Home

Awesome

WebDB

A database that reads and writes records on dat:// websites. How it works

Example

Instantiate:

// in the browser
const WebDB = require('@beaker/webdb')
var webdb = new WebDB('webdb-example')

// in nodejs
const DatArchive = require('node-dat-archive')
const WebDB = require('@beaker/webdb')
var webdb = new WebDB('./webdb-example', {DatArchive})

Define your table:

webdb.define('people', {
  // validate required attributes before indexing
  validate(record) {
    assert(record.firstName && typeof record.firstName === 'string')
    assert(record.lastName && typeof record.lastName === 'string')
    return true
  },

  // secondary indexes for fast queries (optional)
  index: ['lastName', 'lastName+firstName', 'age'],

  // files to index
  filePattern: [
    '/person.json',
    '/people/*.json'
  ]
})

Then open the DB:

await webdb.open()

Next we add archives to be indexed into the database.

await webdb.indexArchive('dat://alice.com')
await webdb.indexArchive(['dat://bob.com', 'dat://carla.com'])

Now we can begin querying the database for records.

// get any person record where lastName === 'Roberts'
var mrRoberts = await webdb.people.get('lastName', 'Roberts')

// response attributes:
console.log(mrRoberts.lastName)          // => 'Roberts'
console.log(mrRoberts)                   // => {lastName: 'Roberts', ...}
console.log(mrRoberts.getRecordURL())    // => 'dat://foo.com/bar.json'
console.log(mrRoberts.getRecordOrigin()) // => 'dat://foo.com'
console.log(mrRoberts.getIndexedAt())    // => 1511913554723

// get any person record named Bob Roberts
var mrRoberts = await webdb.people.get('lastName+firstName', ['Roberts', 'Bob'])

// get all person records with the 'Roberts' lastname
var robertsFamily = await webdb.people
  .where('lastName')
  .equalsIgnoreCase('roberts')
  .toArray()

// get all person records with the 'Roberts' lastname
// and a firstname that starts with 'B'
// - this uses a compound index
var robertsFamilyWithaBName = await webdb.people
  .where('lastName+firstName')
  .between(['Roberts', 'B'], ['Roberts', 'B\uffff'])
  .toArray()

// get all person records on a given origin
// - `:origin` is an auto-generated attribute
var personsOnBobsSite = await webdb.people
  .where(':origin')
  .equals('dat://bob.com')
  .toArray()

// get the 30 oldest people indexed
var oldestPeople = await webdb.people
  .orderBy('age')
  .reverse() // oldest first
  .limit(30)
  .toArray()

// count the # of young people
var oldestPeople = await webdb.people
  .where('age')
  .belowOrEqual(18)
  .count()

We can also use WebDB to create, modify, and delete records (and their matching files).

// set the record
await webdb.people.put('dat://bob.com/person.json', {
  firstName: 'Bob',
  lastName: 'Roberts',
  age: 31
})

// update the record if it exists
await webdb.people.update('dat://bob.com/person.json', {
  age: 32
})

// update or create the record
await webdb.people.upsert('dat://bob.com/person.json', {
  age: 32
})

// delete the record
await webdb.people.delete('dat://bob.com/person.json')

// update the spelling of all Roberts records
await webdb.people
  .where('lastName')
  .equals('Roberts')
  .update({lastName: 'Robertos'})

// increment the age of all people under 18
var oldestPeople = await webdb.people
  .where('age')
  .belowOrEqual(18)
  .update(record => {
    record.age = record.age + 1
  })

// delete the 30 oldest people
var oldestPeople = await webdb.people
  .orderBy('age')
  .reverse() // oldest first
  .limit(30)
  .delete()

Table of Contents

<!-- START doctoc generated TOC please keep comment here to allow auto update --> <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE --> <!-- END doctoc generated TOC please keep comment here to allow auto update -->

How to use WebDB

Table definitions

Use the define() method to define your tables, and then call webdb.open() to create them.

