weave-typescript 0.28.0 → 0.29.0

package/dist/weavesql/weavedb/ingestion_sql.js

@@ -1,13 +1,15 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.upsertFacetConfigQuery = exports.listActiveFacetConfigsQuery = exports.listFacetConfigsQuery = exports.syncOrganizationIngestionReasoningDefaultsQuery = exports.upsertOrganizationIngestionSettingsQuery = exports.getOrganizationIngestionSettingsQuery = exports.ensureOrganizationIngestionSettingsQuery = exports.listEntitiesByThreadQuery = exports.upsertEntityRelationshipQuery = exports.upsertThreadEntityQuery = exports.upsertEntityQuery = exports.listThreadActivityByThreadQuery = exports.createThreadActivityQuery = exports.listThreadReconciliationByPendingThreadQuery = exports.upsertThreadReconciliationQuery = exports.deleteThreadReconciliationByPendingThreadQuery = exports.getIngestionRunThreadCountsQuery = exports.listThreadsByIDsQuery = exports.listThreadsBySupersedesQuery = exports.markThreadsSupersededQuery = exports.bulkRejectThreadsQuery = exports.rejectThreadQuery = exports.bulkApproveThreadsQuery = exports.approveThreadQuery = exports.editPendingThreadQuery = exports.updateThreadClassificationQuery = exports.promoteThreadNextEmbeddingsByOrganizationQuery = exports.setThreadNextEmbeddingQuery = exports.updateThreadEmbeddingQuery = exports.listPendingThreadsByOrganizationQuery = exports.listPendingThreadsByRunQuery = exports.listThreadsByIngestionRunQuery = exports.getThreadByOrganizationAndIDQuery = exports.createThreadQuery = exports.listDocumentTagsByDocumentIDQuery = exports.createDocumentTagQuery = exports.deleteDocumentTagsByDocumentIDQuery = exports.listIngestionHistoryQuery = exports.listIngestionRunsByDocumentQuery = exports.listIngestionRunsByOrganizationQuery = exports.updateIngestionRunStateQuery = exports.getIngestionRunByOrganizationAndIDQuery = exports.createIngestionRunQuery = exports.getDatabaseConfigQuery = exports.listDocumentsByOrganizationQuery = exports.setDocumentStatusQuery = exports.setDocumentParseResultQuery = exports.getDocumentByOrganizationAndFingerprintQuery = exports.getDocumentByOrganizationAndIDQuery = exports.createDocumentQuery = void 0;
-exports.deactivateFacetConfigQuery = void 0;
+exports.listEntitiesByThreadQuery = exports.upsertEntityRelationshipQuery = exports.upsertThreadEntityQuery = exports.upsertEntityQuery = exports.listThreadActivityByThreadQuery = exports.createThreadActivityQuery = exports.listThreadReconciliationByPendingThreadQuery = exports.upsertThreadReconciliationQuery = exports.deleteThreadReconciliationByPendingThreadQuery = exports.getIngestionRunThreadCountsQuery = exports.listThreadsByIDsQuery = exports.listThreadsBySupersedesQuery = exports.markThreadsSupersededQuery = exports.bulkRejectThreadsQuery = exports.rejectThreadQuery = exports.bulkApproveThreadsQuery = exports.approveThreadQuery = exports.deleteDocumentThreadByOrganizationAndIDQuery = exports.editDocumentThreadQuery = exports.editPendingThreadQuery = exports.updateThreadClassificationQuery = exports.promoteThreadNextEmbeddingsByOrganizationQuery = exports.setThreadNextEmbeddingQuery = exports.updateThreadEmbeddingQuery = exports.listPendingThreadsByOrganizationQuery = exports.listPendingThreadsByRunQuery = exports.listThreadsByDocumentQuery = exports.listThreadsByIngestionRunQuery = exports.getThreadByOrganizationAndIDQuery = exports.createThreadQuery = exports.deleteDocumentByOrganizationAndIDQuery = exports.listDocumentTagsByDocumentIDQuery = exports.createDocumentTagQuery = exports.deleteDocumentTagByDocumentIDAndTagQuery = exports.deleteDocumentTagsByDocumentIDQuery = exports.listIngestionHistoryQuery = exports.listIngestionRunsByDocumentQuery = exports.listIngestionRunsByOrganizationQuery = exports.updateIngestionRunStateQuery = exports.getIngestionRunByOrganizationAndIDQuery = exports.createIngestionRunQuery = exports.getDatabaseConfigQuery = exports.getAccessibleDocumentSummaryQuery = exports.listAccessibleDocumentSummariesQuery = exports.listDocumentsByOrganizationQuery = exports.setDocumentStatusQuery = exports.setDocumentParseResultQuery = exports.getDocumentByOrganizationAndFingerprintQuery = exports.getDocumentByOrganizationAndIDQuery = exports.createDocumentQuery = void 0;
