node-type-registry 0.24.0 → 0.26.0

package/authz/authz-file-path.d.ts

@@ -0,0 +1,2 @@
+import type { NodeTypeDefinition } from '../types';
+export declare const AuthzFilePath: NodeTypeDefinition;

package/authz/authz-file-path.js

@@ -0,0 +1,58 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.AuthzFilePath = void 0;
+exports.AuthzFilePath = {
+    name: 'AuthzFilePath',
+    slug: 'authz_file_path',
+    category: 'authz',
+    display_name: 'File Path Share',
+    description: 'Path-scoped file sharing via ltree containment. Grants access when a path_shares row matches the current user, bucket, and an ancestor path with the required permission.',
+    parameter_schema: {
+        type: 'object',
+        properties: {
+            shares_schema: {
+                type: 'string',
+                description: 'Schema of the path_shares table'
+            },
+            shares_table: {
+                type: 'string',
+                description: 'Name of the path_shares table'
+            },
+            files_schema: {
+                type: 'string',
+                description: 'Schema of the files table (used to qualify column references inside the EXISTS subquery)'
+            },
+            files_table: {
+                type: 'string',
+                description: 'Name of the files table (used to qualify column references inside the EXISTS subquery)'
+            },
+            permission_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Boolean column on the path_shares table that grants the required permission (e.g. can_read, can_write)'
+            },
+            bucket_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Column on the files table referencing the bucket',
+                default: 'bucket_id'
+            },
+            path_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Ltree column on the files table representing the file path',
+                default: 'path'
+            }
+        },
+        required: [
+            'shares_schema',
+            'shares_table',
+            'files_table',
+            'permission_field'
+        ]
+    },
+    tags: [
+        'storage',
+        'authz'
+    ]
+};

package/authz/index.d.ts

@@ -2,6 +2,7 @@ export { AuthzAllowAll } from './authz-allow-all';
 export { AuthzAppMembership } from './authz-app-membership';
 export { AuthzComposite } from './authz-composite';
 export { AuthzDenyAll } from './authz-deny-all';
+export { AuthzFilePath } from './authz-file-path';
 export { AuthzDirectOwner } from './authz-direct-owner';
 export { AuthzDirectOwnerAny } from './authz-direct-owner-any';
 export { AuthzEntityMembership } from './authz-entity-membership';

package/authz/index.js

@@ -1,6 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.AuthzTemporal = exports.AuthzRelatedPeerOwnership = exports.AuthzRelatedMemberList = exports.AuthzRelatedEntityMembership = exports.AuthzPublishable = exports.AuthzPeerOwnership = exports.AuthzOrgHierarchy = exports.AuthzNotReadOnly = exports.AuthzMemberList = exports.AuthzEntityMembership = exports.AuthzDirectOwnerAny = exports.AuthzDirectOwner = exports.AuthzDenyAll = exports.AuthzComposite = exports.AuthzAppMembership = exports.AuthzAllowAll = void 0;
+exports.AuthzTemporal = exports.AuthzRelatedPeerOwnership = exports.AuthzRelatedMemberList = exports.AuthzRelatedEntityMembership = exports.AuthzPublishable = exports.AuthzPeerOwnership = exports.AuthzOrgHierarchy = exports.AuthzNotReadOnly = exports.AuthzMemberList = exports.AuthzEntityMembership = exports.AuthzDirectOwnerAny = exports.AuthzDirectOwner = exports.AuthzFilePath = exports.AuthzDenyAll = exports.AuthzComposite = exports.AuthzAppMembership = exports.AuthzAllowAll = void 0;
 var authz_allow_all_1 = require("./authz-allow-all");
 Object.defineProperty(exports, "AuthzAllowAll", { enumerable: true, get: function () { return authz_allow_all_1.AuthzAllowAll; } });
 var authz_app_membership_1 = require("./authz-app-membership");
@@ -9,6 +9,8 @@ var authz_composite_1 = require("./authz-composite");
 Object.defineProperty(exports, "AuthzComposite", { enumerable: true, get: function () { return authz_composite_1.AuthzComposite; } });
 var authz_deny_all_1 = require("./authz-deny-all");
 Object.defineProperty(exports, "AuthzDenyAll", { enumerable: true, get: function () { return authz_deny_all_1.AuthzDenyAll; } });
+var authz_file_path_1 = require("./authz-file-path");
+Object.defineProperty(exports, "AuthzFilePath", { enumerable: true, get: function () { return authz_file_path_1.AuthzFilePath; } });
 var authz_direct_owner_1 = require("./authz-direct-owner");
 Object.defineProperty(exports, "AuthzDirectOwner", { enumerable: true, get: function () { return authz_direct_owner_1.AuthzDirectOwner; } });
 var authz_direct_owner_any_1 = require("./authz-direct-owner-any");

