Merge remote-tracking branch 'origin/anybaseocr-layout-analysis'

Author: kba

ocrd/ocrd/resource_list.yml

@@ -133,19 +133,23 @@ ocrd-anybaseocr-block-segmentation:
     description: block segmentation model for anybaseocr
     size: 256139800
 ocrd-anybaseocr-layout-analysis:
-  - url: https://ocr-d-repo.scc.kit.edu/models/dfki/layoutAnalysis/structure_analysis.h5
-    name: structure_analysis.h5
+  - url: https://ocr-d.kba.cloud/structure_analysis.tar.gz
+    name: structure_analysis
     description: structure analysis model for anybaseocr
-    size: 31477056
+    type: tarball
+    path_in_archive: 'structure_analysis'
+    size: 29002514
   - url: https://ocr-d-repo.scc.kit.edu/models/dfki/layoutAnalysis/mapping_densenet.pickle
     name: mapping_densenet.pickle
     description: mapping model for anybaseocr
     size: 374
 ocrd-anybaseocr-tiseg:
-  - url: https://ocr-d-repo.scc.kit.edu/models/dfki/tiseg/seg_model.hdf5
-    name: seg_model.hdf5
+  - url: https://ocr-d.kba.cloud/seg_model.tar.gz
+    name: seg_model
     description: text image segmentation model for anybaseocr
-    size: 66080688
+    type: tarball
+    path_in_archive: 'seg_model'
+    size: 61388872
 ocrd-kraken-segment:
   - url: https://github.com/mittagessen/kraken/raw/master/kraken/blla.mlmodel
     description: Pretrained baseline segmentation model

ocrd/ocrd/resource_manager.py

@@ -249,17 +249,17 @@ def download(
             else:
                 self._copy_impl(url, fpath, progress_cb)
         elif resource_type == 'tarball':
-            with pushd_popd(tempdir=True):
+            with pushd_popd(tempdir=True) as tempdir:
                 if is_url:
                     self._download_impl(url, 'download.tar.xx', progress_cb, size)
                 else:
                     self._copy_impl(url, 'download.tar.xx', progress_cb)
                 Path('out').mkdir()
                 with pushd_popd('out'):
-                    log.info("Extracting tarball")
+                    log.info("Extracting tarball to %s/out" % tempdir)
                     with open_tarfile('../download.tar.xx', 'r:*') as tar:
                         tar.extractall()
-                    log.info("Copying '%s' from tarball to %s" % (path_in_archive, fpath))
+                    log.info("Copying '%s' from extracted tarball %s/out to %s" % (path_in_archive, tempdir, fpath))
                     copytree(path_in_archive, str(fpath))
         # TODO
         # elif resource_type == 'github-dir':