tesseract.xml

<?xml version="1.0" encoding="UTF-8"?>
<toolspec model="0.1" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="../src/main/resources/toolspec.xsd">
    <id>tesseract</id>
    <name>Tesseract OCR</name>
    <homepage>http://code.google.com/p/tesseract-ocr/</homepage>
    <version>3-0-2</version>
    <installation>
        <os type="linux">
            Available under most distributions via package management.
        </os>
        <os type="windows">
            Binaries for Windows are available from the project homepage. THE FULL PATH TO THE EXECUTABLE MUST NOT CONTAIN WHITESPACES!
        </os>
    </installation>
    <services>
        <service sid="1" name="Tesseract" type="migrate" servicepackage="eu.impact_project.iif.service" contextpathprefix="/impactservices">
            <description>Services for text recognition using Tesseract</description>
            <operations>
                <operation oid="1" name="tesseract">
                    <description>Perform text recognition on image file</description>
                    <!--
                    Usage:
                    tesseract[.exe] <input> <output> -l <lang> [+hocr]

                    Where:
                    <input> 	Input image file in TIF format
                    <output> 	OCR result file in TXT format
                    <lang> 	    Language model used for recognition
                    [+hocr] 	Optional: write hOCR output file (with coordinates)
                    -->
                    <command>tesseract ${input} ${output} -l ${langmod}</command>
                    <inputs>
                        <input name="input">
                            <Datatype>xsd:anyURI</Datatype>
                            <Required>true</Required>
                            <CliMapping>input</CliMapping>
                            <Documentation>URL reference to input file</Documentation>
                            <Default>http://github.com/impactcentre/iif-testfiles/raw/master/testfiles/image.bw.tif</Default>
                        </input>
                        <input name="langmod">
                            <Datatype>xsd:string</Datatype>
                            <Required>true</Required>
                            <CliMapping>langmod</CliMapping>
                            <Documentation>Language module used by Tesseract</Documentation>
                            <Default>deu-frak</Default>
                            <Restriction>
                                <Value>afr</Value>
                                <Value>ara</Value>
                                <Value>aze</Value>
                                <Value>bel</Value>
                                <Value>ben</Value>
                                <Value>bul</Value>
                                <Value>cat</Value>
                                <Value>ces</Value>
                                <Value>chi_sim</Value>
                                <Value>chi_tra</Value>
                                <Value>chr</Value>
                                <Value>dan</Value>
                                <Value>dan-frak</Value>
                                <Value>deu</Value>
                                <Value>deu-frak</Value>
                                <Value>ell</Value>
                                <Value>eng</Value>
                                <Value>enm</Value>
                                <Value>epo</Value>
                                <Value>equ</Value>
                                <Value>est</Value>
                                <Value>eus</Value>
                                <Value>fin</Value>
                                <Value>fra</Value>
                                <Value>frk</Value>
                                <Value>frm</Value>
                                <Value>glg</Value>
                                <Value>grc</Value>
                                <Value>heb</Value>
                                <Value>hin</Value>
                                <Value>hrv</Value>
                                <Value>hun</Value>
                                <Value>ind</Value>
                                <Value>isl</Value>
                                <Value>ita</Value>
                                <Value>ita-old</Value>
                                <Value>jpn</Value>
                                <Value>kan</Value>
                                <Value>kor</Value>
                                <Value>lat-lid</Value>
                                <Value>lav</Value>
                                <Value>lit</Value>
                                <Value>mal</Value>
                                <Value>mkd</Value>
                                <Value>mlt</Value>
                                <Value>msa</Value>
                                <Value>nld</Value>
                                <Value>nor</Value>
                                <Value>osd</Value>
                                <Value>pol</Value>
                                <Value>por</Value>
                                <Value>ron</Value>
                                <Value>rus</Value>
                                <Value>slk</Value>
                                <Value>slk-frak</Value>
                                <Value>slv</Value>
                                <Value>spa</Value>
                                <Value>spa-old</Value>
                                <Value>srp</Value>
                                <Value>swe</Value>
                                <Value>swe-frak</Value>
                                <Value>tha</Value>
                                <Value>tgl</Value>
                                <Value>tur</Value>
                                <Value>ukr</Value>
                                <Value>vie</Value>
                            </Restriction>
                        </input>
                    </inputs>
                    <outputs>
                        <output name="output">
                            <Datatype>xsd:anyURI</Datatype>
                            <Required>false</Required>
                            <CliMapping>output</CliMapping>
                            <Documentation>URL reference to output file</Documentation>
                            <Extension>txt</Extension>
                            <AutoExtension>true</AutoExtension>
                        </output>
                    </outputs>
                </operation>
            </operations>
            <deployto>
                <deployref default="true" ref="local"/>
            </deployto>
        </service>
    </services>
    <deployments>
        <deployment id="local">
            <identifier>http://localhost:8080/impact/instances/tomcat1</identifier>
            <host>localhost</host>
            <ports>
                <port type="http">8080</port>
                <port type="https">8043</port>
            </ports>
            <manager>
                <user>tomcat</user>
                <password>tomcat</password>
                <path>manager</path>
            </manager>
            <!--
            Full path to the directory where the tool has been installed.
            THE PATH MUST NOT CONTAIN WHITESPACES!
            If you want to use backslashes (\) then you must repeat them twice,
            e.g. c:\\foo\\bar
            or you can use just slashes,
            e.g. c:/foo/bar
            -->
            <toolsbasedir>C:\\TESSOCR\\</toolsbasedir>
            <dataexchange>
                <accessdir>../webapps/ROOT/impact/tmp/</accessdir>
                <accessurl>http://localhost:8080/impact/tmp/</accessurl>
            </dataexchange>
        </deployment>
    </deployments>
</toolspec>