Indexing sites

Use indexArchive() and unindexArchive() to control which sites will be indexed. Indexed data will persist in the database until unindexArchive() is called. However, indexArchive() should always be called on load to get the latest data.

If you only want to index the current state of a site, and do not want to watch for updates, call indexArchive() with the {watch: false} option.

You can index and de-index individual files using indexFile() and unindexFile().

Creating queries

Queries are created with a chained function API. You can create a query from the table object using .query(), .where(), or .orderBy(). The where() method returns an object with multiple filter functions that you can use.

var myQuery = webdb.query().where('foo').equals('bar')
var myQuery = webdb.where('foo').equals('bar') // equivalent
var myQuery = webdb.where('foo').startsWith('ba')
var myQuery = webdb.where('foo').between('bar', 'baz', {includeLower: true, includeUpper: false})

Each query has a primary key. By default, this is the url attribute, but it can be changed using .where() or .orderBy(). In this example, the primary key becomes 'foo':

var myQuery = webdb.orderBy('foo')

At this time, the primary key must be one of the indexed attributes. There are 2 indexes created automatically for every record: url and origin. The other indexes are specified in your table's define() call using the index option.

Applying linear-scan filters

After the primary key index is applied, you can apply additional filters using filter(fn) and until(fn). These methods are called "linear scan" filters because they require each record to be loaded and then filtered out. (Until stops when it hits the first false response.)

var myQuery = webdb.query()
  .where('foo').equals('bar')
  .filter(record => record.beep == 'boop') // additional filter

Applying query modifiers

You can apply the following modifiers to your query to alter the output:

Executing 'read' queries

Once your query has been defined, you can execute and read the results using one of these methods:

Executing 'write' queries

Once your query has been defined, you can execute and modify the results using one of these methods:

If you try to modify rows in archives that are not writable, WebDB will throw an error.

Table helper methods

The following methods exist on the table object for query reads and writes:

Record methods

record.getRecordURL()    // => 'dat://foo.com/bar.json'
record.getRecordOrigin() // => 'dat://foo.com'
record.getIndexedAt()    // => 1511913554723

Every record is emitted in a wrapper object with the following methods:

These attributes can be used in indexes with the following IDs:

For instance:

webdb.define('things', {
  // ...
  index: [
    ':indexedAt', // ordered by time the record was indexed
    ':origin+createdAt' // ordered by origin and declared create timestamp (a record attribute)
  ],
  // ...
})
await webdb.open()
webdb.things.where(':indexedAt').above(Date.now() - ms('1 week'))
webdb.things.where(':origin+createdAt').between(['dat://bob.com', 0], ['dat://bob.com', Infinity])

Handling multiple schemas

Since the Web is a complex place, you'll frequently have to deal with multiple schemas which are slightly different. To deal with this, you can use a definition object to support multiple attribute names under one index.

webdb.define('places', {
  // ...

  index: [
    // a simple index definition:
    'name',

    // an object index definition:
    {name: 'zipCode', def: ['zipCode', 'zip_code']}
  ]

  // ...
})

Now, when you run queries on the 'zipCode' key, you will search against both 'zipCode' and 'zip_code'. Note, however, that the records emitted from the query will not be changed by WebDB and so they may differ.

For example:

webdb.places.where('zipCode').equals('78705').each(record => {
  console.log(record.zipCode) // may be '78705' or undefined
  console.log(record.zip_code) // may be '78705' or undefined
})

To solve this, you can preprocess records.

Preprocessing records

Sometimes, you need to modify records before they're stored in the database. This can be for a number of reasons:

For these cases, you can use the preprocess(record) function in the table definition:

webdb.define('places', {
  // ...

  preprocess(record) {
    // normalize zipCode and zip_code
    if (record.zip_code) {
      record.zipCode = record.zip_code
    }

    // move an attribute to the root object for indexing
    record.title = record.info.title

    // compute an attribute
    record.location = `${record.address} ${record.city}, ${record.state} ${record.zipCode}`

    return record
  }

  // ...
})

These attributes will be stored in the WebDB table.