+exports.deactivateFacetConfigQuery = exports.upsertFacetConfigQuery = exports.listActiveFacetConfigsQuery = exports.listFacetConfigsQuery = exports.syncOrganizationIngestionReasoningDefaultsQuery = exports.upsertOrganizationIngestionSettingsQuery = exports.getOrganizationIngestionSettingsQuery = exports.ensureOrganizationIngestionSettingsQuery = void 0;
 exports.createDocument = createDocument;
 exports.getDocumentByOrganizationAndID = getDocumentByOrganizationAndID;
 exports.getDocumentByOrganizationAndFingerprint = getDocumentByOrganizationAndFingerprint;
 exports.setDocumentParseResult = setDocumentParseResult;
 exports.setDocumentStatus = setDocumentStatus;
 exports.listDocumentsByOrganization = listDocumentsByOrganization;
+exports.listAccessibleDocumentSummaries = listAccessibleDocumentSummaries;
+exports.getAccessibleDocumentSummary = getAccessibleDocumentSummary;
 exports.getDatabaseConfig = getDatabaseConfig;
 exports.createIngestionRun = createIngestionRun;
 exports.getIngestionRunByOrganizationAndID = getIngestionRunByOrganizationAndID;
@@ -21,12 +23,14 @@ exports.listDocumentTagsByDocumentID = listDocumentTagsByDocumentID;
 exports.createThread = createThread;
 exports.getThreadByOrganizationAndID = getThreadByOrganizationAndID;
 exports.listThreadsByIngestionRun = listThreadsByIngestionRun;
+exports.listThreadsByDocument = listThreadsByDocument;
 exports.listPendingThreadsByRun = listPendingThreadsByRun;
 exports.listPendingThreadsByOrganization = listPendingThreadsByOrganization;
 exports.updateThreadEmbedding = updateThreadEmbedding;
 exports.setThreadNextEmbedding = setThreadNextEmbedding;
 exports.updateThreadClassification = updateThreadClassification;
 exports.editPendingThread = editPendingThread;
+exports.editDocumentThread = editDocumentThread;
 exports.approveThread = approveThread;
 exports.bulkApproveThreads = bulkApproveThreads;
 exports.rejectThread = rejectThread;
@@ -381,6 +385,245 @@ async function listDocumentsByOrganization(client, args) {
         };
     });
 }
+exports.listAccessibleDocumentSummariesQuery = `-- name: ListAccessibleDocumentSummaries :many
+WITH latest_runs AS (
+    SELECT DISTINCT ON (r.document_id)
+        r.document_id,
+        r.id AS latest_ingestion_run_id,
+        r.status AS latest_ingestion_status
+    FROM weave.ingestion_runs r
+    WHERE r.organization_id = $1
+    ORDER BY r.document_id, r.started_at DESC
+),
+thread_counts AS (
+    SELECT
+        t.document_id,
+        COUNT(*)::int AS thread_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_PENDING'
+        )::int AS pending_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_APPROVED'
+        )::int AS approved_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_REJECTED'
+        )::int AS rejected_count
+    FROM weave.threads t
+    WHERE t.organization_id = $1
+    GROUP BY t.document_id
+),
+tag_data AS (
+    SELECT
+        dt.document_id,
+        array_agg(dt.tag ORDER BY dt.tag ASC) AS tags,
+        array_agg(dt.confidence ORDER BY dt.tag ASC) AS tag_confidences
+    FROM weave.document_tags dt
+    JOIN weave.documents d ON d.id = dt.document_id
+    WHERE d.organization_id = $1
+    GROUP BY dt.document_id
+)
+SELECT
+    d.id,
+    d.organization_id,
+    d.filename,
+    d.mime_type,
+    d.parser_name,
+    d.status,
+    d.scope,
+    d.sensitivity_tags,
+    COALESCE(tag_data.tags, '{}'::text[]) AS tags,
+    COALESCE(tag_data.tag_confidences, '{}'::real[]) AS tag_confidences,
+    d.uploaded_by_user_id,
+    d.uploaded_at,
+    d.updated_at,
+    d.page_count,
+    d.char_count,
+    d.metadata,
+    COALESCE(latest_runs.latest_ingestion_run_id, '00000000-0000-0000-0000-000000000000'::uuid) AS latest_ingestion_run_id,
+    COALESCE(latest_runs.latest_ingestion_status, 'INGESTION_RUN_STATUS_UNSPECIFIED') AS latest_ingestion_status,
+    COALESCE(thread_counts.thread_count, 0)::int AS thread_count,
+    COALESCE(thread_counts.pending_count, 0)::int AS pending_count,
+    COALESCE(thread_counts.approved_count, 0)::int AS approved_count,
+    COALESCE(thread_counts.rejected_count, 0)::int AS rejected_count
+FROM weave.documents d
+LEFT JOIN latest_runs ON latest_runs.document_id = d.id
+LEFT JOIN thread_counts ON thread_counts.document_id = d.id
+LEFT JOIN tag_data ON tag_data.document_id = d.id
+WHERE d.organization_id = $1
+  AND (
