ATS-675:Add All-In-One transformer (#200)

* ATS-695/ATS-675 Add aio boot project

- Added the bare bones of a spring boot project to be used by aio. Currently based loosely on transform-misc.

* ATS-674/ATS-695 Add forms for each transformer.

* ATS-675/ATS-695 add empty test to pass build during dev

* ATS-695 remove maven profile to fix build

* ATS-675 Define interface and the aio transformer

* Fix formatting and rename the module as per review comments

* ATS-675/ATS-695 Add ProbeTestTransformation

Currenly uses MiscController implementation.

* ATS-675/ATS-695 Add logger method,

This will be code repeated in the local transform method and the processTransform method

* ATS-675/ATS-695 Implement local transform method

Minimum implementation for transform method.

* ATS-675/ATS-695  Implement processTransform

* ATS-675/ATS-695 Rename project to alfresco-transform-core-aio-boot

Add alfresco-transform-core-aio dependencies

* ATS-675/ATS-695 Fix build

Update project location
Update imports and variable declarations in TODOs
Add error handling.
Formatting.

* ATS-693: Update transform-misc Dockerfile with newly reserved uid

* Revert "ATS-691: Combine the win/linux pathToFile logic"

This reverts commit 61fe4820

* ATS-693: Update transform-misc Dockerfile with newly reserved uid

* "ATS-693: Add Dockerfile to aio-boot module"

* ATS-675/ATS-695 Add resource required for ProbeTestTrasform

* ATS-675/ATS-695 Remove test resources, to be added in test implementation

* ATS-693: Fix path to jar resources

* ATS-675/ATS-703 Moved Options builder to non boot jar.

* ATS-675/ATS-703 Rename OptionsBuilder to PdfRendererOptionsBuilder

This is to avoid confilct with OptionsBuilders in other T-engines.

* ATS-675/ATS-695 Added PdfRendererApadpter.java

Added dependency to pom.xml
Required transformation of String to Long, method added to Util.java

* ATS-675/ ATS-704

Implemented LibreOfficeAdapter

* ATS-675 Parity with base aio naming convention

* ATS-675/ATS-705 Implemented ImageMagickAdapter

Moved and renamed OptionsBuilder. Moved to alfresco-transform-imagemagick, renamed ImageMagickOptionsBuilder.
Added dependencies to pom.xml

* ATS-693: Implement maven docker build

* Initial tests
* Add initial tests for config aggregation
* Update AbstractTransformerControllerTest to use the new engine config names

* Fix up controller

* Fix travis tests  (#205)

* Fix engine specific properties for engine config location
* Temporarily add engine configs to test resources for the boot modules.  Will need to fix this properly

* Resolve some review comments

* ATS-675 - Move static strings to util class

* Refactor classes for simpler design (#210)

* ATS-702 Fix error handling

(cherry picked from commit e30cb5fda6)

* ATS-675 Rename test class (fixes typo)

* ATS-675: Add aio transformer to static scan
This commit is contained in:
eknizat
2020-04-08 17:40:34 +01:00
committed by GitHub
parent 5c6782a3aa
commit af77d429e7
56 changed files with 4900 additions and 77 deletions

View File

@@ -1,2 +1,5 @@
queue:
engineRequestQueue: ${TRANSFORM_ENGINE_REQUEST_QUEUE:org.alfresco.transform.engine.tika.acs}
engineRequestQueue: ${TRANSFORM_ENGINE_REQUEST_QUEUE:org.alfresco.transform.engine.tika.acs}
transform:
config:
location: classpath:tika_engine_config.json

View File

@@ -116,6 +116,7 @@ import org.springframework.test.web.servlet.request.MockMvcRequestBuilders;
@WebMvcTest(TikaController.class)
public class TikaControllerTest extends AbstractTransformerControllerTest
{
private static final String ENGINE_CONFIG_NAME = "tika_engine_config.json";
private static final String EXPECTED_XHTML_CONTENT_CONTAINS = "<p>The quick brown fox jumps over the lazy dog</p>";
private static final String EXPECTED_TEXT_CONTENT_CONTAINS = "The quick brown fox jumps over the lazy dog";
private static final String EXPECTED_MSG_CONTENT_CONTAINS = "Recipients\n" +
@@ -146,6 +147,12 @@ public class TikaControllerTest extends AbstractTransformerControllerTest
targetExtension = "txt";
}
@Override
public String getEngineConfigName()
{
return ENGINE_CONFIG_NAME;
}
@Override
protected void mockTransformCommand(String sourceExtension,
String targetExtension, String sourceMimetype,

