Merge branch 'master' into fix-rotation

.dockerignore

@@ -1,7 +1,5 @@
 *
-!ocrd
+!ocrd*
 !Makefile
-!setup.py
-!requirements.txt
 !LICENSE
-!README.rst
+!README.md

CHANGELOG.md

@@ -6,6 +6,9 @@ Versioned according to [Semantic Versioning](http://semver.org/).
 ## Unreleased
 
 * image_from_page etc: allow filling with background or transparency
+
+## [1.0.0] - 2019-10-18
+
 * Workspace validation: Validate that files mentioned in pc:Page/@imageFilename exist in METS and on FS, #309
 * `ocrd ocrd-tool parse-params` has the string-or-filepath logic for -p/--parameter as for the [CLI](https://ocr-d.github.io/cli#-p---parameter-param_json)
 
@@ -588,6 +591,9 @@ Fixed
 Initial Release
 
 <!-- link-labels -->
+[1.0.0]: ../../compare/v1.0.0...v1.0.0b19
+[1.0.0b19]: ../../compare/v1.0.0b19...v1.0.0b18
+[1.0.0b18]: ../../compare/v1.0.0b18...v1.0.0b17
 [1.0.0b17]: ../../compare/v1.0.0b17...v1.0.0b16
 [1.0.0b16]: ../../compare/v1.0.0b16...v1.0.0b15
 [1.0.0b15]: ../../compare/v1.0.0b15...v1.0.0b14

Dockerfile

@@ -1,4 +1,4 @@
-FROM ubuntu:18.04
+FROM ubuntu:19.10
 MAINTAINER OCR-D
 ENV DEBIAN_FRONTEND noninteractive
 ENV PYTHONIOENCODING utf8
@@ -7,23 +7,25 @@ ENV LANG C.UTF-8
 
 WORKDIR /build-ocrd
 COPY ocrd ./ocrd
+COPY ocrd_modelfactory ./ocrd_modelfactory/
+COPY ocrd_models ./ocrd_models
+COPY ocrd_utils ./ocrd_utils
+COPY ocrd_validators/ ./ocrd_validators
 COPY Makefile .
-COPY setup.py .
-COPY requirements.txt .
-COPY README.rst .
+COPY README.md .
 COPY LICENSE .
 RUN apt-get update && \
     apt-get -y install --no-install-recommends \
     ca-certificates \
     make \
     sudo \
     git \
-    libglib2.0.0 \
-    libsm6 \
-    libxrender1 \
-    libxext6
+    libglib2.0.0
+    # libxext6
+    # libsm6 \
+    # libxrender1 \
 RUN make deps-ubuntu
 RUN pip3 install --upgrade pip
-RUN make deps install
+RUN make deps-ubuntu install
 
 ENTRYPOINT ["/usr/local/bin/ocrd"]

Makefile

@@ -16,6 +16,9 @@ BUILD_ORDER = ocrd_utils ocrd_models ocrd_modelfactory ocrd_validators ocrd
 
 FIND_VERSION = grep version= ocrd_utils/setup.py|grep -Po "([0-9ab]+\.?)+"
 
+# Additional arguments to docker build. Default: '$(DOCKER_ARGS)'
+DOCKER_ARGS = 
+
 # BEGIN-EVAL makefile-parser --make-help Makefile
 
 help:
@@ -44,6 +47,7 @@ help:
 	@echo "  Variables"
 	@echo ""
 	@echo "    PAGE_VERSION  PAGE schema version to use. Default: '$(PAGE_VERSION)'"
+	@echo "    DOCKER_ARGS   Additional arguments to docker build. Default: '$(DOCKER_ARGS)'"
 	@echo "    DOCKER_TAG    Docker tag."
 	@echo "    PIP_INSTALL   pip install command. Default: $(PIP_INSTALL)"
 
@@ -57,7 +61,7 @@ PIP_INSTALL = pip install
 
 # Dependencies for deployment in an ubuntu/debian linux
 deps-ubuntu:
-	sudo apt install -y python3 python3-pip
+	sudo apt-get install -y python3 python3-pip
 
