1.4.3 release

[serhii73]

Release notes and version bump for dateparser 1.4.3, following the same two-step approach used for 1.4.2 (#1361) and 1.4.1 (#1337):

1. 1.4.3 release notes — add the 1.4.3 section to HISTORY.rst, summarizing every PR merged since v1.4.2.
2. Bump version: 1.4.2 → 1.4.3 — produced by bump-my-version bump patch --no-tag, updating pyproject.toml, dateparser/__init__.py, and stamping the HISTORY.rst release date. SECURITY.md is unchanged because it tracks only major.minor (still 1.4.x).

A patch bump: this cycle is three bug fixes plus one internal thread-safety refactor and one test-infra addition — no new setting, no new public API, no removal and no changed default.

Changelog

Fixes:

• Make parsing thread-safe: parsing from several threads no longer raises an intermittent KeyError from the shared language caches, and a DATE_ORDER or RELATIVE_BASE value meant for one parse no longer leaks into the settings that other parses read, where it could make them return a wrong date (Make thread-safe #1346)
• Do not share the language detector and the detected locale between search_dates() calls, so concurrent searches over text in different languages no longer return None or a date read in the wrong locale (Avoid shared language_detector and locale cache in search_dates #1371)
• Resolve the BST and HDT timezone abbreviations to the offsets the tz database gives them, UTC+1 (British Summer Time) and UTC-9 (Hawaii-Aleutian Daylight Time), instead of +11 and -9:30, which no zone goes by those names today; abbreviations that the tz database maps to more than one offset, such as CST and IST, keep their current offset. Text carrying these abbreviations keeps its wall clock but moves by 10 hours for BST and 30 minutes for HDT, which can put the parsed instant on a different day (Prefer tz database offsets for conflicting timezone abbreviations #1366)
• Reject a %j (day of year) value that the parsed year does not have, instead of rolling it over into the next year, so 1999366 with date_formats=["%Y%j"] returns None rather than 2000-01-01. A format with no year directive is checked against the year strptime defaults to, 1900, which is not a leap year, so 366 with date_formats=["%j"] now returns None where it used to return January 1 (Reject a %j day of year that the parsed year does not have (#271) #1370)

Cleanups and internal improvements:

• Add CodSpeed benchmarks and a workflow that runs them, so a performance regression such as the quadratic backtracking fixed in 1.4.1 is reported on the pull request that introduces it (tests: add CodSpeed benchmarks #1365)

Notes on the changelog

• Both Prefer tz database offsets for conflicting timezone abbreviations #1366 and Reject a %j day of year that the parsed year does not have (#271) #1370 change what previously-parsing input returns, so each bullet spells out the before and after. They sit under Fixes rather than a Breaking changes section, matching this file's precedent: the two historical Breaking changes sections (1.4.1, 1.0.0) were API and feature removals, while output corrections such as Do not read the day as a two-digit year after the year position #1358 and Fix ISO8601 parsing with explicit languages #1352 shipped under Fixes in 1.4.2.
• tests: add CodSpeed benchmarks #1365 is test and CI infrastructure only, listed under Cleanups and internal improvements — the section that has historically carried this kind of work (OSSFuzz integration OSSFuzz Initial Integration #1198, "Extended test cases" Add 11 sept. 2014 test case for september not sept/seven #1191, "CI updates" [GitHub Actions] Update workflow versions #1088). Only trivia has been omitted in past cycles (Update Read the Docs config to use ubuntu-24.04 #1320, Fix merge issue #1333), and there is none this cycle.

Validation

• The diff is exactly three files, +38/-2, with no code changes — the same shape as the 1.4.1 and 1.4.2 release commits. git diff --check reports no whitespace errors.
• Version is consistent across pyproject.toml, dateparser/__init__.py and the imported dateparser.__version__; no stale 1.4.2 remains outside the changelog, and the (unreleased) marker is gone.
• Full test suite green: 24234 passed, 18 skipped, 1 xfailed.
• python -m build + twine check pass for both the sdist and the wheel; the wheel metadata reads Version: 1.4.3 and the long_description carries the new changelog section.
• sphinx-build -W (the docs tox environment) builds cleanly, so the new RST is valid where docs/history.rst includes it.
• pre-commit run passes on the changed files.
• No generated file needed regenerating: nothing merged this cycle touched dateparser_data/ or dateparser/data/, and tests/test_dateparser_data_integrity.py passes.

After merge

The v1.4.3 git tag is intentionally not included in this branch (a tag on a pre-merge branch would point at the wrong commit). After the squash merge, tag the squash commit on master and then create the GitHub Release:

git fetch origin
git tag v1.4.3 origin/master && git push origin v1.4.3
gh release create v1.4.3 --title "1.4.3" --notes-file <markdown notes>

The GitHub Release is the step that publishes to PyPI: publish.yml triggers on release: types: [created] only, and GitHub does not fire that activity type for draft releases — so the release must be created non-draft in one action, or the upload never runs.

🤖 Generated with Claude Code

[codspeed-hq]

Merging this PR will not alter performance

✅ 7 untouched benchmarks

---

_{Comparing release-1.4.3 (1e638ca) with master (27182cd)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.18%. Comparing base (27182cd) to head (1e638ca).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1376   +/-   ##
=======================================
  Coverage   97.18%   97.18%           
=======================================
  Files         236      236           
  Lines        3057     3057           
=======================================
  Hits         2971     2971           
  Misses         86       86           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.