waterline-utils/lib/query/tokenizer.js

Various utilities for working with Waterline queries and adapters.

/* eslint-disable no-use-before-define */
//  ████████╗ ██████╗ ██╗  ██╗███████╗███╗   ██╗██╗███████╗███████╗██████╗
//  ╚══██╔══╝██╔═══██╗██║ ██╔╝██╔════╝████╗  ██║██║╚══███╔╝██╔════╝██╔══██╗
//     ██║   ██║   ██║█████╔╝ █████╗  ██╔██╗ ██║██║  ███╔╝ █████╗  ██████╔╝
//     ██║   ██║   ██║██╔═██╗ ██╔══╝  ██║╚██╗██║██║ ███╔╝  ██╔══╝  ██╔══██╗
//     ██║   ╚██████╔╝██║  ██╗███████╗██║ ╚████║██║███████╗███████╗██║  ██║
//     ╚═╝    ╚═════╝ ╚═╝  ╚═╝╚══════╝╚═╝  ╚═══╝╚═╝╚══════╝╚══════╝╚═╝  ╚═╝
//
// The tokenizer is responsible for taking a nested Waterline statement and
// turning it into a flat set of keys. This allows the query to more easily be
// parsed and prevents further recusion as the query progresses to eventually
// end up as a native query.
//
// In most cases this will not be implemented by adapter authors but will be used
// inside a database driver's `compileStatement` machine.

var _ = require('@sailshq/lodash');


//  ╔═╗╦═╗╔═╗╔═╗╔═╗╔═╗╔═╗╔═╗╦═╗  ╔═╗╦ ╦╔╗╔╔═╗╔╦╗╦╔═╗╔╗╔╔═╗
//  ╠═╝╠╦╝║ ║║  ║╣ ╚═╗╚═╗║ ║╠╦╝  ╠╣ ║ ║║║║║   ║ ║║ ║║║║╚═╗
//  ╩  ╩╚═╚═╝╚═╝╚═╝╚═╝╚═╝╚═╝╩╚═  ╚  ╚═╝╝╚╝╚═╝ ╩ ╩╚═╝╝╚╝╚═╝

// These are the identifiers used in RQL for various keys
var identifiers = {
  'select': 'SELECT',
  'from': 'FROM',
  'or': 'OR',
  'and': 'AND',
  'not': 'NOT',
  'nin': 'NOTIN',
  'in': 'IN',
  'distinct': 'DISTINCT',
  'count': 'COUNT',
  'min': 'MIN',
  'max': 'MAX',
  'sum': 'SUM',
  'avg': 'AVG',
  'limit': 'LIMIT',
  'skip': 'SKIP',
  'groupBy': 'GROUPBY',
  'orderBy': 'ORDERBY',
  'where': 'WHERE',
  'insert': 'INSERT',
  'into': 'INTO',
  'update': 'UPDATE',
  'using': 'USING',
  'del': 'DELETE',
  'join': 'JOIN',
  'innerJoin': 'JOIN',
  'outerJoin': 'JOIN',
  'crossJoin': 'JOIN',
  'leftJoin': 'JOIN',
  'leftOuterJoin': 'JOIN',
  'rightJoin': 'JOIN',
  'rightOuterJoin': 'JOIN',
  'fullOuterJoin': 'JOIN',
  'union': 'UNION',
  'unionAll': 'UNIONALL',
  'as': 'AS',
  '>': 'OPERATOR',
  '<': 'OPERATOR',
  '<>': 'OPERATOR',
  '<=': 'OPERATOR',
  '>=': 'OPERATOR',
  '!=': 'OPERATOR',
  'like': 'OPERATOR',
  'opts': 'OPTS',
  'returning': 'RETURNING'
};

// If these identifiers are found within a WHERE clause, treat them as regular keys.
var WHERE_EXEMPT = [
  'from',
  'distinct',
  'count',
  'min',
  'max',
  'sum',
  'avg',
  'insert',
  'union',
  'as',
  'returning',
  'join'
];

// These are the Data Manipulation Identifiers that denote a subquery
var DML_IDENTIFIERS = [
  'select',
  'insert',
  'update',
  'del'
];

//  ╔╦╗╔═╗╦╔═╔═╗╔╗╔╦╔═╗╔═╗  ┌─┐┌┐  ┬┌─┐┌─┐┌┬┐
//   ║ ║ ║╠╩╗║╣ ║║║║╔═╝║╣   │ │├┴┐ │├┤ │   │
//   ╩ ╚═╝╩ ╩╚═╝╝╚╝╩╚═╝╚═╝  └─┘└─┘└┘└─┘└─┘ ┴
// @obj {Object} - the token obj being processed
// @processor {Object} - a value to insert between each key in the array
var tokenizeObject = function tokenizeObject(obj, processor, parent, isSubQuery, results) {
  // If this obj represent a sub-query, add a sub query token
  if (isSubQuery) {
    results.push({
      type: 'SUBQUERY',
      value: null
    });
  }