package/data/data-file-embedding.d.ts

@@ -0,0 +1,2 @@
+import type { NodeTypeDefinition } from '../types';
+export declare const DataFileEmbedding: NodeTypeDefinition;

package/data/data-file-embedding.js

@@ -0,0 +1,186 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.DataFileEmbedding = void 0;
+exports.DataFileEmbedding = {
+    name: 'DataFileEmbedding',
+    slug: 'data_file_embedding',
+    category: 'data',
+    display_name: 'File Embedding',
+    description: 'Generic, MIME-scoped embedding node for file tables. Supports two modes: ' +
+        'direct (whole-file to single vector, e.g. CLIP for images) when extraction ' +
+        'is omitted, or extract (file to text to chunks to per-chunk vectors) when ' +
+        'extraction config is provided. Composes SearchVector + DataJobTrigger ' +
+        'internally. Multiple instances can coexist on the same table with different ' +
+        'MIME scopes, field names, and embedding strategies.',
+    parameter_schema: {
+        type: 'object',
+        properties: {
+            // ── Vector config (passed through to SearchVector) ─────────────
+            field_name: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Name of the vector embedding column',
+                default: 'embedding'
+            },
+            dimensions: {
+                type: 'integer',
+                description: 'Vector dimensions (e.g. 512 for CLIP, 768 for nomic, 1536 for ada-002)',
+                default: 768
+            },
+            index_method: {
+                type: 'string',
+                enum: ['hnsw', 'ivfflat'],
+                description: 'Index type for similarity search',
+                default: 'hnsw'
+            },
+            metric: {
+                type: 'string',
+                enum: ['cosine', 'l2', 'ip'],
+                description: 'Distance metric',
+                default: 'cosine'
+            },
+            index_options: {
+                type: 'object',
+                description: 'Index-specific options. HNSW: {m, ef_construction}. IVFFlat: {lists}.',
+                default: {}
+            },
+            // ── MIME scoping ───────────────────────────────────────────────
+            mime_patterns: {
+                type: 'array',
+                items: { type: 'string' },
+                description: 'MIME type LIKE patterns to match. Multiple patterns are OR\'d together. ' +
+                    'Examples: [\'image/%\'], [\'application/pdf\', \'text/%\'], [\'audio/%\'].',
+                default: ['image/%']
+            },
+            // ── Job routing ────────────────────────────────────────────────
+            task_identifier: {
+                type: 'string',
+                description: 'Job task identifier for the worker. In direct mode this is the ' +
+                    'embedding worker; in extract mode this is the extraction worker.',
+                default: 'process_file_embedding'
+            },
+            events: {
+                type: 'array',
+                items: { type: 'string', enum: ['INSERT', 'UPDATE'] },
+                description: 'Trigger events that fire the job',
+                default: ['INSERT']
+            },
+            payload_custom: {
+                type: 'object',
+                additionalProperties: { type: 'string', format: 'column-ref' },
+                description: 'Custom payload key-to-column mapping for the job trigger',
+                default: {
+                    file_id: 'id',
+                    key: 'key',
+                    mime_type: 'mime_type',
+                    bucket_id: 'bucket_id'
+                }
+            },
+            trigger_conditions: {
+                description: 'Additional compound conditions beyond MIME filtering. ' +
+                    'Merged with the auto-generated MIME conditions via AND. ' +
+                    'Use this to add status checks, field guards, etc.',
+                'x-codegen-type': 'TriggerCondition | TriggerCondition[]',
+                oneOf: [
+                    { $ref: '#/$defs/triggerCondition' },
+                    { type: 'array', items: { $ref: '#/$defs/triggerCondition' } }
+                ]
+            },
+            // ── Extraction config (optional — enables extract mode) ────────
+            extraction: {
+                type: 'object',
+                description: 'Text extraction configuration. When present, the generator creates ' +
+                    'extraction output fields on the table and configures SearchVector with ' +
+                    'source_fields + stale tracking. When absent, the node operates in direct ' +
+                    'mode (single vector per file, no text extraction).',
+                properties: {
+                    text_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Field to store extracted text/markdown',
+                        default: 'extracted_text'
+                    },
+                    metadata_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'JSONB field for extraction metadata (page count, language, etc.)',
+                        default: 'extracted_metadata'
+                    },
+                    status_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Extraction lifecycle status field',
+                        default: 'extraction_status'
+                    }
+                }
+            },
+            // ── Chunking config (optional — creates embedding_chunks) ──────
+            chunks: {
+                type: 'object',
+                description: 'Chunking configuration. Creates an embedding_chunks record that drives ' +
+                    'automatic text splitting and per-chunk embedding. Only meaningful when ' +
+                    'extraction is also provided.',
+                properties: {
+                    content_field_name: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Name of the text content column in the chunks table',
+                        default: 'content'
+                    },
+                    chunk_size: {
+                        type: 'integer',
+                        description: 'Maximum number of characters per chunk',
+                        default: 1000
+                    },
+                    chunk_overlap: {
+                        type: 'integer',
+                        description: 'Number of overlapping characters between consecutive chunks',
+                        default: 200
+                    },
+                    chunk_strategy: {
+                        type: 'string',
+                        enum: ['fixed', 'sentence', 'paragraph', 'semantic'],
+                        description: 'Strategy for splitting text into chunks',
+                        default: 'paragraph'
+                    },
+                    metadata_fields: {
+                        type: 'object',
+                        description: 'Metadata fields from parent to copy into chunks'
+                    },
+                    enqueue_chunking_job: {
+                        type: 'boolean',
+                        description: 'Whether to auto-enqueue a chunking job on insert/update',
+                        default: true
+                    },
+                    chunking_task_name: {
+                        type: 'string',
+                        description: 'Task identifier for the chunking job queue',
+                        default: 'generate_chunks'
+                    }
+                }
+            },
+            // ── Stale tracking (meaningful in extract mode) ────────────────
+            stale_strategy: {
+                type: 'string',
+                enum: ['column', 'null', 'hash'],
+                description: 'Strategy for tracking embedding staleness when extraction is enabled. ' +
+                    'column: embedding_stale boolean. null: set embedding to NULL. hash: md5 hash.',
+                default: 'column'
+            },
+            include_stale_field: {
+                type: 'boolean',
+                description: 'Whether to include the embedding_stale boolean field (extract mode)',
+                default: true
+            }
+        }
+    },
+    tags: [
+        'embedding',
+        'vector',
+        'ai',
+        'composition',
+        'jobs',
+        'multimodal',
+        'files'
+    ]
+};

