v0.0.15 - Atomic file writes

Bug Fix

• Replaced Path.write_text() with atomic writes across all FileSystemJsonTables operations (_insert, _update, _delete, add_table, add_column, remove_column, _save_table_metadata, _remove_table_metadata, _ensure_metadata_table)
  • Writes now go to a temporary file first, then os.replace() atomically swaps it into place
  • Prevents file corruption caused by non-atomic truncate-and-write under concurrent I/O load
  • Fixes an issue where JSON data files could end up with null bytes or partial content when multiple operations write to the same file rapidly

Full Changelog: v0.0.14...v0.0.15

Accepting global permissions

Full Changelog: v0.0.13...v0.0.14

Adding Permissions

Now you can use our library to statically check permissions from SQL queries:

from abstra_json_sql.authorization import Permissions

p = Permissions(default=False)
p.grant("SELECT", "orders", "status = 'completed' AND total > 100")

p.allowed("select * from orders where status = 'pending'") # False
p.allowed("select * from orders where total <= 100") # False
p.allowed("select * from orders where status = 'completed' AND total > 100") # True

p.revoke("SELECT", "orders", "customer_id = 42")
p.allowed("select * from orders where customer_id = 42 AND status = 'completed' AND total > 100") # False

v0.0.12 - Parser Fixes and CI/CD Pipeline

🐛 Bug Fixes

Parser and Lexer Improvements

• Fixed lexer keyword matching: Keywords are no longer incorrectly identified within valid identifiers

  • Example: inventory is now correctly parsed instead of being split into IN + ventory
  • This resolves parsing errors for table/column names containing SQL keywords

• Fixed operator precedence: Expression parsing now correctly respects operator precedence

  • Logical operators (AND, OR) now have lower precedence than comparison operators
  • price > 10 AND stock > 0 is now correctly parsed as (price > 10) AND (stock > 0)
  • Fixes "Unsupported types for AND" errors

• Added pydantic to requirements.txt: Resolves missing dependency for Pydantic model support

✨ New Features

CI/CD Pipeline

• GitHub Actions workflow with automated quality checks on every PR
• Multi-version testing: Tests run on Python 3.9, 3.10, 3.11, and 3.12
• Code linting: Ruff enforces consistent code style and PEP 8 compliance
• Type checking: Pyright catches type-related bugs
• Coverage reporting: Optional Codecov integration

Development Tools

• Added requirements-dev.txt with pytest, pytest-cov, ruff, and pyright
• Added ruff.toml configuration for linting and formatting
• Added pyrightconfig.json for type checking

🔧 Configuration Changes

• Minimum Python version: Set to 3.9+ (was 3.8+)

  • Required due to modern type hint syntax in the codebase

• Branch protection: Configured for main branch

  • Blocks direct pushes (PRs only)
  • Requires 7 status checks to pass before merge
  • Auto-deletes branches after merge

📊 Test Coverage

• ✅ 190 tests passing across 4 Python versions
• ✅ All code passes linting (Ruff)
• ✅ All code passes type checking (Pyright)

🔗 Related PRs

• #1 - Fix operator precedence and lexer keyword matching
• #2 - Add CI/CD pipeline with pytest, ruff, and pyright

📦 Installation

pip install abstra-json-sql==0.0.12

🙏 Credits

🤖 Generated with Claude Code

Co-Authored-By: Claude noreply@anthropic.com

Test build 11

Full Changelog: v0.0.10...v0.0.11

Test build 10

Full Changelog: v0.0.9...v0.0.10

Test build 9

Full Changelog: v0.0.8...v0.0.9

Test build 8

Full Changelog: v0.0.7...v0.0.8

Test build 7

Full Changelog: v0.0.6...v0.0.7

Test build 6

Full Changelog: v0.0.5...v0.0.6