+      d.scope = 'DOCUMENT_SCOPE_ORG'
+      OR (d.scope = 'DOCUMENT_SCOPE_PRIVATE'
+          AND d.uploaded_by_user_id = $2)
+      OR (d.scope = 'DOCUMENT_SCOPE_RESTRICTED'
+          AND EXISTS (
+              SELECT 1
+              FROM weave.user_sensitivity_clearances c
+              WHERE c.organization_id = $1
+                AND c.user_id = $2
+                AND cardinality(d.sensitivity_tags) = 1
+                AND c.sensitivity_tag_slug = d.sensitivity_tags[1]
+          ))
+  )
+  AND (cardinality($3::text[]) = 0 OR d.status = ANY($3::text[]))
+  AND (cardinality($4::text[]) = 0 OR latest_runs.latest_ingestion_status = ANY($4::text[]))
+  AND (cardinality($5::text[]) = 0 OR d.scope = ANY($5::text[]))
+  AND (cardinality($6::text[]) = 0 OR tag_data.tags && $6::text[] OR d.sensitivity_tags && $6::text[])
+  AND (cardinality($7::text[]) = 0 OR d.sensitivity_tags && $7::text[])
+  AND ($8::timestamptz IS NULL OR d.uploaded_at >= $8::timestamptz)
+  AND ($9::timestamptz IS NULL OR d.uploaded_at <= $9::timestamptz)
+ORDER BY d.uploaded_at DESC, d.id DESC
+LIMIT $11 OFFSET $10`;
+async function listAccessibleDocumentSummaries(client, args) {
+    const result = await client.query({
+        text: exports.listAccessibleDocumentSummariesQuery,
+        values: [args.organizationId, args.actingUserId, args.documentStatuses, args.ingestionStatuses, args.scopes, args.tags, args.restrictedTags, args.uploadedAfter, args.uploadedBefore, args.pageOffset, args.pageSize],
+        rowMode: "array"
+    });
+    return result.rows.map(row => {
+        return {
+            id: row[0],
+            organizationId: row[1],
+            filename: row[2],
+            mimeType: row[3],
+            parserName: row[4],
+            status: row[5],
+            scope: row[6],
+            sensitivityTags: row[7],
+            tags: row[8],
+            tagConfidences: row[9],
+            uploadedByUserId: row[10],
+            uploadedAt: row[11],
+            updatedAt: row[12],
+            pageCount: row[13],
+            charCount: row[14],
+            metadata: row[15],
+            latestIngestionRunId: row[16],
+            latestIngestionStatus: row[17],
+            threadCount: row[18],
+            pendingCount: row[19],
+            approvedCount: row[20],
+            rejectedCount: row[21]
+        };
+    });
+}
+exports.getAccessibleDocumentSummaryQuery = `-- name: GetAccessibleDocumentSummary :one
+WITH latest_runs AS (
+    SELECT DISTINCT ON (r.document_id)
+        r.document_id,
+        r.id AS latest_ingestion_run_id,
+        r.status AS latest_ingestion_status
+    FROM weave.ingestion_runs r
+    WHERE r.organization_id = $1
+      AND r.document_id = $2
+    ORDER BY r.document_id, r.started_at DESC
+),
+thread_counts AS (
+    SELECT
+        t.document_id,
+        COUNT(*)::int AS thread_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_PENDING'
+        )::int AS pending_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_APPROVED'
+        )::int AS approved_count,
+        COUNT(*) FILTER (
+            WHERE t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_REJECTED'
+        )::int AS rejected_count
+    FROM weave.threads t
+    WHERE t.organization_id = $1
+      AND t.document_id = $2
+    GROUP BY t.document_id
+),
+tag_data AS (
+    SELECT
+        dt.document_id,
+        array_agg(dt.tag ORDER BY dt.tag ASC) AS tags,
+        array_agg(dt.confidence ORDER BY dt.tag ASC) AS tag_confidences
+    FROM weave.document_tags dt
+    JOIN weave.documents d ON d.id = dt.document_id
+    WHERE d.organization_id = $1
+      AND dt.document_id = $2
+    GROUP BY dt.document_id
+)
+SELECT
+    d.id,
+    d.organization_id,
+    d.filename,
+    d.mime_type,
+    d.parser_name,
+    d.status,
+    d.scope,
+    d.sensitivity_tags,
+    COALESCE(tag_data.tags, '{}'::text[]) AS tags,
+    COALESCE(tag_data.tag_confidences, '{}'::real[]) AS tag_confidences,
+    d.uploaded_by_user_id,
+    d.uploaded_at,
+    d.updated_at,
+    d.page_count,
+    d.char_count,
+    d.metadata,
+    COALESCE(latest_runs.latest_ingestion_run_id, '00000000-0000-0000-0000-000000000000'::uuid) AS latest_ingestion_run_id,
+    COALESCE(latest_runs.latest_ingestion_status, 'INGESTION_RUN_STATUS_UNSPECIFIED') AS latest_ingestion_status,
