1#!/usr/bin/env bash 2# 3# Re-downloads the test sample files from the Apache Tika test corpus. 4# 5# Usage: ./download.sh 6# Requires: curl. 7 8set -euo pipefail 9 10DATA="$(cd "$(dirname "$0")" && pwd)" 11BASE="https://raw.githubusercontent.com/apache/tika/main/tika-parsers/tika-parsers-standard" 12MS="tika-parsers-standard-modules/tika-parser-microsoft-module/src/test/resources/test-documents" 13MISC="tika-parsers-standard-modules/tika-parser-miscoffice-module/src/test/resources/test-documents" 14PDF="tika-parsers-standard-modules/tika-parser-pdf-module/src/test/resources/test-documents" 15IMG="tika-parsers-standard-modules/tika-parser-image-module/src/test/resources/test-documents" 16XMP="tika-parsers-standard-modules/tika-parser-xmp-commons/src/test/resources/test-documents" 17TXT="tika-parsers-standard-integration-tests/src/test/resources/test-documents" 18 19# fetch <upstream-relative-path> <local-filename> 20fetch() { 21 curl -sSL --fail -o "$DATA/$2" "$BASE/$1" 22 echo " fetched $2" 23} 24 25echo "Downloading samples from Apache Tika..." 26fetch "$MS/testWORD.docx" tika-sample.docx 27fetch "$MS/testWORD_various.docx" tika-various.docx 28fetch "$MS/testEXCEL.xlsx" tika-sample.xlsx 29fetch "$MS/testPPT.pptx" tika-sample.pptx 30fetch "$MS/testPPT_various.pptx" tika-various.pptx 31fetch "$MISC/testOpenOffice2.odt" tika-sample.odt 32fetch "$MISC/versions/LibreOfficeCalc_ods_1.3.ods" tika-sample.ods 33fetch "$MISC/versions/LibreOfficeImpress_odp_1.3.odp" tika-sample.odp 34fetch "$PDF/testPDF.pdf" tika-sample.pdf 35fetch "$XMP/testJPEG_commented_pspcs2mac.jpg" tika-meta.jpg 36fetch "$IMG/testJPEG.jpg" tika-plain.jpg 37fetch "$IMG/testTIFF.tif" tika-meta.tiff 38fetch "$TXT/testTXTNonASCIIUTF8.txt" tika-sample.txt 39 40echo "Done." 41