Spaces:
Runtime error
Runtime error
Commit History
-
Add fuzzy near-match suggestions for numeric columnsβ
Jakub +1: Claude Opus 4.6When unmatched rows exist and numeric columns have a configured margin, find candidate rows from the other side where non-numeric columns match exactly and numeric values differ by at most the margin. Displayed in CLI output and a new Streamlit "Near matches" tab. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>fe0d1b4 -
Add bidirectional matching and row numbers to resultsβ
Jakub +1: Claude Opus 4.6Compare both AβB and BβA so users can see unmatched rows from either table. All result views now include 1-based row numbers from the original file position. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>7bd9057 -
Fix missing decimal separator in streamlit ui
Jakub2e0a6b2 -
Improve number parsing
Jakub0c971d1 -
Add tmp
Jakubea855fb -
Add comprehensive tests for negative number supportβ
Jakub +1: Claude Sonnet 4.5Verified that the app correctly handles negative numbers in all scenarios: - Simple negative number matching - Negative numbers with rounding (e.g., -150.25 β -150) - Negative numbers as strings (e.g., "-100" β -100.0) - Mixed positive and negative numbers (no cross-matching) - Negative numbers in composite keys - Zero vs negative zero (IEEE 754 equality) All tests pass without any code changes - pandas' to_numeric() and round() already handle negative values correctly. Added 6 new tests to TestNumberMatching class. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>2b6c637 -
Add support for 2-digit year date formatsβ
Jakub +1: Claude Sonnet 4.5Added date format patterns with %y (2-digit year) to support formats like dd.mm.yy, dd/mm/yy, mm/dd/yy, and dd-mm-yy. These are now available in both the CLI and Streamlit format picker dropdowns. When using %y, pandas interprets years 00-68 as 2000-2068 and 69-99 as 1969-1999. The tests verify that 2-digit formats parse correctly and match correctly against 4-digit year formats for the same date. Changes: - config.py: Added 4 new entries to DATE_FORMAT_EXAMPLES - README.md: Updated date format reference table - tests/test_comparator.py: Added 4 tests for 2-digit year parsing Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>b712d55 -
Fix duplicate matching: replace isin() with count-aware multiset matchβ
Jakub +1: Claude Sonnet 4.5Series.isin() is a set-membership check β it reports a match whenever a key exists anywhere in Table B, regardless of how many copies are available. When Table A has more rows with a given composite key than Table B, the excess rows were incorrectly marked as matched. Replaced with a groupby.cumcount() / value_counts() approach: rows in A are numbered 0, 1, 2 β¦ within each duplicate group, and a row matches only when its index is less than the number of copies in B. Still O(n + m), fully vectorised. Updated the one test that explicitly asserted the old (wrong) behaviour, and documented the partial-key / duplicate-handling workflow in the README. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>4f3d4cf -
Fix sdk_version: use three-part version for HFβ
Jakub +1: Claude Sonnet 4.5Hugging Face parses sdk_version as major.minor.patch; "1.35" only has two parts and caused "unexpected end of input" on the Space. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>7383d52 -
Add .gitattributes from Hugging Faceβ
Jakub +1: Claude Sonnet 4.5LFS filter rules generated by HF; required by the platform. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>2eb3c18 -
Add dockerfile
Jakub62cb411 -
Add Hugging Face Spaces deploymentβ
Jakub +1: Claude Sonnet 4.5- app.py β root entry-point HF expects; adds src/ to sys.path and delegates to table_comparator.app.main() - requirements.txt β runtime deps for HF's pip installer - README.md β YAML front-matter (sdk: streamlit, app_file: app.py) that HF parses to configure the Space Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>6e5d768 -
Initial commit: table-comparatorβ
Jakub +1: Claude Sonnet 4.5CLI + Streamlit app that checks which rows in one table exist in another. Column mapping is fully user-driven: each pair can be typed as text, number, or date with independent format / rounding / case options. - src/table_comparator/ β config, loader, comparator, click CLI, streamlit UI - sample_data/ β two CSVs with deliberately mismatched columns + a ready-to-run JSON config - tests/ β 69 pytest tests covering validation, serialisation, every column type, null handling, edge cases, and the sample-data integration scenarios - fixed int64/float64 stringification asymmetry in the number normaliser Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>531d507