[BUG] DatabricksQueryTool skips row_limit when an identifier merely contains "limit"

Description

DatabricksQueryTool decides whether to append the row_limit clause with a substring check, so any table or column name that merely contains the letters "limit" silently disables the row cap:

# lib/crewai-tools/src/crewai_tools/tools/databricks_query_tool/databricks_query_tool.py (line ~67, main @ 3d72c707d5)
if self.row_limit and "limit" not in self.query.lower():
    self.query = f"{self.query.rstrip(';')} LIMIT {self.row_limit};"

SELECT * FROM limited_orders contains limit as a substring, so no LIMIT clause is ever appended and the tool fetches unbounded rows — the opposite of the guard's intent.

Steps to Reproduce

from crewai_tools import DatabricksQueryTool

tool = DatabricksQueryTool(catalog=..., schema=...)  # any valid connection
tool.run(query="SELECT * FROM limited_orders")       # default row_limit=1000

The generated SQL has no LIMIT 1000 appended, because "limit" in "select * from limited_orders" is True.

Expected behavior

A table/column name that happens to contain "limit" should not suppress the row cap; SELECT * FROM limited_orders should get LIMIT 1000 appended. Only an actual LIMIT/FETCH clause in the query should suppress it.

Screenshots/Code snippets

SELECT * FROM limited_orders;        -- "limit" is a substring of the identifier -> no cap appended (bug)
SELECT * FROM orders LIMIT 5;        -- real clause -> correctly left unchanged
SELECT * FROM orders;                -- no clause -> correctly gets LIMIT 1000 appended

A clause-aware check (e.g. regex for LIMIT n, LIMIT ALL, FETCH FIRST/NEXT n ROWS) instead of a substring check fixes the class.

Operating System

Ubuntu 24.04

Python Version

3.12

crewAI Version

1.15.18 (latest release)

crewAI Tools Version

main @ 3d72c707d5 (bug verified present there)

Virtual Environment

Venv

Evidence

TABLE_HAS_LIMIT: SELECT * FROM limited_orders

(no LIMIT clause appended, default row_limit=1000 in effect)

Regression tests covering LIMIT/FETCH detection and identifier substring cases: lib/crewai-tools/tests/tools/test_databricks_query_tool.py (9 passed against a candidate fix).

Possible Solution

Replace the substring check with clause detection before appending row_limit:

import re
_HAS_ROW_LIMIT = re.compile(r"\bLIMIT\s+\d+\b|\bLIMIT\s+ALL\b|\bFETCH\s+(FIRST|NEXT)\s+\d+\s+ROWS\b", re.IGNORECASE)

if self.row_limit and not _HAS_ROW_LIMIT.search(self.query):
    self.query = f"{self.query.rstrip(';')} LIMIT {self.row_limit};"

Additional context

I have a fix ready in PR #7121 (reopening it referencing this issue per the first-contribution gate). Analysis was assisted by an AI coding agent, disclosed there as well.

Start in lib/crewai-tools/src/crewai_tools/tools/databricks_query_tool/databricks_query_tool.py around the row_limit check, then run lib/crewai-tools/tests/tools/test_databricks_query_tool.py. Done means identifiers containing "limit" still receive the row cap, while actual LIMIT or FETCH clauses do not receive an additional cap.