v2024-03-08

Removed:

• tesserocr and tesseract (which are now submodules of ocrd_tesserocr)

core f54b002..c5b5580

Release: v2.63.3

• 📦 v2.63.3
• 📝 changelog
• OcrdMets.add_file: fix finding existing el_pagediv
• 📝 changelog
• expose uninstall-workaround
• 📦 v2.63.2
• 📦 v2.63.1
• Merge branch 'fix-get-physical-pages'
• make coverage: omit generateDS code

cor-asv-ann 0a4f684..f7ebb74

• compare/evaluate: ensure lines are shown verbatim on single file level
• compare/evaluate: do not record worst lines verbatim
• compare/evaluate: also record 1% worst lines

ocrd_keraslm 759805b..472197f

Release: v0.4.2

• update assets
• 📦 v0.4.2
• generators: do not require input length > window size
• train: allow passing directory for training data, too
• add checkpointing, allow continuing from ckpt
• suppress TF gibberish

ocrd_tesserocr 08a020f..ed73d96

Release: v0.18.1

• :package v0.18.1
• 📝 changelog
• update tesseract/tesserocr to most recent
• update/improve readme
• simplify dockerfile
• CI: add make test
• add repo/assets as proper submodule, rename -clean → clean-
• make Tesseract build configurable
• also install minimal needed models
• explify dependencies install ← install-tesserocr ← install-tesseract
• test: download files of data assets
• test: also set up logging system during tests

workflow-configuration 8418b3f..bd149f8

Release: 0.1.3

• unprefix regex paths by directory argument, if any
• forgot to convert exit to continue in 6fe3c6b3
• no need to close logger FDs on exit
• ocrd-import: use coproc instead of handcrafted FIFOs for loggers
• ocrd-import: rewrite (no parallel jobs, but parallel logging)…
• ocrd-import: add option --basename, default to using directory as well
• ocrd-import: simplify+speedup…

v2024-02-01

Added:

• https://github.com/OCR-D/ocrd_froc

Removed

• https://github.com/qurator-spk/page2tsv/ (temporarily)

cor-asv-fst 076e04e..4211371

• Merge pull request #4 from stweil/master

core ac1f15b..b94b185

Release: v2.62.0

• reenable circle because docker failed to build on ghcr.io
• 📦 v2.62.0
• {pypi,build}-workaround: missed a commit, s/get_distribution(...).version/dist_version(...)
• Merge branch 'circle-to-gha'
• 📝 changelog
• Merge branch 'master' into ocrd-tool-json-root
• expose ocrd-tool.json for ocrd-dummy in root like processors
• 📦 v2.60.3
• 📝 changelog
• Fix --editable install for setuptools>=64, setuptools#3548
• 📦 v2.60.2
• 📝 changelog
• Merge pull request #1161 from OCR-D/logging-downgrade-level
• Merge pull request #1160 from OCR-D/is-oai-content-loglevel-debug

dinglehopper f077ce2..f8e3108

Release: v0.9.4

• 🚧 GitLab CI Test: Push after pulling
• 🚧 GitLab CI Test: Trigger only on default branch (and do not hardcode it)
• 🚧 GitLab CI Test
• 🔍 ruff: Remove ignore configuration, we use multimethods in a compatible way now
• ⚙ pre-commit: Update hooks
• 🚧 GitLab CI Test
• 🔍 mypy: Use an almost strict mypy configuration, and fix any issues
• 🔍 mypy: Use a compatible syntax for multimethod
• 🔍 mypy: Remove ExtractedText.segments converter
• 🔍 mypy: Avoid using check() for all attr validators
• 🔍 mypy: Make cli.process() typed so mypy checks it (and issues no warning)
• Merge branch 'pr103'
• ⚙ Update ruff+mypy dependencies
• ⚙ pre-commit: Update hooks
• ⬆ Move on to supporting Python >= 3.8 only
• 🐛 Use typing.List instead of list, for Python <3.9
• 🐛 Use Optional instead of | none, for Python <3.10
• ⚙ pre-commit: Update hooks
• 🐛 Fix generating word differences
• ⚙ pre-commit: Update hooks
• Merge branch 'master' of https://github.com/qurator-spk/dinglehopper
• Merge branch 'master' into performance
• ⬆ Update uniseg dependency
• ❎ Make joining grapheme clusters more robust by checking joiner and handling an empty joiner
• 🐛 Fix score_hint call in cli_line_dirs
• 🐛 Fix docstring of distance() for grapheme clusters
• 🐛 Fix calculation of score_hint for edge cases, e.g. when CER is infinite
• 🕸Do not use deprecated ID, pageId options
• ✔ Add mets:FLocat's @LOCTYPE/OTHERLOCTYPE to test data
• ⬆ Update multimethod dependency
• 🐛 Update tests for ExtractedText
• use uniseg again
• update rapidfuzz version
• replace uniseg with uniseg2
• apply black
• move grapheme clusters to ExtractedText
• remove python2.7 futures
• remove unused includes
• only call words_normalized once

eynollah 706433c..032a99e

Release: v0.2.0

• adapt to OcrdFile.local_filename now :Path
• adapt to ocrd>=2.54 url vs local_filename

ocrd_fileformat c5f0c52..ba79de9

Release: v0.10.0

• 📦 v0.10.0
• 📝 changelog
• Update ocr-fileformat to include UB-Mannheim/ocr-fileformat#172
• Merge branch 'fix-textract2page'
• update ocr-fileformat to latest
• Update ocr-fileformat to v0.6.0

