@ -2,21 +2,62 @@
@@ -2,21 +2,62 @@
* Built - in “ faux spells ” : same NoteList + filters as kind - 777 spells . The Spells page uses live
* ` subscribeTimeline ` ( same as Following ) so the first relay results stream in immediately instead of
* waiting for every relay to EOSE on a one - shot query .
*
* * * Why faux feeds can feel slow : * * each timeline shard opens live REQs over the prioritized relay
* stack ( see { @link applyFauxSpellCapsToSubRequests } ) . The * * interests * * spell uses * * one * * shard : all subscribed
* topics go in a single ` #t ` filter ( NIP - 01 OR semantics ) . The notifications spell uses a narrow
* kind list vs full profile kinds .
* /
import { ExtendedKind , PROFILE_FEED_KINDS , READ_ONLY_RELAY_URLS } from '@/constants'
import {
extractHashtagsFromContent ,
extractTTagsFromEvent ,
normalizeTopic
} from '@/lib/discussion-topics'
import { getImetaInfosFromEvent } from '@/lib/event'
import { normalizeTopic } from '@/lib/discussion-topics'
import { normalizeUrl } from '@/lib/url'
import type { TFeedSubRequest } from '@/types'
import { type Event , type Filter , kinds } from 'nostr-tools'
const NOTIFICATION_LIMIT = 500
const DISCUSSION_LIMIT = 500
const MAX_BOOKMARK_IDS = 250
/** Default caps for every faux spell feed (relays per subrequest, events per REQ). */
export const FAUX_SPELL_MAX_RELAYS = 6
export const FAUX_SPELL_EVENT_LIMIT = 200
/ * *
* Trim relay lists and filter limits ( and bookmark ` ids ` ) so faux feeds stay cheap to open .
* /
export function applyFauxSpellCapsToSubRequests ( requests : TFeedSubRequest [ ] ) : TFeedSubRequest [ ] {
return requests . map ( ( r ) = > {
const urls = r . urls . slice ( 0 , FAUX_SPELL_MAX_RELAYS )
const f = { . . . r . filter }
const prevLimit = f . limit
f . limit =
typeof prevLimit === 'number' && prevLimit > 0
? Math . min ( prevLimit , FAUX_SPELL_EVENT_LIMIT )
: FAUX_SPELL_EVENT_LIMIT
if ( Array . isArray ( f . ids ) && f . ids . length > FAUX_SPELL_EVENT_LIMIT ) {
f . ids = f . ids . slice ( 0 , FAUX_SPELL_EVENT_LIMIT )
}
return { urls , filter : f }
} )
}
/ * *
* Mention / notification - shaped kinds only ( aligned with ` NotificationProvider ` , plus zap receipts ) .
* Not full { @link PROFILE_FEED_KINDS } — that asked relays for huge multi - kind slices per ` #p ` .
* /
export const NOTIFICATION_SPELL_KINDS = [
kinds . ShortTextNote ,
kinds . Repost ,
kinds . Reaction ,
kinds . Zap ,
ExtendedKind . COMMENT ,
ExtendedKind . POLL_RESPONSE ,
ExtendedKind . VOICE_COMMENT ,
ExtendedKind . POLL ,
ExtendedKind . PUBLIC_MESSAGE ,
ExtendedKind . ZAP_RECEIPT
] as const
/ * *
* Max distinct ` t ` tag values in one filter ( very long ` #t ` arrays can hit relay limits ) .
* /
const INTERESTS_MAX_TOPICS = 80
/ * *
* Append { @link READ_ONLY_RELAY_URLS } ( e . g . aggr ) after the curated set so every faux REQ includes them unless blocked .
@ -40,6 +81,7 @@ export function appendCuratedReadOnlyRelays(curated: string[], blockedRelays: st
@@ -40,6 +81,7 @@ export function appendCuratedReadOnlyRelays(curated: string[], blockedRelays: st
return out
}
/** NIP-style native media kinds only (picture, video, short video, voice). */
export const MEDIA_SPELL_KINDS = [
ExtendedKind . PICTURE ,
ExtendedKind . VIDEO ,
@ -47,104 +89,12 @@ export const MEDIA_SPELL_KINDS = [
@@ -47,104 +89,12 @@ export const MEDIA_SPELL_KINDS = [
ExtendedKind . VOICE
] as const
/** Kinds shown in the Media faux spell: native media + kind 1 notes filtered by {@link mediaSpellExtraShouldHideEvent}. */
export const MEDIA_SPELL_SHOW_KINDS = [
kinds . ShortTextNote ,
. . . MEDIA_SPELL_KINDS
] as const
/ * *
* Topic roots for kind 1 in the Media spell : a note must also match one of these via ` t ` tag or ` #hashtag `
* ( after { @link normalizeTopic } ) , * * and * * carry media ( imeta / media URL / image | video | audio tag ) .