Serializing records

When records are updated by WebDB, they are published to a Dat site as a file. Since these files are distributed on the Web, it's wise to avoid adding noise to the record.

To control the exact record that will be published, you can set the serialize(record) function in the table definition:

webdb.define('places', {
  // ...

  serialize(record) {
    // write the following object to the dat site:
    return {
      info: record.info,
      city: record.city,
      state: record.state,
      zipCode: record.zipCode
    }
  }

  // ...
})

Using JSON-Schema to validate

The default way to validate records is to provide a validator function. If the function throws an error or returns falsy, the record will not be indexed.

It can be tedious to write validation functions, so you might want to use JSON-Schema:

const Ajv = require('ajv')
webdb.define('people', {
  validate: (new Ajv()).compile({
    type: 'object',
    properties: {
      firstName: {
        type: 'string'
      },
      lastName: {
        type: 'string'
      },
      age: {
        description: 'Age in years',
        type: 'integer',
        minimum: 0
      }
    },
    required: ['firstName', 'lastName']
  }),

  // ...
})

Helper tables

Sometimes you need internal storage to help you maintain application state. This may be for interfaces, or data which is private, or for special kinds of indexes.

For instance, in the Fritter app, we needed an index for notifications. This index was conditional: it needed to contain posts which were replies to the user, or likes which were on the user's post. For cases like this, you can use a "helper table."

webdb.define('notifications', {
  helperTable: true,
  index: ['createdAt'],
  preprocess (record) {
    record.createdAt = Date.now()
  }
})

When the helperTable attribute is set to true in a table definition, the table will not be used to index Dat archives. Instead, it will exist purely in the local data cache, and only contain data which is .put() there. In all other respects, it behaves like a normal table.

In Fritter, the helper table is used with events to track notifications:

// track reply notifications
webdb.posts.on('put', async ({record, url, origin}) => {
  if (origin === userUrl) return // dont index the user's own posts
  if (isAReplyToUser(record) === false) return // only index replies to the user
  if (await isNotificationIndexed(url)) return // don't index if already indexed
  await db.notifications.put(url, {type: 'reply', url})
})
webdb.posts.on('del', async ({url}) => {
  if (await isNotificationIndexed(url)) {
    await db.notifications.delete(url)
  }
})

Class: WebDB

new WebDB([name, opts])

var webdb = new WebDB('mydb')

Create a new WebDB instance. The given name will control where the indexes are saved. You can specify different names to run multiple WebDB instances at once.

WebDB.delete([name])

await WebDB.delete('mydb')

Deletes the indexes and metadata for the given WebDB.

Instance: WebDB

webdb.open()

await webdb.open()

Runs final setup for the WebDB instance. This must be run after .define() to create the table instances.

webdb.close()

await webdb.close()

Closes the WebDB instance.

webdb.delete()

await webdb.delete()

Closes and destroys all indexes in the WebDB instance.

You can .delete() and then .open() a WebDB to recreate its indexes.

await webdb.delete()
await webdb.open()

webdb.define(name, definition)

Creates a new table on the webdb object. The table will be set at webdb.{name} and be the WebDBTable type. This method must be called before open()

Indexed attributes may either be defined as a keypath string or an object definition. The object definition has the following values:

If the value of def is an array, it supports each definition. This is useful when supporting multiple schemas (learn more here).

In the index definition, you can specify compound indexes with a + separator in the keypath. You can also index each value of an array using the * sigil at the start of the name. Some example index definitions:

a simple index           - 'firstName'
as an object def         - {name: 'firstName', def: 'firstName'}
a compound index         - 'firstName+lastName'
index an array's values  - '*favoriteFruits'
many keys                - {name: 'firstName', def: ['firstName', 'first_name']}
many keys, compound      - {name: 'firstName+lastName', def: ['firstName+lastName', 'first_name+last_name']}

You can specify which files should be processed into the table using the filePattern option. If unspecified, it will default to all json files on the site ('*.json').