ocrd_repair_inconsistencies cf879c1..94c482f

• 🕸 README: Mention archival of the project

opencv-python 7cfd1ee..8ad8ec1

Release: 80

• Merge branch 'as/4.9.0-readme-update' into 4.x
• Merge pull request #941 from asmorkalov/as/mac_m1_venv_for_test
• Merge pull request #940 from asmorkalov:as/donation
• Merge pull request #938 from asmorkalov"as/4.9.0-pre
• Merge pull request #934 from asmorkalov/as/native_mac_m1_runner
• Merge pull request #936 from asmorkalov:as/mac_intel_update
• Merge pull request #932 from asmorkalov/as/pre-4.9.0_linux_upgrade
• Merge pull request #931 from asmorkalov:as/ipp_icv_license
• Merge pull request #904 from asmorkalov:as/python_3.12
• Merge pull request #927 from dkurt:try_enable_dependents

sbb_binarization f3c6ac8..b89ec49

Release: v0.1.0

• Merge pull request #65 from rettinghaus/update-tests

tesseract ea0b245..8ee020e

Release: 5.3.4

• Create new release 5.3.4
• Set User-Agent: header field in HTTP request for curl downloads
• Merge pull request #4178 from sadra-barikbin/patch-1
• Merge pull request #4174 from stweil/warnings

workflow-configuration cbc3234..f54c91a

Release: 0.1.3

• ocrd-page-transform: local_filename instead of url
• ocrd-page-transform: fix unbound variable

v2023-12-15

core 742906e..ac1f15b

Release: v2.60.1

• 📦 v2.60.1
• docker: we need .git during build for setuptools_scm
• 📝 changelog
• Merge branch 'git-versioning'
• 📝 changelog
• defaults for mets_basename and mets_server_url
• Merge branch 'master' of https://github.com/OCR-D/core
• 📝 changelog
• ocrd workpace list-page: ignore files without pageId, fix #1148

v2023-12-07

core d8c5813..742906e

Release: v2.59.1

• 📦 v2.59.1
• 📝 changelog
• Merge pull request #1142 from OCR-D/fix-ws-locks
• 📝 changelog
• fix ranges, use numpy
• 📦 v2.59.0
• 📝 changelog
• Merge branch 'network-workflow-api'
• 📝 changelog
• Merge pull request #1139 from OCR-D/bagger-filegrp-filter
• 📝 changelog
• Merge branch 'list-page-extended'
• 📝 changelog
• Merge pull request #1136 from OCR-D/network-processor-api
• 📝 changelog
• bagger: fallback to ID-derived filename to avoid filename conflicts
• test bagger: sample workflow with conflicting basenames
• bagger test: rewrite as pytest
• 📝 changelog
• Merge pull request #1134 from OCR-D/page-update
• 📝 changelog
• Merge branch 'update-apidocs'
• OcrdMets: remove dead code exit method, #1130

dinglehopper dbaccdd..f077ce2

Release: v0.9.4

• 🐛 dinglehopper-summarize: Handle reports without difference stats
• Merge pull request #97 from qurator-spk/clean-remove-six-dep-again
• Merge pull request #96 from qurator-spk/test-on-pr-but-really
• Merge pull request #94 from qurator-spk/test-on-pr
• Merge pull request #93 from qurator-spk/update-dep-multimethod
• Merge pull request #92 from qurator-spk/update-pre-commit
• Merge pull request #91 from qurator-spk/test-remove-circleci
• Merge pull request #90 from qurator-spk/test-on-python-3.12
• ✔ Add mets:FLocat's @LOCTYPE/OTHERLOCTYPE to test data

format-converters 9615db1..fa8b4b5

• Merge pull request #24 from stweil/orientation
• Merge pull request #18 from stweil/rotate
• Merge pull request #23 from stweil/TextEquivNone
• Merge pull request #22 from stweil/image
• Merge pull request #19 from stweil/page_version
• Merge pull request #21 from stweil/image_dimensions
• Merge pull request #20 from stweil/close_statement
• Merge pull request #17 from stweil/update

ocrd_calamari c0a4dfd..ad8febd

Release: v1.0.6

• ⚙ pre-commit: update config (using pre-commit-update hook)
• 🎨 Remove extra whitespace in CircleCI confg
• ✔ CircleCI: Update codecov orb
• Merge pull request #102 from OCR-D/review-test-base
• Merge pull request #106 from OCR-D/setup-update-description
• ✔ CircleCI: Try to fix caching of pip cache
• Merge pull request #107 from OCR-D/circleci-pip-cache-1
• Merge pull request #101 from OCR-D/use-ruff-config-like-dinglehoppers
• Merge pull request #100 from OCR-D/fix/set-default-MODEL-for-bare-pytest
• 🐛 Fix importing + exporting test.base.assets
• 🐛 pre-commit/mypy: Install missing types-setuptools depedency
• 🧹 Remove unused exports (done by ruff --fix .)
• 🛠Add pre-commit configuration
• 🐛 Require ocrd >= 2.54.0 for tf_disable_interactive_logs
• ✒ README: Minor changes
• ✒ README: Fix and simplify example instructions
• 🎨 Reformat using Black + ruff
• 🧹 Remove protobuf restriction (→ calamari-ocr 1.0.6 has it now)
• Merge pull request #78 from mikegerber/test-python-3.11
• 💩 Add a script fix-calamari1-model to fix regexen in 1.0 models
• ✒ README: Wrap long line
• 🧹 Remove workaround for fixed np.str problem in Calamari

