Attachment ingest processor: add resource_name field (#64389)

plugins/ingest-attachment/build.gradle

@@ -93,6 +93,7 @@ forbiddenPatterns {
   exclude '**/*.pdf'
   exclude '**/*.epub'
   exclude '**/*.vsdx'
+  exclude '**/text-cjk-*.txt'
 }
 
 thirdPartyAudit {

plugins/ingest-attachment/src/main/java/org/elasticsearch/ingest/attachment/AttachmentProcessor.java

@@ -56,16 +56,18 @@ public final class AttachmentProcessor extends AbstractProcessor {
     private final int indexedChars;
     private final boolean ignoreMissing;
     private final String indexedCharsField;
+    private final String resourceName;
 
     AttachmentProcessor(String tag, String description, String field, String targetField, Set<Property> properties,
-                        int indexedChars, boolean ignoreMissing, String indexedCharsField) {
+                        int indexedChars, boolean ignoreMissing, String indexedCharsField, String resourceName) {
         super(tag, description);
         this.field = field;
         this.targetField = targetField;
         this.properties = properties;
         this.indexedChars = indexedChars;
         this.ignoreMissing = ignoreMissing;
         this.indexedCharsField = indexedCharsField;
+        this.resourceName = resourceName;
     }
 
     boolean isIgnoreMissing() {
@@ -77,7 +79,10 @@ public IngestDocument execute(IngestDocument ingestDocument) {
         Map<String, Object> additionalFields = new HashMap<>();
 
         byte[] input = ingestDocument.getFieldValueAsBytes(field, ignoreMissing);
-
+        String resourceNameInput = null;
+        if (resourceName != null) {
+            resourceNameInput = ingestDocument.getFieldValue(resourceName, String.class, true);
+        }
         if (input == null && ignoreMissing) {
             return ingestDocument;
         } else if (input == null) {
@@ -96,6 +101,9 @@ public IngestDocument execute(IngestDocument ingestDocument) {
         }
 
         Metadata metadata = new Metadata();
+        if (resourceNameInput != null) {
+            metadata.set(Metadata.RESOURCE_NAME_KEY, resourceNameInput);
+        }
         String parsedContent = "";
         try {
             parsedContent = TikaImpl.parse(input, metadata, indexedChars);
@@ -197,6 +205,7 @@ public static final class Factory implements Processor.Factory {
         public AttachmentProcessor create(Map<String, Processor.Factory> registry, String processorTag,
                                           String description, Map<String, Object> config) throws Exception {
             String field = readStringProperty(TYPE, processorTag, config, "field");
+            String resourceName = readOptionalStringProperty(TYPE, processorTag, config, "resource_name");
             String targetField = readStringProperty(TYPE, processorTag, config, "target_field", "attachment");
             List<String> propertyNames = readOptionalList(TYPE, processorTag, config, "properties");
             int indexedChars = readIntProperty(TYPE, processorTag, config, "indexed_chars", NUMBER_OF_CHARS_INDEXED);
@@ -219,7 +228,7 @@ public AttachmentProcessor create(Map<String, Processor.Factory> registry, Strin
             }
 
             return new AttachmentProcessor(processorTag, description, field, targetField, properties, indexedChars, ignoreMissing,
-                indexedCharsField);
+                indexedCharsField, resourceName);
         }
     }
 

plugins/ingest-attachment/src/test/java/org/elasticsearch/ingest/attachment/AttachmentProcessorTests.java

@@ -55,7 +55,7 @@ public class AttachmentProcessorTests extends ESTestCase {
     @Before
     public void createStandardProcessor() {
         processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
-            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 10000, false, null);
+            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 10000, false, null, null);
     }
 
     public void testEnglishTextDocument() throws Exception {
@@ -88,7 +88,7 @@ public void testHtmlDocumentWithRandomFields() throws Exception {
             selectedProperties.add(AttachmentProcessor.Property.DATE);
         }
         processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
-            "target_field", selectedProperties, 10000, false, null);
+            "target_field", selectedProperties, 10000, false, null, null);
 
         Map<String, Object> attachmentData = parseDocument("htmlWithEmptyDateMeta.html", processor);
         assertThat(attachmentData.keySet(), hasSize(selectedFieldNames.length));
@@ -160,7 +160,7 @@ public void testLegacyWordDocumentWithVisioSchema() throws Exception {
     public void testPdf() throws Exception {
         Map<String, Object> attachmentData = parseDocument("test.pdf", processor);
         assertThat(attachmentData.get("content"),
-                is("This is a test, with umlauts, from München\n\nAlso contains newlines for testing.\n\nAnd one more."));
+            is("This is a test, with umlauts, from München\n\nAlso contains newlines for testing.\n\nAnd one more."));
         assertThat(attachmentData.get("content_type").toString(), is("application/pdf"));
         assertThat(attachmentData.get("content_length"), is(notNullValue()));
     }
@@ -247,15 +247,17 @@ public void testNullValueWithIgnoreMissing() throws Exception {
         IngestDocument originalIngestDocument = RandomDocumentPicks.randomIngestDocument(random(),
             Collections.singletonMap("source_field", null));
         IngestDocument ingestDocument = new IngestDocument(originalIngestDocument);
-        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field", "randomTarget", null, 10, true, null);
+        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
+            "randomTarget", null, 10, true, null, null);
         processor.execute(ingestDocument);
         assertIngestDocument(originalIngestDocument, ingestDocument);
     }
 
