Still looking at option #1- Here is the stack that leads to the delimiter defined in the DefaultIdentifierConfiguration to always be used: (As in, this code path is not currently configurable) http://pastebin.com/y8buS70f
Wouldn't it make more sense for the delimiter eventually used by SchemaGenerator.generateTables(...) to be the delimiter defined by whichever DBDictionary (which implements IdentifierConfiguration) is provided? Thanks, Aron On Wed, Feb 19, 2014 at 11:52 AM, Aron Lurie <a...@cambridgesemantics.com>wrote: > Apologies for the slew of emails, but I just realized that my changes > failed multiple tests > (testTableOps(org.apache.openjpa.jdbc.sql.identifier.TestDBIdentifiers), > testSchemaOps(org.apache.openjpa.jdbc.sql.identifier.TestDBIdentifiers), > testDBIdentifierOps(org.apache.openjpa.jdbc.sql.identifier.TestDBIdentifiers), > testColumnOps(org.apache.openjpa.jdbc.sql.identifier.TestDBIdentifiers)). > > I will keep working on a solution. > > Aron > > > > On Wed, Feb 19, 2014 at 11:42 AM, Aron Lurie > <a...@cambridgesemantics.com>wrote: > >> This should make it easier to read the code in my previous email: >> >> http://pastebin.com/0Q19a4XY >> >> >> On Wed, Feb 19, 2014 at 11:33 AM, Aron Lurie <a...@cambridgesemantics.com >> > wrote: >> >>> Ok - Here is my proposed patch along the lines of option #1 - >>> >>> Index: >>> openjpa-jdbc/src/main/java/org/apache/openjpa/jdbc/identifier/DefaultIdentifierConfiguration.java >>> =================================================================== >>> --- >>> openjpa-jdbc/src/main/java/org/apache/openjpa/jdbc/identifier/DefaultIdentifierConfiguration.java >>> (revision 1569020) >>> +++ >>> openjpa-jdbc/src/main/java/org/apache/openjpa/jdbc/identifier/DefaultIdentifierConfiguration.java >>> (working copy) >>> @@ -52,7 +52,7 @@ >>> } >>> >>> public String getLeadingDelimiter() { >>> - return IdentifierUtil.DOUBLE_QUOTE; >>> + return IdentifierUtil.ESCAPED_DOUBLE_QUOTE; >>> } >>> >>> public String getIdentifierDelimiter() { >>> @@ -73,7 +73,7 @@ >>> } >>> >>> public String getTrailingDelimiter() { >>> - return IdentifierUtil.DOUBLE_QUOTE; >>> + return IdentifierUtil.ESCAPED_DOUBLE_QUOTE; >>> } >>> >>> public boolean getSupportsDelimitedIdentifiers() { >>> Index: >>> openjpa-lib/src/main/java/org/apache/openjpa/lib/identifier/IdentifierUtil.java >>> =================================================================== >>> --- >>> openjpa-lib/src/main/java/org/apache/openjpa/lib/identifier/IdentifierUtil.java >>> (revision 1569020) >>> +++ >>> openjpa-lib/src/main/java/org/apache/openjpa/lib/identifier/IdentifierUtil.java >>> (working copy) >>> @@ -24,6 +24,7 @@ >>> */ >>> public interface IdentifierUtil { >>> public static final String DOUBLE_QUOTE = "\""; >>> + public static final String ESCAPED_DOUBLE_QUOTE = "\\\""; >>> public static final String DOT = "."; >>> public static final String UNDERSCORE = "_"; >>> public static final String SPACE = " "; >>> >>> >>> >>> >>> >>> On Mon, Feb 17, 2014 at 5:48 PM, Rick Curtis <curti...@gmail.com> wrote: >>> >>>> I would start with option #1 >>>> >>>> Thanks, >>>> Rick >>>> >>>> >>>> On Mon, Feb 17, 2014 at 11:31 AM, Aron Lurie < >>>> a...@cambridgesemantics.com>wrote: >>>> >>>> > Hi All, >>>> > >>>> > Running on trunk, I'm currently facing a problem very similar to >>>> > OPENJPA-1540. When I run the ReverseMappingTool on an Oracle database >>>> with >>>> > a lowercase column name, I end up with syntactically incorrect source >>>> code. >>>> > Whereas I should get output like this: >>>> > >>>> > @Column(name="\"foobar\"") >>>> > >>>> > I instead get un-compileable output like this: >>>> > >>>> > @Column(name=""foobar"") >>>> > >>>> > The reason why the problem only affects lowercase column names is >>>> because >>>> > Oracle generally requires names in all upper case. If a name is >>>> specified >>>> > with quotes around it, then it can be allowed to contain lowercase >>>> > characters. >>>> > >>>> > I generally see two ways I can fix this: >>>> > >>>> > 1) Change the way the DBIdentifier's are initialized to surround the >>>> > identifier with a delimiter that includes the escape \. >>>> > >>>> > 2) Change the serialization stack (i.e. >>>> > AnnotationPersistenceMetaDataSerializer.serialize and >>>> > AnnotationEntry.toString) to run some character escape utility (like >>>> > StringEscapeUtils.escapeJava) at String construction time. >>>> > >>>> > Which approach here is best? Or is there a better way I haven't >>>> considered? >>>> > >>>> > Thanks, >>>> > Aron >>>> > >>>> >>>> >>>> >>>> -- >>>> *Rick Curtis* >>>> >>> >>> >> >