ocrd_neat 0f64f07..2f7d01c

Release: v0.0.1

• fix alto2tsv bug
• Merge pull request #15 from stweil/typo
• Merge pull request #14 from stweil/gitignore
• add alto2tsv

ocrd_pagetopdf 7368f51..24f77d3

Release: v1.1.0

• Use first bash from PATH (allows running on macOS)
• Merge pull request #23 from kba/mets-server-url-support

tesseract dc228ed..ea0b245

Release: 5.3.3

• Update status badge for GitHub workflow sw (add missing line break)
• Update status badge for GitHub workflow sw
• Correct indefinite articles before vowels
• Update issue-bug.yml
• Merge pull request #4162 from tfmorris/3000-tcp-scrollview
• Avoid conversions from std::string to char* to std::string
• Remove unnecessary conversions from std::string to C string
• Remove whitespace at line endings
• Update sw.yml
• Update README.md
• Update sw.yml
• Fail on curl download errors
• Add new parameter curl_cookiefile for curl_easy_setopt
• ci: Fix clang build for Ubuntu 22.04
• ci: Allow manual trigger of unittest
• Move bail_out function before libtoolize check
• Merge pull request #4150 from tfmorris/4149-directory-to-stdout

workflow-configuration d0d208c..cbc3234

• add page-ensure-textequiv-conf.xsl
• add page-rename-id-clashes.xsl
• PAGE CLIs: allow combining --pretty and --diff
• new pair of XSLTs: un/flatten table cells
• require ocrd >= 2.58.1
• require ocrd >= 2.58
• ocrd-page-transform: support --mets-server-url

v2023-10-20

core caa7ac3..d8c5813

Release: v2.58.1

• 📦 v2.58.1
• 📝 changelog
• bashlib: set empty string as default value for ocrd__argv[mets_server_url], OCR-D/olena#95
• 📦 v2.58.0
• 📝 changelog
• Merge pull request #1127 from OCR-D/fix-bashlib-log3
• 📝 changelog
• ocrd workspace bulk-add now supports --mets-server-url
• bashlib: short option -U for --mets-server-url
• 📝 changelog
• Merge remote-tracking branch 'origin/fix-bashlib-log2'
• 📝 changelog
• Merge branch 'mets-server-reload'
• 📝 changelog
• Merge pull request #1121 from OCR-D/fix-bashlib-log

ocrd_fileformat cfb24ec..7364265

Release: v0.9.1

• 📦 v0.9.1
• 📝 changelog
• require ocrd >= 2.58.1
• 📦 v0.9.0
• 📝 changelog
• require ocrd >= 2.58
• Support --mets-server-url; get local_filename not url

ocrd_im6convert 105697f..db18917

Release: v0.1.1

• 📦 v0.1.1
• 📝 changelog
• require ocrd >= 2.58.1
• 📦 v0.1.0
• 📝 changelog
• minversion must be MAJOR.MINOR.PATCH
• require ocrd >= 2.58
• add --mets-server-url support

ocrd_olena a2e2520..c1f7cab

Release: v1.5.0

• 📦 v1.5.0
• 📝 changelog
• require ocrd >= 2.58.1
• debug
• minversion must be MAJOR.MINOR.PATCH
• require ocrd >= 2.58
• support --mets-server-url

ocrd_pagetopdf 4f4a330..7368f51

Release: v1.0.0

• require ocrd >= 2.58.1
• require ocrd >= 2.58
• Use local_filename, not url
• restore $zeros assignment
• Support --mets-server-url

workflow-configuration f574f82..d0d208c

Release: 0.1.3

• require ocrd >= 2.58
• ocrd-page-transform: support --mets-server-url

v2023-10-17

cor-asv-ann 006a70e..4216c16

Release: v0.1.14

• adapt to Numpy deprecations
• CI: use ocrd/core-cuda as base image
• CI: dummy venv
• CI: use proper tab character
• CI: clone first
• CI: mkdir first
• CI: chdir to tmp location
• CI: use /tmp for aux clone of ocrd_all
• try getting tensorflow-gpu from Nvidia
• use proper URLs for submodules
• Merge pull request #6 from kba/init-report-dict
• evaluate: skip pages with no results

