Documentation
¶
Overview ¶
Package engine is a fork of github.com/ipni/index-provider/engine with some modifications to make it work for the current Frisbii usecase. It needs to be reconciled back into that package and removed from here.
Package engine provides a reference implementation of the provider.Interface in order to advertise the availability of a list of multihashes to indexer nodes such as "storetheindex". See: https://github.com/ipni/storetheindex
The advertisements are published as a chan of diffs that signal the list of multihashes that are added or removed represented as an IPLD DAG. Walking the chain of advertisements would then provide the latest state of the total multihashes provided by the engine. The list of multihashes are paginated as a collection of interlinked chunks. For the complete advertisement IPLD schema, see:
The engine internally uses "go-libipni/dagsync" to sync the IPLD DAG of advertisements. See: https://github.com/ipni/go-libipni/tree/main/dagsync
Index ¶
- Variables
- type Engine
- func (e *Engine) GetAdv(_ context.Context, adCid cid.Cid) (*schema.Advertisement, error)
- func (e *Engine) GetLatestAdv(ctx context.Context) (cid.Cid, *schema.Advertisement, error)
- func (e *Engine) GetPublisherHttpFunc() (http.HandlerFunc, error)
- func (e *Engine) LinkSystem() *ipld.LinkSystem
- func (e *Engine) NotifyPut(ctx context.Context, provider *peer.AddrInfo, contextID []byte, ...) (cid.Cid, error)
- func (e *Engine) NotifyRemove(ctx context.Context, provider peer.ID, contextID []byte) (cid.Cid, error)
- func (e *Engine) Publish(ctx context.Context, adv schema.Advertisement) (cid.Cid, error)
- func (e *Engine) PublishLatest(ctx context.Context) (cid.Cid, error)
- func (e *Engine) PublishLatestHTTP(ctx context.Context, announceURLs ...*url.URL) (cid.Cid, error)
- func (e *Engine) PublishLocal(ctx context.Context, adv schema.Advertisement) (cid.Cid, error)
- func (e *Engine) RegisterMultihashLister(mhl provider.MultihashLister)
- func (e *Engine) Shutdown() error
- func (e *Engine) Start(ctx context.Context) error
- type Option
- func WithChainedEntries(chunkSize int) Option
- func WithDataTransfer(dt datatransfer.Manager) Option
- func WithDatastore(ds datastore.Batching) Option
- func WithDirectAnnounce(announceURLs ...string) Option
- func WithEntriesCacheCapacity(s int) Option
- func WithExtraGossipData(extraData []byte) Option
- func WithHamtEntries(hashAlg multicodec.Code, bitWidth, bucketSize int) Option
- func WithHost(h host.Host) Option
- func WithHttpPublisherAnnounceAddr(addr string) Option
- func WithHttpPublisherHandlerPath(handlerPath string) Option
- func WithHttpPublisherListenAddr(addr string) Option
- func WithHttpPublisherWithoutServer() Option
- func WithPrivateKey(key crypto.PrivKey) Option
- func WithProvider(provider peer.AddrInfo) Option
- func WithPublisherKind(k PublisherKind) Option
- func WithPurgeCacheOnStart(p bool) Option
- func WithRetrievalAddrs(addrs ...string) Option
- func WithSyncPolicy(syncPolicy *policy.Policy) Option
- func WithTopic(t *pubsub.Topic) Option
- func WithTopicName(t string) Option
- type PublisherKind
Constants ¶
This section is empty.
Variables ¶
var ( // ErrEntriesLinkMismatch signals that the link generated from chunking the mulithashes returned by provider.MultihashLister does not match the previously generated link. This error is most likely caused by the lister returning inconsistent multihashes for the same key. ErrEntriesLinkMismatch = errors.New("regenerated link from multihash lister did not match the original link; multihashes returned by the lister for the same key are not consistent") )
Functions ¶
This section is empty.
Types ¶
type Engine ¶
type Engine struct {
// contains filtered or unexported fields
}
Engine is an implementation of the core reference provider interface.
func New ¶
New creates a new index provider Engine as the default implementation of provider.Interface. It provides the ability to advertise the availability of a list of multihashes associated to a context ID as a chain of linked advertisements as defined by the indexer node protocol implemented by "go-libipni".
Engine internally uses "go-libipni/dagsync", a protocol for propagating and synchronizing changes an IPLD DAG, to publish advertisements. See:
Published advertisements are signed using the given private key. The retAddrs corresponds to the endpoints at which the data block associated to the advertised multihashes can be retrieved. If no retAddrs are specified, then use the listen addresses of the given libp2p host.
The engine also provides the ability to generate advertisements via Engine.NotifyPut and Engine.NotifyRemove as long as a provider.MultihashLister is registered. See: provider.MultihashLister, Engine.RegisterMultihashLister.
