[
https://issues.apache.org/jira/browse/DIRSERVER-2109?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15091041#comment-15091041
]
Stefan Seelmann commented on DIRSERVER-2109:
--------------------------------------------
I digged a bit deeper. I added two more tests to AddIT that add entries with
backlashes in RDN attribute value:
http://svn.apache.org/viewvc?rev=1723924&view=rev
In AddOperationContext line 135 we create a schema-aware clone of the incoming
entry, we clone the entry, the DN, the attributes, the values, and normalize
everything. And deep down in PrepareString line 5187 we *unescape* the value.
The problem is that we unescape the value for 2 different cases: when
normalizing the DN/RDN *and* when normalizing the attribute value itself. In
the latter case the unescape is wrong.
For example let's have this entry (LDAP and Java):
{code}
dn: cn=a\\b,ou=users,ou=system
objectClass: top
objectClass: person
cn: a\b
sn: test
Entry entry = new DefaultEntry( new Dn( "cn=a\\\\b,ou=users,ou=system" ) );
entry.add( "objectClass", "top", "person" );
entry.add( "cn", "a\\b" );
entry.add( "sn", "test" );
{code}
When normalizing the DN the value needs to be unescaped. But when normalizing
the entry's attribute value no unescape must be done.
Here are the stacktraces when unescape is called:
Normalizing of DN: unescape is required
{code}
PrepareString.unescape(char[], int) line: 5195
PrepareString.insignifiantSpacesStringAscii(String, boolean) line: 5187
PrepareString.normalize(String, PrepareString$StringType) line: 253
DeepTrimToLowerNormalizer.normalize(String) line: 103
CachingDeepTrimToLowerNormalizer(CachingNormalizer).normalize(String) line: 124
StringValue(AbstractValue<T>).apply(AttributeType) line: 175
StringValue.<init>(AttributeType, String) line: 122
Dn.atavOidToName(Ava, SchemaManager) line: 1101
Dn.rdnOidToName(Rdn, SchemaManager) line: 1162
Rdn.apply(SchemaManager) line: 481
Dn.apply(SchemaManager, boolean) line: 1222
Dn.apply(SchemaManager) line: 1301
DefaultEntry.normalizeDN(Dn) line: 521
DefaultEntry.<init>(SchemaManager, Entry) line: 291
AddOperationContext.<init>(CoreSession, AddRequest) line: 136
DefaultCoreSession.add(AddRequest, LogChange) line: 244
DefaultCoreSession.add(AddRequest) line: 235
LdapCoreSessionConnection.add(AddRequest) line: 189
LdapCoreSessionConnection.add(Entry) line: 220
AddIT.test_DIRSERVER_2109_2() line: 181
{code}
Noralizing of attribute value: unescape is wrong!
{code}
PrepareString.unescape(char[], int) line: 5195
PrepareString.insignifiantSpacesStringAscii(String, boolean) line: 5187
PrepareString.normalize(String, PrepareString$StringType) line: 253
DeepTrimToLowerNormalizer.normalize(String) line: 103
CachingDeepTrimToLowerNormalizer(CachingNormalizer).normalize(String) line: 124
StringValue(AbstractValue<T>).apply(AttributeType) line: 175
StringValue.<init>(AttributeType, String) line: 122
DefaultAttribute.<init>(AttributeType, Attribute) line: 478
DefaultEntry.<init>(SchemaManager, Entry) line: 315
AddOperationContext.<init>(CoreSession, AddRequest) line: 136
DefaultCoreSession.add(AddRequest, LogChange) line: 244
DefaultCoreSession.add(AddRequest) line: 235
LdapCoreSessionConnection.add(AddRequest) line: 189
LdapCoreSessionConnection.add(Entry) line: 220
AddIT.test_DIRSERVER_2109_2() line: 181
{code}
> Apply LDIF with special crafted DN creates two CN attributes
> ------------------------------------------------------------
>
> Key: DIRSERVER-2109
> URL: https://issues.apache.org/jira/browse/DIRSERVER-2109
> Project: Directory ApacheDS
> Issue Type: Bug
> Affects Versions: 2.0.0-M21
> Reporter: Stefan Seelmann
>
> In Studio we have a test that injects an entry with a special crafted DN (cn
> starts with #, contains plus, comma, quotes). This entry is injected using
> the {{@ApplyLdifFiles}} annotation. The injected entry then contains two CN
> attributes, one with proper unescaped value, one with still escaped value.
> Here is the LDIF:
> {code}
> # cn=\#\\\+\, \"öé\",ou=users,ou=system
> dn:: Y249XCNcXFwrXCwgXCLDtsOpXCIsb3U9dXNlcnMsb3U9c3lzdGVt
> objectClass: inetOrgPerson
> objectClass: organizationalPerson
> objectClass: person
> objectClass: top
> cn:: I1wrLCAiw7bDqSI=
> sn:: I1wrLCAiw7bDqSI=
> givenName:: I1wrLCAiw7bDqSI=
> uid:: I1wrLCAiw7bDqSI=
> {code}
> Here the sysout of the injected entry with the two CN attributes:
> {code}
> dn[n]: cn=\#\\\+\, \"öé\",ou=users,ou=system
> objectclass: organizationalPerson
> objectclass: person
> objectclass: inetOrgPerson
> objectclass: top
> givenname: #\+, "öé"
> uid: #\+, "öé"
> cn: #\+, "öé"
> cn: \#\\\+\, \"öé\"
> sn: #\+, "öé"
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
