Oracle Formatter and ORA-00904

Since Oracle interprets quoted and non-quoted identifiers differently, the database throws ORA-00904 (“invalid identifier”) when Knex’s formatter wraps my identifiers in quotes.

For example:

var tableName = 'employees';
var predicate = {name: 'Johnson'};

knex.select('*').from(tableName).where(predicate); // throws ORA-00904

var newTableName = tableName.toUpperCase();
var newPredicate = {};
for (property in predicate) {
  if (predicate.hasOwnProperty(property)) {
    newPredicate[property.toUpperCase()] = predicate[property];
  }
}

knex.select('*').from(newTableName).where(newPredicate); // works fine

In order to avoid this error, I either find myself .toUpperCase()-ing all my variables containing identifiers before passing them to Knex or rewriting Formatter_Oracle.prototype.wrapValue in lib/dialects/oracle/formatter.js to do it for me:

// Wraps a value (column, tableName) with the correct ticks.
Formatter_Oracle.prototype.wrapValue = function(value) {
  // jthomm adding `.toUpperCase()`
  return (value !== '*' ? '"' + value.toUpperCase() + '"' : '*');
};

I’m not suggesting you guys implement it this way but I was wondering if there is an idiomatic way for me override Formatter_Oracle.prototype.wrapValue with my own function when I instantiate the client or something. Right now I’m literally changing the code in my local Knex installation every time I want to connect to an Oracle database.

[mhofer117]

For anyone finding this issue on google. Here’s the solution using wrapIdentifier with the latest release (0.14.4):

var knex = require("knex")({
  "client"     : "oracledb",
  "connection" : { your connection string },
  // convert all identifiers to uppercase (Oracle standard)
  "wrapIdentifier": (value, origImpl, queryContext) => origImpl(value.toUpperCase())
});

[elhigu]

In knex 0.14 one will be able to override formatters for identifiers https://github.com/tgriesser/knex/pull/2217