feat: support geohash prefix matching for #g filters

Author: kanishka0411

.changeset/geohash-prefix-filters.md

@@ -0,0 +1,9 @@
+---
+"nostream": patch
+---
+
+Implement geohash wildcard/prefix behavior for `#g` filters (closes #265): a
+criterion ending in `*` matches any event `g` tag whose value starts with the
+prefix before `*`; exact matching (no `*`) is unchanged. Only normal geohash
+prefixes are intended as input. This is a Nostream extension, not part of
+NIP-12.

src/repositories/event-repository.ts

@@ -58,6 +58,11 @@ const groupByLengthSpec = groupBy<string, 'exact' | 'even' | 'odd'>(
 
 const logger = createLogger('event-repository')
 
+const isGeohashPrefixCriterion = (filterName: string, criterion: string): boolean =>
+  filterName === '#g' && criterion.endsWith('*')
+
+const stripGeohashPrefixWildcard = (criterion: string): string => criterion.slice(0, -1)
+
 export class EventRepository implements IEventRepository {
   public constructor(
     private readonly masterDbClient: DatabaseClient,
@@ -193,8 +198,21 @@ export class EventRepository implements IEventRepository {
             isEmpty,
             () => andWhereRaw('1 = 0', bd),
             forEach(
-              (criterion: string) =>
-                void orWhereRaw('event_tags.tag_name = ? AND event_tags.tag_value = ?', [filterName[1], criterion], bd),
+              (criterion: string) => {
+                if (isGeohashPrefixCriterion(filterName, criterion)) {
+                  return void orWhereRaw(
+                    'event_tags.tag_name = ? AND event_tags.tag_value LIKE ?',
+                    [filterName[1], `${stripGeohashPrefixWildcard(criterion)}%`],
+                    bd,
+                  )
+                }
+
+                return void orWhereRaw(
+                  'event_tags.tag_name = ? AND event_tags.tag_value = ?',
+                  [filterName[1], criterion],
+                  bd,
+                )
+              },
             ),
           )(criteria)
         })

src/utils/event.ts

@@ -40,6 +40,18 @@ export const isEventMatchingFilter =
   (filter: SubscriptionFilter) =>
   (event: Event): boolean => {
     const startsWith = (input: string) => (prefix: string) => input.startsWith(prefix)
+    const isMatchingGenericTagCriterion = (key: string, criterion: string) => (tag: Tag): boolean => {
+      const [, tagName] = key
+      if (tag[0] !== tagName) {
+        return false
+      }
+
+      if (key === '#g' && criterion.endsWith('*')) {
+        return tag[1].startsWith(criterion.slice(0, -1))
+      }
+
+      return tag[1] === criterion
+    }
 
     // NIP-01: Basic protocol flow description
 
@@ -84,7 +96,7 @@ export const isEventMatchingFilter =
       Object.entries(filter)
         .filter(([key, criteria]) => isGenericTagQuery(key) && Array.isArray(criteria))
         .some(([key, criteria]) => {
-          return !event.tags.some((tag) => tag[0] === key[1] && criteria.includes(tag[1]))
+          return !event.tags.some((tag) => criteria.some((criterion) => isMatchingGenericTagCriterion(key, criterion)(tag)))
         })
     ) {
       return false

test/unit/repositories/event-repository.spec.ts

@@ -320,6 +320,28 @@ describe('EventRepository', () => {
         })
       })
 
+      describe('#g', () => {
+        it('selects geohash tags by prefix when criterion ends with wildcard', () => {
+          const filters = [{ '#g': ['u4pruyd*'] }]
+
+          const query = repository.findByFilters(filters).toString()
+
+          expect(query).to.equal(
+            'select "events".* from "events" left join "event_tags" on "events"."event_id" = "event_tags"."event_id" where (event_tags.tag_name = \'g\' AND event_tags.tag_value LIKE \'u4pruyd%\') order by "event_created_at" asc, "event_id" asc limit 500',
+          )
+        })
+
+        it('keeps geohash tags exact when criterion has no wildcard', () => {
+          const filters = [{ '#g': ['u4pruyd'] }]
+
+          const query = repository.findByFilters(filters).toString()
+
+          expect(query).to.equal(
+            'select "events".* from "events" left join "event_tags" on "events"."event_id" = "event_tags"."event_id" where (event_tags.tag_name = \'g\' AND event_tags.tag_value = \'u4pruyd\') order by "event_created_at" asc, "event_id" asc limit 500',
+          )
+        })
+      })
+
       describe('#p', () => {
         it('selects no events given empty list of #p tags', () => {
           const filters = [{ '#p': [] }]

test/unit/utils/event.spec.ts

@@ -302,6 +302,33 @@ describe('NIP-12', () => {
       expect(isEventMatchingFilter({ '#r': ['something else'] })(event)).to.be.false
     })
   })
+
+  describe('#g filter', () => {
+    beforeEach(() => {
+      event = {
+        id: 'cf8de9db67a1d7203512d1d81e6190f5e53abfdc0ac90275f67172b65a5b09a0',
+        pubkey: 'e8b487c079b0f67c695ae6c4c2552a47f38adfa2533cc5926bd2c102942fdcb7',
+        created_at: 1645030752,
+        kind: 1,
+        tags: [['g', 'u4pruydqqvj']],
+        content: 'g',
+        sig: '53d12018d036092794366283eca36df4e0cabd014b6e91bbf684c8bb9bbbe9dedafa77b6b928587e11e05e036227598dded8713e8da17d55076e12242b361542',
+      }
+    })
+
+    it('returns true if #g filter contains a matching geohash prefix wildcard', () => {
+      expect(isEventMatchingFilter({ '#g': ['u4pruyd*'] })(event)).to.be.true
+    })
+
+    it('returns false if #g filter contains a non-matching geohash prefix wildcard', () => {
+      expect(isEventMatchingFilter({ '#g': ['u4pruz*'] })(event)).to.be.false
+    })
+
+    it('keeps #g filter exact when criterion has no wildcard', () => {
+      expect(isEventMatchingFilter({ '#g': ['u4pruyd'] })(event)).to.be.false
+      expect(isEventMatchingFilter({ '#g': ['u4pruydqqvj'] })(event)).to.be.true
+    })
+  })
 })
 
 describe('NIP-16', () => {