package/data/data-image-embedding.d.ts

@@ -1,2 +1,14 @@
 import type { NodeTypeDefinition } from '../types';
+/**
+ * Image-specific preset of DataFileEmbedding.
+ *
+ * At the SQL layer, data_image_embedding delegates entirely to
+ * data_file_embedding, merging image-specific defaults before forwarding.
+ * The parameter schema here is intentionally identical to DataFileEmbedding;
+ * only the defaults differ (dimensions: 512, task: process_image_embedding,
+ * mime_patterns: ['image/%']).
+ *
+ * Kept as a separate node type for backward compatibility — existing
+ * blueprints that reference DataImageEmbedding continue to work unchanged.
+ */
 export declare const DataImageEmbedding: NodeTypeDefinition;

package/data/data-image-embedding.js

@@ -1,15 +1,31 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.DataImageEmbedding = void 0;
+/**
+ * Image-specific preset of DataFileEmbedding.
+ *
+ * At the SQL layer, data_image_embedding delegates entirely to
+ * data_file_embedding, merging image-specific defaults before forwarding.
+ * The parameter schema here is intentionally identical to DataFileEmbedding;
+ * only the defaults differ (dimensions: 512, task: process_image_embedding,
+ * mime_patterns: ['image/%']).
+ *
+ * Kept as a separate node type for backward compatibility — existing
+ * blueprints that reference DataImageEmbedding continue to work unchanged.
+ */
 exports.DataImageEmbedding = {
     name: 'DataImageEmbedding',
     slug: 'data_image_embedding',
     category: 'data',
     display_name: 'Image Embedding',
-    description: 'Composition wrapper that creates a vector embedding field with HNSW/IVFFlat index (via SearchVector) and a job trigger with compound conditions (via DataJobTrigger) that fires on INSERT for image files matching mime_type patterns. Designed for storage file tables.',
+    description: 'Image-specific preset of DataFileEmbedding. Delegates to DataFileEmbedding ' +
+        'with image-oriented defaults: dimensions=512 (CLIP), mime_patterns=[\'image/%\'], ' +
+        'task_identifier=\'process_image_embedding\', direct mode (no extraction). ' +
+        'Accepts all DataFileEmbedding parameters — any overrides are forwarded through.',
     parameter_schema: {
         type: 'object',
         properties: {
+            // ── Vector config (passed through to DataFileEmbedding) ──────────
             field_name: {
                 type: 'string',
                 format: 'column-ref',
@@ -18,47 +34,48 @@ exports.DataImageEmbedding = {
             },
             dimensions: {
                 type: 'integer',
-                description: 'Vector dimensions',
+                description: 'Vector dimensions (default 512 for CLIP-style image embeddings)',
                 default: 512
             },
             index_method: {
                 type: 'string',
-                enum: [
-                    'hnsw',
-                    'ivfflat'
-                ],
+                enum: ['hnsw', 'ivfflat'],
                 description: 'Index type for similarity search',
                 default: 'hnsw'
             },
             metric: {
                 type: 'string',
-                enum: [
-                    'cosine',
-                    'l2',
-                    'ip'
-                ],
+                enum: ['cosine', 'l2', 'ip'],
                 description: 'Distance metric',
                 default: 'cosine'
             },
+            index_options: {
+                type: 'object',
+                description: 'Index-specific options. HNSW: {m, ef_construction}. IVFFlat: {lists}.',
+                default: {}
+            },
+            // ── MIME scoping ───────────────────────────────────────────────
+            mime_patterns: {
+                type: 'array',
+                items: { type: 'string' },
+                description: 'MIME type LIKE patterns to match. Multiple patterns are OR\'d together.',
+                default: ['image/%']
+            },
+            // ── Job routing ────────────────────────────────────────────────
             task_identifier: {
                 type: 'string',
-                description: 'Job task identifier for the embedding worker',
+                description: 'Job task identifier for the image embedding worker',
                 default: 'process_image_embedding'
