Jump to: navigation, search

Difference between revisions of "SMILA/Documentation/Bundle org.eclipse.smila.processing.pipelets"

(PushRecordsPipelet)
Line 754: Line 754:
  
 
This pipelet can be used to write the content of an attachment to a file.
 
This pipelet can be used to write the content of an attachment to a file.
 +
 +
If the attachment does not exist a warning is logged, but the record will not be dropped.
  
 
=== Configuration ===
 
=== Configuration ===
Line 781: Line 783:
  
 
=== Example ===
 
=== Example ===
 +
 +
This example saves all bytes of the attachment "content" to the file "/use/home/example.txt".
  
 
<source lang="xml">
 
<source lang="xml">
Line 790: Line 794:
 
       <rec:Val key="pathAttribute">/usr/home/example.txt</rec:Val>
 
       <rec:Val key="pathAttribute">/usr/home/example.txt</rec:Val>
 
       <rec:Val key="contentAttachment">content</rec:Val>
 
       <rec:Val key="contentAttachment">content</rec:Val>
 +
    </proc:configuration>
 +
  </proc:invokePipelet>
 +
</extensionActivity>
 +
</source>
 +
 +
== org.eclipse.smila.processing.pipelets.PushRecordsPipelet ==
 +
 +
=== Description ===
 +
 +
Sends all current records to another (asynchronous) job.
 +
 +
The records are not removed from the pipeline - thus a following pipelet in the current pipeline will process the records as well.
 +
 +
=== Configuration ===
 +
 +
{| border="1"
 +
!Property
 +
!Type
 +
!Read Type
 +
!Description
 +
|-
 +
|''type''
 +
|String
 +
|init
 +
|The name of the target job.
 +
|-
 +
|}
 +
 +
=== Example ===
 +
 +
This example sends all current records to the job "TheOtherJob".
 +
 +
<source lang="xml">
 +
<extensionActivity>
 +
  <proc:invokePipelet name="callJob">
 +
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.PushRecordsPipelet" />
 +
    <proc:variables input="request" />
 +
    <proc:configuration>
 +
      <rec:Val key="job">TheOtherJob</rec:Val>
 
     </proc:configuration>
 
     </proc:configuration>
 
   </proc:invokePipelet>
 
   </proc:invokePipelet>

Revision as of 08:34, 20 April 2012

This page describes the SMILA pipelets provided by bundle org.eclipse.smila.processing.pipelets.

Contents

General

All pipelets in this bundle support the configurable error handling as described in SMILA/Development_Guidelines/How_to_write_a_Pipelet#Implementation. When used in jobmanager workflows, records causing errors are dropped.

Read Type

  • runtime: Parameters are read when processing records. Parameter value can be set per Record.
  • init: Parameters are read once from Pipelet configuration when initializing the Pipelet. Parameter value can not be overwritten in Record.

org.eclipse.smila.processing.pipelets.CommitRecordsPipelet

Description

Commits each record in the input variable on the blackboard to the storages. Can be used to save the records immediately during the workflow instead of only when a workflow has been finished.

Configuration

none.

org.eclipse.smila.processing.pipelets.AddValuesPipelet

Adds something to an attribute in the processed records. If the attribute does not contain a sequence already, the current value will be wrapped in one before the new values are added.

Configuration

Property Type Read Type Description
outputAttribute string runtime The name of the attribute to add values to
valuesToAdd Anything, usually a value or a sequence of values runtime The values to add

Example

From a test pipeline: This adds two string values to whatever already exists in attribute "out" of the processed records.

<proc:invokePipelet name="addValuesToNonExistingAttribute">
  <proc:pipelet class="org.eclipse.smila.processing.pipelets.AddValuesPipelet" />
  <proc:variables input="request" />
  <proc:configuration>
  <rec:Val key="outputAttribute">out</rec:Val>
    <rec:Seq key="valuesToAdd">
      <rec:Val>value1</rec:Val>
      <rec:Val>value2</rec:Val>
    </rec:Seq>
  </proc:configuration>
</proc:invokePipelet>

org.eclipse.smila.processing.pipelets.SetValuePipelet

Sets a value for an attribute in every processed records. If the attribute exists already, it is not change by default. Useful for initializations of required attributes.

Configuration

Property Type Read Type Description
outputAttribute string runtime The name of the attribute to set the value for
value anything runtime The constant value to set for the attribute (a map or sequence is possible, too)
overwrite boolean runtime Indicates to overwrite any value that the attribute contains already (optional, defaults to false)

Example

This sets a map containing two values into attribute1, even if there is already a value in that attribute.