+    COALESCE(thread_counts.thread_count, 0)::int AS thread_count,
+    COALESCE(thread_counts.pending_count, 0)::int AS pending_count,
+    COALESCE(thread_counts.approved_count, 0)::int AS approved_count,
+    COALESCE(thread_counts.rejected_count, 0)::int AS rejected_count
+FROM weave.documents d
+LEFT JOIN latest_runs ON latest_runs.document_id = d.id
+LEFT JOIN thread_counts ON thread_counts.document_id = d.id
+LEFT JOIN tag_data ON tag_data.document_id = d.id
+WHERE d.organization_id = $1
+  AND d.id = $2
+  AND (
+      d.scope = 'DOCUMENT_SCOPE_ORG'
+      OR (d.scope = 'DOCUMENT_SCOPE_PRIVATE'
+          AND d.uploaded_by_user_id = $3)
+      OR (d.scope = 'DOCUMENT_SCOPE_RESTRICTED'
+          AND EXISTS (
+              SELECT 1
+              FROM weave.user_sensitivity_clearances c
+              WHERE c.organization_id = $1
+                AND c.user_id = $3
+                AND cardinality(d.sensitivity_tags) = 1
+                AND c.sensitivity_tag_slug = d.sensitivity_tags[1]
+          ))
+  )`;
+async function getAccessibleDocumentSummary(client, args) {
+    const result = await client.query({
+        text: exports.getAccessibleDocumentSummaryQuery,
+        values: [args.organizationId, args.documentId, args.actingUserId],
+        rowMode: "array"
+    });
+    if (result.rows.length !== 1) {
+        return null;
+    }
+    const row = result.rows[0];
+    return {
+        id: row[0],
+        organizationId: row[1],
+        filename: row[2],
+        mimeType: row[3],
+        parserName: row[4],
+        status: row[5],
+        scope: row[6],
+        sensitivityTags: row[7],
+        tags: row[8],
+        tagConfidences: row[9],
+        uploadedByUserId: row[10],
+        uploadedAt: row[11],
+        updatedAt: row[12],
+        pageCount: row[13],
+        charCount: row[14],
+        metadata: row[15],
+        latestIngestionRunId: row[16],
+        latestIngestionStatus: row[17],
+        threadCount: row[18],
+        pendingCount: row[19],
+        approvedCount: row[20],
+        rejectedCount: row[21]
+    };
+}
 exports.getDatabaseConfigQuery = `-- name: GetDatabaseConfig :one
 SELECT
     id,
@@ -761,6 +1004,13 @@ async function deleteDocumentTagsByDocumentID(client, args) {
         rowMode: "array"
     });
 }
+exports.deleteDocumentTagByDocumentIDAndTagQuery = `-- name: DeleteDocumentTagByDocumentIDAndTag :execrows
+DELETE FROM weave.document_tags dt
+USING weave.documents d
+WHERE dt.document_id = d.id
+  AND d.organization_id = $1
+  AND dt.document_id = $2
+  AND dt.tag = $3`;
 exports.createDocumentTagQuery = `-- name: CreateDocumentTag :one
 INSERT INTO weave.document_tags (
     document_id,
@@ -809,6 +1059,10 @@ async function listDocumentTagsByDocumentID(client, args) {
         };
     });
 }
+exports.deleteDocumentByOrganizationAndIDQuery = `-- name: DeleteDocumentByOrganizationAndID :execrows
+DELETE FROM weave.documents
+WHERE organization_id = $1
+  AND id = $2`;
 exports.createThreadQuery = `-- name: CreateThread :one
 INSERT INTO weave.threads (
     id,
@@ -1020,6 +1274,68 @@ async function listThreadsByIngestionRun(client, args) {
         };
     });
 }
+exports.listThreadsByDocumentQuery = `-- name: ListThreadsByDocument :many
+SELECT
+    t.id,
+    t.organization_id,
+    t.document_id,
+    d.filename AS document_name,
+    t.ingestion_run_id,
+    t.text,
+    t.text_original,
+    t.knowledge_type,
+    t.facet_fields,
+    t.confidence,
+    t.temporal_validity,
+    t.classification,
+    t.retrieval_status,
+    t.source_location,
+    t.superseded_by,
+    t.supersedes,
+    t.cluster_id,
+    t.extracted_at,
+    t.approved_at,
+    t.approved_by_user_id,
+    t.edited,
+    t.metadata
+FROM weave.threads t
+JOIN weave.documents d ON d.id = t.document_id
+WHERE t.organization_id = $1
+  AND t.document_id = $2
+ORDER BY t.extracted_at ASC, t.id ASC`;
+async function listThreadsByDocument(client, args) {
+    const result = await client.query({
+        text: exports.listThreadsByDocumentQuery,
+        values: [args.organizationId, args.documentId],
+        rowMode: "array"
+    });
+    return result.rows.map(row => {
+        return {
+            id: row[0],
+            organizationId: row[1],
+            documentId: row[2],