             },
-            mime_patterns: {
+            events: {
                 type: 'array',
-                items: {
-                    type: 'string'
-                },
-                description: 'MIME type LIKE patterns to match (e.g., image/%, video/%). Multiple patterns are OR\'d together.',
-                default: ['image/%']
+                items: { type: 'string', enum: ['INSERT', 'UPDATE'] },
+                description: 'Trigger events that fire the job',
+                default: ['INSERT']
             },
             payload_custom: {
                 type: 'object',
-                additionalProperties: {
-                    type: 'string',
-                    format: 'column-ref'
-                },
+                additionalProperties: { type: 'string', format: 'column-ref' },
                 description: 'Custom payload key-to-column mapping for the job trigger',
                 default: {
                     file_id: 'id',
@@ -66,6 +83,76 @@ exports.DataImageEmbedding = {
                     mime_type: 'mime_type',
                     bucket_id: 'bucket_id'
                 }
+            },
+            trigger_conditions: {
+                description: 'Additional compound conditions beyond MIME filtering. ' +
+                    'Merged with the auto-generated MIME conditions via AND.',
+                'x-codegen-type': 'TriggerCondition | TriggerCondition[]',
+                oneOf: [
+                    { $ref: '#/$defs/triggerCondition' },
+                    { type: 'array', items: { $ref: '#/$defs/triggerCondition' } }
+                ]
+            },
+            // ── Extraction config (optional — enables extract mode) ────────
+            extraction: {
+                type: 'object',
+                description: 'Text extraction configuration. Forwarded to DataFileEmbedding. ' +
+                    'When present, enables extract mode (e.g., OCR for images).',
+                properties: {
+                    text_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Field to store extracted text',
+                        default: 'extracted_text'
+                    },
+                    metadata_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'JSONB field for extraction metadata',
+                        default: 'extracted_metadata'
+                    },
+                    status_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Extraction lifecycle status field',
+                        default: 'extraction_status'
+                    }
+                }
+            },
+            // ── Chunking config (optional — forwarded to DataFileEmbedding) ─
+            chunks: {
+                type: 'object',
+                description: 'Chunking configuration. Forwarded to DataFileEmbedding. ' +
+                    'Only meaningful when extraction is also provided.',
+                properties: {
+                    content_field_name: {
+                        type: 'string',
+                        format: 'column-ref',
+                        default: 'content'
+                    },
+                    chunk_size: { type: 'integer', default: 1000 },
+                    chunk_overlap: { type: 'integer', default: 200 },
+                    chunk_strategy: {
+                        type: 'string',
+                        enum: ['fixed', 'sentence', 'paragraph', 'semantic'],
+                        default: 'paragraph'
+                    },
+                    metadata_fields: { type: 'object' },
+                    enqueue_chunking_job: { type: 'boolean', default: true },
+                    chunking_task_name: { type: 'string', default: 'generate_chunks' }
+                }
+            },
+            // ── Stale tracking (forwarded to DataFileEmbedding) ────────────
+            stale_strategy: {
+                type: 'string',
+                enum: ['column', 'null', 'hash'],
+                description: 'Strategy for tracking embedding staleness in extract mode',
+                default: 'column'
+            },
+            include_stale_field: {
+                type: 'boolean',
+                description: 'Whether to include the embedding_stale boolean field',
+                default: true
             }
         }
     },