core de08453..3b0307a

Release: v2.56.0

• 📦 v2.56.0
• 📝 changelog
• Merge branch 'network-log-refactoring'
• 📦 v2.55.2
• 📝 changelog
• 🐛 pydantic fields must not start with underscore
• 📦 v2.55.1
• 📝 changelog
• Merge pull request #1113 from OCR-D/bulk-add-url-local-filename
• 📦 v2.55.0
• 📝 changelog
• Merge branch 'workflow-endpoint'
• 📝 changelog
• generate_page_range: verify single-page range based on start
• generate_page_range: warn, not raise, if start==end, fix #1106
• 📝 changelog
• logging: remove custom logging in ocrd_network, use explicit logger name
• logging remove hard-coded setLevel in decorators/ocrd_network
• helpers.ruin_processor: setOverrideLogLevel if log_level is provided
• ocrd log: default to ocrd.log_cli logger name
• METS Server: add basic logging of operations
• logging: use ocrd.{utils,models,exif} not ocrd_{utils,models,exif}
• ocrd_utils.logging.getLogger: no more initLogging
• 📝 changelog
• Merge branch 'master' into mets-server-fixes-2023-09-15
• mets server: remove socket file on shutdown
• mets server: do the chmod before server start, not before connection
• 📝 changelog
• METS server: make socket world-readable/-writable
• 📦 v2.54.0
• Merge pull request #1095 from OCR-D/run-cli-mets-server-url
• 📝 changelog
• Merge branch 'revise-logging'
• Merge pull request #1093 from OCR-D/create-default-queue
• Merge pull request #1080 from OCR-D/revise-logging
• bashlib: fix --help output
• 📝 changelog
• Merge branch 'keep-remote-links'
• 📝 changelog
• downgrade ValueError to log.warning about inconsistent pageId for processor calls
• Merge branch 'master' into warn-empty-page
• raise ValueError if --page-id is provided but leads to empty result
• 📝 changelog
• Merge pull request #1069 from OCR-D/processing_server_ext_1046
• Merge branch 'mets-server'
• 📝 changelog
• ci: localhost -> 127.0.0.1
• pin requests < 2.30, OCR-D/core#1082
• mets server: forbid local/remote workspace with different directories
• mets server: allow both local_filename and url to be None
• Merge branch 'master' into mets-server
• mets server: test both UDS and TCP variant
• ClientSideOcrdFile et al need url too
• pass mets_server_url from run_processor
• typo: -{,-}mets-server-url
• move ClientSideOcrd{Agent,File} to ocrd_models
• METS server: support -U for processor options
• workspace server start: pass workspace context
• mets server: single option --mets-server-url/-U
• mets server will never pass content to workspace.add_file
• mets server: no content will pass through it
• mets server: clean up is_remote muddle
• mets server: support unique_identifier
• mets server: str handlers
• mets server: provide fallback for non-wrapped OcrdFile methods
• mets server: remove XXX HACK comments, they are not;
• mets server: improve docs
• mets server: add stop
• Update ocrd/ocrd/cli/workspace.py
• ocrd workspace CLI: reference METS server option
• METS server: consistently use local_filename
• Update ocrd/ocrd/cli/workspace.py
• METS Server: equivalent functionality to files for agents
• finish implementation / test mets server
• Merge remote-tracking branch 'origin/master' into mets-server
• workspace: save content to file only if not remote
• Merge branch 'mets-server' of https://github.com/kba/ocrd-core into mets-server
• mets_server: file search/adding on /file not /
• mets_server: missed mimetype kwarg
• mets_server: different loggers for socket/host-port
• Merge branch 'mets-server' of https://github.com/kba/ocrd-core into mets-server
• mets_server: replace Model constructor with static create calls
• --port must be int
• resolver: shorten mets_server_{host,port} check
• mets_server: only save_mets on PUT and DELETE
• OcrdWorkspace.is_remote should be a bool
• ClientSideOcrdMets: fix signature of self.file_groups
• mets-server: bashlib should take same args
• remove noise from makefile
• slowly but determinedly
• getting there
• .
• wip

dinglehopper 0fd4ea1..dbaccdd

Release: v0.9.4

• ✒ README: Minor whitespace cleanup
• ✒ README: Recommend installing via pip and from PyPI
• 📦 v0.9.4
• 🎨 editorconfig: *.json should have a final newline
• 🧹 pyproject: Remove extra *.json
• 🧹 Remove empty setup.cfg
• 📦 v0.9.3
• 🐛 Remove MANIFEST.in workaround, now that setuptools_ocrd is fixed
• 📦 v0.9.2
• 🧹 .gitignore dist/
• 🐛 Workaround sdist not containing top-level ocrd-tool.json
• ⚙ GitHub Actions: Call test workflow when (before) deploying
• 🎨 Release: Make installing setuptools-ocrd conditional on ocrd-tool.json
• 🐛 Release: Try fixing getting the version (install setuptools-ocrd)
• 📦 v0.9.1
• ✒ README: Update badges
• Revert "🚧 GitHub Actions: Try testing on Python 3.12"
• 🚧 GitHub Actions: Try testing on Python 3.12
• 🚧 GitHub Actions/CircleCI: Remove testing from CircleCI config
• 🚧 GitHub Actions: Do no try installing ruff on Python 3.6
• 🚧 GitHub Actions: Do no try installing pytest-ruff on Python 3.6
• 🚧 GitHub Actions: Avoid compiling OpenCV and NumPy on Python 3.6
• 🚧 GitHub Actions: Fix testing for Python 3.6
• 🚧 GitHub Actions: Disable matrix fail-fast
• 🚧 GitHub Actions: Test on multiple Python versions
• 🚧 GitHub Actions: Test report
• 🚧 GitHub Actions: Try shell for loop to install from all requirements*.txt
• 🚧 GitHub Actions: Rework test, run in src/
• 🚧 GitHub Actions: Allow running test manually
• 🚧 GitHub Actions: Rename test workflow, also run on schedule
• 🚧 GitHub Actions: Add test worklow
• 🚧 GitHub Actions: Add release workflow
• 🧹 Make dinglehopper.* exports explicit
• ⚙ ruff: Ignore F811 (no redefinitions) for now, as ruff considers the multimethods redefinitions
• 🎨 Reformat comments + strings manually (not auto-fixed by Black)
• ⬆ Use f-strings
• 🎨 Reformat using Black
• 🎨 Sort imports (auto-fixed by ruff)
• ⚙ Add pre-commit
• 🛠 Replace flake8 + pylint with ruff
• ⚙ Move mypy settings to pyproject.toml
• ⚙ pytest.ini → pyproject.toml
• 🐛 Detect encoding (incl BOM) when reading files
• 🐛 Move source into src/ to fix install
• ⚙ Migrate to pyproject.toml
• 🚧 CircleCI: Run black
• Merge pull request #83 from INL/feat/batch-processing
• Merge pull request #82 from CircleCI-config-suggestions-bot/StoreTestResults
• 🧹 .gitignore .python-version (for pyenv)
• 🧹 Remove qurator. namespace prefix
• 🐛 Fix installing by calling find_namespace_packages in setup.py
• 🕸Do not use deprecated ID, pageId options
• 🔧 Remove explicit namespace_packages
• ✔ CircleCI: Explicitly install binary opencv-python-headless (dep of OCR-D?) to avoid compilation
• 🐛 Remove deprecated declare_namespace call

