<?xml version="1.0" encoding="utf-8"?>
<!-- Licensed Materials - Property of IBM

     IBM SPSS Products: Text Analytics

     (C) Copyright IBM Corp. 2003, 2011

     US Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP
     Schedule Contract with IBM Corp.  -->
<Extension version="1.0" debug="false">
  <ExtensionDetails id="text_mining_web_feed" label="Text Mining Web Feed" version="1.0" providerTag="SPSS Inc." copyright="(c) 2006 SPSS Inc." description="SPSS Web Feed Node" />
  <Resources>
  	<HelpInfo id="help" type="native" path="/ta_guide_ddita/"/>
	<Bundle id="translations.webfeed" type="properties" path="messages"/>    
    <SharedLibrary id="TMWBWebFeed" path="spss.TMWBServer/bin/WebFeed" />
    <JarFile id="RSSDialog" path="./lib/rssDialog.jar" />
    <JarFile id="TableLayout-bin-jdk1.5-2009-08-26" path="../spss.TMWBClient/lib/TableLayout-bin-jdk1.5-2009-08-26.jar"/>
    <JarFile id="NekoHTML" path="./lib/nekohtml.jar" />
    <JarFile id="Commons" path=".lib/commons-collections-3.1.jar" />
    <JarFile id="Jdom" path="./lib/jdom.jar" />
    <JarFile id="Rome" path="./lib/rome-1.0.jar" />
    <JarFile id="WebServerAxis2" path="../spss.TMWBClient/lib/WebServerAxis2.jar" />
  </Resources>
  <Node id="webfeed" type="dataReader" customPalette="text_mining" relativePosition="after" relativeTo="filelist" label="Web Feed" labelKey="webfeednode.LABEL" helpLink="textmining/tmfc/webfeednode_general.htm" description="Reads data from an RSS or HTML URL" descriptionKey="webfeednode_tooltip.LABEL">
    <Properties>
      <!-- =========== Field name localisation =========== -->    
      <Property name="title_field"             valueType="string" defaultValueKey="titleField.LABEL" />
      <Property name="short_description_field" valueType="string" defaultValueKey="shortDescField.LABEL" />
      <Property name="description_field"       valueType="string" defaultValueKey="descriptionField.LABEL" />
      <Property name="authors_field"           valueType="string" defaultValueKey="authorsField.LABEL" />
      <Property name="contributors_field"      valueType="string" defaultValueKey="contributorsField.LABEL" />
      <Property name="published_date_field"    valueType="string" defaultValueKey="publishedDateField.LABEL" />
      <Property name="modified_date_field"     valueType="string" defaultValueKey="modifiedDateField.LABEL" />

      <!-- =========== Input tab =========== -->
      <Property name="urls" valueType="string" />
      <Property name="recent_entries" valueType="boolean" defaultValue="true" />
      <Property name="use_previous" valueType="boolean" defaultValue="false" />
      <Property name="use_previous_label" valueType="string" />
      <Property name="limit_entries" valueType="integer" defaultValue="500" />
      <Property name="temp_update_urls" valueType="boolean" defaultValue="true" />
      <Property name="combo_box_data" valueType="string" />
      <Property name="rss_feed" valueType="string" />

      <!-- =========== Records tab =========== -->
      <Property name="title" valueType="string" />
      <Property name="short_description" valueType="string" />
      <Property name="description" valueType="string" />
      <Property name="authors" valueType="string" />
      <Property name="contributors" valueType="string" />
      <Property name="published_date" valueType="string" />
      <Property name="modified_date" valueType="string" />
      <Property name="start_record" valueType="string" />
      
