mirror of
https://github.com/qurator-spk/ocrd-galley.git
synced 2025-10-30 02:34:13 +01:00
94 lines
2.3 KiB
Bash
Executable file
94 lines
2.3 KiB
Bash
Executable file
#!/bin/sh
|
|
|
|
|
|
|
|
DATA_SUBDIR=data
|
|
|
|
get_from_annex() {
|
|
annex_get 'calamari-models/GT4HistOCR/2019-07-22T15:49+0200/*.ckpt*'
|
|
annex_get 'tesseract-models/GT4HistOCR/*.traineddata'
|
|
annex_get 'textline_detection/*.h5'
|
|
}
|
|
|
|
get_from_web() {
|
|
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/model.tar.xz' 'calamari-models/GT4HistOCR/2019-07-22T15:49+0200'
|
|
download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar' 'tesseract-models/GT4HistOCR'
|
|
# FIXME needs update download_to 'https://file.spk-berlin.de:8443/textline_detection/models.tar.gz' 'textline_detection'
|
|
}
|
|
|
|
|
|
|
|
check_data_subdir() {
|
|
result=0
|
|
|
|
if git submodule status $DATA_SUBDIR | grep -q '^-'; then
|
|
echo "$DATA_SUBDIR/ is not an initialized submodule"; result=1
|
|
fi
|
|
if ! [ -e $DATA_SUBDIR/.git/annex ]; then
|
|
echo "$DATA_SUBDIR/ is not a git annex repository"; result=1
|
|
fi
|
|
if ! (cd $DATA_SUBDIR && git annex version | grep -q 'local repository version: 7'); then
|
|
echo "$DATA_SUBDIR/ is not a git annex repository version 7"; result=1
|
|
fi
|
|
if ! (cd $DATA_SUBDIR && git remote | grep -q '^nfs$'); then
|
|
echo "$DATA_SUBDIR/ has no git remote 'nfs'"; result=1
|
|
fi
|
|
|
|
return $result
|
|
}
|
|
|
|
suggest_commands() {
|
|
echo "Suggested commands:"
|
|
echo
|
|
echo "git submodule update --init"
|
|
echo "(cd $DATA_SUBDIR && git annex init --version=7)"
|
|
echo "(cd $DATA_SUBDIR && git remote add nfs /<... path to ...>/GitNX-Repository/qurator/qurator-data)"
|
|
}
|
|
|
|
annex_get() {
|
|
file_pattern="$1"
|
|
|
|
(
|
|
cd data
|
|
git annex get $file_pattern
|
|
|
|
# fsck seems to be necessary to fix the files if we're in a submodule
|
|
git annex fsck $file_pattern
|
|
)
|
|
}
|
|
|
|
|
|
download_to() {
|
|
download_source="$1"
|
|
unpack_to="$2"
|
|
|
|
(
|
|
cd data
|
|
tmpf=`mktemp 'tmp.XXXXX'`
|
|
wget -O $tmpf "$download_source"
|
|
mkdir -p "$unpack_to"
|
|
# XXX Unpacking relies on tar -a unpacking any tar compression, might not work everywhere?
|
|
tar -C "$unpack_to" -af $tmpf -xv
|
|
rm -f $tmpf
|
|
)
|
|
}
|
|
|
|
|
|
set -e
|
|
|
|
|
|
if ! check_data_subdir; then
|
|
select choice in "Abort to manually fix $DATA_SUBDIR submodule" "Download data files from the web"; do
|
|
if [ $REPLY = 1 ]; then
|
|
suggest_commands
|
|
exit
|
|
else
|
|
get_from_web
|
|
break
|
|
fi
|
|
done
|
|
else
|
|
get_from_annex
|
|
fi
|
|
|
|
docker build -t my_ocrd_workflow .
|