test: point the search-parity oracle at wss://search.brainstorm.world

Switch fetch_fixtures.py's default relay and the parity tests' docs from
search-staging to the production search relay, and re-record
tools/search-parity/fixture.json from it. The new corpus keeps the same
shape: 8 cases x 8 events, every event satisfies its NIP-01 filter, and
8 of 64 carry no literal occurrence of their search term, so the KDoc's
numbers still hold; its quoted examples are swapped for ones in the new
recording. Also fixes the script's usage line, which pointed at a stale
output path.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BvR9ox7m91GLFUARuqj8o5

commons/src/jvmTest/kotlin/com/vitorpamplona/amethyst/commons/search/SearchParityFixture.kt

@@ -30,7 +30,7 @@ import kotlinx.serialization.json.jsonPrimitive
 import java.io.File
 
 /**
- * Real answers from a live vespa-relay (`search-staging.brainstorm.world`), recorded once by
+ * Real answers from a live vespa-relay (`search.brainstorm.world`), recorded once by
  * `tools/search-parity/fetch_fixtures.py` and read from disk here.
  *
  * Recorded rather than fetched so `./gradlew test` stays offline and deterministic — a suite that

commons/src/jvmTest/kotlin/com/vitorpamplona/amethyst/commons/search/SearchRelayParityTest.kt

@@ -31,7 +31,7 @@ import kotlin.test.assertTrue
  *
  * ## What this can and cannot prove
  *
- * The oracle is `search-staging.brainstorm.world`, a vespa-relay — the same software Amethyst's
+ * The oracle is `search.brainstorm.world`, a vespa-relay — the same software Amethyst's
  * token language was ported from. It is a **valid oracle for NIP-01 filtering** and **not one for
  * NIP-50 text matching**, and the difference is the whole design of this file.
  *
@@ -42,8 +42,8 @@ import kotlin.test.assertTrue
  * pins it.
  *
  * NIP-50 is not closed. This relay retrieves topically, not lexically: asked for `bitcoin` it
- * returns a block-height summary that never says "bitcoin", and asked for `nostr` it returns "made
- * my display name refer to my npub's last characters". Eight of the 64 recorded events — 12% —
+ * returns a liquidation-levels heatmap that only ever says "BTC", and asked for `"open source"` it
+ * returns "FOSS you say will save you". Eight of the 64 recorded events — 12% —
  * contain no literal occurrence of the term that fetched them.
  *
  * The fixture is recorded with the `include:spam` lens, which waives the web-of-trust gate. That

tools/search-parity/fetch_fixtures.py

@@ -6,7 +6,7 @@ Run by hand when the fixture needs refreshing; the test that consumes its output
 touches the network, so `./gradlew test` stays hermetic and offline.
 
     ./gradlew :cli:installDist
-    tools/search-parity/fetch_fixtures.py > commons/src/jvmTest/resources/search-parity-fixture.json
+    tools/search-parity/fetch_fixtures.py > tools/search-parity/fixture.json
 
 Each case pins the filter FIELDS (kinds/tags/window) as flags rather than folding them into
 the search string, so the test can rebuild the identical NIP-01 Filter and assert our matcher
@@ -18,7 +18,7 @@ import shlex
 import subprocess
 import sys
 
-RELAY = os.environ.get("RELAY", "wss://search-staging.brainstorm.world")
+RELAY = os.environ.get("RELAY", "wss://search.brainstorm.world")
 AMY = os.environ.get("AMY", os.path.join(os.getcwd(), "cli/build/install/amy/bin/amy"))
 # The relay gates and ranks results through the searcher's web of trust, and answers an
 # anonymous query with nothing at all. `include:spam` waives that gate, which is what makes
@@ -26,9 +26,9 @@ AMY = os.environ.get("AMY", os.path.join(os.getcwd(), "cli/build/install/amy/bin
 # to one account's trust graph and re-records differently as that graph moves.
 #
 # It does NOT make the relay lexical. Measured on this corpus, results still include events
-# carrying no literal occurrence of the term — asked for `nostr` it returns "made my display
-# name refer to my npub's last characters". Retrieval is semantic; see the parity test for why
-# that means text results are not compared.
+# carrying no literal occurrence of the term — asked for `"open source"` it returns "FOSS you say
+# will save you". Retrieval is semantic; see the parity test for why that means text results are
+# not compared.
 LENS = os.environ.get("LENS", "include:spam")
 LIMIT = os.environ.get("LIMIT", "8")
 TIMEOUT = os.environ.get("TIMEOUT", "30")