Example:

webdb.define('people', {
  validate(record) {
    assert(record.firstName && typeof record.firstName === 'string')
    assert(record.lastName && typeof record.lastName === 'string')
    return true
  },
  index: ['lastName', 'lastName+firstName', 'age'],
  filePattern: [
    '/person.json',
    '/people/*.json'
  ]
})

await webdb.open()
// the new table will now be defined at webdb.people

webdb.indexArchive(url[, opts])

await webdb.indexArchive('dat://foo.com')

Add one or more dat:// sites to be indexed. The method will return when the site has been fully indexed. This will add the given archive to the "sources" list.

webdb.unindexArchive(url)

await webdb.unindexArchive('dat://foo.com')

Remove a dat:// site from the dataset. The method will return when the site has been fully de-indexed. This will remove the given archive from the "sources" list.

webdb.indexFile(archive, filepath)

await webdb.indexFile(fooArchive, '/bar.json')

Add a single file to the index. The method will return when the file has been indexed.

This will not add the file or its archive to the "sources" list. Unlike indexArchive, WebDB will not watch the file after this call.

webdb.indexFile(url)

await webdb.indexFile('dat://foo.com/bar.json')

Add a single file to the index. The method will return when the file has been indexed.

This will not add the file or its archive to the "sources" list.

webdb.unindexFile(archive, filepath)

await webdb.unindexFile(fooArchive, '/bar.json')

Remove a single file from the dataset. The method will return when the file has been de-indexed.

webdb.unindexFile(url)

await webdb.unindexFile('dat://foo.com')

Remove a single file from the dataset. The method will return when the file has been de-indexed.

webdb.listSources()

var urls = await webdb.listSources()

Lists the URLs of the dat:// sites which are included in the dataset.

webdb.isSource(url)

var urls = await webdb.isSource('dat://foo.com')

Is the given dat:// URL included in the dataset?

Event: 'open'

webdb.on('open', () => {
  console.log('WebDB is ready for use')
})

Emitted when the WebDB instance has been opened using open().

Event: 'open-failed'

webdb.on('open-failed', (err) => {
  console.log('WebDB failed to open', err)
})

Emitted when the WebDB instance fails to open during open().

Event: 'indexes-reset'

webdb.on('indexes-reset', () => {
  console.log('WebDB detected a change in schemas and reset all indexes')
})

Emitted when the WebDB instance detects a change in the schemas and has to reindex the dataset. All indexes are cleared and will be reindexed as sources are added.

Event: 'indexes-updated'

webdb.on('indexes-updated', (url, version) => {
  console.log('Tables were updated for', url, 'at version', version)
})

Emitted when the WebDB instance has updated the stored data for a archive.

Event: 'source-indexing'

webdb.on('source-indexing', (url, startVersion, targetVersion) => {
  console.log('Tables are updating for', url, 'from version', startVersion, 'to', targetVersion)
})

Emitted when the WebDB instance has started to index the given archive.

Event: 'source-index-progress'

webdb.on('source-index-progress', (url, tick, total) => {
  console.log('Update for', url, 'is', Math.round(tick / total * 100), '% complete')
})

Emitted when an update has been applied during an indexing process.

Event: 'source-indexed'

webdb.on('source-indexed', (url, version) => {
  console.log('Tables were updated for', url, 'at version', version)
})

Emitted when the WebDB instance has indexed the given archive. This is similar to 'indexes-updated', but it fires every time a source is indexed, whether or not it results in updates to the indexes.

Event: 'source-missing'

webdb.on('source-missing', (url) => {
  console.log('WebDB couldnt find', url, '- now searching')
})

Emitted when a source's data was not locally available or found on the network. When this occurs, WebDB will continue searching for the data, and emit 'source-found' on success.

Event: 'source-found'

webdb.on('source-found', (url) => {
  console.log('WebDB has found and indexed', url)
})

Emitted when a source's data was found after originally not being found during indexing. This event will only be emitted after 'source-missing' is emitted.

