Add intelligent SQL auto-completion with prioritization

[tobias-fire]

Summary

Adds intelligent SQL auto-completion to the REPL with context-aware prioritization and usage tracking.

Features

1. SQL Auto-Completion

• Tables: Completes table names (both short and schema-qualified)
• Columns: Completes column names (both short and table-qualified)
• Schemas: Completes schema names with trailing dot for exploration
• Functions: Completes SQL function names with opening parenthesis

2. Intelligent Prioritization (5-tier system)

• Priority 5000: Columns from tables in current query (highest)
• Priority 4000: Tables and schemas
• Priority 3000: Qualified columns from other tables
• Priority 2000: Unqualified columns from other tables
• Priority 1000: Functions
• Priority 0-990: System schemas (information_schema, pg_catalog)

3. Context-Aware Suggestions

• Analyzes current SQL statement to extract referenced tables
• Prioritizes columns from tables already mentioned in the query
• Adjusts priorities based on query context

4. Usage Tracking

• Tracks last 10 queries to learn usage patterns
• Frequently used items get higher priority within their class
• Session-only tracking (not persisted)

5. Schema Exploration

• Type schema_name. to see all tables in that schema
• Type table_name. to see all columns in that table
• Intelligent filtering based on partial input

Architecture

New Components

• src/completion/usage_tracker.rs - Tracks table/column/function usage frequency
• src/completion/priority_scorer.rs - Calculates priority scores (5-tier system)
• src/completion/context_analyzer.rs - Extracts tables from SQL statements

Modified Components

• src/completion/mod.rs - Integrated scoring and sorting
• src/completion/schema_cache.rs - Added schemas and functions support
• src/main.rs - Initialize UsageTracker
• src/query.rs - Track successful queries
• src/context.rs - Store usage tracker reference

Testing

• 24 new unit tests for prioritization system
• All 157 tests passing
• Tested with information_schema exploration
• Tested with function completion

Examples

No table context - typing SELECT u<Tab>:

[users, public.users, user_logs, users.user_id, user_id, ...]

Tables appear first, then columns.

With table context - typing SELECT * FROM users WHERE u<Tab>:

[user_id, username, updated_at, users, user_logs, ...]

Columns from users appear first.

Schema exploration - typing information_schema.<Tab>:

[information_schema.tables, information_schema.columns, ...]

Function completion - typing len<Tab>:

[length(, ...]

🤖 Generated with Claude Code