package/data/index.d.ts

@@ -1,6 +1,7 @@
 export { DataCompositeField } from './data-composite-field';
 export { DataDirectOwner } from './data-direct-owner';
 export { DataEntityMembership } from './data-entity-membership';
+export { DataFileEmbedding } from './data-file-embedding';
 export { DataFeatureFlag } from './data-feature-flag';
 export { DataForceCurrentUser } from './data-force-current-user';
 export { DataId } from './data-id';

package/data/index.js

@@ -1,12 +1,14 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.TableUserSettings = exports.TableUserProfiles = exports.TableOrganizationSettings = exports.SearchVector = exports.SearchUnified = exports.SearchTrgm = exports.SearchSpatialAggregate = exports.SearchSpatial = exports.SearchFullText = exports.SearchBm25 = exports.DataTimestamps = exports.DataTags = exports.DataStatusField = exports.DataSoftDelete = exports.DataSlug = exports.DataPublishable = exports.DataPeoplestamps = exports.DataOwnershipInEntity = exports.DataOwnedFields = exports.DataJsonb = exports.DataLimitCounter = exports.DataJobTrigger = exports.DataInheritFromParent = exports.DataInflection = exports.DataImmutableFields = exports.DataImageEmbedding = exports.DataId = exports.DataForceCurrentUser = exports.DataFeatureFlag = exports.DataEntityMembership = exports.DataDirectOwner = exports.DataCompositeField = void 0;
+exports.TableUserSettings = exports.TableUserProfiles = exports.TableOrganizationSettings = exports.SearchVector = exports.SearchUnified = exports.SearchTrgm = exports.SearchSpatialAggregate = exports.SearchSpatial = exports.SearchFullText = exports.SearchBm25 = exports.DataTimestamps = exports.DataTags = exports.DataStatusField = exports.DataSoftDelete = exports.DataSlug = exports.DataPublishable = exports.DataPeoplestamps = exports.DataOwnershipInEntity = exports.DataOwnedFields = exports.DataJsonb = exports.DataLimitCounter = exports.DataJobTrigger = exports.DataInheritFromParent = exports.DataInflection = exports.DataImmutableFields = exports.DataImageEmbedding = exports.DataId = exports.DataForceCurrentUser = exports.DataFeatureFlag = exports.DataFileEmbedding = exports.DataEntityMembership = exports.DataDirectOwner = exports.DataCompositeField = void 0;
 var data_composite_field_1 = require("./data-composite-field");
 Object.defineProperty(exports, "DataCompositeField", { enumerable: true, get: function () { return data_composite_field_1.DataCompositeField; } });
 var data_direct_owner_1 = require("./data-direct-owner");
 Object.defineProperty(exports, "DataDirectOwner", { enumerable: true, get: function () { return data_direct_owner_1.DataDirectOwner; } });
 var data_entity_membership_1 = require("./data-entity-membership");
 Object.defineProperty(exports, "DataEntityMembership", { enumerable: true, get: function () { return data_entity_membership_1.DataEntityMembership; } });
+var data_file_embedding_1 = require("./data-file-embedding");
+Object.defineProperty(exports, "DataFileEmbedding", { enumerable: true, get: function () { return data_file_embedding_1.DataFileEmbedding; } });
 var data_feature_flag_1 = require("./data-feature-flag");
 Object.defineProperty(exports, "DataFeatureFlag", { enumerable: true, get: function () { return data_feature_flag_1.DataFeatureFlag; } });
 var data_force_current_user_1 = require("./data-force-current-user");