+            documentName: row[3],
+            ingestionRunId: row[4],
+            text: row[5],
+            textOriginal: row[6],
+            knowledgeType: row[7],
+            facetFields: row[8],
+            confidence: row[9],
+            temporalValidity: row[10],
+            classification: row[11],
+            retrievalStatus: row[12],
+            sourceLocation: row[13],
+            supersededBy: row[14],
+            supersedes: row[15],
+            clusterId: row[16],
+            extractedAt: row[17],
+            approvedAt: row[18],
+            approvedByUserId: row[19],
+            edited: row[20],
+            metadata: row[21]
+        };
+    });
+}
 exports.listPendingThreadsByRunQuery = `-- name: ListPendingThreadsByRun :many
 SELECT
     id,
@@ -1407,6 +1723,111 @@ async function editPendingThread(client, args) {
         metadata: row[20]
     };
 }
+exports.editDocumentThreadQuery = `-- name: EditDocumentThread :one
+WITH updated AS (
+    UPDATE weave.threads AS t
+    SET
+        text_original = CASE
+            WHEN t.text_original IS NULL OR t.text_original = '' THEN t.text
+            ELSE t.text_original
+        END,
+        text = $1,
+        source_location = COALESCE($2::jsonb, t.source_location),
+        metadata = t.metadata || COALESCE($3::jsonb, '{}'::jsonb),
+        edited = TRUE
+    WHERE t.id = $4
+      AND t.organization_id = $5
+      AND t.document_id = $6
+      AND t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_PENDING'
+    RETURNING
+        id,
+        organization_id,
+        document_id,
+        ingestion_run_id,
+        text,
+        text_original,
+        knowledge_type,
+        facet_fields,
+        confidence,
+        temporal_validity,
+        classification,
+        retrieval_status,
+        source_location,
+        superseded_by,
+        supersedes,
+        cluster_id,
+        extracted_at,
+        approved_at,
+        approved_by_user_id,
+        edited,
+        metadata
+)
+SELECT
+    updated.id,
+    updated.organization_id,
+    updated.document_id,
+    d.filename AS document_name,
+    updated.ingestion_run_id,
+    updated.text,
+    updated.text_original,
+    updated.knowledge_type,
+    updated.facet_fields,
+    updated.confidence,
+    updated.temporal_validity,
+    updated.classification,
+    updated.retrieval_status,
+    updated.source_location,
+    updated.superseded_by,
+    updated.supersedes,
+    updated.cluster_id,
+    updated.extracted_at,
+    updated.approved_at,
+    updated.approved_by_user_id,
+    updated.edited,
+    updated.metadata
+FROM updated
+JOIN weave.documents d ON d.id = updated.document_id`;
+async function editDocumentThread(client, args) {
+    const result = await client.query({
+        text: exports.editDocumentThreadQuery,
+        values: [args.text, args.sourceLocation, args.metadata, args.id, args.organizationId, args.documentId],
+        rowMode: "array"
+    });
+    if (result.rows.length !== 1) {
+        return null;
+    }
+    const row = result.rows[0];
+    return {
+        id: row[0],
+        organizationId: row[1],
+        documentId: row[2],
+        documentName: row[3],
+        ingestionRunId: row[4],
+        text: row[5],
+        textOriginal: row[6],
+        knowledgeType: row[7],
+        facetFields: row[8],
+        confidence: row[9],
+        temporalValidity: row[10],
+        classification: row[11],
+        retrievalStatus: row[12],
+        sourceLocation: row[13],
+        supersededBy: row[14],
+        supersedes: row[15],
+        clusterId: row[16],
+        extractedAt: row[17],
+        approvedAt: row[18],
+        approvedByUserId: row[19],
+        edited: row[20],
+        metadata: row[21]
+    };
+}
+exports.deleteDocumentThreadByOrganizationAndIDQuery = `-- name: DeleteDocumentThreadByOrganizationAndID :execrows
+DELETE FROM weave.threads
+WHERE organization_id = $1
+  AND document_id = $2
+  AND id = $3
+  AND retrieval_status = 'THREAD_RETRIEVAL_STATUS_PENDING'`;
 exports.approveThreadQuery = `-- name: ApproveThread :one
 UPDATE weave.threads
 SET

package/dist/weavesql/weavedb/retrieval_sql.d.ts

@@ -53,6 +53,37 @@ export interface SearchApprovedThreadsSemanticByUserPermissionsRow {
     conceptCluster: string;
 }