Event: 'source-error'

webdb.on('source-error', (url, err) => {
  console.log('WebDB failed to index', url, err)
})

Emitted when a source fails to load.

Instance: WebDBTable

table.count()

var numRecords = await webdb.mytable.count()

Count the number of records in the table.

table.delete(url)

await webdb.mytable.delete('dat://foo.com/bar.json')

Delete the record at the given URL.

table.each(fn)

await webdb.mytable.each(record => {
  console.log(record)
})

Iterate over all records in the table with the given function.

table.filter(fn)

var records = await webdb.mytable.filter(record => {
  return (record.foo == 'bar')
})

Start a new query and apply the given filter function to the resultset.

table.get(url)

var record = await webdb.mytable.get('dat://foo.com/myrecord.json')

Get the record at the given URL.

table.get(key, value)

var record = await webdb.mytable.get('foo', 'bar')

Get the record first record to match the given key/value query.

table.isRecordFile(url)

var isRecord = webdb.mytable.isRecordFile('dat://foo.com/myrecord.json')

Tells you whether the given URL matches the table's file pattern.

table.limit(n)

var query = webdb.mytable.limit(10)

Creates a new query with the given limit applied.

table.listRecordFiles(url)

var recordFiles = await webdb.mytable.listRecordFiles('dat://foo.com')

Lists all files on the given URL which match the table's file pattern.

table.name

The name of the table.

table.offset(n)

var query = webdb.mytable.offset(5)

Creates a new query with the given offset applied.

table.orderBy(key)

var query = webdb.mytable.orderBy('foo')

Creates a new query ordered by the given key.

table.put(url, record)

await webdb.mytable.put('dat://foo.com/myrecord.json', {foo: 'bar'})

Replaces or creates the record at the given URL with the record.

table.query()

var query = webdb.mytable.query()

Creates a new query.

table.reverse()

var query = webdb.mytable.reverse()

Creates a new query with reverse-order applied.

table.schema

The schema definition for the table.

table.toArray()

var records = await webdb.mytable.toArray()

Returns an array of all records in the table.

table.update(url, updates)

var wasUpdated = await webdb.mytable.update('dat://foo.com/myrecord.json', {foo: 'bar'})

Updates the target record with the given key values, if it exists.

table.update(url, fn)

var wasUpdated = await webdb.mytable.update('dat://foo.com/myrecord.json', record => {
  record.foo = 'bar'
  return record
})

Updates the target record with the given function, if it exists.

table.upsert(url, updates)

var didCreateNew = await webdb.mytable.upsert('dat://foo.com/myrecord.json', {foo: 'bar'})

If a record exists at the target URL, will update it with the given key values. If a record does not exist, will create the record.

table.upsert(url, fn)

var didCreateNew = await webdb.mytable.upsert('dat://foo.com/myrecord.json', record => {
  if (record) {
    // update
    record.foo = 'bar'
    return record
  }
  // create
  return {foo: 'bar'}
})

Updates the target record with the given function, if it exists. If a record does not exist, will give a falsy value to the method.

table.where(key)

var whereClause = webdb.mytable.where('foo')

Creates a new where-clause using the given key.

Event: 'put-record'

webdb.mytable.on('put-record', ({url, origin, indexedAt, record}) => {
  console.log('Table was updated for', url, '(origin:', origin, ') at ', indexedAt)
  console.log('Record data:', record)
})

Emitted when the table has updated the stored data for a record.

Event: 'del-record'

webdb.mytable.on('del-record', ({url, origin, indexedAt}) => {
  console.log('Table was updated for', url, '(origin:', origin, ') at ', indexedAt)
})

Emitted when the table has deleted the stored data for a record. This can happen because the record has been deleted, or because a new version of the record fails validation.

Instance: WebDBQuery

query.clone()

var query = webdb.mytable.query().clone()

Creates a copy of the query.

query.count()

var numRecords = await webdb.mytable.query().count()

Gives the count of records which match the query.

query.delete()

var numDeleted = await webdb.mytable.query().delete()

Deletes all records which match the query.

query.each(fn)

await webdb.mytable.query().each(record => {
  console.log(record)
})

Calls the given function with all records which match the query.

query.eachKey(fn)

await webdb.mytable.query().eachKey(url => {
  console.log('URL =', url)
})

Calls the given function with the value of the query's primary key for each matching record.