* /
export const MEDIA_SPELL_TOPIC_SEEDS = [
'vlog' ,
'video' ,
'reel' ,
'gallery' ,
'podcast' ,
'photography' ,
'photo' ,
'music' ,
'screencast'
] as const
const MEDIA_SPELL_TOPIC_KEYWORDS = new Set (
MEDIA_SPELL_TOPIC_SEEDS . map ( ( t ) = > normalizeTopic ( t ) ) . filter ( Boolean )
)
function hasMediaSpellTopicTag ( event : Event ) : boolean {
for ( const topic of extractTTagsFromEvent ( event ) ) {
if ( topic && MEDIA_SPELL_TOPIC_KEYWORDS . has ( topic ) ) return true
}
for ( const topic of extractHashtagsFromContent ( event . content ) ) {
if ( topic && MEDIA_SPELL_TOPIC_KEYWORDS . has ( topic ) ) return true
}
return false
}
function imetaTagsIndicateMedia ( event : Event ) : boolean {
for ( const im of getImetaInfosFromEvent ( event ) ) {
const mime = im . m ? . toLowerCase ( ) ? ? ''
if ( mime . startsWith ( 'image/' ) || mime . startsWith ( 'video/' ) || mime . startsWith ( 'audio/' ) ) {
return true
}
const u = im . url ? ? ''
if (
/\.(jpe?g|png|gif|webp|heic|mp4|webm|m4v|mov|mkv|avi|mp3|m4a|aac|ogg|opus|wav|flac)(\?|#|$)/i . test (
u
)
) {
return true
}
}
return false
}
function hasImageOrStreamTag ( event : Event ) : boolean {
for ( const t of event . tags ) {
const name = t [ 0 ] ? . toLowerCase ( )
if ( name === 'image' && t [ 1 ] ? . startsWith ( 'http' ) ) return true
if ( ( name === 'video' || name === 'audio' || name === 'stream' ) && t [ 1 ] ? . startsWith ( 'http' ) ) {
return true
}
}
return false
}
const CONTENT_MEDIA_FILE_EXT_RE =
/https?:\/\/[^\s<>"')]+\.(?:jpe?g|png|gif|webp|svg|bmp|heic|mp4|webm|m4v|mov|mkv|avi|mp3|m4a|aac|ogg|opus|wav|flac)(?:[\w#./?&=%~+-]*)/i
/** Embed-style hosts (excludes GIF sticker sites like Giphy/Tenor). */
const CONTENT_MEDIA_HOST_RE =
/https?:\/\/(?:(?:[\w-]+\.)*(?:spotify\.com|fountain\.fm)\/|(?:www\.)?(?:youtube\.com\/(?:watch|embed|shorts)|youtu\.be\/|vimeo\.com\/|twitch\.tv\/|instagram\.com\/|(?:i\.)?imgur\.com\/|soundcloud\.com\/|(?:www\.)?tiktok\.com\/|rumble\.com\/|odysee\.com\/))/i
function contentHasMediaUrl ( content : string ) : boolean {
return CONTENT_MEDIA_FILE_EXT_RE . test ( content ) || CONTENT_MEDIA_HOST_RE . test ( content )
}
function hasKind1MediaPayload ( event : Event ) : boolean {
return imetaTagsIndicateMedia ( event ) || hasImageOrStreamTag ( event ) || contentHasMediaUrl ( event . content )
}
/** Kind 1: require {@link MEDIA_SPELL_TOPIC_SEEDS} match **and** imeta / media URL / image|video|audio tag. */
export function isKind1MediaSpellEligible ( event : Event ) : boolean {
if ( event . kind !== kinds . ShortTextNote ) return false
return hasMediaSpellTopicTag ( event ) && hasKind1MediaPayload ( event )
}
/** NoteList `extraShouldHideEvent`: hide kind 1 notes that fail the combined topic + media check. */
export function mediaSpellExtraShouldHideEvent ( evt : Event ) : boolean {
if ( evt . kind !== kinds . ShortTextNote ) return false
return ! isKind1MediaSpellEligible ( evt )
}