nmalign cf7c60f..7832c90

Release: v0.0.3

• adapt to Numpy deprecations

ocrd_calamari 3a029ca..c0a4dfd

Release: v1.0.6

• Merge pull request #90 from OCR-D/tf_disable_interactive_logs
• ✒ README: Use backtick syntax for code block
• ✔ CircleCI: Do not test on Python 3.6 anymore (EOL since 2021-12-23)
• v1.0.6
• 🐛 Fix installation by keeping protobuf < 4.0

ocrd_cis 43a356a..fcc02fd

Release: v0.1.5

• adapt to Numpy and Pillow deprecations
• segment: fix baseline extraction
• segment: adapt to OpenCV changes
• resegment (baseline/ccomps): improve handling of fg conflicts
• resegment: add param baseline_only
• check_page/region/line: skip assumptions on number of components
• adapt to Shapely 2.0 deprecations
  ...

v2023-06-28

Unreleased

v2023-06-28

Changed:

• Bash prompt excludes user name undefined in docker container, #376, #366
• Docker: ocrd-all-tool.json is built during container build, #379
• Docker: XDG_CONFIG_HOME is set to XDG_DATA_HOME/ocrd-resources, so resources.yml is in usually-mounted location, #377, #252
• Docker: /data is world-writeable now, so log files can be written there, #377, #252

core 6708624..552cfcd

Release: v2.52.0

• Dockerfile: install wheel before make install
• 📦 v2.52.0
• 📝 changelog
• ci: debug macos
• Makefile: PIP_INSTALL from environment via ?=
• ci: try fixing macos
• Merge branch 'master' into improve-packaging
• Makefile: trailing whitespace
• update gitignore
• add ocrd.processor.builtin.dummy pkg (needed for resource discovery)
• add requirements.txt to manifest (so it's available at build time)
• make pypi: use build module instead of setuptools CLI
• make uninstall: run in reverse BUILD_ORDER
• make install: run conjunctively in BUILD_ORDER

ocrd_calamari 3a029ca..ed7a926

Release: v1.0.6

• v1.0.6
• 🐛 Fix installation by keeping protobuf < 4.0

ocrd_cis a0ea0a2..43a356a

Release: v0.1.5

• postcorrect: improve/update OCR-D wrapper…
• ocropy-train: improve/update OCR-D wrapper…
• ocrd-tool: rm old ocrd-cis-ocropy-rec (gone in 9e20991)
  Merge branch 'kba:typo' #91 into fix-alpha-shape
  Merge branch 'kba:double-page-max-size' #96 into fix-alpha-shape
  Merge branch 'kba:resolve-resources' #83 into fix-alpha-shape

ocrd_detectron2 04bf4c6..5f8bdcb

Release: v0.1.7

• update model URLs to GH release archive
• requirements.txt: avoid pkg names in comments
• requirements.txt: add pycocotools as explicit dependency
• require wheel (since pip does not pull it anymore)
• deps: no build isolation so Detectron2 compilation can use Torch
• Docker: add badges and basic description
• Docker username: vars, not env
• Docker: use underscore in tagname, alright
• add Dockerfile and GH Action to publish at Dockerhub and GHCR

ocrd_kraken b13dd8a..1e71324

Release: v0.3.0

• recognize: ignore 'one_channel_mode' unless model has 1 input channel
• segment/recognize: warn if no GPU available

ocrd_repair_inconsistencies c898d6c..cf879c1

• Merge pull request #13 from stweil/master

opencv-python 474a1cc..b534ea2

Release: 72

• Merge pull request #853 from asmorkalov/as/add_pyi_to_package

sbb_binarization 010ec99..f3c6ac8

Release: v0.1.0

• Update test.yml

tesseract 1569e50..bb8803a

Release: 5.3.1

• Update .mailmap
• Create config.yml
• Remove old broken GitHub action vcpkg-4.1.1 (fixes issue #4078)
• cmake: check if leptonica was build with tiff support
• cmake: provide info about disabled LibArchive and CURL
• cmake: allow to disable tiff (-DDISABLE_TIFF=ON)
• Merge pull request #4073 from stweil/osd
• Merge pull request #4071 from stweil/clean
• Merge pull request #4066 from stweil/lstmtraining
• Merge pull request #4068 from stweil/sprintf
• Remove unused code in function fix_rep_char
• Merge pull request #4067 from stweil/misc
• Support for Sgaw and W Pwo Karen languages in the Myanmar validator. (#4065)
• issue-bug.yml: Windows versions 7, 8, 8.1 are not supported anymore
• snap: Update from leptonica 1.74.2 to latest 1.83.1
• fix: Fix snap package building
• Create new release 5.3.1
• Remove whitespace at line endings
• Fix issue #4010 (#4041)
• cmake: add missing HAVE_NEON to config_auto.h
• Merge branch 'main' of https://github.com/tesseract-ocr/tesseract
• cmake: adjust build to autotool settings
• Merge branch 'main' of https://github.com/tesseract-ocr/tesseract
• cmake: improve NEON build

tesserocr e184c62..3c9519b

Release: v2.6.0

• add github-workflow building wheels

v2023-06-14

Changed:

