Wrapper script for Apache Tika, optionally enable Tesseract OCR for a single invocation

tika

#!/bin/bash
##
##  Simple wrapper script for Apache tika
##
##  Usage: download the latest `tika-app*.jar` from [1] to your ~/bin
##  and copy this script alongside. Make executable with
##
##    chmod a+x ~/bin/tika`
##
##  Test with `tika --help`.
##
##  Author:   Kevin Ernst (ernstki -at- mail.uc.edu)
##  Date:     15 Dec 2023
##  License:  WTFPL
##
##  [1]: https://tika.apache.org/download.html

# set TRACE=1 in the environment to enable execution tracing
(( TRACE )) && set -x
set -u

MYDIR=$(cd "$(dirname "$BASH_SOURCE")" && pwd)
# Yes, yes, I know XDG, ~/.config, ~/Library/Application Settings and all that.
CONFIG="$MYDIR"/tika_no_ocr.xml

if [[ $1 =~ ^-(o|-ocr)$ ]]; then
    # default mode of operation, use Tesseract if installed
    if ! which tesseract &>/dev/null; then
        echo "WARNING: Tesseract not found in search path. Tika won't try to OCR file contents." >&2
    fi
    shift
else
    if [[ ! -s "$CONFIG" ]]; then
        # source: https://cwiki.apache.org/confluence/display/TIKA/TikaOCR#TikaOCR-disable-ocr
        echo "Writing new parser "$CONFIG" to exclude TesseractOCRParser." >&2
        cat <<END_OF_XML >"$CONFIG"
<?xml version="1.0" encoding="UTF-8"?>
<properties>
  <parsers>
    <parser class="org.apache.tika.parser.DefaultParser">
      <parser-exclude class="org.apache.tika.parser.ocr.TesseractOCRParser"/>
    </parser>
  </parsers>
</properties>
END_OF_XML
    fi

    echo "Tesseract OCR is disabled for this run; try '-o' / '--ocr' to enable it." >&2
    set -- --config="$CONFIG" "$@"
fi

java -jar "$MYDIR"/tika-app-*.jar "$@"