  // Determine whether we're currently processing a WHERE clause.
  // We use a stack of counters to do this, adding and removing from the stack
  // when subqueries are started / ended, and incrementing/decrementing the
  // first counter in the stack when WHERE clauses are started and ended.
  var inWhere = !!_.reduce(results, function(memo, item) {
    if (item.type === 'IDENTIFIER' || item.value === 'WHERE') {
      memo[0]++;
    }
    else if (item.type === 'ENDIDENTIFIER' || item.value === 'WHERE') {
      memo[0]--;
    }
    else if (item.type === 'SUBQUERY') {
      memo.unshift(0);
    }
    else if (item.type === 'ENDSUBQUERY') {
      memo.unshift();
    }
    return memo;
  }, [0])[0];

  _.each(_.keys(obj), function tokenizeKey(key, idx) {
    // Check if the key is a known identifier
    var isIdentitifier = identifiers[key];

    // If so, look ahead at it's value to determine what to do next.
    if (isIdentitifier) {

      //  ╦ ╦╦ ╦╔═╗╦═╗╔═╗  ╔═╗═╗ ╦╔═╗╔╦╗╔═╗╔╦╗
      //  ║║║╠═╣║╣ ╠╦╝║╣───║╣ ╔╩╦╝║╣ ║║║╠═╝ ║
      //  ╚╩╝╩ ╩╚═╝╩╚═╚═╝  ╚═╝╩ ╚═╚═╝╩ ╩╩   ╩

      // If we're currently inside a WHERE clause and we encounter a key that is normally an identifier
      // but is in the WHERE_EXEMPT list, process it as a regular key.
      if (inWhere && _.contains(WHERE_EXEMPT, key)) {
        results.push({
          type:  'KEY',
          value: key
        });
        if (_.isObject(obj[key])) {
          tokenizeObject(obj[key], undefined, undefined, undefined, results);
          return;
        }
        results.push({
          type: 'VALUE',
          value: obj[key]
        });
        return;
      }


      //  ╔═╗╔═╗╔═╗╦═╗╔═╗╔╦╗╔═╗╦═╗  ╔═╗╦═╗╔═╗╔╦╗╦╔═╗╔═╗╔╦╗╔═╗╔═╗
      //  ║ ║╠═╝║╣ ╠╦╝╠═╣ ║ ║ ║╠╦╝  ╠═╝╠╦╝║╣  ║║║║  ╠═╣ ║ ║╣ ╚═╗
      //  ╚═╝╩  ╚═╝╩╚═╩ ╩ ╩ ╚═╝╩╚═  ╩  ╩╚═╚═╝═╩╝╩╚═╝╩ ╩ ╩ ╚═╝╚═╝

      // If the identifier is an OPERATOR, add it's tokens
      if (identifiers[key] === 'OPERATOR') {
        // If there is a parent and the previous key in the results isn't
        // a KEY add it's key first. This is used when a key has multiple
        // criteria. EX: { values: { '>': 100, '<': 200 }}
        if (parent && _.last(results).type !== 'KEY') {
          results.push({
            type: 'KEY',
            value: parent
          });
        }

        processOperator(key, obj[key], results);
        return;
      }

      // If the identifier is an IN
      if (identifiers[key] === 'IN') {
        processIn(obj[key], undefined, results);
        return;
      }

      // If the identifier is an OR, start a group and add each token.
      if (identifiers[key] === 'OR') {
        processOr(obj[key], results);
        return;
      }

      // If the identifier is an AND, start a group and add each token.
      if (identifiers[key] === 'AND') {
        processAnd(obj[key], results);
        return;
      }

      // If the identifier is a NOT
      if (identifiers[key] === 'NOT') {
        processNot(obj[key], results);
        return;
      }

      // If the identifier is a NOTIN
      if (identifiers[key] === 'NOTIN') {
        processIn(obj[key], true, results);
        return;
      }

      //  ╔═╗ ╦ ╦╔═╗╦═╗╦╔═╗╔═╗
      //  ║═╬╗║ ║║╣ ╠╦╝║║╣ ╚═╗
      //  ╚═╝╚╚═╝╚═╝╩╚═╩╚═╝╚═╝

      // If the identifier is a FROM, add it's token
      if (identifiers[key] === 'FROM') {
        processFrom(obj[key], results);
        return;
      }

      // If the identifier is a WHERE, add it's token and process it's values
      if (identifiers[key] === 'WHERE') {
        processWhere(obj[key], results);
        return;
      }

      // If the identifier is a GROUP BY aggregation
      if (identifiers[key] === 'GROUPBY') {
        processGroupBy(obj[key], results);
        return;
      }

      // If the identifier is an ORDER BY, add the sort options
      if (identifiers[key] === 'ORDERBY') {
        processOrderBy(obj[key], results);
        return;
      }

