Merged in feature/textextract (pull request #108)

Start adding Textract + UUID changes * base * startclient * ts * short * tests
2025-03-20 11:06:41 +00:00
parent 1587da9d11
commit 53ea7d34e6
196 changed files with 25972 additions and 1456 deletions
@@ -0,0 +1,110 @@
+-- name: CreateClient :one
+INSERT INTO clients (externalId, name) VALUES ($1, $2) RETURNING id;
+
+-- name: GetClient :one
+SELECT * FROM fullClients WHERE id = $1;
+
+-- name: GetClientByExternalId :one
+SELECT * FROM fullClients WHERE externalId = $1;
+
+-- name: UpdateClient :exec
+UPDATE clients SET name = $1 WHERE id = $2;
+
+-- name: AddClientCanSync :exec
+INSERT INTO clientCanSync (canSync, clientId) VALUES ($1, $2);
+
+-- name: IsClientSynced :one
+WITH
+docs AS (
+    -- Get all documents for this client
+    SELECT id, clientId FROM documents WHERE clientId = $1
+),
+doc_clean_entries as (
+    -- Documents with their current clean entries
+    SELECT
+        d.id AS document_id,
+        d.clientId as client_id,
+        cte.id AS clean_entry_id,
+        cte.fail as clean_fail
+    FROM
+        docs d
+        LEFT JOIN currentCleanEntries cte ON cte.documentId = d.id
+),
+doc_text_entries AS (
+    -- Documents with their current text entries
+    SELECT
+        d.document_id,
+        cte.id AS text_entry_id,
+        d.clean_entry_id,
+        d.clean_fail
+    FROM
+        doc_clean_entries d
+        LEFT JOIN currentTextEntries cte ON cte.cleanEntryId = d.clean_entry_id
+),
+required_results AS (
+    -- All required document-query-version combinations
+    SELECT
+        d.document_id,
+        cqdt.queryId,
+        cqdt.queryVersion,
+        dte.text_entry_id
+    FROM
+        doc_clean_entries d
+        JOIN collectorQueryDependencyTree cqdt ON cqdt.clientId = d.client_id
+        JOIN doc_text_entries dte ON dte.document_id = d.document_id
+            and dte.text_entry_id IS NOT NULL
+    where d.clean_fail is null
+),
+existing_results AS (
+    -- Valid results that exist
+    SELECT
+        rr.document_id AS document_id,
+        r.queryId,
+        r.id AS result_id
+    FROM
+        results r
+        JOIN required_results rr ON
+            r.queryId = rr.queryId AND
+            r.queryVersion = rr.queryVersion AND
+            r.textEntryId = rr.text_entry_id
+),
+missing_results AS (
+    -- Find missing results
+    SELECT rr.queryId
+    FROM required_results rr
+    LEFT JOIN existing_results er on er.queryId = rr.queryId
+    WHERE er.result_id is null
+),
+dependency_check AS (
+    -- Check for missing dependencies
+    SELECT DISTINCT er.queryId, er.result_id, qcri.queryId, rd.resultId
+    FROM existing_results er
+    JOIN queryCurrentRequiredIds qcri on qcri.requiredQueryId = er.queryId
+    LEFT JOIN resultDependencies rd on rd.requiredResultId = er.result_id
+    WHERE rd.resultId is null
+)
+SELECT (
+    -- No documents means client is synced
+    NOT EXISTS (SELECT 1 FROM docs)
+
+    OR
+
+    -- Documents with no text entries
+    (
+        NOT EXISTS (
+            SELECT 1 FROM doc_text_entries
+            where clean_entry_id is null or
+            (clean_fail is null and text_entry_id is null)
+        )
+
+        and
+
+        -- Documents with missing results
+        NOT EXISTS (SELECT 1 FROM missing_results)
+
+        and
+
+        -- Documents with missing dependencies
+        NOT EXISTS (SELECT 1 FROM dependency_check)
+    )::bool
+)::bool as is_synced;