      <!-- =========== HTML Settings tab =========== -->
      <Property name="html_alg"           valueType="string"  defaultValue="0"/>
      <Property name="discard_lines"      valueType="boolean" defaultValue="true"/>
      <Property name="discard_words"      valueType="boolean" defaultValue="true"/>
      <Property name="discard_scw"        valueType="boolean" defaultValue="true"/>
      <Property name="min_words"	  valueType="integer" defaultValue="10"/>
      <Property name="min_avg_len"        valueType="integer" defaultValue="3"/>
      <Property name="max_scw"            valueType="integer" defaultValue="33"/>
      <Property name="discard_tags"       valueType="boolean" defaultValue="true"/>
      <Property name="discard_spec_words" valueType="boolean" defaultValue="true"/>
      <Property name="tags"               valueType="string"  defaultValue="script,style,span,li,iframe,a"/>
      <Property name="words"              valueType="string"  defaultValue="copyright,&#x00A9;"/>

    </Properties>
    <UserInterface>
      <Icons>
        <Icon type="smallNode" imagePath="sm_webfeed.gif" />
        <Icon type="standardNode" imagePath="lg_webfeed.gif" />
      </Icons>
      <Tabs>
        <Tab label="Input" labelKey="inputTab.LABEL" helpLink="textmining/tmfc/webfeednode_input.htm">
          <ExtensionObjectPanel id="panel1" panelClass="com.spss.rssFeed.ui.InputPanel" />
        </Tab>
        <Tab label="Records" labelKey="recordsTab.LABEL" helpLink="textmining/tmfc/webfeednode_records.htm">
          <ExtensionObjectPanel id="panel2" panelClass="com.spss.rssFeed.ui.RecordsPanel" />
		</Tab>
		<Tab label="Content Filter" labelKey="htmlTab.LABEL" helpLink="textmining/tmfc/webfeednode_content.htm">
          <ExtensionObjectPanel id="panel3" panelClass="com.spss.rssFeed.ui.HTMLSettingsPanel" />
        </Tab>
      </Tabs>
    </UserInterface>
    <Execution>
      <Module libraryId="TMWBWebFeed" name="TMWebFeed">
        <StatusCodes>
          <StatusCode code="0" status="error" message="Initialization Error" messageKey="initError.LABEL"/>
          <StatusCode code="1" status="error" message="Internal Error" messageKey="internalError.LABEL"/>
          <StatusCode code="2" status="error" message="Invalid Metadata" messageKey="invalidMetadataError.LABEL"/>
          <StatusCode code="3" status="error" message="Invalid input data" messageKey="invalidInputDataError.LABEL"/>
          <StatusCode code="4" status="error" message="Invalid number of field parameters" messageKey="invalidNoFieldParametersError.LABEL"/>
          <StatusCode code="5" status="error" message="No data can be found" messageKey="noDataCanBeFoundError.LABEL"/>
          <StatusCode code="6" status="error" message="Error returned from gSoap" messageKey="gSoapError.LABEL"/>
	  

	  <StatusCode code="7"  status="warning" message="Thread canceled"    messageKey="threadCancelled.LABEL"/>
          <StatusCode code="8"  status="success" message="Processing URL"     messageKey="processingURL.LABEL"/>
          <StatusCode code="9"  status="success" message="Processing URL from feed" messageKey="processingURLFromFeed.LABEL"/>
          <StatusCode code="10" status="error"   message="Malformed RSS"      messageKey="MalformedRSS.LABEL"/>
          <StatusCode code="11" status="error"   message="Malformed HTML"     messageKey="MalformedHTML.LABEL"/>
          <StatusCode code="12" status="error"   message="General error"      messageKey="GeneralError.LABEL"/>
          <StatusCode code="13" status="error"   message="Unable to connect to URL" messageKey="UnableConnectURL.LABEL"/>



        </StatusCodes>
      </Module>
    </Execution>
    <!-- This tells Clementine that the node's output data model is based on the input data model  with some extra fields added -->
    <OutputDataModel mode="extend" method="sharedLibrary" libraryId="TMWBWebFeed">
    </OutputDataModel>
  </Node>
</Extension>