• All docker images now contain git checkouts and retain /build, i.e. behave like the -git variants
• No more git updates within docker build, but fix git module dependency outside
• Reduce docker image size (by reinstating all-in-one layer, removing cache, avoiding duplicate CUDA libraries...)
• Use git submodule update --single-branch on CI to reduce docker image size

Added:

• make deps-cuda: non-intrusively support CUDA system dependencies (in docker or native)
• make ocrd-all-tool.json: Generate and upload a combination of all processors' ocrd-tool.json, #362
• make test-workflow: Run a workflow with most processors as a general smoke test
• make test-cuda: to test whether CUDA properly set up and has GPU available
• make test-core: Run OCR-D/core unit tests

Fixed:

• dependencies between modules, esp. with custom OCRD_MODULES selection
• editable mode (pip install -e)
• OpenCV build
• get tesserocr from PyPI if disabled as a module
• get ocrd from PyPI if core disabled as a module
• consistent interoperable module versions (esp. Numpy/OpenCV/Shapely/Protobuf/Torch/TF Python dependencies)

cor-asv-ann 006a70e..2c4b1ff

Release: v0.1.14

• CI: use ocrd/core-cuda as base image
• CI: dummy venv
• CI: use proper tab character
• CI: clone first
• CI: mkdir first
• CI: chdir to tmp location
• CI: use /tmp for aux clone of ocrd_all
• try getting tensorflow-gpu from Nvidia
• use proper URLs for submodules
• Merge pull request #6 from kba/init-report-dict
• evaluate: skip pages with no results

core de08453..6708624

Release: v2.51.0

• Merge pull request #1055 from bertsky/deps-cuda
• ci: disable upterm for gh actions
• readme: remove dockerhub/travis badge, add GH actions badge
• debug gh actions
• test bashlib: /usr/bin/env bash instead of /bin/bash
• test_workspace_bagger: use ocr-d.de instead of google.com for testing
• disable logging tests until properly fixed
• docker-image: reuse local ghcr.io image instead of docker.io
• 📦 v2.51.0
• 📝 changelog
• make help: improve description
• Revert "Merge remote-tracking branch 'hnesk/no-more-pkg_resources' into release-2.36.0"
• remove out-dated processor resources
• docker-cuda: improve (reduce size) again…
• docker-cuda: rewrite…
• core-cuda: use same CUDA libs as needed for Torch anyway
• Merge branch 'pr-1008' into reduce-cuda
• Merge branch 'master' of https://github.com/OCR-D/core into reduce-cuda
• make install on py36: revert to prefer-binary via install
• make install on py36: fix prefer-binary syntax
• make install on py36: prefer binary OpenCV/Numpy via pip config instead of preinstall
• core-cuda: install more CUDA libs via pip and ld.so.conf, simplify Dockerfile for that
• core-cuda: use CUDA 11.8, install cuDNN via pip and make available system-wide via ld.so.conf
• reinstate workaround for shapely, but more robust
• docker-cuda: change base image, no multi-CUDA runtimes
• keep gcc, no autoremove
• rehash after pip upgrade
• give up workaround for shapely-CUDA issue

dinglehopper 0fd4ea1..35be58c

• Merge pull request #83 from INL/feat/batch-processing
• Merge pull request #82 from CircleCI-config-suggestions-bot/StoreTestResults
• 🧹 .gitignore .python-version (for pyenv)
• 🧹 Remove qurator. namespace prefix
• 🐛 Fix installing by calling find_namespace_packages in setup.py
• 🕸Do not use deprecated ID, pageId options
• 🔧 Remove explicit namespace_packages
• ✔ CircleCI: Explicitly install binary opencv-python-headless (dep of OCR-D?) to avoid compilation
• 🐛 Remove deprecated declare_namespace call

eynollah ea792d1..706433c

Release: v0.2.0

• Revert "Merge pull request #97 from qurator-spk/420-namespace-package"
• Merge pull request #100 from bertsky/patch-2
• Merge pull request #97 from qurator-spk/420-namespace-package

ocrd_cis c90b29f..a0ea0a2

Release: v0.1.5

• Merge branch 'kba:typo' #91 into fix-alpha-shape
• Merge branch 'kba:double-page-max-size' #96 into fix-alpha-shape
• Merge branch 'kba:resolve-resources' #83 into fix-alpha-shape
• segment: adapt to OpenCV changes
• resegment (baseline/ccomps): improve handling of fg conflicts
• resegment: add param baseline_only
• check_page/region/line: skip assumptions on number of components
• adapt to Shapely 2.0 deprecations
• adapt to Numpy 1.24 dtypes
• resegment: list instead of generator
• re/segment: improve polygon simplification
• re/segment: join_baselines: skip lines outside of polygon
• re/segment: join_baselines: for complex subtypes, apply recursively
• re/segment: join_polygons: connect touching neighbours, too