The key is determined by the index being used. By default, this is the url attribute, but it can be changed by using where() or orderBy().

Example:

await webdb.mytable.orderBy('age').eachKey(age => {
  console.log('Age =', age)
})

query.eachUrl(fn)

await webdb.mytable.query().eachUrl(url => {
  console.log('URL =', url)
})

Calls the given function with the URL of each matching record.

query.filter(fn)

var query = webdb.mytable.query().filter(record => {
  return record.foo == 'bar'
})

Applies an additional filter on the query.

query.first()

var record = await webdb.mytable.query().first()

Returns the first result in the query.

query.keys()

var keys = await webdb.mytable.query().keys()

Returns the value of the primary key for each matching record.

The key is determined by the index being used. By default, this is the url attribute, but it can be changed by using where() or orderBy().

var ages = await webdb.mytable.orderBy('age').keys()

query.last()

var record = await webdb.mytable.query().last()

Returns the last result in the query.

query.limit(n)

var query = webdb.mytable.query().limit(10)

Limits the number of matching record to the given number.

query.offset(n)

var query = webdb.mytable.query().offset(10)

Skips the given number of matching records.

query.orderBy(key)

var query = webdb.mytable.query().orderBy('foo')

Sets the primary key and sets the resulting order to match its values.

query.put(record)

var numWritten = await webdb.mytable.query().put({foo: 'bar'})

Replaces each matching record with the given value.

query.urls()

var urls = await webdb.mytable.query().urls()

Returns the url of each matching record.

query.reverse()

var query = webdb.mytable.query().reverse()

Reverses the order of the results.

query.toArray()

var records = await webdb.mytable.query().toArray()

Returns the value of each matching record.

query.uniqueKeys()

var keys = await webdb.mytable.query().uniqueKeys()

Returns the value of the primary key for each matching record, with duplicates filtered out.

The key is determined by the index being used. By default, this is the url attribute, but it can be changed by using where() or orderBy().

Example:

var ages = await webdb.mytable.orderBy('age').uniqueKeys()

query.until(fn)

var query = webdb.mytable.query().until(record => {
  return record.foo == 'bar'
})

Stops emitting matching records when the given function returns true.

query.update(updates)

var numUpdated = await webdb.mytable.query().update({foo: 'bar'})

Updates all matching record with the given values.

query.update(fn)

var numUpdated = await webdb.mytable.query().update(record => {
  record.foo = 'bar'
  return record
})

Updates all matching record with the given function.

query.where(key)

var whereClause = webdb.mytable.query().where('foo')

Creates a new where clause.

Instance: WebDBWhereClause

where.above(value)

var query = webdb.mytable.query().where('foo').above('bar')
var query = webdb.mytable.query().where('age').above(18)

where.aboveOrEqual(value)

var query = webdb.mytable.query().where('foo').aboveOrEqual('bar')
var query = webdb.mytable.query().where('age').aboveOrEqual(18)

where.anyOf(values)

var query = webdb.mytable.query().where('foo').anyOf(['bar', 'baz'])

Does not work on compound indexes.

where.anyOfIgnoreCase(values)

var query = webdb.mytable.query().where('foo').anyOfIgnoreCase(['bar', 'baz'])

Does not work on compound indexes.

where.below(value)

var query = webdb.mytable.query().where('foo').below('bar')
var query = webdb.mytable.query().where('age').below(18)

where.belowOrEqual(value)

var query = webdb.mytable.query().where('foo').belowOrEqual('bar')
var query = webdb.mytable.query().where('age').belowOrEqual(18)

where.between(lowerValue, upperValue[, options])

var query = webdb.mytable.query().where('foo').between('bar', 'baz', {includeUpper: true, includeLower: true})
var query = webdb.mytable.query().where('age').between(18, 55, {includeLower: true})

where.equals(value)

var query = webdb.mytable.query().where('foo').equals('bar')

where.equalsIgnoreCase(value)

var query = webdb.mytable.query().where('foo').equalsIgnoreCase('bar')

