[ https://issues.apache.org/jira/browse/DDLUTILS-246?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Tony Rippy updated DDLUTILS-246: -------------------------------- Attachment: Issue246.patch Here's a patch that properly escapes table names before calling JDBC's DatabaseMetaData functions. I have verified that this fixes the issue, at least with PostgreSQL. Still needs testing on other databases. > JdbcModelReader merges metadata between tables > ---------------------------------------------- > > Key: DDLUTILS-246 > URL: https://issues.apache.org/jira/browse/DDLUTILS-246 > Project: DdlUtils > Issue Type: Bug > Components: Core (No specific database) > Affects Versions: 1.0 > Environment: Java 1.5, Oracle 8/9/10/11, DB2 8/9, MSSQL 2000/2005, > possibly others > Reporter: Frank Jakop > Assignee: Thomas Dudziak > Priority: Critical > Attachments: Issue246.patch > > > When reading the columns for a table, JdbcModelReader relies on metadata > given by DatabaseMetaDataWrapper. > Given two tables > TABLE FOO_MATIC (COL1 INTEGER) > TABLE FOOTMATIC (COL2 INTEGER, COL3 INTEGER) > the returned columns for table FOO_MATIC is (COL1, COL2, COL3). No need to > say, that subsequent queries on FOO_MATIC fail due to nonexistent columns > COL2 and COL3. > The reason for this behavior is that the underscore is interpreted as a > wildcard character in DatabaseMetaData.getColumns(String,String), so both > tables match the pattern FOO_MATIC. > The workaround I use is to re-check the table name in > JdbcModelReader.readColumns(DatabaseMetaDataWrapper,String) > while (columnData.next()) > { > if (columnData.getString("TABLE_NAME").equals(tableName)) > { > Map values = readColumns(columnData, getColumnsForColumn()); > columns.add(readColumn(metaData, values)); > } > } -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.