⚙️ Get tessdata_best from git-annex/our mirror to make downloads more robust

Dockerfile

@@ -63,10 +63,7 @@ RUN pip3 install --no-cache-dir --upgrade pip && \
 RUN mkdir -p /var/lib/calamari-models
 COPY data/calamari-models/GT4HistOCR /var/lib/calamari-models/GT4HistOCR
 RUN mkdir -p $TESSDATA_PREFIX
-RUN curl -sSL -O https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/$TESSDATA_BEST_VERSION.tar.gz && \
+ADD data/mirror/github.com/tesseract-ocr/tessdata_best/archive/${TESSDATA_BEST_VERSION}-repacked.tar.gz $TESSDATA_PREFIX/
-    tar xzf $TESSDATA_BEST_VERSION.tar.gz && \
-    mv tessdata_best-$TESSDATA_BEST_VERSION/* $TESSDATA_PREFIX/ && \
-    rm -rf $TESSDATA_BEST_VERSION.tar.gz
 COPY data/tesseract-models/GT4HistOCR/GT4HistOCR_2000000.traineddata $TESSDATA_PREFIX/
 COPY data/textline_detection /var/lib/textline_detection
 

build

@@ -9,11 +9,13 @@ 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'
+  annex_get 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
 }
 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'
   download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz'     'textline_detection'
+  download_to --no-unpack 'https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' 'mirror/github.com/tesseract-ocr/tessdata_best/archive/'
 }
 . $self_dir/qurator_data_lib.sh
 handle_data