 export declare function searchApprovedThreadsSemanticByUserPermissions(client: Client, args: SearchApprovedThreadsSemanticByUserPermissionsArgs): Promise<SearchApprovedThreadsSemanticByUserPermissionsRow[]>;
+export declare const searchAccessibleDocumentsSemanticQuery = "-- name: SearchAccessibleDocumentsSemantic :many\nWITH tag_data AS (\n    SELECT\n        dt.document_id,\n        array_agg(dt.tag ORDER BY dt.tag ASC) AS tags,\n        array_agg(dt.confidence ORDER BY dt.tag ASC) AS tag_confidences\n    FROM weave.document_tags dt\n    JOIN weave.documents d ON d.id = dt.document_id\n    WHERE d.organization_id = $4\n    GROUP BY dt.document_id\n),\nmatches AS (\n    SELECT\n        t.id AS thread_id,\n        t.text,\n        t.classification,\n        t.document_id,\n        d.filename AS document_name,\n        d.mime_type,\n        d.scope,\n        d.sensitivity_tags,\n        d.uploaded_at,\n        COALESCE(tag_data.tags, '{}'::text[]) AS tags,\n        COALESCE(tag_data.tag_confidences, '{}'::real[]) AS tag_confidences,\n        t.source_location,\n        t.approved_at,\n        CAST(\n            1.0::double precision\n            - (t.embedding <=> $5::vector)::double precision\n            AS double precision\n        ) AS score\n    FROM weave.threads t\n    JOIN weave.documents d ON d.id = t.document_id\n    LEFT JOIN tag_data ON tag_data.document_id = d.id\n    WHERE t.organization_id = $4\n      AND t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_APPROVED'\n      AND t.embedding IS NOT NULL\n      AND (\n          d.scope = 'DOCUMENT_SCOPE_ORG'\n          OR (d.scope = 'DOCUMENT_SCOPE_PRIVATE'\n              AND d.uploaded_by_user_id = $6)\n          OR (d.scope = 'DOCUMENT_SCOPE_RESTRICTED'\n              AND EXISTS (\n                  SELECT 1\n                  FROM weave.user_sensitivity_clearances c\n                  WHERE c.organization_id = $4\n                    AND c.user_id = $6\n                    AND cardinality(d.sensitivity_tags) = 1\n                    AND c.sensitivity_tag_slug = d.sensitivity_tags[1]\n              ))\n      )\n      AND (cardinality($7::text[]) = 0 OR tag_data.tags && $7::text[] OR d.sensitivity_tags && $7::text[])\n      AND (cardinality($8::text[]) = 0 OR d.sensitivity_tags && $8::text[])\n      AND (cardinality($9::text[]) = 0 OR d.scope = ANY($9::text[]))\n      AND ($10::timestamptz IS NULL OR d.uploaded_at >= $10::timestamptz)\n      AND ($11::timestamptz IS NULL OR d.uploaded_at <= $11::timestamptz)\n)\nSELECT\n    document_id,\n    document_name,\n    mime_type,\n    scope,\n    sensitivity_tags,\n    tags,\n    tag_confidences,\n    uploaded_at,\n    MAX(score)::double precision AS score,\n    COUNT(*)::int AS matched_thread_count,\n    (array_agg(thread_id ORDER BY score DESC, approved_at DESC))[1] AS best_thread_id,\n    (array_agg(text ORDER BY score DESC, approved_at DESC))[1] AS best_text,\n    (array_agg(classification ORDER BY score DESC, approved_at DESC))[1] AS best_classification,\n    (array_agg(source_location ORDER BY score DESC, approved_at DESC))[1] AS best_source_location\nFROM matches\nWHERE score >= $1::double precision\nGROUP BY\n    document_id,\n    document_name,\n    mime_type,\n    scope,\n    sensitivity_tags,\n    tags,\n    tag_confidences,\n    uploaded_at\nORDER BY score DESC, uploaded_at DESC\nLIMIT $3 OFFSET $2";
+export interface SearchAccessibleDocumentsSemanticArgs {
+    minSimilarity: number;
+    pageOffset: string;
+    resultLimit: string;
+    organizationId: string;
+    queryEmbedding: string;
+    actingUserId: string;
+    tags: string[];
+    restrictedTags: string[];
+    scopes: string[];
+    uploadedAfter: Date | null;
+    uploadedBefore: Date | null;
+}
+export interface SearchAccessibleDocumentsSemanticRow {
+    documentId: string;
+    documentName: string;
+    mimeType: string;
+    scope: string;
+    sensitivityTags: string[];
+    tags: string;
+    tagConfidences: string;
+    uploadedAt: Date;
+    score: number;
+    matchedThreadCount: number;
+    bestThreadId: string | null;
+    bestText: string | null;
+    bestClassification: string | null;
+    bestSourceLocation: string | null;
+}
+export declare function searchAccessibleDocumentsSemantic(client: Client, args: SearchAccessibleDocumentsSemanticArgs): Promise<SearchAccessibleDocumentsSemanticRow[]>;
