v2.47.1

Fixed:

• Docker: install python3-venv, do not install python3-pip, #1003, #1004, OCR-D/ocrd_all#352

v2.47.0

Fixed:

• ocrd resmgr: handle namespaces packages gracefully for Python <= 3.6, #917, #985
• ocrd resmgr: guess media type with filetype.py in addition to MIME_TO_EXT, #991
• OcrdMets: Insert mets:agent in a schema-compliant way, #976, #977
• ocrd_cli_wrap_processor: remove unnecessary ocrd_tool kwarg, #998, #999
• Docker base image builds again, except CUDA, #986

Added:

• ocrd resmgr: support Google Drive links, #993, #992

Changed:

• Use the new importlib.resources.files API, #995, #996
• ocrd resmgr: resources for ocrd_anybaseocr removed from central list, provided by the project, #989, OCR-D/ocrd_anybaseocr#100

v2.46.0

Changed:

• WorkspaceValidator: an OcrdFile without a pageId is not an error, but a document-wide file, #485, #979
• WorkspaceBackupManager: add snapshot on init if enabled, #981
• 🔥 end-of-life for python 3.6, test from 3.7 to 3.11, #956
• 🔥 update base image to Ubuntu 20.04, #956

Fixed:

• bashlib: Handle empty list of input files, #978
• OcrdMets.find_files: don't override the @LOCTYPE of file candidates, #980
• ocrd resmgr: replace libmagic with simple lookup by suffix, #982, #984

Added:

• helpers: get_cached_processor to get instances of a processor in preparation for #974, #972

v2.45.1

Fixed:

• ocrd resmgr: insert new entries first, so dedup works as expected, #971

v2.45.0

Fixed:

• ocrd resmgr download --overwrite now works properly for both directories and files, #690, #797
• ocrd resmgr: archive resources can now also be ZIP files and reference files and folders in them, #967
• ocrd-dummy: can now be used to create PAGE-XML for images without copying, #803, #814

v2.44.0

Added:

• ocrd zip update command to update checksums for an OCRD-ZIP after changing it, #363, #951

Removed:

• ocrd zip bag does no longer support the long-broken --in-place option, #964, #363

v2.43.0

Added:

• OcrdMets.refresh_caches to update caches after changes to XML outside of OcrdMets, #957, #960

v2.42.1

Fixed:

• Regressions from PR #875, #957, #958
• Missing import in ocrd workspace merge, #956

v2.42.0

Fixed:

• Symlinks in workspaces are properly resolved now, #802, #954

Added:

• Optional caching of access to METS, configured via environment variable OCRD_METS_CACHING, #875
• CPU and memory profiling , configured via environment variable OCRD_PROFILE and OCRD_PROFILE_FILE, #678

Changed:

• ocrd workspace find: supports comma-separated regexes, ranges and literal values for --page-id, #955
• ocrd workspace find: ranges are generated with last number in string, #955

v2.41.0

Fixed:

• ocrd workspace list-installed should not create spurious entries for moduledir files, #940
• OcrdResourceManager.download does not need to query size via HTTP Content-Length in most cases, #924, #939
• make install: Reinstall shapely to work around shapely/shapely#1598, #947

Changed:

• ocrd workspace bulk-add: Generate file_id consistent with conventions from filename if no --file-id given, #943