Redis Databases Wrapped in Sheets

Redis has relicensed away from open source!

The original BSD-3-Clause still applies to version 7.2.4.

This demo has been tested with KeyDB and other servers that support the "Redis serialization protocol" (RESP).

KeyDB is a Redis-compatible in-memory data store. It is capable of storing sets, lists and other simple data structures.

SheetJS is a JavaScript library for reading and writing data from spreadsheets.

This demo defines a schema for storing Redis databases in spreadsheets. We'll explore how to use SheetJS and Redis NodeJS connector modules to pull data from XLSX files to a Redis database and to serialize a database to a workbook.

Tested Deployments

This demo was tested in the following environments:

Server	Connector Module	Date
KeyDB `6.3.4`	`redis` (`4.6.13`)	2024-04-23
Redis `6.2.14`	`redis` (`4.6.13`)	2024-04-23
Redis `7.2.4`	`redis` (`4.6.13`)	2024-04-23
Valkey `7.2.5`	`redis` (`4.6.13`)	2024-04-23

Overview

Redis has 5 core data types: "String", List", "Set", "Sorted Set", and "Hash". Since the keys and values are limited to simple strings (and numbers), it is possible to store complete databases in a single worksheet.

SheetJSRedis.xlsx

Integration Details

SheetJSRedis.mjs exports the methods:

redis_to_ws creates a SheetJS worksheet by querying a redis client
ws_to_redis creates an array of query objects from the SheetJS worksheet

Array of Arrays

The shared data representation is an "array of arrays"¹. Each array within the structure corresponds to one row.

The Redis to SheetJS converter generates an array of arrays of the data by running queries to fetch data from the database. The SheetJS aoa_to_sheet and sheet_add_aoa² methods build up worksheets from arrays of arrays. Once the worksheet is created, it can be added to a SheetJS workbook object³ and exported using writeFile⁴.

The SheetJS to Redis converter works in reverse. Workbook files are parsed with the SheetJS readFile method⁵ and the desired worksheet is pulled from the workbook object. An array of arrays can be created with the sheet_to_json⁶ utility function. The data structure can be scanned to generate Redis queries.

Appending Columns

Since the data is column-oriented, the goal is to add the data starting on the first row of the column after the data.

To calculate the starting point for writing data, SheetJS provides encode_cell and decode_range utility functions for working with addresses and ranges⁷.

The following snippet takes an array of arrays of values and writes the values to a worksheet starting from the column after the worksheet range:

function add_aoa_to_next_column(worksheet, aoa) {
  /* get range of worksheet */
  const range = XLSX.utils.decode_range(worksheet["!ref"])
  /* the origin to write new data will start in the column after the range */
  const origin = XLSX.utils.encode_cell({
    r: 0, // start on first row
    c: range.e.c + 1 // column after end
  });
  /* add data */
  XLSX.utils.sheet_add_aoa(worksheet, aoa, { origin });
}

Strings

Strings can be stored in a unified String table. The first column holds keys and the second column holds values:

XXX|    A    |   B   |
---+---------+-------+
 1 | Strings |       |
 2 |         |       |
 3 | Hello   | World |
 4 | Sheet   | JS    |

The SheetJS array-of-arrays representation of the string table is an array of key/value pairs.

The pairs can be generated from Redis by querying for all of the keys using the KEYS⁸ method, testing if the corresponding value is a string using the TYPE⁹ method, and fetching string values using the GET¹⁰ method:

const aoa = ["Strings"]; aoa.length = 2; // [ "Strings", empty ]
const keys = await client.KEYS("*");
for(let key of keys) {
  const type = await client.TYPE(key);
  if(type == "string") aoa.push([key, await client.GET(key)]);
}

Lists

Lists are unidimensional and can be stored in their own columns.

XXX|    C    |
---+---------+
 1 | List    |
 2 | List1   |
 3 | List1V1 |
 4 | List1V2 |

The SheetJS array-of-arrays representation of lists is a column of values.

Redis LRANGE¹¹ returns a simple array of values. sheet_add_aoa interprets the result as one row, so the data should be transposed.

The code transposes the result with values.map(v => [v]).

const values = await client.LRANGE(key, 0, -1);
const aoa = [ ["List"], [key] ].concat(values.map(v => [v]));

Sets

Sets are unidimensional and can be stored in their own columns.

XXX|   D   |
---+-------+
 1 | Set   |
 2 | Set1  |
 3 | Set1A |
 4 | Set1B |

The SheetJS array-of-arrays representation of sets is a column of values.

Redis SMEMBERS¹² returns an array of values. sheet_add_aoa interprets the result as one row, so the data should be transposed.

The code transposes the result with values.map(v => [v]).

const values = await client.SMEMBERS(key);
const aoa = [ ["Set"], [key] ].concat(values.map(v => [v]));

Sorted Sets

Sorted Sets have an associated score which can be stored in the second column.

XXX|    E    | F |
---+---------+---+
 1 | Sorted  |   |
 2 | ZSet1   |   |
 3 | Key1    | 1 |
 4 | Key2    | 2 |

The SheetJS array-of-arrays representation is an array of key/score pairs.

