23 lines
2.1 KiB
JavaScript
23 lines
2.1 KiB
JavaScript
|
// Match repository.WikiRecoveryFingerprints: UTF-8 fields framed by an unsigned
|
||
|
|
// 64-bit big-endian byte length. Only digests leave the database.
|
||
|
|
const framed = value => `int8send(octet_length(convert_to(coalesce(${value},''),'UTF8'))::bigint)||convert_to(coalesce(${value},''),'UTF8')`
|
||
|
|
const framedArray = expression => `(SELECT string_agg(${framed('field')}, ''::bytea ORDER BY ordinal) FROM unnest(${expression}) WITH ORDINALITY AS fields(field,ordinal))`
|
||
|
|
const textArray = fields => `ARRAY[${fields.join(',')}]::text[]`
|
||
|
|
|
||
|
|
function recoveryFingerprintsSQL(source = 'd', kb = 'k', doc = 'n') {
|
||
|
|
const fields = textArray(["'wiki-recovery-source-v1'", `${source}.tenant_id::text`, `${source}.scope_id`, `${source}.related_id`, `${doc}.title`])
|
||
|
|
const chunkFields = textArray(['c.id', 'c.content', "coalesce(c.content_revision,0)::text", "coalesce(c.chunk_index,0)::text",
|
||
|
|
"coalesce(c.start_at,0)::text", "coalesce(c.end_at,0)::text", 'c.chunk_type', 'c.parent_chunk_id', 'c.image_info', "coalesce(c.is_enabled,false)::text"])
|
||
|
|
const configFields = textArray(["'wiki-recovery-model-v1'", `${kb}.wiki_config::text`, `${kb}.indexing_strategy::text`])
|
||
|
|
const modelFields = textArray(['m.id', 'm.name', 'm.type', 'm.source', 'm.status', 'm.parameters::text'])
|
||
|
|
return {
|
||
|
|
source: `encode(sha256(${framedArray(fields)} || coalesce((SELECT string_agg(${framedArray(chunkFields)}, ''::bytea ORDER BY c.id)
|
||
|
|
FROM chunks c WHERE c.tenant_id=${source}.tenant_id AND c.knowledge_base_id=${source}.scope_id AND c.knowledge_id=${source}.related_id AND c.deleted_at IS NULL),''::bytea)),'hex')`,
|
||
|
|
model: `encode(sha256(${framedArray(configFields)} || coalesce((SELECT string_agg(${framedArray(modelFields)}, ''::bytea ORDER BY wanted.ordinal)
|
||
|
|
FROM unnest(ARRAY[coalesce(nullif(${kb}.wiki_config->>'synthesis_model_id',''),${kb}.summary_model_id),${kb}.embedding_model_id]) WITH ORDINALITY wanted(id,ordinal)
|
||
|
|
JOIN models m ON m.id=wanted.id AND m.tenant_id=${source}.tenant_id AND m.deleted_at IS NULL WHERE wanted.id<>''),''::bytea)),'hex')`,
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
module.exports = { recoveryFingerprintsSQL }
|