The engine must be started via Engine.Start before use and discarded via Engine.Shutdown when no longer needed.
func (*Engine) GetAdv ¶
GetAdv gets the advertisement associated to the given cid c. The context is not used.
func (*Engine) GetLatestAdv ¶
GetLatestAdv gets the latest advertisement by the provider. If there are no previously published advertisements, then cid.Undef is returned as the advertisement CID.
func (*Engine) GetPublisherHttpFunc ¶
func (e *Engine) GetPublisherHttpFunc() (http.HandlerFunc, error)
GetPublisherHttpFunc gets the http.HandlerFunc that can be used to serve advertisements over HTTP. The returned handler is only valid if the PublisherKind is HttpPublisher and the HttpPublisherWithoutServer option is set.
func (*Engine) LinkSystem ¶
func (e *Engine) LinkSystem() *ipld.LinkSystem
LinkSystem gets the link system used by the engine to store and retrieve advertisement data.
func (*Engine) NotifyPut ¶
func (e *Engine) NotifyPut(ctx context.Context, provider *peer.AddrInfo, contextID []byte, md metadata.Metadata) (cid.Cid, error)
NotifyPut publishes an advertisement that signals the list of multihashes associated to the given contextID is available by this provider with the given metadata. A provider.MultihashLister is required, and is used to look up the list of multihashes associated to a context ID.
Note that prior to calling this function a provider.MultihashLister must be registered.
See: Engine.RegisterMultihashLister, Engine.Publish.
func (*Engine) NotifyRemove ¶
func (e *Engine) NotifyRemove(ctx context.Context, provider peer.ID, contextID []byte) (cid.Cid, error)
NotifyRemove publishes an advertisement that signals the list of multihashes associated to the given contextID is no longer available by this provider.
Note that prior to calling this function a provider.MultihashLister must be registered.
See: Engine.RegisterMultihashLister, Engine.Publish.
func (*Engine) Publish ¶
Publish stores the given advertisement locally via Engine.PublishLocal first, then publishes a message onto the gossipsub to signal the change in the latest advertisement by the provider to indexer nodes.
The publication mechanism uses dagsync.Publisher internally. See: https://github.com/ipni/go-libipni/tree/main/dagsync
func (*Engine) PublishLatest ¶
PublishLatest re-publishes the latest existing advertisement to pubsub.
func (*Engine) PublishLatestHTTP ¶
PublishLatestHTTP publishes the latest existing advertisement to the specific indexers.
func (*Engine) PublishLocal ¶
PublishLocal stores the advertisement in the local link system and marks it locally as the latest advertisement.
The context is used for storing internal mapping information onto the datastore.
See: Engine.Publish.
func (*Engine) RegisterMultihashLister ¶
func (e *Engine) RegisterMultihashLister(mhl provider.MultihashLister)
RegisterMultihashLister registers a provider.MultihashLister that is used to look up the list of multihashes associated to a context ID. At least one such registration must be registered before calls to Engine.NotifyPut and Engine.NotifyRemove.
Note that successive calls to this function will replace the previous registration. Only a single registration is supported.
See: provider.Interface
func (*Engine) Shutdown ¶
Shutdown shuts down the engine and discards all resources opened by the engine. The engine is no longer usable after the call to this function.
func (*Engine) Start ¶
Start starts the engine by instantiating the internal storage and joining the configured gossipsub topic used for publishing advertisements.
The context is used to instantiate the internal LRU cache storage. See: Engine.Shutdown, chunker.NewCachedEntriesChunker, dtsync.NewPublisherFromExisting
type Option ¶
type Option func(*options) error
Option sets a configuration parameter for the provider engine.
func WithChainedEntries ¶
WithChainedEntries sets format of advertisement entries to chained Entry Chunk with the given chunkSize as the maximum number of multihashes per chunk.
If unset, advertisement entries are formatted as chained Entry Chunk with default maximum of 16384 multihashes per chunk.
To use HAMT as the advertisement entries format, see: WithHamtEntries. For caching configuration: WithEntriesCacheCapacity, chunker.CachedEntriesChunker
func WithDataTransfer ¶
func WithDataTransfer(dt datatransfer.Manager) Option
WithDataTransfer sets the instance of datatransfer.Manager to use. If unspecified a new instance is created automatically.
Note that this option only takes effect if the PublisherKind is set to DataTransferPublisher. See: WithPublisherKind.
func WithDatastore ¶
WithDatastore sets the datastore that is used by the engine to store advertisements. If unspecified, an ephemeral in-memory datastore is used. See: datastore.NewMapDatastore.
func WithDirectAnnounce ¶
WithDirectAnnounce sets indexer URLs to send direct HTTP announcements to.
func WithEntriesCacheCapacity ¶
WithEntriesCacheCapacity sets the maximum number of advertisement entries DAG to cache. The cached DAG may be in chained Entry Chunk or HAMT format. See WithChainedEntries and WithHamtEntries to select the ad entries DAG format.