      //  ╔╦╗╔╦╗╦    ╔═╗╔═╗╔╦╗╔╦╗╔═╗╔╗╔╔╦╗╔═╗
      //   ║║║║║║    ║  ║ ║║║║║║║╠═╣║║║ ║║╚═╗
      //  ═╩╝╩ ╩╩═╝  ╚═╝╚═╝╩ ╩╩ ╩╩ ╩╝╚╝═╩╝╚═╝

      // If the identifier is a SELECT, add it's token
      if (identifiers[key] === 'SELECT') {
        processSelect(obj[key], results);
        return;
      }

      // If the identifier is an INSERT, add it's token
      if (identifiers[key] === 'INSERT') {
        processInsert(obj[key], results);
        return;
      }

      // If the identifier is an UPDATE, add it's token
      if (identifiers[key] === 'UPDATE') {
        processUpdate(obj[key], results);
        return;
      }

      // If the identifier is a DELETE, add it's token
      if (identifiers[key] === 'DELETE') {
        processDelete(results);
        return;
      }

      // If the identifier is an INTO, add it's token
      if (identifiers[key] === 'INTO') {
        processInto(obj[key], results);
        return;
      }

      // If the identifier is an USING, add it's token
      if (identifiers[key] === 'USING') {
        processUsing(obj[key], results);
        return;
      }

      //  ╔═╗╔═╗╔═╗╦═╗╔═╗╔═╗╔═╗╔╦╗╔═╗╔═╗
      //  ╠═╣║ ╦║ ╦╠╦╝║╣ ║ ╦╠═╣ ║ ║╣ ╚═╗
      //  ╩ ╩╚═╝╚═╝╩╚═╚═╝╚═╝╩ ╩ ╩ ╚═╝╚═╝

      // If the identifier is a AVG
      if (identifiers[key] === 'AVG') {
        processAggregations(obj[key], 'AVG', results);
        return;
      }

      // If the identifier is a SUM
      if (identifiers[key] === 'SUM') {
        processAggregations(obj[key], 'SUM', results);
        return;
      }

      // If the identifier is a MIN
      if (identifiers[key] === 'MIN') {
        processAggregations(obj[key], 'MIN', results);
        return;
      }

      // If the identifier is a MAX
      if (identifiers[key] === 'MAX') {
        processAggregations(obj[key], 'MAX', results);
        return;
      }

      // If the identifier is a COUNT
      if (identifiers[key] === 'COUNT') {
        processAggregations(obj[key], 'COUNT', results);
        return;
      }

      //  ╔═╗╔╦╗╦ ╦╔═╗╦═╗
      //  ║ ║ ║ ╠═╣║╣ ╠╦╝
      //  ╚═╝ ╩ ╩ ╩╚═╝╩╚═

      // If the identifier is a LIMIT
      if (identifiers[key] === 'LIMIT') {
        processPagination(obj[key], 'LIMIT', results);
        return;
      }

      // If the indetifier is an SKIP
      if (identifiers[key] === 'SKIP') {
        processPagination(obj[key], 'SKIP', results);
        return;
      }

      // AS is only available on sub queries
      if (identifiers[key] === 'AS') {
        if (!isSubQuery) {
          return;
        }
        processAs(obj[key], results);
        return;
      }

      // If the indetifier is an RETURNING
      if (identifiers[key] === 'RETURNING') {
        processReturning(obj[key], results);
        return;
      }

      //   ╦╔═╗╦╔╗╔╔═╗
      //   ║║ ║║║║║╚═╗
      //  ╚╝╚═╝╩╝╚╝╚═╝

      // If the identifier is a JOIN, add it's token and process the joins
      if (identifiers[key] === 'JOIN') {
        processJoin(obj[key], key, results);
        return;
      }

      //  ╦ ╦╔╗╔╦╔═╗╔╗╔╔═╗
      //  ║ ║║║║║║ ║║║║╚═╗
      //  ╚═╝╝╚╝╩╚═╝╝╚╝╚═╝

      // If the identifier is a UNION
      if (identifiers[key] === 'UNION') {
        processUnion(obj[key], 'UNION', results);
        return;
      }

      // If the identifier is a UNIONALL
      if (identifiers[key] === 'UNIONALL') {
        processUnion(obj[key], 'UNIONALL', results);
        return;
      }

      //  ╔═╗╔═╗╔╦╗╔═╗
      //  ║ ║╠═╝ ║ ╚═╗
      //  ╚═╝╩   ╩ ╚═╝

      // Handle any known values in the opts. Opts must be a dictionary.
      if (identifiers[key] === 'OPTS') {
        if (!_.isPlainObject(obj[key])) {
          return;
        }

        _.each(obj[key], function processOpt(val, key) {
          // Handle PG schema values
          if (key === 'schema') {
            return processSchema(val, results);
          }
        });

        return;
      }

      // Add the identifier
      results.push({
        type: identifiers[key],
        value: key
      });

      // If the identifier is an array, loop through each item and tokenize
      if (_.isArray(obj[key])) {
        _.each(obj[key], function tokenizeJoinPiece(expr) {
          tokenizeObject(expr, undefined, undefined, undefined, results);
        });

        return;
      }

      // If the identifier is an object, continue tokenizing it
      if (_.isPlainObject(obj[key])) {
        tokenizeObject(obj[key], undefined, key, undefined, results);
        return;
      }