/** Notifications spell: same kind set as profile-style feeds, restricted to `#p` = you on the relay. */
/** Notifications faux spell: `#p` = you, narrow kinds — see module docstring. */
export function buildMentionsSpellFilter ( pubkey : string ) : Filter {
const pk = /^[0-9a-f]{64}$/i . test ( pubkey . trim ( ) ) ? pubkey . trim ( ) . toLowerCase ( ) : pubkey . trim ( )
return {
kinds : [ . . . PROFILE_FEED _KINDS] ,
limit : NOTIFICATION _LIMIT,
kinds : [ . . . NOTIFICATION_SPELL_KINDS ] ,
limit : FAUX_SPELL_EVENT_LIMIT ,
'#p' : [ pk ]
}
}
@ -152,22 +102,25 @@ export function buildMentionsSpellFilter(pubkey: string): Filter {
@@ -152,22 +102,25 @@ export function buildMentionsSpellFilter(pubkey: string): Filter {
export function buildDiscussionFilter ( ) : Filter {
return {
kinds : [ ExtendedKind . DISCUSSION ] ,
limit : DISCUSSION _LIMIT
limit : FAUX_SPELL_EVENT _LIMIT
}
}
export function buildMediaSpellFilter ( ) : Filter {
return { kinds : [ . . . MEDIA_SPELL_SHOW_ KINDS ] , limit : 500 }
return { kinds : [ . . . MEDIA_SPELL_KINDS ] , limit : FAUX_SPELL_EVENT_LIMIT }
}
export function buildCalendarSpellFilter ( ) : Filter {
return {
kinds : [ ExtendedKind . CALENDAR_EVENT_DATE , ExtendedKind . CALENDAR_EVENT_TIME ] ,
limit : 200
limit : FAUX_SPELL_EVENT_LIMIT
}
}
/** One subrequest per topic (OR). Uses same kind set as the main profile/favorites feed. */
/ * *
* One subrequest for all interests : NIP - 01 treats multiple ` #t ` values as OR ( any topic matches ) .
* Same relay set as before , but a single timeline shard instead of one per hashtag .
* /
export function buildInterestsSubRequests (
relayUrls : string [ ] ,
rawTopics : string [ ] ,
@ -176,16 +129,18 @@ export function buildInterestsSubRequests(
@@ -176,16 +129,18 @@ export function buildInterestsSubRequests(
if ( ! relayUrls . length || ! rawTopics . length || ! kindsList . length ) return [ ]
const topics = Array . from (
new Set ( rawTopics . map ( ( t ) = > normalizeTopic ( t ) ) . filter ( ( t ) = > t . length > 0 ) )
)
) . slice ( 0 , INTERESTS_MAX_TOPICS )
if ( ! topics . length ) return [ ]
return topics . map ( ( topic ) = > ( {
return [
{
urls : relayUrls ,
filter : {
kinds : kindsList ,
'#t' : [ topic ] ,
limit : 400
'#t' : topics ,
limit : FAUX_SPELL_EVENT_LIMIT
}
}
} ) )
]
}
/** Bookmark list e-tags only (hex ids); addressable (a-tag) bookmarks need separate fetches. */
@ -195,5 +150,7 @@ export function buildBookmarksSubRequests(bookmarkListEvent: Event | null, urls:
@@ -195,5 +150,7 @@ export function buildBookmarksSubRequests(bookmarkListEvent: Event | null, urls:
. filter ( ( t ) = > t [ 0 ] === 'e' && t [ 1 ] && /^[a-f0-9]{64}$/i . test ( t [ 1 ] ) )
. map ( ( t ) = > t [ 1 ] as string )
if ( ! ids . length ) return [ ]
return [ { urls , filter : { ids : ids.slice ( 0 , MAX_BOOKMARK_IDS ) , limit : MAX_BOOKMARK_IDS } } ]
const cap = FAUX_SPELL_EVENT_LIMIT
const slice = ids . slice ( 0 , cap )
return [ { urls , filter : { ids : slice , limit : slice.length } } ]
}