Does not work on compound indexes.

where.noneOf(values)

var query = webdb.mytable.query().where('foo').noneOf(['bar', 'baz'])

Does not work on compound indexes.

where.notEqual(value)

var query = webdb.mytable.query().where('foo').notEqual('bar')

Does not work on compound indexes.

where.startsWith(value)

var query = webdb.mytable.query().where('foo').startsWith('ba')

Does not work on compound indexes.

where.startsWithAnyOf(values)

var query = webdb.mytable.query().where('foo').startsWithAnyOf(['ba', 'bu'])

Does not work on compound indexes.

where.startsWithAnyOfIgnoreCase(values)

var query = webdb.mytable.query().where('foo').startsWithAnyOfIgnoreCase(['ba', 'bu'])

Does not work on compound indexes.

where.startsWithIgnoreCase(value)

var query = webdb.mytable.query().where('foo').startsWithIgnoreCase('ba')

Does not work on compound indexes.

How it works

WebDB abstracts over the DatArchive API to provide a simple database-like interface. It's inspired by Dexie.js and built using LevelDB. (In the browser, it runs on IndexedDB using level.js.

WebDB scans a set of source Dat archives for files that match a path pattern. Web DB caches and indexes those files so they can be queried easily and quickly. WebDB also provides a simple interface for adding, editing, and removing records from archives.

WebDB sits on top of Dat archives. It duplicates ingested data into IndexedDB, which acts as a throwaway cache. The cached data can be reconstructed at any time from the source Dat archives.

WebDB treats individual files in the Dat archive as individual records in a table. As a result, there's a direct mapping for each table to a folder of JSON files. For instance, if you had a posts table, it might map to the /posts/*.json files. WebDB's mutators, e.g., put, add, update, simply writes records as JSON files in the posts/ directory. WebDB's readers and query-ers, like get() and where(), read from the IndexedDB cache.

WebDB watches its source archives for changes to the JSON files that compose its records. When the files change, it syncs and reads the changes, then updates IndexedDB, keeping query results up-to-date. Roughly, the flow is: put() -> archive/posts/12345.json -> indexer -> indexeddb -> get().

Why not put all records in one file?

Storing records in one file—posts.json for example—is an intuitive way to manage data on the peer-to-peer Web, but putting each record in an individual file is a much better choice for performance and linkability.

Performance

The dat:// protocol doesn't support partial updates at the file-level, which means that with multiple records in a single file, every time a user adds a record, anyone who follows that user must sync and re-download the entire file. As the file continues to grow, performance will degrade. Putting each record in an individual file is much more efficient: when a record is created, peers in the network will only download the newly-created file.

Linkability

Putting each record in an individual file also makes each record linkable! This isn't as important as performance, but it's a nice feature to have. See Dog Legs McBoot's status update as an example:

dat://232ac2ce8ad4ed80bd1b6de4cbea7d7b0cad1441fa62312c57a6088394717e41/posts/0jbdviucy.json

Changelog

A quick overview of the notable changes to WebDB:

4.1.0

Added "helper tables," which make it possible to track private state and build more sophisticated indexes.

4.0.0

Replaced JSON-Schema validation with an open validate function. This was done to reduce the output bundle size (by 200kb!) and to improve overall flexibility (JSON-Schema and JSON-LD do not work together very well).

3.0.0

The addSource() and removeSource() methods were replaced with indexArchive(), indexFile(), unindexArchive(), and unindexFile(). The indexArchive() method also provides an option to disable watching.

This change was made as we found controlling the index was an important part of using WebDB. Frequently we'd want to index an archive temporarily, for instance to view a user's profile on first visit.

This new API gives better control for those use-cases, and no longer assumes you want to continue watching an archive after indexing it once.