<extensionActivity>
  <proc:invokePipelet name="setMapForExistingAttribute">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.SetValuePipelet" />
    <proc:variables input="request" />
    <proc:configuration>
      <rec:Val key="outputAttribute">attribute1</rec:Val>
      <rec:Val key="overwrite" type="boolean">true</rec:Val>
      <rec:Map key="value">
        <rec:Val key="key1">value1</rec:Val>
        <rec:Val key="key2">value2</rec:Val>
      </rec:Map>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.RemoveAttributePipelet

Removes an attribute from each record.

Configuration

The configuration property is either read from the _parameters attribute of a record or from the pipelet configuration. If not set at all, the record remains unchanged.

Property Type Read Type Description
removeAttribute A string value runtime The name of the attribute to remove

Example

To remove the complete structure in attribute _parameters, use:

<extensionActivity>
  <proc:invokePipelet name="removeParameters">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.RemoveAttributePipelet" />
    <proc:variables input="result" output="result" />
    <proc:configuration>
     <rec:Val key="removeAttribute">_parameters</rec:Val>
    </proc:configuration>        
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.FilterPipelet

Copies only those record IDs to the result which match a configurable regular expression in a configurable single-valued attribute. This is useful for conditional processing while at the same time pushing multiple records through the pipeline in a single request: Instead of using BPEL conditions use a FilterPipelet to select only the matching records in a new variable and use the this variable as the input variable for the next pipelets. You can still use the original BPEL variable in the BPEL <reply> activity at the end of the pipeline to return all records as the final result.

Configuration

The configuration properties are read either from the _parameters attribute of each record or from the pipelet configuration.

Property Type Read Type Description
filterAttribute A string value runtime The name of the attribute to match
filterExpression A string value runtime The regular expression to match the attribute value against

Example

To get only those records in the textRecords BPEL variable that have a MimeType starting with text something like this could be used:

<extensionActivity>
  <proc:invokePipelet name="invokeFilterPipelet">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.FilterPipelet" />
    <proc:variables input="request" output="textRecords" />
    <proc:configuration>
      <rec:Val key="filterAttribute">MimeType</rec:Val>
      <rec:Val key="filterExpression">text/.+</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.HtmlToTextPipelet

Description

Extract plain text and metadata from an HTML document from an attribute or attachment of each record and writes the results to configurable attributes or attachments.

The pipelet uses the CyberNeko HTML parser NekoHTML to parse HTML documents.

Configuration

Property Type Read Type Description
inputType String : ATTACHMENT, ATTRIBUTE runtime Defines whether the HTML input is found in an attachment or in an attribute of the record
outputType String : ATTACHMENT, ATTRIBUTE runtime Defines whether the plain text should be stored in an attachment or in an attribute of the record
inputName String runtime Name of input attachment or path to input attribute (process literals of attribute)
outputName String runtime Name of output attachment or path to output attribute for plain text (store result as literals of attribute)
defaultEncoding String runtime Optional, default encoding to apply to documents when not specified in the documents themselves
removeContentTags String runtime Comma-separated list of HTML tags (case-insensitive) for which the complete content should be removed from the resulting plain text. If not set, it defaults to "applet,frame,object,script,style". If the value is set, you must add the default tags explicitly to have their contents removed, too.
meta:<name> String: attribute path init Store the content of the <META> tag with name="<name>" (case insensitive) to the attribute named as the value of the property. E.g. a property named "meta:author" with value "authors" causes the content attributes of <META name="author" content="..."> tags to be stored in the attribute authors of the respective record.
tag:title String: attribute path init Store the content of the <TITLE> tag with to the attribute named as the value of the property.

Example

This configuration extracts plain text from the HTML document in attachment "html" and stores the results to the attribute "text". It removes the complete content of heading tags <h1>, ..., <h4>. In addition to that, it looks for <meta> tags with names "author" and "keywords" and stores their contents in attributes "authors" and "keywords", respectively:

<extensionActivity>
  <proc:invokePipelet name="invokeHtml2Txt">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.HtmlToTextPipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="inputType">ATTACHMENT</rec:Val>
      <rec:Val key="outputType">ATTRIBUTE</rec:Val>
      <rec:Val key="inputName">html</rec:Val>
      <rec:Val key="outputName">text</rec:Val>
      <rec:Val key="defaultEncoding">UTF-8</rec:Val>
      <rec:Val key="meta:author">author</rec:Val>
      <rec:Val key="meta:keywords">keywords</rec:Val>
      <rec:Val key="meta:title">title</rec:Val>
      <rec:Val key="removeContentTags">h1,h2,h3,h4</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.CopyPipelet