ZRANGE_WITHSCORES¹³ returns an array of objects which can be reshaped.

const values = await client.ZRANGE_WITHSCORES(key, 0, -1);
const aoa = [ ["Sorted"], [key] ].concat(values.map(v => [v.value, v.score]));

Hashes

Hashes are stored like the string table, with key and value columns in order.

XXX|   G   |   H   |
---+-------+-------+
 1 | Hash  |       |
 2 | Hash1 |       |
 3 | Key1  | Val1  |
 4 | Key2  | Val2  |

The SheetJS array-of-arrays representation is an array of key/value pairs.

HGETALL¹⁴ returns an object which can be converted using Object.entries:

const values = await client.HGETALL(key);
const aoa = [ ["Hash"], [key] ].concat(Object.entries(values));

Complete Example

The most recent version of the redis node module does not work with most versions of NodeJS. It is "ESM-only", requiring NodeJS 18 or later. As a result, this demo also requires NodeJS version 18 or later.

Set up and start a local Redis-compatible server.

Redis-compatible servers (click to show)

This demo was last tested on macOS.

KeyDB

KeyDB was installed with:

brew install keydb@6.3.4

The following command started the server process:

keydb-server --protected-mode no

Valkey

Valkey was installed with:

brew install valkey

The Valkey formula will create symlinks to redis-server and other commands. This conflicts with the main redis package. redis must be unlinked:

brew unlink redis
brew link valkey

The following command started the server process:

redis-server /usr/local/etc/redis.conf

Redis 6

Redis 6 was installed with:

brew install redis@6.2

The following command started the server process:

redis-server /usr/local/etc/redis.conf

Redis 7

Redis 7 was installed with:

brew install redis@7.2

The following command started the server process:

redis-server /usr/local/etc/redis.conf

When the demo was last tested, Redis 7.2.4 was installed. The output from the Redis server will display the version number:

69385:C 23 Apr 2024 13:24:48.520 * Redis version=7.2.4, bits=64, commit=00000000, modified=0, pid=69385, just started

Please raise an issue with Homebrew if a later version is installed.

Download the following scripts:

curl -LO https://docs.sheetjs.com/nosql/SheetJSRedis.mjs
curl -LO https://docs.sheetjs.com/nosql/SheetJSRedisTest.mjs

Install dependencies:

npm i --save https://cdn.sheetjs.com/xlsx-0.20.3/xlsx-0.20.3.tgz redis@4.6.13

Run the test script:

node SheetJSRedisTest.mjs

Testing

Inspect the output and compare with the data in SheetJSRedisTest.mjs.

Expected Output (click to show)

SET [ 'baz', '0' ]
SET [ 'foo', 'bar' ]
HSET [
  'user:1000',
  {
    name: 'John Smith',
    email: 'john.smith@example.com',
    password: 's3cret',
    visits: '1'
  }
]
HSET [
  'user:1001',
  {
    name: 'Mary Jones',
    email: 'mjones@example.com',
    password: 'hunter2'
  }
]
SADD [ 'superpowers', [ 'flight', 'x-ray vision' ] ]
SADD [ 'birdpowers', [ 'flight', 'pecking' ] ]
RPUSH [ 'friends', [ 'sam', 'alice', 'bob' ] ]
ZADD [
  'hackers',
  [
    { value: 'Grace Hopper', score: 1906 },
    { value: 'Alan Turing', score: 1912 },
    { value: 'Claude Shannon', score: 1916 },
    { value: 'Alan Kay', score: 1940 },
    { value: 'Richard Stallman', score: 1953 },
    { value: 'Sophie Wilson', score: 1957 },
    { value: 'Yukihiro Matsumoto', score: 1965 },
    { value: 'Linus Torvalds', score: 1969 }
  ]
]

Open SheetJSRedis.xlsx and verify the columns have the correct data. The sheet should have the same structure as the screenshot at the top of this page.

See "Array of Arrays" in "Utility Functions" ↩
See "Array of Arrays Input" in "Utility Functions". ↩
See "Workbook Helpers" in "Utility Functions" ↩
See writeFile in "Writing Files" ↩
See readFile in "Reading Files" ↩
See sheet_to_json in "Utilities" ↩
See "Utilities" in "Addresses and Ranges" ↩
See KEYS in the Redis documentation. ↩
See TYPE in the Redis documentation. ↩
See GET in the Redis documentation. ↩
See LRANGE in the Redis documentation. ↩
See SMEMBERS in the Redis documentation. ↩
The official command is ZRANGE. ZRANGE_WITHSCORES is a special command supported by the NodeJS wrapper. ↩
See HGETALL in the Redis documentation. ↩

Overview​

Integration Details​

Array of Arrays​

Appending Columns​

Strings​

Lists​

Sets​

Sorted Sets​

Hashes​

Complete Example​

Footnotes​

Overview

Integration Details

Array of Arrays

Appending Columns

Strings

Lists

Sets

Sorted Sets

Hashes

Complete Example

Footnotes