.stop();
}
// pass record to next command in chain:
return super.doProcess(record);
}
@Override
protected boolean doProcess(Record record) {
if (preserveExisting
&& record
.getFields()
.containsKey(
Fields.ATTACHMENT_MIME_TYPE)) {; // we must preserve the existing MIME type
} else {
List attachments = record.get(Fields.ATTACHMENT_BODY);
if (attachments.size() > 0) {
Object attachment = attachments.get(0);
Preconditions.checkNotNull(attachment);
InputStream stream;
if (attachment instanceof byte[]) {
stream = new ByteArrayInputStream((byte[]) attachment);
} else {
stream = (InputStream) attachment;
}
Metadata metadata = new Metadata();
// If you specify the resource name (the filename, roughly) with this
// parameter, then Tika can use it in guessing the right MIME type
String resourceName = (String) record.getFirstValue(Fields.ATTACHMENT_NAME);
if (resourceName != null) {
metadata.add(Metadata.RESOURCE_NAME_KEY, resourceName);
}
// Provide stream's charset as hint to Tika for better auto detection
String charset = (String) record.getFirstValue(Fields.ATTACHMENT_CHARSET);
if (charset != null) {
metadata.add(Metadata.CONTENT_ENCODING, charset);
}
if (includeMetaData) {
for (Entry<String, Object> entry : record.getFields().entries()) {
metadata.add(entry.getKey(), entry.getValue().toString());
}
}
String mimeType = getMediaType(stream, metadata, excludeParameters);
record.replaceValues(Fields.ATTACHMENT_MIME_TYPE, mimeType);
}
}
return super.doProcess(record);
}
@Override
protected boolean doProcess(Record record, InputStream inputStream) {
Parser parser = detectParser(record);
if (parser == null) {
return false;
}
ParseContext parseContext = new ParseContext();
parseContext.set(Locale.class, locale);
Metadata metadata = new Metadata();
for (Entry<String, Object> entry : record.getFields().entries()) {
metadata.add(entry.getKey(), entry.getValue().toString());
}
SolrContentHandler handler =
solrContentHandlerFactory.createSolrContentHandler(metadata, solrParams, schema);
try {
inputStream = TikaInputStream.get(inputStream);
ContentHandler parsingHandler = handler;
// String xpathExpr = "/xhtml:html/xhtml:body/xhtml:div/descendant:node()";
if (xpathExpr != null) {
Matcher matcher = PARSER.parse(xpathExpr);
parsingHandler = new MatchingContentHandler(parsingHandler, matcher);
}
try {
parser.parse(inputStream, parsingHandler, metadata, parseContext);
} catch (IOException e) {
throw new MorphlineRuntimeException("Cannot parse", e);
} catch (SAXException e) {
throw new MorphlineRuntimeException("Cannot parse", e);
} catch (TikaException e) {
throw new MorphlineRuntimeException("Cannot parse", e);
}
} finally {
if (inputStream != null) {
Closeables.closeQuietly(inputStream);
}
}
SolrInputDocument doc = handler.newDocument();
LOG.debug("solr doc: {}", doc);
Record outputRecord = toRecord(doc);
return getChild().process(outputRecord);
}