 export declare const searchApprovedThreadsSemanticByKnowledgeTypesQuery = "-- name: SearchApprovedThreadsSemanticByKnowledgeTypes :many\nSELECT\n    t.id AS thread_id,\n    t.text,\n    t.knowledge_type,\n    t.facet_fields,\n    similarity.score,\n    t.classification,\n    t.document_id,\n    d.filename AS document_name,\n    t.source_location,\n    t.approved_at,\n    t.temporal_validity,\n    COALESCE(entity_data.entities, '{}'::text[]) AS entities,\n    COALESCE(rs.reinforcement_count, 0) AS reinforcement_count,\n    COALESCE(cc.name, '') AS concept_cluster\nFROM weave.threads t\nJOIN weave.documents d ON d.id = t.document_id\nCROSS JOIN LATERAL (\n    SELECT CAST(\n        1.0::double precision\n        - (t.embedding <=> $1::vector)::double precision\n        AS double precision\n    ) AS score\n) similarity\nLEFT JOIN weave.thread_reinforcement_scores rs ON rs.thread_id = t.id\nLEFT JOIN weave.concept_clusters cc ON cc.id = t.cluster_id\nLEFT JOIN LATERAL (\n    SELECT array_agg(DISTINCT e.canonical ORDER BY e.canonical) AS entities\n    FROM weave.thread_entities te\n    JOIN weave.entities e ON e.id = te.entity_id\n    WHERE te.thread_id = t.id\n) entity_data ON TRUE\nWHERE t.organization_id = $2\n  AND t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_APPROVED'\n  AND t.embedding IS NOT NULL\n  AND t.knowledge_type = ANY($3::text[])\n  AND similarity.score >= $4::double precision\nORDER BY score DESC, t.approved_at DESC\nLIMIT $5";
 export interface SearchApprovedThreadsSemanticByKnowledgeTypesArgs {
     queryEmbedding: string;

package/dist/weavesql/weavedb/retrieval_sql.js

@@ -1,9 +1,10 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.listApprovedReconcilePoolByOrganizationQuery = exports.deleteSessionRetrievalContextQuery = exports.listSessionRetrievalContextQuery = exports.insertSessionRetrievalContextQuery = exports.searchConceptClusterThreadsBySimilarityAndUserPermissionsQuery = exports.searchConceptClusterThreadsBySimilarityQuery = exports.searchLeafConceptClustersBySimilarityQuery = exports.searchApprovedThreadsByEntitiesAndUserPermissionsQuery = exports.searchApprovedThreadsByEntitiesQuery = exports.searchApprovedThreadsSemanticByKnowledgeTypesAndUserPermissionsQuery = exports.searchApprovedThreadsSemanticByKnowledgeTypesQuery = exports.searchApprovedThreadsSemanticByUserPermissionsQuery = exports.searchApprovedThreadsSemanticQuery = exports.refreshThreadReinforcementScoresQuery = void 0;
+exports.listApprovedReconcilePoolByOrganizationQuery = exports.deleteSessionRetrievalContextQuery = exports.listSessionRetrievalContextQuery = exports.insertSessionRetrievalContextQuery = exports.searchConceptClusterThreadsBySimilarityAndUserPermissionsQuery = exports.searchConceptClusterThreadsBySimilarityQuery = exports.searchLeafConceptClustersBySimilarityQuery = exports.searchApprovedThreadsByEntitiesAndUserPermissionsQuery = exports.searchApprovedThreadsByEntitiesQuery = exports.searchApprovedThreadsSemanticByKnowledgeTypesAndUserPermissionsQuery = exports.searchApprovedThreadsSemanticByKnowledgeTypesQuery = exports.searchAccessibleDocumentsSemanticQuery = exports.searchApprovedThreadsSemanticByUserPermissionsQuery = exports.searchApprovedThreadsSemanticQuery = exports.refreshThreadReinforcementScoresQuery = void 0;
 exports.refreshThreadReinforcementScores = refreshThreadReinforcementScores;
 exports.searchApprovedThreadsSemantic = searchApprovedThreadsSemantic;
 exports.searchApprovedThreadsSemanticByUserPermissions = searchApprovedThreadsSemanticByUserPermissions;
+exports.searchAccessibleDocumentsSemantic = searchAccessibleDocumentsSemantic;
 exports.searchApprovedThreadsSemanticByKnowledgeTypes = searchApprovedThreadsSemanticByKnowledgeTypes;
 exports.searchApprovedThreadsSemanticByKnowledgeTypesAndUserPermissions = searchApprovedThreadsSemanticByKnowledgeTypesAndUserPermissions;
 exports.searchApprovedThreadsByEntities = searchApprovedThreadsByEntities;
@@ -166,6 +167,116 @@ async function searchApprovedThreadsSemanticByUserPermissions(client, args) {
         };
     });
 }
+exports.searchAccessibleDocumentsSemanticQuery = `-- name: SearchAccessibleDocumentsSemantic :many
+WITH tag_data AS (
+    SELECT
+        dt.document_id,