 # Install test python deps via pip
 deps-test:
@@ -181,7 +185,7 @@ pyclean:
 
 # Build docker image
 docker:
-	docker build -t $(DOCKER_TAG) .
+	docker build -t $(DOCKER_TAG) $(DOCKER_ARGS) .
 
 #
 # bash library

ocrd/ocrd/cli/workspace.py

@@ -47,7 +47,7 @@ def workspace_cli(ctx, directory, mets_basename, backup):
 ''')
 @pass_workspace
 @click.option('-a', '--download', is_flag=True, help="Download all files")
-@click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url']))
+@click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url']))
 @click.option('--page-strictness', help="How strict to check PAGE consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
 @click.argument('mets_url')
 def validate_workspace(ctx, mets_url, download, skip, page_strictness):

ocrd/ocrd/cli/zip.py

@@ -77,7 +77,7 @@ def spill(dest, src):
     """
     resolver = Resolver()
     workspace_bagger = WorkspaceBagger(resolver)
-    workspace = workspace_bagger.spill(src, directory)
+    workspace = workspace_bagger.spill(src, dest)
     print(workspace)
 
 # ----------------------------------------------------------------------

ocrd/ocrd/workspace_bagger.py

@@ -1,5 +1,5 @@
 from datetime import datetime
-from os import makedirs, chdir, walk, getcwd
+from os import makedirs, chdir, walk
 from os.path import join, isdir, basename, exists, relpath
 from shutil import make_archive, rmtree, copyfile, move
 from tempfile import mkdtemp
@@ -16,9 +16,12 @@
     is_local_filename,
     unzip_file_to_dir,
 
+    MIMETYPE_PAGE,
     VERSION,
 )
 from ocrd_validators.constants import BAGIT_TXT, TMP_BAGIT_PREFIX, OCRD_BAGIT_PROFILE_URL
+from ocrd_modelfactory import page_from_file
+from ocrd_models.ocrd_page import to_xml
 
 from .workspace import Workspace
 
@@ -58,9 +61,11 @@ def _log_or_raise(self, msg):
 
     def _bag_mets_files(self, workspace, bagdir, ocrd_manifestation_depth, ocrd_mets, processes):
         mets = workspace.mets
+        changed_urls = {}
 
         # TODO allow filtering by fileGrp@USE and such
         with pushd_popd(workspace.directory):
+            # URLs of the files before changing
             for f in mets.find_files():
                 log.info("Resolving %s (%s)", f.url, ocrd_manifestation_depth)
                 if is_local_filename(f.url):
@@ -77,15 +82,40 @@ def _bag_mets_files(self, workspace, bagdir, ocrd_manifestation_depth, ocrd_mets
                 file_grp_dir = join(bagdir, 'data', f.fileGrp)
                 if not isdir(file_grp_dir):
                     makedirs(file_grp_dir)
-                self.resolver.download_to_directory(file_grp_dir, f.url, basename="%s%s" % (f.ID, f.extension))
-                f.url = join(f.fileGrp, f.ID + f.extension)
+
+                _basename = "%s%s" % (f.ID, f.extension)
+                _relpath = join(f.fileGrp, _basename)
+                self.resolver.download_to_directory(file_grp_dir, f.url, basename=_basename)
+                changed_urls[f.url] = _relpath
+                f.url = _relpath
 
             # save mets.xml
             with open(join(bagdir, 'data', ocrd_mets), 'wb') as f:
                 f.write(workspace.mets.to_xml())
 
+        # Walk through bagged workspace and fix the PAGE
+        # Page/@imageFilename and
+        # AlternativeImage/@filename
+        bag_workspace = Workspace(self.resolver, directory=join(bagdir, 'data'))
+        with pushd_popd(bag_workspace.directory):
+            for page_file in bag_workspace.mets.find_files(mimetype=MIMETYPE_PAGE):
+                pcgts = page_from_file(page_file)
+                changed = False
+                #  page_doc.set(imageFileName
+            #  for old, new in changed_urls:
+                for old, new in changed_urls.items():
+                    if pcgts.get_Page().imageFilename == old:
+                        pcgts.get_Page().imageFilename = new
+                        changed = True
+                    # TODO replace AlternativeImage, recursively...
+                if changed:
+                    with open(page_file.url, 'w') as out:
+                        out.write(to_xml(pcgts))
+                    #  log.info("Replace %s -> %s in %s" % (old, new, page_file))
+
             chdir(bagdir)
             total_bytes, total_files = make_manifests('data', processes, algorithms=['sha512'])
+            log.info("New vs. old: %s" % changed_urls)
         return total_bytes, total_files
 
     def _set_bag_info(self, bag, total_bytes, total_files, ocrd_identifier, ocrd_manifestation_depth, ocrd_base_version_checksum):

ocrd_utils/setup.py

@@ -5,7 +5,7 @@
 
 setup(
     name='ocrd_utils',
-    version='1.0.0b19',
+    version='1.0.0',
     description='OCR-D framework - shared code, helpers, constants',
     long_description=open('README.md').read(),
     long_description_content_type='text/markdown',