2023-08-17 18:34:15 +02:00
|
|
|
#!/bin/sh
|
|
|
|
|
set -ex
|
|
|
|
|
|
2023-08-17 18:37:22 +02:00
|
|
|
test_id=`basename $0`
|
|
|
|
|
cd `mktemp -d /tmp/$test_id-XXXXX`
|
2023-08-17 18:34:15 +02:00
|
|
|
|
2024-05-06 23:40:55 +02:00
|
|
|
OCRD_CIS_OCROPY_MODEL=fraktur.pyrnn.gz
|
|
|
|
|
|
|
|
|
|
# Prepare processors
|
|
|
|
|
ocrd resmgr download ocrd-cis-ocropy-recognize $OCRD_CIS_OCROPY_MODEL
|
|
|
|
|
|
2023-08-17 18:34:15 +02:00
|
|
|
# Prepare test workspace
|
2024-05-06 23:40:55 +02:00
|
|
|
wget https://qurator-data.de/examples/actevedef_718448162.first-page.zip
|
|
|
|
|
unzip actevedef_718448162.first-page.zip
|
|
|
|
|
cd actevedef_718448162.first-page
|
|
|
|
|
|
|
|
|
|
# XXX ocrd-cis-ocropy-segment wasn't happy with the binarized input (no
|
|
|
|
|
# "binarized" AlternativeImage?!), so we do it here again
|
|
|
|
|
ocrd-skimage-binarize -I OCR-D-IMG -O OCR-D-IMG-BIN
|
2023-08-17 18:34:15 +02:00
|
|
|
|
|
|
|
|
# Run tests
|
2024-05-06 23:40:55 +02:00
|
|
|
ocrd-cis-ocropy-segment \
|
|
|
|
|
-I OCR-D-IMG-BIN -O TEST-CIS-OCROPY-SEG-LINE \
|
|
|
|
|
-P level-of-operation page
|
|
|
|
|
test "$(grep TextLine TEST-CIS-OCROPY-SEG-LINE/*.xml | wc -l)" -gt 50
|
|
|
|
|
|
|
|
|
|
ocrd-cis-ocropy-recognize \
|
|
|
|
|
-I TEST-CIS-OCROPY-SEG-LINE -O TEST-CIS-OCROPY-OCR \
|
|
|
|
|
-P model $OCRD_CIS_OCROPY_MODEL
|
|
|
|
|
test "$(grep Unicode TEST-CIS-OCROPY-OCR/*.xml | wc -l)" -gt 50
|