      // Otherwise WTF?
      return;
    }

    // Otherwise add the token for the key
    results.push({
      type: 'KEY',
      value: key
    });

    // If the value is an object, recursively parse it unless it matches as
    // a sub query
    if (_.isPlainObject(obj[key])) {
      // Check if the value is a subquery first
      var subQuery = checkForSubquery(obj[key], results);
      if (subQuery) {
        return;
      }

      // Otherwise parse the object
      tokenizeObject(obj[key], undefined, key, undefined, results);
      return;
    }

    // If the value is a primitive add it's token
    results.push({
      type: 'VALUE',
      value: obj[key]
    });

    // If there is a processor and we are not on the last key, add it as well.
    // This is used for things like:
    // {
    //   not: {
    //     firstName: 'foo',
    //     lastName: 'bar'
    //   }
    // }
    // Where we need to insert a NOT statement between each key
    if (processor && (_.keys(obj).length > idx + 1)) {
      results.push(processor);
    }
  });

  // If this obj represent a sub-query, close the sub query token
  if (isSubQuery) {
    results.push({
      type: 'ENDSUBQUERY',
      value: null
    });
  }
};


//  ╔═╗╦ ╦╔═╗╔═╗╦╔═  ╔═╗╔═╗╦═╗  ╔═╗╦ ╦╔╗ ╔═╗ ╦ ╦╔═╗╦═╗╦ ╦
//  ║  ╠═╣║╣ ║  ╠╩╗  ╠╣ ║ ║╠╦╝  ╚═╗║ ║╠╩╗║═╬╗║ ║║╣ ╠╦╝╚╦╝
//  ╚═╝╩ ╩╚═╝╚═╝╩ ╩  ╚  ╚═╝╩╚═  ╚═╝╚═╝╚═╝╚═╝╚╚═╝╚═╝╩╚═ ╩
var checkForSubquery = function checkForSubquery(value, results) {
  var isSubquery = false;

  // Check if the object has any top level DML identifiers
  _.each(value, function checkForIdentifier(val, key) {
    if (_.indexOf(DML_IDENTIFIERS, key) < 0) {
      return;
    }
    isSubquery = true;
  });

  // If this is a sub query, tokenize it as such
  if (isSubquery) {
    tokenizeObject(value, undefined, undefined, isSubquery, results);
    return isSubquery;
  }

  return isSubquery;
};


//  ╔═╗╔═╗╔═╗╦═╗╔═╗╔╦╗╔═╗╦═╗╔═╗
//  ║ ║╠═╝║╣ ╠╦╝╠═╣ ║ ║ ║╠╦╝╚═╗
//  ╚═╝╩  ╚═╝╩╚═╩ ╩ ╩ ╚═╝╩╚═╚═╝
var processOperator = function processOperator(operator, value, results) {
  // Add the operator to the results
  results.push({
    type: 'OPERATOR',
    value: operator
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  // Add the operator to the results
  results.push({
    type: 'ENDOPERATOR',
    value: operator
  });
};


//  ╔═╗╔═╗╦  ╔═╗╔═╗╔╦╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ╚═╗║╣ ║  ║╣ ║   ║   ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╚═╝╚═╝╩═╝╚═╝╚═╝ ╩   ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processSelect = function processSelect(value, results) {
  // Check if a distinct or other key is being used
  if (_.isPlainObject(value) && !_.isArray(value)) {
    if (value.distinct) {
      // Add the distinct to the results
      results.push({
        type: 'IDENTIFIER',
        value: 'DISTINCT'
      });

      // Add the value to the results
      results.push({
        type: 'VALUE',
        value: value.distinct
      });

      // Add the enddistinct to the results
      results.push({
        type: 'ENDIDENTIFIER',
        value: 'DISTINCT'
      });

      return;
    }
  }

  // If the value is not an array or object, add the value
  if (!_.isPlainObject(value) && !_.isArray(value)) {
    // Add the SELECT to the results
    results.push({
      type: 'IDENTIFIER',
      value: 'SELECT'
    });

    // Add the value to the results
    results.push({
      type: 'VALUE',
      value: value
    });

    // Add the ENDSELECT to the results
    results.push({
      type: 'ENDIDENTIFIER',
      value: 'SELECT'
    });

    return;
  }

  // If the value is not an array, make it one so that we can process each
  // element.
  if (!_.isArray(value)) {
    value = [value];
  }

  // Process each item in there SELECT statement and process subqueries as
  // needed.
  _.each(value, function processSelectKey(val) {
    // Add the SELECT to the results
    results.push({
      type: 'IDENTIFIER',
      value: 'SELECT'
    });

    // If the value isn't an object, no need to process it further
    if (!_.isPlainObject(val)) {
      results.push({
        type: 'VALUE',
        value: val
      });
    }

