Package io.outofprintmagazine.corpus.batch

Class CorpusBatchStep

java.lang.Object

io.outofprintmagazine.corpus.batch.CorpusBatchStep

All Implemented Interfaces:

ICorpusBatchStep

Direct Known Subclasses:

Analyze, CleanText, CoreNLPMyersBriggsScores, CoreNLPTfidfScores, CoreNLPZScores, CorpusAggregate, CorpusWord2Vec, DocumentWord2Vec, ExtractText, FilterOutput, GenerateDocID, HttpDownload, ImportDirectory, ParseArchive, ParseBodyStory, ParseStory, ParseStory, ParseStory, ParseTableTOC, ParseTOC, ParseTOC, ParseTOC, Polly, Stamper, ThreadedCorpusBatchStep, ThreadedCorpusBatchStepTask, TomcatExport

public abstract class CorpusBatchStep
extends Object
implements ICorpusBatchStep

Field Summary

Fields

Modifier and Type	Field	Description
protected List<String>	dictionaryPOS

Constructor Summary

Constructors

Constructor	Description
CorpusBatchStep()

Method Summary

Modifier and Type	Method	Description
protected com.fasterxml.jackson.databind.node.ObjectNode	copyInputToOutput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
protected void	copyInputToOutput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getAuthor(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getAuthor(org.jsoup.nodes.Document doc)
CorpusBatchStepModel	getData()
protected String	getDate(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getDate(org.jsoup.nodes.Document doc)
protected SimpleDateFormat	getDateFormat()
com.fasterxml.jackson.databind.node.ObjectNode	getDefaultProperties()
protected String	getDocID(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getExtensionFromMimeType(String mimeType)
protected com.fasterxml.jackson.databind.JsonNode	getJsonNodeFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
protected com.fasterxml.jackson.databind.JsonNode	getJsonNodeFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem, String property)
protected org.jsoup.nodes.Document	getJsoupDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
protected org.jsoup.nodes.Document	getJsoupDocumentFromStorageNormalized(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
protected String	getLink(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected com.fasterxml.jackson.databind.ObjectMapper	getMapper()
protected String	getMimeTypeFromExtension(String extension)
protected String	getOutputScratchFilePath(String fileName)
protected String	getOutputScratchFilePath(String fileName, String extension)
protected String	getOutputScratchFilePathFromInput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem, String extension)
protected io.outofprintmagazine.util.IParameterStore	getParameterStore()
protected IScratchStorage	getStorage()
protected String	getStorageLink(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getText(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getText(org.jsoup.nodes.Element element)
protected String	getTextDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
protected String	getTextDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem, String property)
protected String	getTextWithSelector(org.jsoup.nodes.Element element, String selector)
protected String	getThumbnail(org.jsoup.nodes.Document doc)
protected String	getTitle(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected String	getTitle(org.jsoup.nodes.Document doc)
protected boolean	isDictionaryWord(String pos)
com.fasterxml.jackson.databind.node.ArrayNode	run(com.fasterxml.jackson.databind.node.ArrayNode input)
abstract com.fasterxml.jackson.databind.node.ArrayNode	runOne(com.fasterxml.jackson.databind.node.ObjectNode input)
protected void	setAuthor(String author, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setAuthor(org.jsoup.nodes.Document doc, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
void	setData(CorpusBatchStepModel data)
protected void	setDate(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setDate(String date, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setDate(org.jsoup.nodes.Document doc, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setDocID(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem, String docID)
protected void	setLink(String link, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
void	setParameterStore(io.outofprintmagazine.util.IParameterStore parameterStore)
void	setStorage(IScratchStorage storage)
protected void	setStorageLink(String storage, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setThumbnail(String thumbnail, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setThumbnail(org.jsoup.nodes.Document doc, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setTitle(String title, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
protected void	setTitle(org.jsoup.nodes.Document doc, com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

dictionaryPOS

protected List<String> dictionaryPOS

Constructor Detail

CorpusBatchStep

public CorpusBatchStep()

Method Detail

getMapper

protected com.fasterxml.jackson.databind.ObjectMapper getMapper()

getParameterStore

protected io.outofprintmagazine.util.IParameterStore getParameterStore()
                                                                throws IOException

Throws:

IOException

setParameterStore

public void setParameterStore(io.outofprintmagazine.util.IParameterStore parameterStore)

Specified by:

setParameterStore in interface ICorpusBatchStep

setStorage

public void setStorage(IScratchStorage storage)

Specified by:

setStorage in interface ICorpusBatchStep

getStorage

protected IScratchStorage getStorage()
                              throws IOException

Throws:

