[
https://issues.apache.org/jira/browse/HIVE-27432?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Oliver Schiller updated HIVE-27432:
-----------------------------------
Description:
A database name that ends with '#' is not properly considered by the parsing
logic in method parseDBName:
{code:java}
public static String[] parseDbName(String dbName, Configuration conf) throws
MetaException {
if (dbName == null)
{ return Arrays.copyOf(nullCatalogAndDatabase, nullCatalogAndDatabase.length); }
if (hasCatalogName(dbName)) {
if (dbName.endsWith(CATALOG_DB_SEPARATOR)) {
// This means the DB name is null
return new String[]
{dbName.substring(1, dbName.length() - 1), null}
;
} else if (dbName.endsWith(DB_EMPTY_MARKER)) {
// This means the DB name is empty
return new String[]
{dbName.substring(1, dbName.length() - DB_EMPTY_MARKER.length() - 1), ""}
;
}
String[] names = dbName.substring(1).split(CATALOG_DB_SEPARATOR, 2);
{code}
The method checks whether the given string in dbName ends with '#', and if it
does, it assumes that db name is null. This is inconsistent to how the split is
performed. The split operation splits the db name at the first occurrence of a
'#'. That means "@foo#bar#" would be considered as having a null db name, but
the split operation would split it into ["foo", "bar#"|#"].
The right solution appears to be that the first occurrence of '#' is searched
for. If the first occurrence of '#' is the last character in dbName, then and
only then the db name should be considered null.
was:
A database name that ends with '#' is not properly considered by the parsing
logic in method parseDBName:
public static String[] parseDbName(String dbName, Configuration conf) throws
MetaException \{
if (dbName == null) {
return Arrays.copyOf(nullCatalogAndDatabase,
nullCatalogAndDatabase.length);
}
if (hasCatalogName(dbName)) \{
if (dbName.endsWith(CATALOG_DB_SEPARATOR)) {
// This means the DB name is null
return new String[] {dbName.substring(1, dbName.length() - 1), null};
} else if (dbName.endsWith(DB_EMPTY_MARKER)) \{
// This means the DB name is empty
return new String[] {dbName.substring(1, dbName.length() -
DB_EMPTY_MARKER.length() - 1), ""};
}
String[] names = dbName.substring(1).split(CATALOG_DB_SEPARATOR, 2);
The method checks whether the given string in dbName ends with '#', and if it
does, it assumes that db name is null. This is inconsistent to how the split is
performed. The split operation splits the db name at the first occurrence of a
'#'. That means "@foo#bar#" would be considered as having a null db name, but
the split operation would split it into ["foo", "bar#"].
The right solution appears to be that the first occurrence of '#' is searched
for. If the first occurrence of '#' is the last character in dbName, then and
only then the db name should be considered null.
> Database name that ends with '#' is not properly considered when splitting
> catalog and db names.
> ------------------------------------------------------------------------------------------------
>
> Key: HIVE-27432
> URL: https://issues.apache.org/jira/browse/HIVE-27432
> Project: Hive
> Issue Type: Bug
> Components: Hive, Metastore, Standalone Metastore
> Reporter: Oliver Schiller
> Priority: Minor
>
> A database name that ends with '#' is not properly considered by the parsing
> logic in method parseDBName:
> {code:java}
> public static String[] parseDbName(String dbName, Configuration conf) throws
> MetaException {
> if (dbName == null)
> { return Arrays.copyOf(nullCatalogAndDatabase,
> nullCatalogAndDatabase.length); }
> if (hasCatalogName(dbName)) {
> if (dbName.endsWith(CATALOG_DB_SEPARATOR)) {
> // This means the DB name is null
> return new String[]
> {dbName.substring(1, dbName.length() - 1), null}
> ;
> } else if (dbName.endsWith(DB_EMPTY_MARKER)) {
> // This means the DB name is empty
> return new String[]
> {dbName.substring(1, dbName.length() - DB_EMPTY_MARKER.length() - 1), ""}
> ;
> }
> String[] names = dbName.substring(1).split(CATALOG_DB_SEPARATOR, 2);
> {code}
> The method checks whether the given string in dbName ends with '#', and if it
> does, it assumes that db name is null. This is inconsistent to how the split
> is performed. The split operation splits the db name at the first occurrence
> of a '#'. That means "@foo#bar#" would be considered as having a null db
> name, but the split operation would split it into ["foo", "bar#"|#"].
> The right solution appears to be that the first occurrence of '#' is searched
> for. If the first occurrence of '#' is the last character in dbName, then and
> only then the db name should be considered null.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