    // Check if the object is a sub-query
    if (_.isPlainObject(val)) {
      var isSubquery = checkForSubquery(val, results);

      // If it's not, add it's value
      if (!isSubquery) {
        results.push({
          type: 'VALUE',
          value: val
        });
      }
    }

    // Add the ENDSELECT to the results
    results.push({
      type: 'ENDIDENTIFIER',
      value: 'SELECT'
    });
  });
};


//  ╔═╗╦═╗╔═╗╔╦╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ╠╣ ╠╦╝║ ║║║║  ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╚  ╩╚═╚═╝╩ ╩  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processFrom = function processFrom(value, results) {
  // Check if a schema is being used
  if (_.isObject(value) && !_.isFunction(value) && !_.isArray(value)) {
    // Add the FROM identifier
    results.push({
      type: 'IDENTIFIER',
      value: 'FROM'
    });

    // Check if a subquery is being used
    var isSubQuery = checkForSubquery(value, results);

    if (!isSubQuery && value.table) {
      results.push({
        type: 'VALUE',
        value: value.table
      });
    }

    results.push({
      type: 'ENDIDENTIFIER',
      value: 'FROM'
    });

    return;
  }

  // Otherwise just add the FROM identifier and value
  results.push({
    type: 'IDENTIFIER',
    value: 'FROM'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'FROM'
  });
};


//  ╔═╗╔═╗╦ ╦╔═╗╔╦╗╔═╗  ╔═╗╔═╗╔╦╗
//  ╚═╗║  ╠═╣║╣ ║║║╠═╣  ║ ║╠═╝ ║
//  ╚═╝╚═╝╩ ╩╚═╝╩ ╩╩ ╩  ╚═╝╩   ╩
var processSchema = function processSchema(value, results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'SCHEMA'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'SCHEMA'
  });
};


//  ╦╔╗╔╔═╗╔═╗╦═╗╔╦╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ║║║║╚═╗║╣ ╠╦╝ ║   ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╩╝╚╝╚═╝╚═╝╩╚═ ╩   ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processInsert = function processInsert(value, results) {
  // Add the insert statment
  results.push({
    type: 'IDENTIFIER',
    value: 'INSERT'
  });

  // Check if an array is being used
  if (_.isArray(value)) {
    _.each(value, function appendInsertEach(record, idx) {
      // Add a group clause
      results.push({
        type: 'GROUP',
        value: idx
      });

      // If the value is a plain object, proccess it
      if (_.isObject(record) && !_.isFunction(record) && !_.isArray(record)) {
        _.each(_.keys(record), function appendInsertValue(key) {
          results.push({
            type: 'KEY',
            value: key
          });

          results.push({
            type: 'VALUE',
            value: record[key]
          });
        });
      }

      // Close the group clause
      results.push({
        type: 'ENDGROUP',
        value: idx
      });
    });
  }

  // Check if a plain object value is being used
  if (_.isObject(value) && !_.isFunction(value) && !_.isArray(value)) {
    _.each(_.keys(value), function appendInsertValue(key) {
      results.push({
        type: 'KEY',
        value: key
      });

      results.push({
        type: 'VALUE',
        value: value[key]
      });
    });
  }

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'INSERT'
  });
};


//  ╦╔╗╔╔╦╗╔═╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ║║║║ ║ ║ ║  ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╩╝╚╝ ╩ ╚═╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processInto = function processInto(value, results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'INTO'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'INTO'
  });
};


//  ╦ ╦╔═╗╔╦╗╔═╗╔╦╗╔═╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ║ ║╠═╝ ║║╠═╣ ║ ║╣   ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╚═╝╩  ═╩╝╩ ╩ ╩ ╚═╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processUpdate = function processUpdate(value, results) {
  // Add the update statment
  results.push({
    type: 'IDENTIFIER',
    value: 'UPDATE'
  });

  // Check if a value is being used
  if (_.isObject(value)) {
    _.each(_.keys(value), function appendUpdateValue(key) {
      results.push({
        type: 'KEY',
        value: key
      });

      results.push({
        type: 'VALUE',
        value: value[key]
      });
    });
  }

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'UPDATE'
  });
};


//  ╦ ╦╔═╗╦╔╗╔╔═╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ║ ║╚═╗║║║║║ ╦  ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╚═╝╚═╝╩╝╚╝╚═╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processUsing = function processUsing(value, results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'USING'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'USING'
  });
};


//  ╔╦╗╔═╗╦  ╔═╗╔╦╗╔═╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//   ║║║╣ ║  ║╣  ║ ║╣   ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ═╩╝╚═╝╩═╝╚═╝ ╩ ╚═╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processDelete = function processDelete(results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'DELETE'
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'DELETE'
  });
};


