[IMPALA-1920] parse_url() behaves incorrectly when URL contains '@' in query portion

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 2.2, Impala 2.1.2
• Fix Version/s: Impala 2.6.0
• Component/s: Backend
• Labels: function sql-language
• Environment: CentOS 6 (2.6.32-431.17.1.el6.centos.plus.x86_64); CDH 5.3.2
• Target Version: Product Backlog
• Assignee: Lars Volker
• Reporter: Charlie Flowers
• Created: 25/Mar/15 22:02
• Updated: 24/Apr/16 13:41
• Resolved: 24/Apr/16 13:41

Description

parse_url() returns incorrect values for HOST and USERINFO when '@' is present in the query string.

Example:

> select parse_url('http://host.com/page.htm?val=@not_the_host', 'HOST');
Query: select parse_url('http://host.com/page.htm?val=@not_the_host', 'HOST')
not_the_host

> select parse_url('http://host.com/page.htm?val=not_the_host', 'HOST');
Query: select parse_url('http://host.com/page.htm?val=not_the_host', 'HOST')
host.com

> select parse_url('http://host.com/page.htm?val=@not_the_host', 'USERINFO');
Query: select parse_url('http://host.com/page.htm?val=@not_the_host', 'USERINFO')
host.com/page.htm?val=

> select parse_url('http://host.com/page.htm?val=not_the_host', 'USERINFO');
Query: select parse_url('http://host.com/page.htm?val=not_the_host', 'USERINFO')
NULL
WARNINGS: UDF WARNING: Could not parse URL: http://host.com/page.htm?val=not_the_host