IOException

getDateFormat

protected SimpleDateFormat getDateFormat()

getData

public CorpusBatchStepModel getData()

Specified by:

getData in interface ICorpusBatchStep

setData

public void setData(CorpusBatchStepModel data)

Specified by:

setData in interface ICorpusBatchStep

getDefaultProperties

public com.fasterxml.jackson.databind.node.ObjectNode getDefaultProperties()

Specified by:

getDefaultProperties in interface ICorpusBatchStep

copyInputToOutput

protected void copyInputToOutput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem,
                                 com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)
                          throws IOException

Throws:

IOException

copyInputToOutput

protected com.fasterxml.jackson.databind.node.ObjectNode copyInputToOutput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
                                                                    throws IOException

Throws:

IOException

run

public com.fasterxml.jackson.databind.node.ArrayNode run(com.fasterxml.jackson.databind.node.ArrayNode input)

Specified by:

run in interface ICorpusBatchStep

runOne

public abstract com.fasterxml.jackson.databind.node.ArrayNode runOne(com.fasterxml.jackson.databind.node.ObjectNode input)
                                                              throws Exception

Specified by:

runOne in interface ICorpusBatchStep

Throws:

Exception

getText

protected String getText(org.jsoup.nodes.Element element)

getText

protected String getText(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getTextWithSelector

protected String getTextWithSelector(org.jsoup.nodes.Element element,
                                     String selector)

getAuthor

protected String getAuthor(org.jsoup.nodes.Document doc)

getAuthor

protected String getAuthor(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setAuthor

protected void setAuthor(String author,
                         com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setAuthor

protected void setAuthor(org.jsoup.nodes.Document doc,
                         com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getTitle

protected String getTitle(org.jsoup.nodes.Document doc)

getTitle

protected String getTitle(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setTitle

protected void setTitle(String title,
                        com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setTitle

protected void setTitle(org.jsoup.nodes.Document doc,
                        com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getThumbnail

protected String getThumbnail(org.jsoup.nodes.Document doc)

setThumbnail

protected void setThumbnail(String thumbnail,
                            com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setThumbnail

protected void setThumbnail(org.jsoup.nodes.Document doc,
                            com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getDate

protected String getDate(org.jsoup.nodes.Document doc)

getDate

protected String getDate(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setDate

protected void setDate(String date,
                       com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setDate

protected void setDate(org.jsoup.nodes.Document doc,
                       com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setDate

protected void setDate(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setLink

protected void setLink(String link,
                       com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getLink

protected String getLink(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setStorageLink

protected void setStorageLink(String storage,
                              com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

setDocID

protected void setDocID(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem,
                        String docID)

getDocID

protected String getDocID(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getStorageLink

protected String getStorageLink(com.fasterxml.jackson.databind.node.ObjectNode outputStepItem)

getJsoupDocumentFromStorage

protected org.jsoup.nodes.Document getJsoupDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
                                                        throws Exception

Throws:

Exception

getJsoupDocumentFromStorageNormalized

protected org.jsoup.nodes.Document getJsoupDocumentFromStorageNormalized(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
                                                                  throws Exception

Throws:

Exception

getTextDocumentFromStorage

protected String getTextDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
                                     throws Exception

Throws:

Exception

getTextDocumentFromStorage

protected String getTextDocumentFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem,
                                            String property)
                                     throws Exception

Throws:

Exception

getJsonNodeFromStorage

protected com.fasterxml.jackson.databind.JsonNode getJsonNodeFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem)
                                                                  throws Exception

Throws:

Exception

getJsonNodeFromStorage

protected com.fasterxml.jackson.databind.JsonNode getJsonNodeFromStorage(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem,
                                                                         String property)
                                                                  throws Exception

Throws:

Exception

getOutputScratchFilePathFromInput

protected String getOutputScratchFilePathFromInput(com.fasterxml.jackson.databind.node.ObjectNode inputStepItem,
                                                   String extension)
                                            throws Exception

Throws:

Exception

getOutputScratchFilePath

protected String getOutputScratchFilePath(String fileName)
                                   throws Exception

Throws:

Exception

getOutputScratchFilePath

protected String getOutputScratchFilePath(String fileName,
                                          String extension)
                                   throws Exception

Throws:

Exception

getMimeTypeFromExtension

protected String getMimeTypeFromExtension(String extension)

getExtensionFromMimeType

protected String getExtensionFromMimeType(String mimeType)
                                   throws org.apache.tika.mime.MimeTypeException

Throws:

org.apache.tika.mime.MimeTypeException

isDictionaryWord

protected boolean isDictionaryWord(String pos)