+        array_agg(dt.tag ORDER BY dt.tag ASC) AS tags,
+        array_agg(dt.confidence ORDER BY dt.tag ASC) AS tag_confidences
+    FROM weave.document_tags dt
+    JOIN weave.documents d ON d.id = dt.document_id
+    WHERE d.organization_id = $4
+    GROUP BY dt.document_id
+),
+matches AS (
+    SELECT
+        t.id AS thread_id,
+        t.text,
+        t.classification,
+        t.document_id,
+        d.filename AS document_name,
+        d.mime_type,
+        d.scope,
+        d.sensitivity_tags,
+        d.uploaded_at,
+        COALESCE(tag_data.tags, '{}'::text[]) AS tags,
+        COALESCE(tag_data.tag_confidences, '{}'::real[]) AS tag_confidences,
+        t.source_location,
+        t.approved_at,
+        CAST(
+            1.0::double precision
+            - (t.embedding <=> $5::vector)::double precision
+            AS double precision
+        ) AS score
+    FROM weave.threads t
+    JOIN weave.documents d ON d.id = t.document_id
+    LEFT JOIN tag_data ON tag_data.document_id = d.id
+    WHERE t.organization_id = $4
+      AND t.retrieval_status = 'THREAD_RETRIEVAL_STATUS_APPROVED'
+      AND t.embedding IS NOT NULL
+      AND (
+          d.scope = 'DOCUMENT_SCOPE_ORG'
+          OR (d.scope = 'DOCUMENT_SCOPE_PRIVATE'
+              AND d.uploaded_by_user_id = $6)
+          OR (d.scope = 'DOCUMENT_SCOPE_RESTRICTED'
+              AND EXISTS (
+                  SELECT 1
+                  FROM weave.user_sensitivity_clearances c
+                  WHERE c.organization_id = $4
+                    AND c.user_id = $6
+                    AND cardinality(d.sensitivity_tags) = 1
+                    AND c.sensitivity_tag_slug = d.sensitivity_tags[1]
+              ))
+      )
+      AND (cardinality($7::text[]) = 0 OR tag_data.tags && $7::text[] OR d.sensitivity_tags && $7::text[])
+      AND (cardinality($8::text[]) = 0 OR d.sensitivity_tags && $8::text[])
+      AND (cardinality($9::text[]) = 0 OR d.scope = ANY($9::text[]))
+      AND ($10::timestamptz IS NULL OR d.uploaded_at >= $10::timestamptz)
+      AND ($11::timestamptz IS NULL OR d.uploaded_at <= $11::timestamptz)
+)
+SELECT
+    document_id,
+    document_name,
+    mime_type,
+    scope,
+    sensitivity_tags,
+    tags,
+    tag_confidences,
+    uploaded_at,
+    MAX(score)::double precision AS score,
+    COUNT(*)::int AS matched_thread_count,
+    (array_agg(thread_id ORDER BY score DESC, approved_at DESC))[1] AS best_thread_id,
+    (array_agg(text ORDER BY score DESC, approved_at DESC))[1] AS best_text,
+    (array_agg(classification ORDER BY score DESC, approved_at DESC))[1] AS best_classification,
+    (array_agg(source_location ORDER BY score DESC, approved_at DESC))[1] AS best_source_location
+FROM matches
+WHERE score >= $1::double precision
+GROUP BY
+    document_id,
+    document_name,
+    mime_type,
+    scope,
+    sensitivity_tags,
+    tags,
+    tag_confidences,
+    uploaded_at
+ORDER BY score DESC, uploaded_at DESC
+LIMIT $3 OFFSET $2`;
+async function searchAccessibleDocumentsSemantic(client, args) {
+    const result = await client.query({
+        text: exports.searchAccessibleDocumentsSemanticQuery,
+        values: [args.minSimilarity, args.pageOffset, args.resultLimit, args.organizationId, args.queryEmbedding, args.actingUserId, args.tags, args.restrictedTags, args.scopes, args.uploadedAfter, args.uploadedBefore],
+        rowMode: "array"
+    });
+    return result.rows.map(row => {
+        return {
+            documentId: row[0],
+            documentName: row[1],
+            mimeType: row[2],
+            scope: row[3],
+            sensitivityTags: row[4],
+            tags: row[5],
+            tagConfidences: row[6],
+            uploadedAt: row[7],
+            score: row[8],
+            matchedThreadCount: row[9],
+            bestThreadId: row[10],
+            bestText: row[11],
+            bestClassification: row[12],
+            bestSourceLocation: row[13]
+        };
+    });
+}
 exports.searchApprovedThreadsSemanticByKnowledgeTypesQuery = `-- name: SearchApprovedThreadsSemanticByKnowledgeTypes :many
 SELECT
     t.id AS thread_id,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "weave-typescript",
-  "version": "0.28.0",
+  "version": "0.29.0",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
   "files": [