Helper to work with cassandra's inverted indeces table. Allows to partition index key to multiple row keys.

gistfile1.js

"use strict";

var slug = require('slug');

var hashFunctions = {
  /**
   * Return string hash, based on first two symbols of string
   * @param {string} str Str to calculate the hash
   */
  "two_symbol_hash": function (str) {
    str = slug(str);
    str = str.toLowerCase().replace(/^www\d*\./i, '').replace(/\.|\d|-/gi, 'z');
    return str.slice(0, 2);
  },

  "two_symbol_hash_range": function twoSymbolHashRange() {
    if (twoSymbolHashRange.result) {
      return twoSymbolHashRange.result;
    }

    var results = [];
    for(var ch = 'a'.charCodeAt(0); ch <= 'z'.charCodeAt(0); ch++ ) {
      for(var sch = 'a'.charCodeAt(0); sch <= 'z'.charCodeAt(0); sch++ ) {
        results.push(String.fromCharCode(ch) + String.fromCharCode(sch));
      }
    }

    twoSymbolHashRange.result = results;
    return results;
  }
};

/**
 * Return a row key for an index table in Cassandra, based on value,
 * @param {string} keyPrefix Key prefix. e.g. 'some_tag', 'another_tag', etc.
 * @param {string} keyValue Value to put on that key
 * @param {string} [hashFunction='two_symbol_hash'] Hash function to use
 * @return {string} Partition key to use in cassandra index table. e.g.
 * 'some_tag_ab', 'another_tag_jj', etc.
 */
function createPartitionKey(keyPrefix, value, hashFunction) {
  hashFunction = hashFunction || 'two_symbol_hash';
  return keyPrefix + '_' + hashFunctions[hashFunction](value);
}

/**
 * Get key range for specified key to lookup in Cassandra index table
 * @param {string} keyPrefix For which index key get key range. 'some_tag',
 * 'another_tag', etc
 * @param {string} [hashFunction='two_symbol_hash'] Which hash function to use
 * @return {Array} array of keys to lookup in cassandra. e.g. ['some_tag_1',
 * 'some_tag_2', 'some_tag_3'...]
 */
function getPartiotionKeyRange(keyPrefix, hashFunction) {
  hashFunction = hashFunction || 'two_symbol_hash';
  var results = [];
  hashFunctions[hashFunction + '_range']().forEach(function (hashKey) {
    results.push("'" + keyPrefix + '_' + hashKey + "'");
  });
  return results;
}

/**
 * Get query to getch key values from inverted index
 * @param {string} indexName Name of the table to query on
 * @param {string} key Name of the key to get from index
 * @param {string} [hashFunction='two_symbol_hash'] Name of the hash function
 * for partitioning keys in index
 * @return {string} Query to get key values from inverted index
 */
function getFromInvertedIndexQuery(indexName, key, hashFunction) {
  var keys = getPartiotionKeyRange(key, hashFunction);
  var query = "SELECT * FROM " + indexName + " WHERE key IN (";
  query += keys.join(',') + ');';
  return query;
}

/**
 * Get query to put value to inverted index for specified key
 * @param {string} indexName Name of the table to put value
 * @param {string} key Name of the key to put
 * @param {string} value Value to put into key
 * @param {string} [hashFunction='two_symbol_hash'] Name of the hash function
 * for partitioning keys in index
 * @return {string} Query to put value into key in inverted index
 */
function putToInvertedIndexQuery(indexName, key, value, hashFunction) {
  key = createPartitionKey(key, value, hashFunction);
  var query = "INSERT INTO " + indexName + " (key, value) VALUES ('" + key + "', '" + value + "');";
  return query;
}

module.exports = {
  getFromInvertedIndexQuery: getFromInvertedIndexQuery,
  putToInvertedIndexQuery: putToInvertedIndexQuery
};