//  ╔╗╔╔═╗╔╦╗  ╔═╗╔═╗╔╗╔╔╦╗╦╔╦╗╦╔═╗╔╗╔
//  ║║║║ ║ ║   ║  ║ ║║║║ ║║║ ║ ║║ ║║║║
//  ╝╚╝╚═╝ ╩   ╚═╝╚═╝╝╚╝═╩╝╩ ╩ ╩╚═╝╝╚╝
var processNot = function processNot(value, results) {
  // Add a condition
  var condition = {
    type: 'CONDITION',
    value: 'NOT'
  };

  results.push(condition);

  // Tokenize the values within the condition
  if (_.isObject(value) && !_.isFunction(value) && !_.isArray(value)) {
    tokenizeObject(value, condition, undefined, undefined, results);
    return;
  }

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDCONDITION',
    value: 'NOT'
  });
};


//  ╦╔╗╔  ╔═╗╔═╗╔╗╔╔╦╗╦╔╦╗╦╔═╗╔╗╔
//  ║║║║  ║  ║ ║║║║ ║║║ ║ ║║ ║║║║
//  ╩╝╚╝  ╚═╝╚═╝╝╚╝═╩╝╩ ╩ ╩╚═╝╝╚╝
var processIn = function processIn(value, negate, results) {
  // Add a condition
  var startCondition;
  var endCondition;

  if (negate) {
    startCondition = {
      type: 'CONDITION',
      value: 'NOTIN'
    };

    endCondition = {
      type: 'ENDCONDITION',
      value: 'NOTIN'
    };
  } else {
    startCondition = {
      type: 'CONDITION',
      value: 'IN'
    };

    endCondition = {
      type: 'ENDCONDITION',
      value: 'IN'
    };
  }

  results.push(startCondition);

  // If the value isn't an object, no need to process it further
  if (!_.isPlainObject(value)) {
    results.push({
      type: 'VALUE',
      value: value
    });
  }

  // Check if the object is a sub-query
  if (_.isObject(value) && !_.isFunction(value) && !_.isArray(value)) {
    var isSubquery = checkForSubquery(value, results);

    // If it's not, add it's value
    if (!isSubquery) {
      results.push({
        type: 'VALUE',
        value: value
      });
    }
  }

  results.push(endCondition);
};


//  ╦ ╦╦ ╦╔═╗╦═╗╔═╗  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗
//  ║║║╠═╣║╣ ╠╦╝║╣   ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║
//  ╚╩╝╩ ╩╚═╝╩╚═╚═╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩
var processWhere = function processWhere(value, results) {
  // Tokenize the where and then call the tokenizer on the where values
  results.push({
    type: 'IDENTIFIER',
    value: 'WHERE'
  });

  tokenizeObject(value, undefined, undefined, undefined, results);

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'WHERE'
  });
};


//  ╔═╗╦═╗  ╔═╗╦═╗╔═╗╦ ╦╔═╗╦╔╗╔╔═╗
//  ║ ║╠╦╝  ║ ╦╠╦╝║ ║║ ║╠═╝║║║║║ ╦
//  ╚═╝╩╚═  ╚═╝╩╚═╚═╝╚═╝╩  ╩╝╚╝╚═╝
var processOr = function processOr(value, results) {
  // Add the Or token
  results.push({
    type: 'CONDITION',
    value: 'OR'
  });

  // For each condition in the OR, add a group token and process the criteria.
  _.forEach(value, function appendOrCrieria(criteria, idx) {
    // Start a group
    results.push({
      type: 'GROUP',
      value: idx
    });

    tokenizeObject(criteria, undefined, undefined, undefined, results);

    // End a group
    results.push({
      type: 'ENDGROUP',
      value: idx
    });
  });

  // Close the condition
  results.push({
    type: 'ENDCONDITION',
    value: 'OR'
  });
};


//  ╔═╗╔╗╔╔╦╗  ╔═╗╦═╗╔═╗╦ ╦╔═╗╦╔╗╔╔═╗
//  ╠═╣║║║ ║║  ║ ╦╠╦╝║ ║║ ║╠═╝║║║║║ ╦
//  ╩ ╩╝╚╝═╩╝  ╚═╝╩╚═╚═╝╚═╝╩  ╩╝╚╝╚═╝
var processAnd = function processAnd(value, results) {
  // Only process grouped AND's if the value is an array
  if (!_.isArray(value)) {
    return;
  }

  // Add the AND token
  results.push({
    type: 'CONDITION',
    value: 'AND'
  });

  // For each condition in the OR, add a group token and process the criteria.
  _.each(value, function appendAndCrieria(criteria, idx) {
    // Start a group
    results.push({
      type: 'GROUP',
      value: idx
    });

    tokenizeObject(criteria, undefined, undefined, undefined, results);

    // End a group
    results.push({
      type: 'ENDGROUP',
      value: idx
    });
  });

  // Close the condition
  results.push({
    type: 'ENDCONDITION',
    value: 'AND'
  });
};