     public void testNonExistentWithIgnoreMissing() throws Exception {
         IngestDocument originalIngestDocument = RandomDocumentPicks.randomIngestDocument(random(), Collections.emptyMap());
         IngestDocument ingestDocument = new IngestDocument(originalIngestDocument);
-        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field", "randomTarget", null, 10, true, null);
+        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
+            "randomTarget", null, 10, true, null, null);
         processor.execute(ingestDocument);
         assertIngestDocument(originalIngestDocument, ingestDocument);
     }
@@ -264,15 +266,17 @@ public void testNullWithoutIgnoreMissing() throws Exception {
         IngestDocument originalIngestDocument = RandomDocumentPicks.randomIngestDocument(random(),
             Collections.singletonMap("source_field", null));
         IngestDocument ingestDocument = new IngestDocument(originalIngestDocument);
-        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field", "randomTarget", null, 10, false, null);
+        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
+            "randomTarget", null, 10, false, null, null);
         Exception exception = expectThrows(Exception.class, () -> processor.execute(ingestDocument));
         assertThat(exception.getMessage(), equalTo("field [source_field] is null, cannot parse."));
     }
 
     public void testNonExistentWithoutIgnoreMissing() throws Exception {
         IngestDocument originalIngestDocument = RandomDocumentPicks.randomIngestDocument(random(), Collections.emptyMap());
         IngestDocument ingestDocument = new IngestDocument(originalIngestDocument);
-        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field", "randomTarget", null, 10, false, null);
+        Processor processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
+            "randomTarget", null, 10, false, null, null);
         Exception exception = expectThrows(Exception.class, () -> processor.execute(ingestDocument));
         assertThat(exception.getMessage(), equalTo("field [source_field] not present as part of path [source_field]"));
     }
@@ -295,9 +299,27 @@ private Map<String, Object> parseDocument(String file, AttachmentProcessor proce
         return attachmentData;
     }
 
+    private Map<String, Object> parseDocument(String file, AttachmentProcessor processor, Map<String, Object> optionalFields,
+                                              boolean includeResourceName)
+        throws Exception {
+        Map<String, Object> document = new HashMap<>();
+        document.put("source_field", getAsBinaryOrBase64(file));
+        if (includeResourceName) {
+            document.put("resource_name", file);
+        }
+        document.putAll(optionalFields);
+
+        IngestDocument ingestDocument = RandomDocumentPicks.randomIngestDocument(random(), document);
+        processor.execute(ingestDocument);
+
+        @SuppressWarnings("unchecked")
+        Map<String, Object> attachmentData = (Map<String, Object>) ingestDocument.getSourceAndMetadata().get("target_field");
+        return attachmentData;
+    }
+
     public void testIndexedChars() throws Exception {
         processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
-            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 19, false, null);
+            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 19, false, null, null);
 
         Map<String, Object> attachmentData = parseDocument("text-in-english.txt", processor);
 
@@ -308,7 +330,7 @@ public void testIndexedChars() throws Exception {
         assertThat(attachmentData.get("content_length"), is(19L));
 
         processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
-            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 19, false, "max_length");
+            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 19, false, "max_length", null);
 
         attachmentData = parseDocument("text-in-english.txt", processor);
 
@@ -333,6 +355,38 @@ public void testIndexedChars() throws Exception {
         assertThat(attachmentData.get("content"), is("\"God Save the Queen\" (alternatively \"God Save the King\""));
         assertThat(attachmentData.get("content_type").toString(), containsString("text/plain"));
         assertThat(attachmentData.get("content_length"), is(56L));
+
+    }
+
+    public void testIndexedCharsWithResourceName() throws Exception {
+        processor = new AttachmentProcessor(randomAlphaOfLength(10), null, "source_field",
+            "target_field", EnumSet.allOf(AttachmentProcessor.Property.class), 100,
+            false, null, "resource_name");
+
+        Map<String, Object> attachmentData = parseDocument("text-cjk-big5.txt", processor, Collections.singletonMap("max_length", 100),
+            true);
+
+        assertThat(attachmentData.keySet(), containsInAnyOrder("language", "content", "content_type", "content_length"));
+        assertThat(attachmentData.get("content").toString(), containsString("碩鼠碩鼠，無食我黍！"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("text/plain"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("charset=Big5"));
+        assertThat(attachmentData.get("content_length"), is(100L));
+
+        attachmentData = parseDocument("text-cjk-gbk.txt", processor, Collections.singletonMap("max_length", 100), true);
+
+        assertThat(attachmentData.keySet(), containsInAnyOrder("language", "content", "content_type", "content_length"));
+        assertThat(attachmentData.get("content").toString(), containsString("硕鼠硕鼠，无食我黍！"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("text/plain"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("charset=GB18030"));
+        assertThat(attachmentData.get("content_length"), is(100L));
+
+        attachmentData = parseDocument("text-cjk-euc-jp.txt", processor, Collections.singletonMap("max_length", 100), true);
+
+        assertThat(attachmentData.keySet(), containsInAnyOrder("language", "content", "content_type", "content_length"));
+        assertThat(attachmentData.get("content").toString(), containsString("碩鼠よ碩鼠よ、\n我が黍を食らう無かれ！"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("text/plain"));
+        assertThat(attachmentData.get("content_type").toString(), containsString("charset=EUC-JP"));
+        assertThat(attachmentData.get("content_length"), is(100L));
     }
 