ocrd_fileformat dacfa50..4e7e0de

Release: v0.7.0

• 📦 v0.7.0
• update ocr-fileformat

ocrd_kraken 802c6b0..b13dd8a

Release: v0.3.0

• segment/recognize: default to device=cuda:0 (now backed by safe fall-back)
• segment/recognize: fall back to CPU if no CUDA device
• fix typo
• update changelog
• recognize: project text upwards in order by concatenation
• recognize: ensure baseline/boundary are consistent
• recognize: ignore invalid baselines
• setup metadata: update/improve
• deps-ubuntu: update
• improve/update readme
• Dockerfile: use CUDA base image, improve labels
• update changelog
• recognize: pass lines in baseline format if any baselines are annotated
• update blla.model URL (master→main)
• recognize: workaround for empty/failed line records
• recognize: workaround for better quality box cuts
• recognize: avoid invalid polygons on single-glyph words
• Revert "recognize: avoid invalid polygons on single-glyph words"
• segment: also show tags/type prediction
• recognize: avoid invalid polygons on single-glyph words
• recognize: use proper data structures of rpred

ocrd_pagetopdf 6155605..4f4a330

Release: v1.0.0

• Merge pull request #22 from bertsky/fix-input-files

ocrd_wrap 63c04d5..2cd800d

Release: v0.1.8

• 📦 0.1.8
• Merge pull request #10 from bertsky/update-numpy

opencv-python 6b73d90..474a1cc

Release: 72

• Merge pull request #849 from asmorkalov/as/python3_for_build
• Fix: numpy version for python 3.11 (#839)
• Merge pull request #852 from asmorkalov:as/ci_check
• Merge pull request #837 from bertsky/fix-py38-build
• Merge pull request #838 from henryiii/patch-2

sbb_binarization 39ef3fd..010ec99

Release: v0.1.0

• 📦 v0.1.0
• Update README.md
• update CI badge
• Merge pull request #59 from bertsky/change-model-url
• Merge pull request #56 from bertsky/non-verbose

workflow-configuration cb923f7..5aff777

• ocrd-import: add option --regex (positive path selector)
• ocrd-import: fix skipping in subshell
• add METS transforms to TOC
• generalise standalone CLI for both PAGE and METS XSL, update documentation
• mets-copy-agents.xsl: make path for other-mets relative to input m...

v2023-03-26

core cbe83ab..de08453

Release: v2.49.0

• :package v2.49.0
• 📝 changelog
• drop eynollah model from resource list, provided by eynollah itself
• 📝 changelog
• rename Docker image (to make work with GHCR)
• 📦 v2.48.1
• 📝 changelog
• core-cuda: CUDA 11.3 instead of 11.2
• ocrd_tool_validator: fix link in comment, fix #1019
• 📦 v2.48.0
• 📝 changelog
• Merge remote-tracking branch 'origin/build36-speedup-without-update'
• 📝 changelog
• Merge branch 'master' into fix-972
• Set ws outside the constructor
• chdir to ws in the beginning
• Undo the revert of getcwd()
• Revert getcwd() location - failing tests
• Raise the error instead of returning it
• Change getcwd() call location
• Chdir before processor.process() calls
• Fix the instance caching
• 📦 v2.47.4
• 📝 changelog
• Merge pull request #1011 from OCR-D/drop-fontgroup-resources
• 📦 v2.47.3
• 📝 changelog
• Dockerfile: reintroduce python3-pip so "pip install -U pip" succeeds again
• 📦 v2.47.2
• 📝 changelog
• Merge branch 'pr/986'
• 📦 v2.47.1
• 📝 changelog
• Merge pull request #1004 from stweil/fix-dockerfile
• Merge pull request #1003 from OCR-D/fix-docker-venv
• 📦 v2.47.0
• require importlib_resources for python <= 3.8, #996
• 📝 changelog
• resmgr: drop anybaseocr models from resource list
• 📝 changelog
• Merge branch 'pr/999'
• 📝 changelog
• ocrd_utils: adapt to newer importlib.resources API
• Merge pull request #994 from OCR-D/fix-scrutinizer
• 📝 changelog
• Merge pull request #977 from OCR-D/fix-add-agent
• 📝 changelog
• Merge branch 'pr/993'
• 📝 changelog
• Merge pull request #991 from bertsky/fix-resmgr-download-mimetype
• 📝 changelog
• Merge pull request #985 from OCR-D/fix-917
• 📦 v2.46.0
• 📝 changelog
• Merge branch 'drop-3.6'
• 📝 changelog
• Merge branch 'drop-mime-magic'
• 📝 changelog
• Merge branch 'pr/980'
• 📝 changelog
• Merge branch 'pr/981'
• 📝 changelog
• Merge pull request #972 from OCR-D/ref-processor-helper
• 📝 changelog
• Merge pull request #979 from OCR-D/workspace-validator-empty-pageid
• 📝 changelog
• Merge pull request #978 from OCR-D/bashlib-inputfiles

dinglehopper c4ab7c9..0fd4ea1 (rewind)

• ✔ Add @cneud's former 40 GB problem files to the test suite
• 🎨 Reformat using Black
• ✔ CircleCI: Test on Python 3.11

eynollah 13bc237..ea792d1

Release: v0.2.0

• 📦 v0.2.0
• 📝 changelog
• Merge remote-tracking branch 'bertsky/savedmodel-format'
• 📦 v0.1.0
• 📝 changelog
• Update config.yml (#90)
• Merge pull request #88 from emresvd/patch-1
• Update config.yml (#89)

ocrd_anybaseocr 94e5037..5978a1f

Release: v1.9.0

• Merge pull request #97 from stweil/master
• 📦 v1.9.0
• 📝 changelog
• move remaining resources to GWDG S3
• block-segmentation: parameter "th" should be number
• add resources to ocrd-tool.json, #98

