main/office/misc/tesseract/pspec.xml

<?xml version="1.0" ?>
<!DOCTYPE PISI SYSTEM "https://pisilinux.org/projeler/pisi/pisi-spec.dtd">
<PISI>
    <Source>
        <Name>tesseract</Name>
        <Homepage>https://github.com/tesseract-ocr/tesseract</Homepage>
        <Packager>
            <Name>PisiLinux Community</Name>
            <Email>admins@pisilinux.org</Email>
        </Packager>
        <PartOf>office.misc</PartOf>
        <License>APACHE</License>
        <IsA>app:console</IsA>
        <Summary>Tesseract OCR Engine</Summary>
        <Description>This package contains an OCR engine - libtesseract and a command line program - tesseract. Tesseract 4 adds a new neural net (LSTM) based OCR engine which is focused on line recognition, but also still supports the legacy Tesseract OCR engine of Tess</Description>
        <Archive sha1sum="eda6a3958352fd6fc7d77a0d387df647de73f34c" type="targz">https://github.com/tesseract-ocr/tesseract/archive/refs/tags/5.3.1.tar.gz</Archive>
        <BuildDependencies>
            <Dependency>leptonica-devel</Dependency>
            <Dependency>cairo-devel</Dependency>
            <Dependency>icu4c-devel</Dependency>
            <Dependency>pango-devel</Dependency>
            <Dependency>tiff-devel</Dependency>
            <Dependency>libgomp</Dependency>
            <!-- dependecies for documentation -->
            <Dependency>libxslt-devel</Dependency>
            <Dependency>asciidoc</Dependency>
        </BuildDependencies>
    </Source>

    <Package>
        <Name>tesseract</Name>
        <Summary>Tesseract OCR Engine</Summary>
        <RuntimeDependencies>
            <Dependency>tiff</Dependency>
            <Dependency>libgcc</Dependency>
            <Dependency>libgomp</Dependency>
            <Dependency>leptonica</Dependency>
        </RuntimeDependencies>
        <Files>
            <Path fileType="library">/usr/lib/libtesseract.so*</Path>
            <Path fileType="executable">/usr/bin/tesseract</Path>
            <Path fileType="data">/usr/share/tessdata</Path>
            <Path fileType="doc">/usr/share/doc</Path>
            <Path fileType="data">/usr/share</Path>
        </Files>
    </Package>

    <Package>
        <Name>tesseract-devel</Name>
        <Summary>Development files for tesseract</Summary>
        <RuntimeDependencies>
            <Dependency>leptonica-devel</Dependency>
            <Dependency release="current">tesseract</Dependency>
        </RuntimeDependencies>
        <Files>
            <Path fileType="library">/usr/lib/pkgconfig/tesseract.pc</Path>
            <Path fileType="header">/usr/include/tesseract</Path>
        </Files>
    </Package>

    <Package>
        <Name>tesseract-tools</Name>
        <Summary>Tools for tesseract</Summary>
        <RuntimeDependencies>
            <Dependency>cairo</Dependency>
            <Dependency>icu4c</Dependency>
            <Dependency>pango</Dependency>
            <Dependency>glib2</Dependency>
            <Dependency>harfbuzz</Dependency>
            <Dependency>leptonica</Dependency>
            <Dependency>fontconfig</Dependency>
            <Dependency release="current">tesseract</Dependency>
        </RuntimeDependencies>
        <Files>
            <Path fileType="executable">/usr/bin/set_unicharset_properties</Path>
            <Path fileType="executable">/usr/bin/language-specific.sh</Path>
            <Path fileType="executable">/usr/bin/unicharset_extractor</Path>
            <Path fileType="executable">/usr/bin/tesstrain_utils.sh</Path>
            <Path fileType="executable">/usr/bin/combine_lang_model</Path>
            <Path fileType="executable">/usr/bin/classifier_tester</Path>
            <Path fileType="executable">/usr/bin/merge_unicharsets</Path>
            <Path fileType="executable">/usr/bin/combine_tessdata</Path>
            <Path fileType="executable">/usr/bin/shapeclustering</Path>
            <Path fileType="executable">/usr/bin/ambiguous_words</Path>
            <Path fileType="executable">/usr/bin/dawg2wordlist</Path>
            <Path fileType="executable">/usr/bin/wordlist2dawg</Path>
            <Path fileType="executable">/usr/bin/tesstrain.sh</Path>
            <Path fileType="executable">/usr/bin/lstmtraining</Path>
            <Path fileType="executable">/usr/bin/cntraining</Path>
            <Path fileType="executable">/usr/bin/mftraining</Path>
            <Path fileType="executable">/usr/bin/text2image</Path>
            <Path fileType="executable">/usr/bin/lstmeval</Path>
        </Files>
    </Package>

    <History>
        <Update release="5">
            <Date>2026-02-19</Date>
            <Version>5.3.1</Version>
            <Comment>Rebuild</Comment>
            <Name>Kamil Atlı</Name>
            <Email>suvari@pisilinux.org</Email>
        </Update>
        <Update release="3">
            <Date>2023-05-09</Date>
            <Version>5.3.1</Version>
            <Comment>Version bump</Comment>
            <Name>Kamil Atlı</Name>
            <Email>suvari@pisilinux.org</Email>
        </Update>
        <Update release="2">
            <Date>2020-01-19</Date>
            <Version>4.1.1</Version>
            <Comment>Version bump.</Comment>
            <Name>Blue Devil</Name>
            <Email>bluedevil@sctzine.com</Email>
        </Update>
        <Update release="1">
            <Date>2019-01-25</Date>
            <Version>4.0.0</Version>
            <Comment>First release</Comment>
            <Name>PisiLinux Community</Name>
            <Email>admins@pisilinux.org</Email>
        </Update>
    </History>
</PISI>