Description

This pipelet can be used to copy or move attribute values to other attributes or to copy or move a string value between attributes and/or attachments. It suppoprts two execution modes:

  • COPY: copy the value from the input attribute/attachment to the output attribute/attachment
  • MOVE: same as COPY, but after that delete the value from the input attribute/attachment

When an attribute is copied to another attribute, the type remains the same. When copying an attachment to an attribute, a string value is created by assuming the the attachment is a text in UTF-8 encoding. When copying an attribute value to an attachment, the attribute must be single value which is interpreted as a string value and converted to a byte array using UTF-8 encoding.

Configuration

Property Type Read Type Description
inputType String : ATTACHMENT, ATTRIBUTE runtime selects if the input is found in an attachment or attribute of the record
outputType String : ATTACHMENT, ATTRIBUTE runtime selects if output should be stored in an attachment or attribute of the record
inputName String runtime name of input attachment or input attribute
outputName String runtime name of output attachment or output attribute
mode String : COPY, MOVE runtime execution mode. Copy the value or move (copy and delete) the value. Default is COPY.

Example

This configuration shows how to copy the value of attachment 'Content' into the attribute 'TextContent':

<!-- copy txt from attachment to attribute -->
<extensionActivity>
  <proc:invokePipelet name="invokeCopyContent">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.CopyPipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="inputType">ATTACHMENT</rec:Val>
      <rec:Val key="outputType">ATTRIBUTE</rec:Val>
      <rec:Val key="inputName">Content</rec:Val>
      <rec:Val key="outputName">TextContent</rec:Val>
      <rec:Val key="mode">COPY</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.SubAttributeExtractorPipelet

Description

Extracts literal values from an attribute that has a nested map. The attributes in the nested map can have nested maps themselves. To address a attribute in the nested structure, a path needs to be specified. The pipelet supports different execution modes:

  • FIRST: selects only the first literal of the specified attribute
  • LAST: selects only the last literal of the specified attribute
  • ALL_AS_LIST: selects all literal values of the specified attribute and returns a list
  • ALL_AS_ONE: selects all literal values of the specified attribute and concatenates them to a single string, using a separator (default is blank)

This pipelet works only on attributes, not on attachments!

Note: If the maps on the path are nested in sequences, the pipelet uses the first element of such a sequence.

Configuration

Property Type Read Type Description
inputPath String runtime the path to the input attribute with Literals
outputPath String runtime the name of the attribute to store the extracted value(s) as Literals in (not a path, only a top-level attribute, currently)
mode String : FIRST, LAST, ALL_AS_LIST, ALL_AS_ONE runtime execution mode. See above for details.
separator String runtime the separation string used for mode ALL_AS_ONE. Default is a blank

Example

This configuration can be applied to records provided by the FeedAgent. It shows how to access the subattribute 'Value' of attribute 'Contents', concatenating all values to one:

<!-- extract content -->
<extensionActivity>
  <proc:invokePipelet name="extract content">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.SubAttributeExtractorPipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="inputPath">Contents/Value</rec:Val>
      <rec:Val key="outputPath">Content</rec:Val>
      <rec:Val key="mode">ALL_AS_ONE</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.ReplacePipelet

Description

Searches for one or more patterns in the literal value of an attribute and substitutes the found occurrences by the configured replacements.

You can choose from different matching types:

  • entity: Every pattern is matched against the whole attribute value (with respect to the ignoreCase property) and the first matching pattern defines the new value of the attribute. If no pattern matches, the result is the current value of the attribute.
  • substring: All patterns that are part of the attribute value are replaced.
  • regexp: Interpret all patterns as regular expression, see Matcher#replaceAll(String)

This pipelet works only on attributes, not on attachments!