ocrd_calamari c7ad6eb..3a029ca

Release: v1.0.5

• ✔ Do not test on Python 3.11 for now (unsupported)
• ✔ CircleCI: Install TF by explicitly invoking pip on Py 3.11
• ✔ Do not delete test workspace when DEBUG env variable is set
• ✔ CircleCI: Test on Python 3.11, too
• ✔ CircleCI: Install binary OpenCV for Python 3.6
• Revert "✔ CircleCI: Do not test on Python 3.6 anymore"
• ✔ CircleCI: Do not test on Python 3.6 anymore
• ✔ Fix tests to use the new filenames
• 🐛 Fix NumPy dependency (hopefully...)
• 🐛 Fix syntax error in setup.py
• 🐛 Require NumPy < 1.24 due to np.str deprecation/error

ocrd_detectron2 e005d3c..04bf4c6

Release: v0.1.7

• publish: commit images, too
• CI: publish test results to gh-pages
• add link for test results
• Delete jekyll-gh-pages.yml
• try to fix gh-pages
• add GH pages
• doc: add CI badge
• CI: use cache instead of artifacts for models
• CI: try to get the damn conditional to work
• CI: use stupid GHA negation syntax
• CI: fix action URL
• CI: fix negation
• CI: cache detectron models via artifacts
• Merge branch 'master' of ssh://github.com/bertsky/ocrd_detectron2
• some workarounds for broken model configs
• 📦 0.1.7
• add CLI test
• add models for magazine layout by Jambo-sudo (PubLayNet+custom GT) and LayoutParser (PRImA Layout GT)
• adapt to numpy v1.24
• 📦 0.1.6
• add fixture for badly written config files (base path)
• add models for table detection with Psarpei/Multi-Type-TD-TSR
• ocrd-tool resources: update/fix
• make deps: add torch deps explicitly
• avoid colon in generated region IDs

ocrd_fileformat 5022408..dacfa50

Release: v0.6.2

• 📦 v0.6.2
• 📝 changelog
• Merge pull request #44 from bertsky/patch-3

ocrd_keraslm 787341d..9c50478

Release: v0.4.1

• deps: hold numpy and h5py

ocrd_tesserocr 515be8d..09d1e13

Release: v0.17.0

• try with lowercase image tag
• rename Docker image
• Update docker-image.yml
• Create docker-image.yml
• 📦 0.17.0
• Merge pull request #191 from bertsky/override-insteadof-wrap
• deps-ubuntu: allow PPA to fail (on newer distributions)
• CI: chmod PPA Tesseract tessdata g+w
• deps-ubuntu: allow PPA to fail (on newer distributions)
• CI: fix conditional step syntax
• CI: seems to require sudo
• CI: speedup Py36 deps
• CI: update+simplify (on cimg/python:* instead of ocrd/core)

ocrd_typegroups_classifier ffa40fc..a78a85f

Release: v0.5.0

• 📦 v0.5.0
• 📝 changelog
• Merge pull request #15 from OCR-D/ocrd-tool-resources
• 📝 changelog
• Merge pull request #14 from OCR-D/textline-level

opencv-python 736b905..6b73d90

Release: 72

• Merge pull request #820 from asmorkalov/as/config_py_path
• Merge pull request #803 from asmorkalov/as/license
• Merge pull request #790 from asmorkalov/as/docs_update
• Merge pull request #787 from asmorkalov/as/migrate_mac_m1
• Merge pull request #744 from peter-kovacs-aimotive/add-vulkan-license
• Merge pull request #768 from AlexeySalmin/patch-1
• Merge pull request #776 from TheCleric/fix/numpy_version_mac

sbb_binarization aeb6804..39ef3fd

Release: v0.0.11

• Merge pull request #54 from qurator-spk/circleci-python37_38

tesseract a6e0aa7..1569e50

Release: 5.3.0

• textord: Catch empty rows in block iterator (fixes #4039)
• cmake: sync with autotools (OPENMP_SIMD...

v2023-02-06

core 6331433..ee92cfc

Release: 2.45.1

• 📦 v2.45.1
• 📝 changelog
• resmgr: insert from tool instead of append

ocrd_detectron2 fde2f3c..f3342a4

Release: 0.1.5

• 📦 0.1.5
• fix debug_img indentation (only once per page/table)
• ocrd-tool.json: fix PubLayNet/jpleorx model specs

ocrd_olahd_client 10d70a9..6bcbb4b

Release: v0.0.2

• 📦 v0.0.2
• 📝 changelog
• Merge branch 'manipulate-mets-agent'

opencv-python ede2269..736b905

Release: 68

• OpenCV package does not distribute zlib (#780)
• OpenCV 4.7.0 release preparation
• Merge pull request #756 from asmorkalov:as/pipelines_update_4.7

tesseract 6a21a74..4142b32

Release: 5.3.0

• Fix some whitespace issues in source code and text files
• Merge pull request #3992 from seupedro/patch-1
• fix "cannot pass non-trivial object of type 'std::string'"
• show out filename on successful created of traineddata (combine_lang_model)
• fix "cannot pass non-trivial object of type 'std::string'"
• unicharset_extractor: - run ReadMemBoxes only for box files - do not write unicharset in case of broken box file
• Update issue-bug.yml
• Create an issue template for a feature request
• Create a new issue template
• Create new release 5.3.0
• Update README.md
• cmake - msvc/openmp: clean&document configuration
• cmake - mscvc: silent warning C4068: unknown pragma 'GCC'
• Create new release 5.3.0-rc1
• Replace MacOS -> macOS