[HADOOP-13051] Add Glob unit test for special characters

• Type: Test
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 3.0.0-alpha1
• Fix Version/s: 3.0.0-alpha1
• Component/s: fs
• Hadoop Flags: Reviewed
• Assignee: Harsh J
• Reporter: Harsh J
• Created: 22/Apr/16 12:55
• Updated: 12/May/16 18:27
• Resolved: 05/May/16 21:23

Description

On branch-2, the below is the (incorrect) behaviour today, where paths with special characters get dropped during globStatus calls:

bin/hdfs dfs -mkdir /foo
bin/hdfs dfs -touchz /foo/foo1
bin/hdfs dfs -touchz $'/foo/foo1\r'
bin/hdfs dfs -ls '/foo/*'
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1^M
bin/hdfs dfs -ls '/foo/*'
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1

Whereas trunk has the right behaviour, subtly fixed via the pattern library change of HADOOP-12436:

bin/hdfs dfs -mkdir /foo
bin/hdfs dfs -touchz /foo/foo1
bin/hdfs dfs -touchz $'/foo/foo1\r'
bin/hdfs dfs -ls '/foo/*'
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1^M
bin/hdfs dfs -ls '/foo/*'
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1
-rw-r--r--   3 harsh supergroup          0 2016-04-22 17:35 /foo/foo1^M

(I've placed a ^M explicitly to indicate presence of the intentional hidden character)

We should still add a simple test-case to cover this situation for future regressions.

Issue Links

• is duplicated by: HADOOP-13099 Glob should return files with special characters in name
• relates to: HADOOP-12436 GlobPattern regex library has performance issues with wildcard characters