     private Object getAsBinaryOrBase64(String filename) throws Exception {

plugins/ingest-attachment/src/test/resources/org/elasticsearch/ingest/attachment/test/sample-files/text-cjk-big5.txt

@@ -0,0 +1,14 @@
+�ӹ��ӹ��A�L���ڶ��I�T���e�k�A���ڪ��U�C�u�N�h�k�A�A���֤g�C�֤g�֤g�A���o�کҡC
+�ӹ��ӹ��A�L���ڳ��I�T���e�k�A���ڪּw�C�u�N�h�k�A�A���ְ�C�ְ�ְ�A���o�ڪ��C
+�ӹ��ӹ��A�L���ڭ]�I�T���e�k�A���ڪֳҡC�u�N�h�k�A�A���֭��C�֭��֭��A�֤��ø��C
+���������A�I�_�����A�����ֵ֡A�����_���A���_���A����n�n�C
+���������A�I�_�����A���������A�O�¬O��A���]���b�A�A���L��C
+���i�v��A���i���k�A�����ڨp�A����ڦ�A�`��`�_�A�k������C
+�������աA���ճ����A�ا��h�H�A�I���P��C
+�F���Z�R�A�ڰ��k誡A�کh�u�������A���H�����h�C
+�F�������A�ڰ��ȶ��A�کh�u�������A���H���öˡC
+�F�����o�A�ڰ��Ȩo�A�ڹ�ڱ�o�A����S�o�C
+�h���u���A��y�_�N�A���h�_���A���餣��A�ܩ��ѮV�A��P���ѡC
+�X�J�_���A���^�_�A�k�l����A�������S�̡A�ݧڽѩh�A�E�ΧB�n�C
+�X�J�_�z�A���^�_���A���׸�ߣ�A�٨����ڡAࡿ��_���A���妳�`�C
+�ګ�άu�A�����üۡA�䶷�P���A�ڤ߱y�y�A�r���X�C�A�H�g�ڼ~�C

plugins/ingest-attachment/src/test/resources/org/elasticsearch/ingest/attachment/test/sample-files/text-cjk-euc-jp.txt

@@ -0,0 +1,29 @@
+���ͤ����ͤ衢
+�椬�Ф򿩤餦̵���졪
+���н��˴Ӥ����ˡ�
+���Τ��Ƹܤߤ�������
+
+�¤��ƾ��˽����ꡢ
+��γ��ڤ�Ŭ����
+���ڤ���ڤ衢
+੤˲椬�������
+
+���ͤ����ͤ衢
+�椬���򿩤餦̵���졪
+���н��˴Ӥ����ˡ�
+������������������
+
+�¤��ƾ��˽����ꡢ
+��γڹ��Ŭ����
+�ڹ��ڹ�衢
+੤˲椬ľ������
+
+���ͤ����ͤ衢
+�椬�Ĥ򿩤餦̵���졪
+���н��˴Ӥ����ˡ�
+���Τ���ϫ��������
+
+�¤��ƾ��˽����ꡢ
+��γڹ٤�Ŭ����
+�ڹ٤�ڹ٤衢
+ï��Ƿ���Ʊʹ椻��

plugins/ingest-attachment/src/test/resources/org/elasticsearch/ingest/attachment/test/sample-files/text-cjk-gbk.txt

@@ -0,0 +1,3 @@
+˶��˶����ʳ���������Ů��Ī�ҿϹˡ��Ž�ȥŮ���ʱ�����������������뼵�������
+˶��˶����ʳ���������Ů��Ī�ҿϵ¡��Ž�ȥŮ���ʱ��ֹ����ֹ��ֹ���뼵���ֱ��
+˶��˶����ʳ���磡�����Ů��Ī�ҿ��͡��Ž�ȥŮ���ʱ��ֽ����ֽ��ֽ���˭֮���š