v0.3.4 - query repair hardening

What's Changed

• Fixed query repair false positives for dotted string literals like 'Foo.Bar'
• Validated sparse arrays using fields across sampled objects instead of only the first object
• Improved nested leaf and alias repair suggestions
• Quoted special-character keys in Smart Inspect suggestions and repair Try: commands
• Added repair validation for from query row contexts and selected expressions

Verification

• Full test suite passed on Python 3.9, 3.10, and 3.11 in CI
• Local package build and twine checks passed

v0.3.3 - query repair suggestions

What's Changed

• Added query repair suggestions for unknown fields in selected fields, filters, sorting, grouping, and aggregations
• Error output now includes nearby field matches and a copy-paste Try: command
• Improved nested field discovery for repair suggestions

Verification

• Full test suite passed on Python 3.9, 3.10, and 3.11 in CI
• Local package build and twine checks passed

v0.3.2 - Smart Inspect mode

What's Changed

• Smart Inspect: running jonq data.json now shows root shape, readable field types, sample values, and suggested queries without adding new CLI flags
• Suggested queries are generated from actual JSON fields for table output, raw scalar output, boolean filters, and group-by counts
• Improved inspect behavior for deeply nested JSON and mixed arrays

Verification

• Full test suite passed on Python 3.9, 3.10, and 3.11 in CI
• Local package build and twine checks passed

v0.3.1 - raw scalar output

What's Changed

• Added raw scalar output for shell pipelines: jonq users.json "select name" -r
• Single-field selections now print plain values one per line in raw mode
• Multi-field selections remain compact JSON objects per line for scriptable output
• Updated README, usage docs, API docs, completions, and changelog

Verification

• Full test suite passed on Python 3.9, 3.10, and 3.11 in CI
• Local package build and twine checks passed

v0.3.0 - CASE/WHEN, table output, COALESCE, follow mode, and 20+ new functions

Added

Output & Display

• Table output: --format table or -t renders results as aligned terminal tables with headers and separators
• YAML output: --format yaml renders results as YAML (uses pyyaml if available, built-in fallback otherwise)
• --explain mode: shows parsed query breakdown (fields, conditions, grouping) and the generated jq filter
• --time flag: prints parse/execute/format timing to stderr
• Auto-detect stdin: curl ... | jonq "select name" works without needing - as source

Query Language

• CASE/WHEN/THEN/ELSE/END: conditional expressions — select case when age > 30 then "senior" else "junior" end as level
• COALESCE: null fallback with nested function support — select coalesce(nickname, name) as display
• || operator: SQL-standard string concatenation — select first || ' ' || last as full_name
• IS NULL / IS NOT NULL: null checks in conditions — select * if email is not null
• Type casting: int(), float(), str() / string(), type()
• Date/time functions: todate(), fromdate(), date(), timestamp() with null-safe execution
• Utility functions: keys(), values(), trim(), ltrim(), rtrim(), tojson(), fromjson(), reverse(), sort(), unique(), flatten(), not_null(), to_entries(), from_entries()

Terminal UX

• Rich REPL: readline with persistent history (~/.jonq_history) and tab completion for field names + keywords
• --follow mode: stream NDJSON from stdin line-by-line — tail -f app.log | jonq --follow "select level, msg if level = 'error'"
• Shell completions: jonq --completions bash|zsh|fish generates completion scripts

Changed

• Version bump to 0.3.0
• Unified all function-name-to-jq-builtin mappings into a single _FUNC_MAP constant (eliminated 3 duplicate copies)
• Expression-only queries now properly map over array elements instead of wrapping the whole input
• Path-explorer schema preview now shows nested JSON paths like orders[].price instead of only shallow field listings
• Reuse bounded sync and async jq worker pools for repeated identical filters
• Main streaming execution path now chunks root-array JSON in memory instead of splitting to temp files
• Repositioned README/package metadata around JSON exploration and extraction

Fixed

• Null-sensitive functions (todate, tonumber) are now guarded against null input instead of crashing
• Schema validator recognizes CASE expressions, || operator, and coalesce — no longer rejects them as unknown fields
• Sync jq runtime errors that only write to stderr no longer hang waiting on stdout
• Async jq workers are cleaned up with their event loop, preventing leftover subprocess transport warnings
• --follow mode skips empty results instead of printing [] for non-matching lines

v0.2.0

[v0.2.0] - 2026-03-08

Added

• DISTINCT: select distinct city returns unique rows
• COUNT DISTINCT: select count(distinct city) as n
• Standalone LIMIT: select * limit 10 (independent of sort)
• IN operator: select * if city in ('New York', 'Chicago')
• NOT operator: select * if not age > 30
• LIKE operator: select * if name like 'Al%' (supports % wildcard for startswith, endswith, contains)
• String functions: upper(), lower(), length()
• Math functions: round(), abs(), ceil(), floor()
• Schema preview: run jonq data.json with no query to inspect fields and types
• Interactive REPL: jonq -i data.json for interactive querying
• Watch mode: --watch flag to re-run query on file change
• URL fetch: jonq https://api.example.com/data "select id"
• Multi-file glob: jonq 'logs/*.json' "select *"
• Auto-detect NDJSON: no flag needed for newline-delimited JSON
• Fuzzy field suggestions: typo correction with Levenshtein distance
• Colorized output: syntax-highlighted JSON when outputting to terminal
• python -m jonq: added __main__.py entry point
• Constants module: centralized all magic numbers and ANSI codes

Fixed

• Dynamic HAVING field mappings instead of hardcoded alias dict
• Removed inline jq comments (# sum) from generated filters
• from clause now correctly iterates arrays (.path[] instead of .path)
• Schema validator recognizes sort, distinct, as, arithmetic operators
• Schema validator skips validation when from clause is used
• Logging level changed from INFO to WARNING (no more filter spam)

Changed

• Version bump to 0.2.0
• Replaced all star imports with explicit imports
• Replaced all hardcoded ANSI codes with _Colors class
• Replaced all magic numbers with named constants
• Removed dead code: handle_error, extract_value_from_quotes, _AGG_FUNCTIONS, parse_condition (jq_filter.py), use_fast parameter
• Executor raises ValueError instead of RuntimeError for jq errors
• Updated all Sphinx docs, README, and SYNTAX.md