v2.1.3

Changed:

• bagit-profile matches changes from spec v3.4.2 (metadata dir)

v2.1.2

Changed:

• have save_mets use UTF-8 encoding for byte-serialization (no entities), #388

Fixed:

• regression from #374, #387

v2.1.1

Added:

• PAGE validator: coordinate self-validity and mutual consistency, #374

Fixed:

• Add more related mime types and fix image/jpeg, #382

v2.1.0

Added:

• Workspace validation will check cardinality of images per file is 1, #243, OCR-D/spec#132

Changed:
..

• bashlib will no longer warn about "non-conformant" file group names, #365
• Invalid file:/ URL will now raise exceptions, #373
• image_from_*: increase tolerance for size mismatch after rotation to 2px, #371

v2.0.2: Merge pull request #348 from kba/improve-param-validation

Changed:

• ocrd process: Validate parameters when validating a task
• Dockerfile: Revert to Ubuntu 18.04 for LTS compatibility, #344
• Parameter validation: Raise exception for unknown parameters
• ocrd ocrd-tool validate: Raise exception for unknown keys in JSON

v2.0.1

Fixed:

• METS CREATEDATE date format now ISO8601, #360
• ocrd workspace find allow outputting file group, #359
• processor decorator: --version should succeed independent of parameters, #358

Changed:

• ocrd process uses the ocrd-tool.json of the tools to check whether output file group necessary, #296
• Dockerfile: Revert to Ubuntu 18.04 for LTS compatibility, #344
• pixel density warnings downgraded further to "notice", #361

v2.0.0

Changed:

• image_from_page etc: allow filling with background or transparency
• 🔥 API changes, #311, #327
• Dockerfile: Omit ENTRYPOINT, OCR-D/spec#130, #340
• Relax pixel density validation errors to warnings, OCR-D/spec#129, #339

v1.0.1

Fixed:

• Add dimension to workspace validation skip list, #329
• Update ocrd-tool.json schema to spec 3.3.0 (no output_file_grp, no syntax restriction on content-type)
• PAGE XML output references xsi:schemaLocation, #331
• Update Pillow to 6.2.0

v1.0.0

Since v1.0.0b19:

Added:
* Workspace validation: Validate that files mentioned in pc:Page/@imageFilename exist in METS and on FS, #309

Fixed:
* ocrd ocrd-tool parse-params has the string-or-filepath logic for -p/--parameter as for the CLI

Since v0.15.2

Added:

• Spec-conformant handling of AlternativeImage, OCR-D/spec#116, OCR-D/ocrd_tesserocr#33, #284
• ocrd workspace list-page to list all page IDs
• ocrd workspace remove to remove files, #275, #245
• ocrd workspace remove-group to remove file groups, #275, #245
• ocrd workspace prune-files
• Workspace validation: Validate that files mentioned in pc:Page/@imageFilename exist in METS and on FS, #309
• utils: MIME_TO_EXT to map mime types to preferred extension
• Validation of imageHeight/imageWidth in PAGE vs. actual image height/width, #229
• image_from_page: allow filtering by feature (@comment), #294
• points_from_y0x0y1x1 for inverted x/y pairs
• many utility methods for image manipulation and coordinate handling, #268, OCR-D/ocrd_tesserocr#49
  • bbox_from_points
  • bbox_from_xywh
  • bbox_from_polygon
  • coordinates_for_segment
  • coordinates_of_segment
  • crop_image
  • membername
  • image_from_polygon
  • points_from_bbox
  • points_from_polygon
  • points_from_xywh
  • polygon_from_bbox
  • polygon_from_x0y0x1y1
  • polygon_from_xywh
  • polygon_mask
  • rotate_coordinates
  • xywh_from_bbox

Fixed:

• Handle TIFF ResolutionUnit not being set #250
• bashlib: --mets-file should be --mets
• ocrd workspace set-id case in argument error
• fix DeprecationWarning for PyYAML 5.1+
• use headless opencv
• Regression with ocrd_page data types, #269
• Segfault issue with Pillow >= 6.0.0, #270
• ocrd ocrd-tool parse-params has the string-or-filepath logic for -p/--parameter as for the CLI
• Workspace: Simplify file download code, add extensions to files
• Processor: chdir to workspace directory on init so relative files resolve properly
• typos in docstrings
• README: 'module' -> 'package'
• workspace.image_from_page: logic with rotation/angle
• Adapted test suite to OCR-D/assets now with file extensions
• Require Pillow == 5.4.1 throughout
• regression in namespace handling of PAGE output, #277
• METS is serialized as Unicode instead of character entities, #279

Changed:

• 🔥 Drop Python2 support
• 🔥 Refactored project into 5 modules with little dependencies each
• Implement 3.2.0 of the spec
• OcrdFile: Default fileGrp to TEMP
• OcrdFile: Accept url constructor arg
• Extended page with TextStyle for Page, , PRImA-Research-Lab/PAGE-XML#8
• -m/--mets is not required anymore, #301
• ocrd workspace prune-files: Throw on error removing non-existant file
• -p/--parameter argument accepts raw JSON as well now, #239
• workspace bagger will create files with extension
• export additional region types from generated code, #241
• save_mets is atomic now, #278, #285
• missing required parameters should raise exception, fix #244 #247
• Improve pixel density logic in OcrdExif, #256, #37, OCR-D/ocrd_tesserocr#54
• 🔥 stop supporting python <= 3.4
• Support only 2019-07-15 PAGE version

Removed:

• 🔥 Move factory methods from OcrdPage and OcrdExif to new module ocrd.model_factory
• Factor out XML constants to ocrd.constants.xml
• 🔥 BaseProcessor.add_output_file removed

v1.0.0b19

Changed:

• image_from_page: allow filtering by feature (@comment), #294