If unset, the default capacity of 1024 is used. This means at most 1024 DAGs will be cached.
The cache is evicted using LRU policy. Note that the capacity dictates the number of complete chains that are cached, not individual entry chunks. This means, the maximum storage used by the cache is a factor of capacity, chunk size and the length of multihashes in each chunk.
As an example, for 128-bit long multihashes the cache with default capacity of 1024, and default chunk size of 16384 can grow up to 256MiB when full.
func WithExtraGossipData ¶
WithExtraGossipData supplies extra data to include in the pubsub announcement. Note that this option only takes effect if the PublisherKind is set to DataTransferPublisher. See: WithPublisherKind.
func WithHamtEntries ¶
func WithHamtEntries(hashAlg multicodec.Code, bitWidth, bucketSize int) Option
WithHamtEntries sets format of advertisement entries to HAMT with the given hash algorithm, bit-width and bucket size.
If unset, advertisement entries are formatted as chained Entry Chunk with default maximum of 16384 multihashes per chunk.
Only multicodec.Identity, multicodec.Sha2_256 and multicodec.Murmur3X64_64 are supported as hash algorithm. The bit-width and bucket size must be at least 3 and 1 respectively. For more information on HAMT data structure, see:
For caching configuration: WithEntriesCacheCapacity, chunker.CachedEntriesChunker
func WithHost ¶
WithHost specifies the host to which the provider engine belongs. If unspecified, a host is created automatically. See: libp2p.New.
func WithHttpPublisherAnnounceAddr ¶
WithHttpPublisherAnnounceAddr sets the address to be supplied in announce messages to tell indexers where to retrieve advertisements.
This option only takes effect if the PublisherKind is set to HttpPublisher.
func WithHttpPublisherHandlerPath ¶
WithHttpPublisherHandlerPath should only be used with WithHttpPublisherWithoutServer
func WithHttpPublisherListenAddr ¶
WithHttpPublisherListenAddr sets the net listen address for the HTTP publisher. If unset, the default net listen address of '0.0.0.0:3104' is used.
Note that this option only takes effect if the PublisherKind is set to HttpPublisher. See: WithPublisherKind.
func WithHttpPublisherWithoutServer ¶
func WithHttpPublisherWithoutServer() Option
WithHttpPublisherWithoutServer sets the HTTP publisher to not start a server. Setting up the handler is left to the user.
func WithPrivateKey ¶
func WithProvider ¶
WithProvider sets the peer and addresses for the provider to put in indexing advertisements. This value overrides `WithRetrievalAddrs`
func WithPublisherKind ¶
func WithPublisherKind(k PublisherKind) Option
WithPublisherKind sets the kind of publisher used to announce new advertisements. If unset, advertisements are only stored locally and no announcements are made. See: PublisherKind.
func WithPurgeCacheOnStart ¶
WithPurgeCacheOnStart sets whether to clear any cached entries chunks when the provider engine starts. If unset, cache is rehydrated from previously cached entries stored in datastore if present. See: WithDatastore.
func WithRetrievalAddrs ¶
WithRetrievalAddrs sets the addresses that specify where to get the content corresponding to an indexing advertisement. If unspecified, the libp2p host listen addresses are used. See: WithHost.
func WithSyncPolicy ¶
func WithTopic ¶
WithTopic sets the pubsub topic on which new advertisements are announced. To use the default pubsub configuration with a specific topic name, use WithTopicName. If both options are specified, WithTopic takes presence.
Note that this option only takes effect if the PublisherKind is set to DataTransferPublisher. See: WithPublisherKind.
func WithTopicName ¶
WithTopicName sets toe topic name on which pubsub announcements are published. To override the default pubsub configuration, use WithTopic.
Note that this option only takes effect if the PublisherKind is set to DataTransferPublisher. See: WithPublisherKind.
type PublisherKind ¶
type PublisherKind string
PublisherKind represents the kind of publisher to use in order to announce a new advertisement to the network. See: WithPublisherKind, NoPublisher, DataTransferPublisher, HttpPublisher.
const ( // NoPublisher indicates that no announcements are made to the network and all advertisements // are only stored locally. NoPublisher PublisherKind = "" // DataTransferPublisher makes announcements over a gossipsub topic and exposes a // datatransfer/graphsync server that allows peers in the network to sync advertisements. DataTransferPublisher PublisherKind = "dtsync" // HttpPublisher exposes a HTTP server that announces published advertisements and allows peers // in the network to sync them over raw HTTP transport. HttpPublisher PublisherKind = "http" )
Source Files