✨ Use dinglehopper's new OCR-D interface

Dockerfile

@@ -20,10 +20,5 @@ RUN pip3 install --no-cache-dir -r /tmp/requirements.txt
 COPY my_ocrd_workflow /usr/bin
 COPY xsd              /usr/bin/xsd
 
-# XXX Hack
-COPY ../experiments/ocr-eval /usr/local/ocr-eval
-RUN pip3 install --no-cache-dir -r /usr/local/ocr-eval/requirements.txt
-
-
 WORKDIR /data
 CMD ["/usr/bin/my_ocrd_workflow"]

my_ocrd_workflow

@@ -155,17 +155,8 @@ do_validate
 
 
 if ocrd workspace list-group | grep -q OCR-D-GT-PAGE; then
-  for g in `ocrd workspace list-page`; do
-    echo "== $g"
-    gt=`ocrd workspace find -G OCR-D-GT-PAGE -g $g`
-    ocr=`ocrd workspace find -G OCR-D-OCR-TESS -g $g`
-
-    if [ -n "$gt" ]; then
-      python3 /usr/local/ocr-eval/cli.py $gt $ocr
-      mv report.json report-$g.json
-      mv report.html report-$g.html
-    fi
-  done
+  remove_filegrp OCR-D-OCR-TESS-EVAL mets.xml
+  ocrd-dinglehopper -m mets.xml -I OCR-D-GT-PAGE,OCR-D-OCR-TESS -O OCR-D-OCR-TESS-EVAL
 fi
 
 # vim:tw=120:

requirements.txt

@@ -6,3 +6,5 @@ https://github.com/mikegerber/ocrd_kraken/archive/fix/pass-down-page-id.tar.gz
 
 tesserocr == 2.3.1  # 2.4.0 fails with Ubuntu 18.04's tesseract
 ocrd_tesserocr
+
+https://github.com/qurator-spk/dinglehopper/archive/0f056b9.tar.gz