Version 0.7.1

Added

• Github action: publish to PyPI

Version 0.7.0

Added

• Add retrieve and pdfs as high-level operations
• Metadata preparation can add records to separate origin feeds
• Initial package manager functionality (registering packages and displaying them in the docs)
• Search: update of records and propagation of changes
• Several SearchSources (including SearchSource query validation)
• Revisions of CLI (verbose mode, user feedback)
• Colrev merge (reconciliation coding when merging git branches)
• dedupe --merge/--unmerge
• Integrated colrev pre-commit hooks
• PRISMA diagram (data endpoint)
• Obsidian (data endpoint)
• Preparation: not-in-toc exception/warning
• Setup of pytests

Changed

• Curated records are now explicitly identified through curation_IDs
• Revise colrev validate (commits, users, properties)
• Detailed advisor (using get_advice() for data endpoints)
• Performance improvements and simplification of status (cli)
• Moved correction functionality to SearchSources (refactored correction path)
• Preparation: simplified preparation rounds (default settings)
• Retrieve TEIs through local_index (if available) instead of recreating it
• Replace pathos by Threadpool
• Revise the documentation
• Revise and extend exceptions

Removed

• Remove persistent colrev-ids
• Remove realtime review
• Dependencies ansiwrap and p-tqdm

Fixed

• **kwargs calls in ReviewManager
• Indexing of non-curated records
• Address special cases in dedupe (active learning)

Version 0.6.0

Added

• Web-based editor for project settings
• Comprehensive architecture refactoring
• Conformance with pylint, mypy, flake8
• Introduced packages
• Updated file and directory structure
• Documentation of modules, classes, and methods
• Github-pages as a data package_endpoint

Changed

• Renamed from colrev_core to colrev (integrated cli)
• Switch to poetry for dependency management
• Renamed scripts to package_endpoints
• PDF-hash generation based on Docker to avoid platform dependency issues
• Switch to Jinja templates (instead of concatenating multiple strings)

Fixed

• Concurrent request session handling
• StatusStats calculations

Version 0.5.0

Added

• Push/pull (including corrections), sync, validate, service operations
• Data provenance model (colrev_data_provenance, colrev_masterdata_provenance)
• Extensible endpoints (search, prep, prescreen, pdf-get, pdf-prep, screen, data)
• Prescreen scope

Changed

• Improvements: prep, dedupe operations
• Performance improvements (e.g., status, bibtexparser > pybtex)
• Extended Record class (e.g., merge and fuse_best_fields)
• LocalIndex: Elasticsearch to Opensearch
• Dedupe: testing and parameter optimization (option to prevent same-source merges)
• Settings.json and validation
• Updated documentation
• Testing and refactoring (e.g., for Windows, prefer keyword arguments in functions, python package type information)

Version 0.4.0

Added

• Extract functionality: ReviewDataset, Process
• Developed LocalIndex, EnvironmentManager, OpenSearch
• Curation model, including Resource installation and a "correction path"
• Search operation (reintegrating paper_feed and local_paper_index)
• Prep exclusion based on languages

Changed

• Object-oriented refactoring of the whole codebase
• Use Zotero translators (instead of bibutils) for imports
• Duplicate identification (add FP safeguards based on LocalIndex, add a procedure for small samples)
• Consistent PDF path handling
• Structured data extraction based on csv

Fixed

• Loggers
• Performance issues in prep and status

Version 0.3.0

Added

• Introduced ReviewManager and integrated hooks/checks
• Fetch metadata from Open Library
• Required fields for misc
• Information on needs_manual_preparation (man_prep_hints)
• Activated mypy hooks
• Introduced custom load scripts
• Documentation
• LocalIndex: hash-table implementation for indexing and retrieval

Changed

• Dedupe: based on active learning (dedupe-io)
• Improved batches
• Pass records instead of BibDatabase
• PDF prep and longer pdf hashes

Removed

• CLI: now in separate colrev repository

Fixed

• Initializing repositories
• Backward search adds two entries to search_details
• Logging (reinitialize after batches/commits)

Version 0.2.0

Added

• Status model (rev_status, md_status, pdf_status)
• Implemented cli interface
• Import formats (bib, ris, endn, pdf, text list of references)
• Docker services for import, ocr, building the paper etc.
• Metadata repositories for record preparation (crossref, dblp, semantic scholar)
• PDF preparation (OCR, metadata validation)
• Commit message reporting
• Check and validation of iteration completeness
• Support for building papers based on pandoc

Changed

• Integrated review process status (including prescreen, screen inclusion vs exclusion) in the references.bib
• Renamed scripts and cli entrypoints
• Refactored code
• Tracing from hash_id to origin links
• Extended and refactored pre-commit hooks

Removed

• R scripts for sample statistics (the goal is to implement them in Python)
• hash_id function, trace_entry, trace_hash_id

Fixed

• Bugs in analysis/combine_individual_search_results.py and in analysis/acquire_pdfs.py
• Catch exceptions and check bad responses in analysis/acquire_pdfs.py
• Bug in git modification check for references.bib in analysis/utils.py
• Exception in anaylsis/screen_2.py (IndexError)
• Global constant conflict with analysis/entry_hash_function.py (nameparser.config/CONSTANTS)

Version 0.1.0

Added

• First version of the pipeline, including status, reformat_bibliography, trace_entry, trace_hash_id, combine_individual_search_results, cleanse_records, screen_sheet, screen_1, acquire_pdfs, screen_2, data_sheet and data_pages
• Environment setup including Dockerfile and Makefiles