@shadowforge0/aquifer-memory 1.5.8 → 1.5.12

package/scripts/sample-bench-queries.sql

@@ -1,75 +0,0 @@
--- Sample real user "first question" turns from sessions for retro recall bench.
--- Output: 30 rows balanced across agent (cc/main/life) × language (zh/en/mixed).
--- Usage:
---   psql $DATABASE_URL -f scripts/sample-bench-queries.sql --csv > queries.csv
---   then convert to JSON via:
---     node -e "const fs=require('fs');const lines=fs.readFileSync('queries.csv','utf8').trim().split('\n').slice(1);const out={version:1,queries:lines.map((l,i)=>{const[id,sid,ag,src,lang,qt]=l.split(','); return {id:'q-sql-'+(i+1),sessionRowId:Number(id),sessionId:sid,agentId:ag,source:src,sourceKind:'sql-sampled',lang,text:qt.replace(/^\"|\"$/g,'').replace(/\"\"/g,'\"')}})};fs.writeFileSync('queries.json',JSON.stringify(out,null,2))"
-
--- Override with `psql -v schema=aquifer -f sample-bench-queries.sql`
-\if :{?schema}
-\else
-  \set schema 'miranda'
-\endif
-
-WITH raw_turns AS (
-  SELECT
-    s.id AS session_row_id,
-    s.session_id,
-    s.agent_id,
-    s.source,
-    s.started_at,
-    m.ordinality AS turn_ordinal,
-    m.msg->>'role' AS role,
-    m.msg->>'content' AS content
-  FROM :"schema".sessions s
-  CROSS JOIN LATERAL jsonb_array_elements(
-    COALESCE(s.messages->'normalized', s.messages)
-  ) WITH ORDINALITY AS m(msg, ordinality)
-  WHERE s.agent_id IN ('main', 'life', 'cc')
-    AND s.user_count > 0
-),
-question_turns AS (
-  SELECT *,
-    CASE
-      WHEN content ~ '[\u4e00-\u9fff]' AND content ~ '[A-Za-z]' THEN 'mixed'
-      WHEN content ~ '[\u4e00-\u9fff]' THEN 'zh'
-      WHEN content ~ '[A-Za-z]' THEN 'en'
-      ELSE 'other'
-    END AS lang_bucket,
-    row_number() OVER (PARTITION BY session_row_id ORDER BY turn_ordinal) AS seq_in_session
-  FROM raw_turns
-  WHERE role = 'user'
-    AND content IS NOT NULL
-    AND length(content) BETWEEN 4 AND 200
-    AND (content ~ '[?？]' OR content ~ '(嗎|呢|怎麼|如何|why|how|what|can you|could you)')
-),
-first_questions AS (
-  SELECT *
-  FROM question_turns
-  WHERE seq_in_session = 1
-    AND lang_bucket IN ('mixed', 'zh', 'en')
-),
-balanced AS (
-  SELECT *,
-    row_number() OVER (PARTITION BY agent_id, lang_bucket ORDER BY random()) AS bucket_rn
-  FROM first_questions
-)
-SELECT
-  session_row_id,
-  session_id,
-  agent_id,
-  source,
-  lang_bucket AS lang,
-  content AS query_text
-FROM balanced
-WHERE
-  (agent_id = 'main' AND lang_bucket = 'mixed' AND bucket_rn <= 4) OR
-  (agent_id = 'main' AND lang_bucket = 'zh'    AND bucket_rn <= 3) OR
-  (agent_id = 'main' AND lang_bucket = 'en'    AND bucket_rn <= 2) OR
-  (agent_id = 'life' AND lang_bucket = 'mixed' AND bucket_rn <= 3) OR
-  (agent_id = 'life' AND lang_bucket = 'zh'    AND bucket_rn <= 2) OR
-  (agent_id = 'life' AND lang_bucket = 'en'    AND bucket_rn <= 2) OR
-  (agent_id = 'cc'   AND lang_bucket = 'mixed' AND bucket_rn <= 4) OR
-  (agent_id = 'cc'   AND lang_bucket = 'zh'    AND bucket_rn <= 3) OR
-  (agent_id = 'cc'   AND lang_bucket = 'en'    AND bucket_rn <= 2)
-ORDER BY agent_id, lang;