//   ╦╔═╗╦╔╗╔  ╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╦╗╔═╗╔╗╔╔╦╗╔═╗
//   ║║ ║║║║║  ╚═╗ ║ ╠═╣ ║ ║╣ ║║║║╣ ║║║ ║ ╚═╗
//  ╚╝╚═╝╩╝╚╝  ╚═╝ ╩ ╩ ╩ ╩ ╚═╝╩ ╩╚═╝╝╚╝ ╩ ╚═╝
var processJoin = function processJoin(value, joinType, results) {
  // Ensure we have an array value
  if (!_.isArray(value)) {
    value = [value];
  }

  _.each(value, function processJoinInstructions(joinInstructions) {
    // Add a JOIN token
    results.push({
      type: 'IDENTIFIER',
      value: joinType.toUpperCase()
    });

    // Ensure the instructions include a FROM and an ON and that the ON
    // is made up of two table keys.
    if (!_.has(joinInstructions, 'from') || !_.has(joinInstructions, 'on')) {
      throw new Error('Invalid join instructions');
    }

    // Check if this is an AND or an OR join statement. An AND statement will
    // just be an array of conditions and an OR statement will have a single
    // OR key as the value.

    // Process AND
    if (_.isArray(joinInstructions.on)) {
      (function andInstructions() {
        var JOIN_TABLE = joinInstructions.from;
        results.push({ type: 'KEY', value: 'TABLE' });
        results.push({ type: 'VALUE', value: JOIN_TABLE });

        _.each(joinInstructions.on, function onSet(set) {
          var PARENT_TABLE = _.first(_.keys(set));
          var CHILD_TABLE = _.keys(set)[1];
          var PARENT_COLUMN = set[_.first(_.keys(set))];
          var CHILD_COLUMN = set[_.keys(set)[1]];

          var setKeys = [
            { type: 'COMBINATOR', value: 'AND' },
            { type: 'KEY', value: 'TABLE_KEY' },
            { type: 'VALUE', value: PARENT_TABLE },
            { type: 'KEY', value: 'COLUMN_KEY' },
            { type: 'VALUE', value: PARENT_COLUMN },
            { type: 'KEY', value: 'TABLE_KEY' },
            { type: 'VALUE', value: CHILD_TABLE },
            { type: 'KEY', value: 'COLUMN_KEY' },
            { type: 'VALUE', value: CHILD_COLUMN }
          ];

          _.each(setKeys, function appendSet(set) {
            results.push(set);
          });
        });
      })();

      // Process OR
    } else if (_.isArray(joinInstructions.on.or)) {
      (function orInstructions() {
        var JOIN_TABLE = joinInstructions.from;
        results.push({ type: 'KEY', value: 'TABLE' });
        results.push({ type: 'VALUE', value: JOIN_TABLE });

        _.each(joinInstructions.on.or, function orSet(set) {
          var PARENT_TABLE = _.first(_.keys(set));
          var CHILD_TABLE = _.keys(set)[1];
          var PARENT_COLUMN = set[_.first(_.keys(set))];
          var CHILD_COLUMN = set[_.keys(set)[1]];

          var setKeys = [
            { type: 'COMBINATOR', value: 'OR' },
            { type: 'KEY', value: 'TABLE_KEY' },
            { type: 'VALUE', value: PARENT_TABLE },
            { type: 'KEY', value: 'COLUMN_KEY' },
            { type: 'VALUE', value: PARENT_COLUMN },
            { type: 'KEY', value: 'TABLE_KEY' },
            { type: 'VALUE', value: CHILD_TABLE },
            { type: 'KEY', value: 'COLUMN_KEY' },
            { type: 'VALUE', value: CHILD_COLUMN }
          ];

          _.each(setKeys, function appendSet(set) {
            results.push(set);
          });
        });
      })();

      // Otherwise ensure that the ON key has two keys
    } else if (!_.isPlainObject(joinInstructions.on) || _.keys(joinInstructions.on).length !== 2) {
      throw new Error('Invalid join instructions');

      // Handle normal, single level joins
    } else {
      (function buildJoinResults() {
        var JOIN_TABLE = joinInstructions.from;
        var PARENT_TABLE = _.first(_.keys(joinInstructions.on));
        var CHILD_TABLE = _.keys(joinInstructions.on)[1];
        var PARENT_COLUMN = joinInstructions.on[_.first(_.keys(joinInstructions.on))];
        var CHILD_COLUMN = joinInstructions.on[_.keys(joinInstructions.on)[1]];

        var joinResults = [
          { type: 'KEY', value: 'TABLE' },
          { type: 'VALUE', value: JOIN_TABLE },
          { type: 'KEY', value: 'TABLE_KEY' },
          { type: 'VALUE', value: PARENT_TABLE },
          { type: 'KEY', value: 'COLUMN_KEY' },
          { type: 'VALUE', value: PARENT_COLUMN },
          { type: 'KEY', value: 'TABLE_KEY' },
          { type: 'VALUE', value: CHILD_TABLE },
          { type: 'KEY', value: 'COLUMN_KEY' },
          { type: 'VALUE', value: CHILD_COLUMN }
        ];

        _.each(joinResults, function appendSet(set) {
          results.push(set);
        });
      })();
    }

    results.push({
      type: 'ENDIDENTIFIER',
      value: joinType.toUpperCase()
    });
  });
};