Configuration

Property Type Read Type Description
inputAttribute String runtime the name of the attribute that contains the literal to search in
outputAttribute String runtime the name of the attribute to store the result value as string, defaults to the input attribute
type String : entity, substring, regexp init Identifies the type of the pattern, see above for details. Defaults to substring.
ignoreCase Boolean init indicates that the case is ignored when matching patterns, defaults to false.
mapping Map init A mapping of multiple patterns and replacements. Each key is a pattern and its value the replacement.
pattern String init the pattern to apply to the literal value (see above for a description of possible types), required if no mapping is given
replacement String init the substitution string used to replace all occurrences of the pattern, defaults to the empty string

Examples

This configuration can be used to map language ids to their label:

<extensionActivity>
  <proc:invokePipelet name="set language label">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.ReplacePipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="inputAttribute">Language</rec:Val>
      <rec:Val key="outputAttribute">LanguageLabel</rec:Val>
      <rec:Val key="type">entity</rec:Val>
      <rec:Val key="ignoreCase" type="boolean">true</rec:Val>
      <rec:Map key="mapping">
        <rec:Val key="de">German</rec:Val>
        <rec:Val key="en">English</rec:Val>
        <rec:Val key="es">Spanish</rec:Val>
        <rec:Val key="fr">French</rec:Val>
        ...
      </rec:Map>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

This configuration can be used to cut the time information from a timestamp:

<extensionActivity>
  <proc:invokePipelet name="cut time">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.ReplacePipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="inputAttribute">ModificationTime</rec:Val>
      <rec:Val key="outputAttribute">ModificationDate</rec:Val>
      <rec:Val key="type">regexp</rec:Val>
      <rec:Val key="pattern">[T ].*</rec:Val>
      <rec:Val key="replacement"></rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.ScriptPipelet

Description

Executes a script for each record.

For execution the Java Scripting API (JSR 223) is responsible - thus any compatible scripting engine can be used. JavaScript is available "out of the box" and the default script language.

The context of the script will contain four variables:

  • blackboard: a reference to the blackboard
  • id: the ID of the current record
  • record: the metadata of the current record
  • results: a slightly modified version of a result collector that provides methods to add a new record id to the list of result ids (results.addResult('...id...')) and to drop the current record from the same list (results.excludeCurrentRecord())

Please be aware that the intention of this pipelet is to write pipelines fast, but not to write fast pipelines - the script is parsed for every record. Don't use it for production environments where performance matters, but use it to develop an algorithm that you can put into your own pipelet.

Configuration

Property Type Read Type Description
type String init the mime type of the scripting language, defaults to "text/javascript"
scriptFile String runtime the path of the file that contains the script - modifications of this file are observed on every execution of the pipelet
script String init The "inline" script, required unless scriptFile is specified (ignored in that case)
resultAttribute String runtime The name of an attribute that will receive the result of the script (usually the result of the last expression)

Examples

This configuration can be used to concatenate the values of two attributes and save the result into a third one:

<extensionActivity>
  <proc:invokePipelet name="create full name">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.ScriptPipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="script">record.get("firstName") + " " + record.get("lastName")</rec:Val>
      <rec:Val key="resultAttribute">fullName</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

This configuration can be used to execute a java script file from $SMILA_PATH$/configuration/example.js:

<extensionActivity>
  <proc:invokePipelet name="execute script">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.ScriptPipelet" />
    <proc:variables input="request" output="request" />
    <proc:configuration>
      <rec:Val key="scriptFile">configuration/example.js</rec:Val>    
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.ExecPipelet

Description

Executes an external program for each record.

This pipelet may be used to integrate native programs into the pipeline.

Attention: This pipelet may lead to security issues! Please be aware that although one can not change the executed command during runtime (as this parameter is only evaluated at initialization time), it is possible to change the arguments and input of the command using values in the processed record. Every "pipeline developer" should ensure that only arguments in the expected value range are processed (especially if the program is allowing files from the file system as arguments).

Configuration

Property Type Read Type Description
command String init The program to execute (including its path in the file system).
directory String runtime The (optional) working directory for the command. The SMILA directory is used if not given.
parameters Sequence of strings runtime The optional parameters given to the program (ignored if the contents of the parameters attribute exists).
parametersAttribute String runtime The optional name of the attribute that contains the sequence of parameters given to the program.
inputAttachment String runtime The optional name of the attachment that contains the bytes to send as input for the program.
outputAttachment String runtime The optional name of the attachment that is filled with the standard output of the program.
exitCodeAttribute String runtime The name of the attribute that is filled with the exit code of the program.
errorAttachment String runtime The optional name of the attachment that is filled with the error output of the program.
failOnError Either a boolean or a sequence of strings runtime Indicates to mark a record as failed if the program returns an error code. Either as a sequence of exit code ranges or as a boolean where "true" means that everything except 0 is an error code. Defaults to false.