package/esm/authz/authz-file-path.d.ts

@@ -0,0 +1,2 @@
+import type { NodeTypeDefinition } from '../types';
+export declare const AuthzFilePath: NodeTypeDefinition;

package/esm/authz/authz-file-path.js

@@ -0,0 +1,55 @@
+export const AuthzFilePath = {
+    name: 'AuthzFilePath',
+    slug: 'authz_file_path',
+    category: 'authz',
+    display_name: 'File Path Share',
+    description: 'Path-scoped file sharing via ltree containment. Grants access when a path_shares row matches the current user, bucket, and an ancestor path with the required permission.',
+    parameter_schema: {
+        type: 'object',
+        properties: {
+            shares_schema: {
+                type: 'string',
+                description: 'Schema of the path_shares table'
+            },
+            shares_table: {
+                type: 'string',
+                description: 'Name of the path_shares table'
+            },
+            files_schema: {
+                type: 'string',
+                description: 'Schema of the files table (used to qualify column references inside the EXISTS subquery)'
+            },
+            files_table: {
+                type: 'string',
+                description: 'Name of the files table (used to qualify column references inside the EXISTS subquery)'
+            },
+            permission_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Boolean column on the path_shares table that grants the required permission (e.g. can_read, can_write)'
+            },
+            bucket_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Column on the files table referencing the bucket',
+                default: 'bucket_id'
+            },
+            path_field: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Ltree column on the files table representing the file path',
+                default: 'path'
+            }
+        },
+        required: [
+            'shares_schema',
+            'shares_table',
+            'files_table',
+            'permission_field'
+        ]
+    },
+    tags: [
+        'storage',
+        'authz'
+    ]
+};

package/esm/authz/index.d.ts

@@ -2,6 +2,7 @@ export { AuthzAllowAll } from './authz-allow-all';
 export { AuthzAppMembership } from './authz-app-membership';
 export { AuthzComposite } from './authz-composite';
 export { AuthzDenyAll } from './authz-deny-all';
+export { AuthzFilePath } from './authz-file-path';
 export { AuthzDirectOwner } from './authz-direct-owner';
 export { AuthzDirectOwnerAny } from './authz-direct-owner-any';
 export { AuthzEntityMembership } from './authz-entity-membership';

package/esm/authz/index.js

@@ -2,6 +2,7 @@ export { AuthzAllowAll } from './authz-allow-all';
 export { AuthzAppMembership } from './authz-app-membership';
 export { AuthzComposite } from './authz-composite';
 export { AuthzDenyAll } from './authz-deny-all';
+export { AuthzFilePath } from './authz-file-path';
 export { AuthzDirectOwner } from './authz-direct-owner';
 export { AuthzDirectOwnerAny } from './authz-direct-owner-any';
 export { AuthzEntityMembership } from './authz-entity-membership';

package/esm/data/data-file-embedding.d.ts

@@ -0,0 +1,2 @@
+import type { NodeTypeDefinition } from '../types';
+export declare const DataFileEmbedding: NodeTypeDefinition;

