[METAMODEL-1191] JdbcDataContext get wrong schema name on MySQL

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 5.0.0
• Labels: jdbc
• Environment: MySQL 8.0
• Assignee: Unassigned
• Reporter: Hao Ling
• Created: 17/Aug/18 06:28
• Updated: 15/Dec/19 10:31

Description

A MySQL database has multiple schemas: "dc", "rdc", "rdc_test" ...

The jdbc connection url is jdbc:mysql://192.168.1.199:3306/rdc_test?useUnicode=true...

When use JdbcDataContext.getDefaultSchemaName() to obtain the default schema,

we expected this function to return "rdc_test"，which is the part of connection url, but it returns "dc". 

After debug, we found the issue is caused by following code:

JdbcDataContext.java  Line 672:

 

for (int i = 0; i < schemaNames.size() && !found; i++) {
    String schemaName = schemaNames.get(i);
    if (lastToken.indexOf(schemaName) != -1) {
        result = schemaName;
        found = true;
    }
}

 

Here the lastToken is "rdc_test?useUnicode=....." which is extracted from url, it's correct. When use indexOf() to match schemaNames array, the first schema "dc" is matched and returned.

The most simplest solution I think is to get the longest schema which is a part of lastToken. Below is the fixed code. We tested it and works fine.

 

for (int i = 0; i < schemaNames.size(); i++) {
    String schemaName = schemaNames.get(i);
    if (lastToken.indexOf(schemaName) != -1) {
        if (result == null) {
            result = schemaName;
        } else {
            result = schemaName.length() > result.length() ? schemaName : result;
        }
        found = true;
    }
}

 

Consider the same string with schema name may also appears in the right part of '?' character, is it better to use startWith() instead of indexOf()? Like this:

if (lastToken.startWith(schemaName)) 

 

 

 

 

 

Issue Links

• links to: GitHub Pull Request #187