Examples

This configuration can be used to execute FFMPEG for transformation of an MP3 input file into a WAV output file:

<extensionActivity>
  <proc:invokePipelet name="ConvertMP3">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.ExecPipelet" />
    <proc:variables input="request" />
    <proc:configuration>
      <rec:Val key="command">.../ffmpeg</rec:Val>
      <rec:Seq key="parameters">
          <rec:Val>-i</rec:Val>
          <rec:Val>.../example.mp3</rec:Val>
          <rec:Val>-ar</rec:Val>
          <rec:Val>16000</rec:Val>
          <rec:Val>.../example.wav</rec:Val>
      </rec:Seq>
      <rec:Val key="failOnError" type="boolean">true</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.MimeTypeIdentifyPipelet

Description

This pipelet is used to identify the MIME type of a document. It uses an org.eclipse.smila.processing.pipelets.mimetype.MimeTypeIdentifier service to perform the actual identification of the MIME type. Depending on the specified properties, the MIME type is detected from the file content, from the file extension, or from both. If the identification does not return a MIME type - and if configured accordingly - the service will search the metadata for this information. The identified MIME type is then stored to an attribute in the record.

Configuration

The pipelet is configured using the <configuration> section inside the <invokePipelet> activity of the corresponding BPEL file. It provides the following properties:

Property Type Read Type Usage Description
FileExtensionAttribute String init Optional Name of the attribute containing the file extension
ContentAttachment String init Optional Name of the attachment containing the file content
MetaDataAttribute String init Optional Name of the attribute containing metadata information, e.g. a Web Crawler returns a response header containing applicable MIME type information
MimeTypeAttribute String init Required Name of the attribute to store the identified MIME type to

Note that at least one of the properties FileExtensionAttribute, ContentAttachment, and MetaDataAttribute must be specified!

Example

The following example is used in the SMILA example application to identify the MIME types of documents that are delivered by the File System Crawler or Web Crawler.

addpipeline.bpel

<extensionActivity>
    <proc:invokePipelet name="detect MimeType">
        <proc:pipelet class="org.eclipse.smila.processing.pipelets.MimeTypeIdentifyPipelet" />
        <proc:variables input="request" output="request" />
        <proc:configuration>
          <rec:Val key="FileExtensionAttribute">Extension</rec:Val>
          <rec:Val key="MetaDataAttribute">MetaData</rec:Val>
          <rec:Val key="MimeTypeAttribute">MimeType</rec:Val>
        </proc:configuration>
    </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.FileReaderPipelet

Description

This pipelet can be used to read content from a file and add it as an attachment.

Configuration

Property Type Read Type Description
pathAttribute String runtime The name of the attribute with the path of the file to read from
contentAttachment String runtime The name of the attachment to store the content

Example

<!-- read from file and add attachment -->
<extensionActivity>
  <proc:invokePipelet name="invokeReadFile">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.FileReaderPipelet" />
    <proc:variables input="request" />
    <proc:configuration>
      <rec:Val key="pathAttribute">path</rec:Val>
      <rec:Val key="contentAttachment">content</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.FileWriterPipelet

Description

This pipelet can be used to write the content of an attachment to a file.

If the attachment does not exist a warning is logged, but the record will not be dropped.

Configuration

Property Type Read Type Description
pathAttribute String runtime The name of the attribute with the path of the target file
contentAttachment String runtime The name of the attachment to write to the file
append Boolean runtime Indicates to append the attachment to the file (if it exists already), defaults to false

Example

This example saves all bytes of the attachment "content" to the file "/use/home/example.txt".

<extensionActivity>
  <proc:invokePipelet name="writeFile">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.FileWriterPipelet" />
    <proc:variables input="request" />
    <proc:configuration>
      <rec:Val key="pathAttribute">/usr/home/example.txt</rec:Val>
      <rec:Val key="contentAttachment">content</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

org.eclipse.smila.processing.pipelets.PushRecordsPipelet

Description

Sends all current records to another (asynchronous) job.

The records are not removed from the pipeline - thus a following pipelet in the current pipeline will process the records as well.

Configuration

Property Type Read Type Description
type String init The name of the target job.

Example

This example sends all current records to the job "TheOtherJob".

<extensionActivity>
  <proc:invokePipelet name="callJob">
    <proc:pipelet class="org.eclipse.smila.processing.pipelets.PushRecordsPipelet" />
    <proc:variables input="request" />
    <proc:configuration>
      <rec:Val key="job">TheOtherJob</rec:Val>
    </proc:configuration>
  </proc:invokePipelet>
</extensionActivity>

LanguageLabel