View File

@@ -0,0 +1,508 @@
{
"transformOptions": {
"tikaOptions": [
{"value": {"name": "targetEncoding"}}
],
"archiveOptions": [
{"value": {"name": "includeContents"}},
{"value": {"name": "targetEncoding"}}
],
"pdfboxOptions": [
{"value": {"name": "notExtractBookmarksText"}},
{"value": {"name": "targetEncoding"}}
]
},
"transformers": [
{
"transformerName": "Archive",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/x-cpio", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-cpio", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-cpio", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-cpio", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/java-archive", "targetMediaType": "text/html"},
{"sourceMediaType": "application/java-archive", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/java-archive", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/java-archive", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-tar", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-tar", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-tar", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-tar", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/zip", "targetMediaType": "text/html"},
{"sourceMediaType": "application/zip", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/zip", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/zip", "targetMediaType": "text/xml"}
],
"transformOptions": [
"archiveOptions"
]
},
{
"transformerName": "OutlookMsg",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/vnd.ms-outlook", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-outlook", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-outlook", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-outlook", "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
},
{
"transformerName": "PdfBox",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/pdf", "targetMediaType": "text/csv"},
{"sourceMediaType": "application/pdf", "targetMediaType": "text/html"},
{"sourceMediaType": "application/pdf", "maxSourceSizeBytes": 26214400, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/pdf", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/pdf", "targetMediaType": "text/xml"}
],
"transformOptions": [
"pdfboxOptions"
]
},
{
"transformerName": "Office",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/msword", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/msword", "priority": 60, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/msword", "priority": 60, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/msword", "priority": 60, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-project", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-project", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-project", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-project", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-outlook", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-outlook", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-outlook", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-outlook", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.visio", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.visio", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.visio", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.visio", "priority": 55, "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
},
{
"transformerName": "Poi",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/vnd.ms-excel", "priority": 55, "targetMediaType": "text/csv"},
{"sourceMediaType": "application/vnd.ms-excel", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 55, "targetMediaType": "text/csv"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 65, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 60, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 60, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 60, "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
},
{
"transformerName": "OOXML",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "priority": 60, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "priority": 55, "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
},
{
"transformerName": "TikaAuto",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/x-cpio", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-cpio", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-cpio", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-cpio", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/java-archive", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/java-archive", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/java-archive", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/java-archive", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-netcdf", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-netcdf", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-netcdf", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-netcdf", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/msword", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/msword", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/msword", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/msword", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-word.document.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document" , "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-word.template.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-gzip", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-gzip", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-gzip", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-gzip", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-hdf", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-hdf", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-hdf", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-hdf", "targetMediaType": "text/xml"},
{"sourceMediaType": "text/html", "targetMediaType": "text/html"},
{"sourceMediaType": "text/html", "priority": 60, "targetMediaType": "text/plain"},
{"sourceMediaType": "text/html", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "text/html", "targetMediaType": "text/xml"},
{"sourceMediaType": "text/x-java-source", "targetMediaType": "text/html"},
{"sourceMediaType": "text/x-java-source", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "text/x-java-source", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "text/x-java-source", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.apple.keynote", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.apple.keynote", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.apple.keynote", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-project", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-project", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-project", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-project", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.apple.numbers", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.apple.numbers", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.apple.numbers", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.chart", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.chart", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.chart", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.chart", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.image", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.image", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.image", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.image", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-master", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-master", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-master", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-master", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/ogg", "targetMediaType": "text/html"},
{"sourceMediaType": "application/ogg", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/ogg", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/ogg", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-web", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-web", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-web", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-web", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation-template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation-template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation-template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.presentation-template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet-template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet-template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet-template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.spreadsheet-template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.oasis.opendocument.text-template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.apple.pages", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.apple.pages", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.apple.pages", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/pdf", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/pdf", "maxSourceSizeBytes": 26214400, "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/pdf", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/pdf", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.template.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.addin.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slideshow.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slideshow", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.presentation.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-rar-compressed", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-rar-compressed", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-rar-compressed", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-rar-compressed", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/rss+xml", "targetMediaType": "text/html"},
{"sourceMediaType": "application/rss+xml", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/rss+xml", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/rss+xml", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/rtf", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/rtf", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/rtf", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/rtf", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-powerpoint.slide.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.presentationml.slide", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.sun.xml.writer", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.sun.xml.writer", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.sun.xml.writer", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.sun.xml.writer", "targetMediaType": "text/xml"},
{"sourceMediaType": "text/plain", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "text/plain", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "text/plain", "priority": 55, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "text/plain", "priority": 55, "targetMediaType": "text/xml"},
{"sourceMediaType": "text/xml", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "text/xml", "priority": 55, "targetMediaType": "text/plain"},
{"sourceMediaType": "text/xml", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "text/xml", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.visio", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.visio", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.visio", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.visio", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/xhtml+xml", "targetMediaType": "text/html"},
{"sourceMediaType": "application/xhtml+xml", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/xhtml+xml", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/xhtml+xml", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.addin.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.binary.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.sheet.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.ms-excel.template.macroenabled.12", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "priority": 55, "targetMediaType": "text/html"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.template", "targetMediaType": "text/xml"},
{"sourceMediaType": "application/x-compress", "targetMediaType": "text/html"},
{"sourceMediaType": "application/x-compress", "targetMediaType": "text/plain"},
{"sourceMediaType": "application/x-compress", "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/x-compress", "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
},
{
"transformerName": "TextMining",
"supportedSourceAndTargetList": [
{"sourceMediaType": "application/msword", "priority": 65, "targetMediaType": "text/html"},
{"sourceMediaType": "application/msword", "priority": 65, "targetMediaType": "text/plain"},
{"sourceMediaType": "application/msword", "priority": 65, "targetMediaType": "application/xhtml+xml"},
{"sourceMediaType": "application/msword", "targetMediaType": "text/xml"}
],
"transformOptions": [
"tikaOptions"
]
}
]
}