//  ╔═╗╦═╗╔═╗╦ ╦╔═╗  ╔╗ ╦ ╦
//  ║ ╦╠╦╝║ ║║ ║╠═╝  ╠╩╗╚╦╝
//  ╚═╝╩╚═╚═╝╚═╝╩    ╚═╝ ╩
var processGroupBy = function processGroupBy(value, results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'GROUPBY'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'GROUPBY'
  });
};


//  ╔═╗╦═╗╔╦╗╔═╗╦═╗  ╔╗ ╦ ╦
//  ║ ║╠╦╝ ║║║╣ ╠╦╝  ╠╩╗╚╦╝
//  ╚═╝╩╚══╩╝╚═╝╩╚═  ╚═╝ ╩
var processOrderBy = function processOrderBy(values, results) {
  // Tokenize the order by and then call the tokenizer on the values
  results.push({
    type: 'IDENTIFIER',
    value: 'ORDERBY'
  });

  if (!_.isArray(values)) {
    values = [values];
  }

  _.each(values, function tokenizeSet(tokenSet) {
    tokenizeObject(tokenSet, undefined, undefined, undefined, results);
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'ORDERBY'
  });
};


//  ╔═╗╔═╗╔═╗╦═╗╔═╗╔═╗╔═╗╔╦╗╦╔═╗╔╗╔╔═╗
//  ╠═╣║ ╦║ ╦╠╦╝║╣ ║ ╦╠═╣ ║ ║║ ║║║║╚═╗
//  ╩ ╩╚═╝╚═╝╩╚═╚═╝╚═╝╩ ╩ ╩ ╩╚═╝╝╚╝╚═╝
var processAggregations = function processAggregations(value, aggregation, results) {
  results.push({
    type: 'IDENTIFIER',
    value: aggregation
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: aggregation
  });
};


//  ╔═╗╔═╗╔═╗╦╔╗╔╔═╗╔╦╗╦╔═╗╔╗╔
//  ╠═╝╠═╣║ ╦║║║║╠═╣ ║ ║║ ║║║║
//  ╩  ╩ ╩╚═╝╩╝╚╝╩ ╩ ╩ ╩╚═╝╝╚╝
var processPagination = function processPagination(value, operator, results) {
  results.push({
    type: 'IDENTIFIER',
    value: operator
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: operator
  });
};


//  ╔═╗╦═╗╔═╗╔═╗╔═╗╔═╗╔═╗  ╦ ╦╔╗╔╦╔═╗╔╗╔
//  ╠═╝╠╦╝║ ║║  ║╣ ╚═╗╚═╗  ║ ║║║║║║ ║║║║
//  ╩  ╩╚═╚═╝╚═╝╚═╝╚═╝╚═╝  ╚═╝╝╚╝╩╚═╝╝╚╝
var processUnion = function processUnion(values, type, results) {
  results.push({
    type: 'UNION',
    value: type
  });

  _.each(values, function processUnionValue(value, idx) {
    // Start each union subquery with an ENDGROUP
    results.push({
      type: 'GROUP',
      value: idx
    });

    // Build the subquery
    checkForSubquery(value, results);

    // Close each subquery with an ENDGROUP token
    results.push({
      type: 'ENDGROUP',
      value: idx
    });
  });

  results.push({
    type: 'ENDUNION',
    value: type
  });
};


//  ╔═╗╦═╗╔═╗╔═╗╔═╗╔═╗╔═╗  ╔═╗╔═╗
//  ╠═╝╠╦╝║ ║║  ║╣ ╚═╗╚═╗  ╠═╣╚═╗
//  ╩  ╩╚═╚═╝╚═╝╚═╝╚═╝╚═╝  ╩ ╩╚═╝
var processAs = function processAs(value, results) {
  results.push({
    type: 'IDENTIFIER',
    value: 'AS'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'AS'
  });
};


//  ╦═╗╔═╗╔╦╗╦ ╦╦═╗╔╗╔╦╔╗╔╔═╗
//  ╠╦╝║╣  ║ ║ ║╠╦╝║║║║║║║║ ╦
//  ╩╚═╚═╝ ╩ ╚═╝╩╚═╝╚╝╩╝╚╝╚═╝
var processReturning = function processReturning(value, results) {
  // Add the RETURNING to the results
  results.push({
    type: 'IDENTIFIER',
    value: 'RETURNING'
  });

  results.push({
    type: 'VALUE',
    value: value
  });

  results.push({
    type: 'ENDIDENTIFIER',
    value: 'RETURNING'
  });
};


module.exports = function tokenizer(expression) {
  if (!expression) {
    throw new Error('Missing expression');
  }

  // Hold the built up results
  var results = [];

  // Kick off recursive parsing of the RQL object
  tokenizeObject(expression, undefined, undefined, undefined, results);

  // Return the tokenenized result set
  return results;
};
/* eslint-enable no-use-before-define */