Fix UTF-8 boundary issues in mysql_top_level_limit, parse_usize_token,
and mysql_keyword_at functions that caused crashes when SQL contains
Chinese characters in string literals.
The issue was introduced in commit 021e1682 which added byte-level
indexing without proper UTF-8 boundary checks. When parsing SQL with
Chinese text (e.g., \"WHERE name LIKE '%批量保存%'\"), the byte index
could point to the middle of a multi-byte UTF-8 character, causing
Rust to panic when slicing the string.
Changes:
- parse_usize_token: Use std::str::from_utf8 to validate slice before parsing
- mysql_keyword_at: Add is_char_boundary checks for safe string slicing
- mysql_top_level_limit: Skip non-UTF-8-boundary bytes after incrementing index
This ensures all byte-level indexing operations stay on valid UTF-8 boundaries,
preventing panics while maintaining the original functionality of detecting
LIMIT clauses for bounded result set collection.
Fixes crash reported by users when executing queries with Chinese conditions.
* fix(redis):format INFO output as plain text in cluster mode
* fix(redis): unify INFO category command output in cluster mode
In Redis cluster mode, the INFO command (with category parameters) returns aggregated results from multiple primaries. This fix formats the output as human-readable plain text, consistent with standalone mode, to improve readability.
Closes#1326
When the LLM calls `list_tables` tool without a `schema` parameter, the
backend receives an empty string. This caused PostgreSQL and MySQL
metadata queries to return no results because:
- PostgreSQL: `WHERE n.nspname = ''` matches nothing
- MySQL: `WHERE TABLE_SCHEMA = ''` matches nothing
Fix by falling back to sensible defaults:
- PostgreSQL `list_tables`/`get_columns`: empty schema → "public"
- MySQL `list_tables`: empty schema → database name (MySQL schema == database)