package/esm/data/data-file-embedding.js

@@ -0,0 +1,183 @@
+export const DataFileEmbedding = {
+    name: 'DataFileEmbedding',
+    slug: 'data_file_embedding',
+    category: 'data',
+    display_name: 'File Embedding',
+    description: 'Generic, MIME-scoped embedding node for file tables. Supports two modes: ' +
+        'direct (whole-file to single vector, e.g. CLIP for images) when extraction ' +
+        'is omitted, or extract (file to text to chunks to per-chunk vectors) when ' +
+        'extraction config is provided. Composes SearchVector + DataJobTrigger ' +
+        'internally. Multiple instances can coexist on the same table with different ' +
+        'MIME scopes, field names, and embedding strategies.',
+    parameter_schema: {
+        type: 'object',
+        properties: {
+            // ── Vector config (passed through to SearchVector) ─────────────
+            field_name: {
+                type: 'string',
+                format: 'column-ref',
+                description: 'Name of the vector embedding column',
+                default: 'embedding'
+            },
+            dimensions: {
+                type: 'integer',
+                description: 'Vector dimensions (e.g. 512 for CLIP, 768 for nomic, 1536 for ada-002)',
+                default: 768
+            },
+            index_method: {
+                type: 'string',
+                enum: ['hnsw', 'ivfflat'],
+                description: 'Index type for similarity search',
+                default: 'hnsw'
+            },
+            metric: {
+                type: 'string',
+                enum: ['cosine', 'l2', 'ip'],
+                description: 'Distance metric',
+                default: 'cosine'
+            },
+            index_options: {
+                type: 'object',
+                description: 'Index-specific options. HNSW: {m, ef_construction}. IVFFlat: {lists}.',
+                default: {}
+            },
+            // ── MIME scoping ───────────────────────────────────────────────
+            mime_patterns: {
+                type: 'array',
+                items: { type: 'string' },
+                description: 'MIME type LIKE patterns to match. Multiple patterns are OR\'d together. ' +
+                    'Examples: [\'image/%\'], [\'application/pdf\', \'text/%\'], [\'audio/%\'].',
+                default: ['image/%']
+            },
+            // ── Job routing ────────────────────────────────────────────────
+            task_identifier: {
+                type: 'string',
+                description: 'Job task identifier for the worker. In direct mode this is the ' +
+                    'embedding worker; in extract mode this is the extraction worker.',
+                default: 'process_file_embedding'
+            },
+            events: {
+                type: 'array',
+                items: { type: 'string', enum: ['INSERT', 'UPDATE'] },
+                description: 'Trigger events that fire the job',
+                default: ['INSERT']
+            },
+            payload_custom: {
+                type: 'object',
+                additionalProperties: { type: 'string', format: 'column-ref' },
+                description: 'Custom payload key-to-column mapping for the job trigger',
+                default: {
+                    file_id: 'id',
+                    key: 'key',
+                    mime_type: 'mime_type',
+                    bucket_id: 'bucket_id'
+                }
+            },
+            trigger_conditions: {
+                description: 'Additional compound conditions beyond MIME filtering. ' +
+                    'Merged with the auto-generated MIME conditions via AND. ' +
+                    'Use this to add status checks, field guards, etc.',
+                'x-codegen-type': 'TriggerCondition | TriggerCondition[]',
+                oneOf: [
+                    { $ref: '#/$defs/triggerCondition' },
+                    { type: 'array', items: { $ref: '#/$defs/triggerCondition' } }
+                ]
+            },
+            // ── Extraction config (optional — enables extract mode) ────────
+            extraction: {
+                type: 'object',
+                description: 'Text extraction configuration. When present, the generator creates ' +
+                    'extraction output fields on the table and configures SearchVector with ' +
+                    'source_fields + stale tracking. When absent, the node operates in direct ' +
+                    'mode (single vector per file, no text extraction).',
+                properties: {
+                    text_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Field to store extracted text/markdown',
+                        default: 'extracted_text'
+                    },
+                    metadata_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'JSONB field for extraction metadata (page count, language, etc.)',
+                        default: 'extracted_metadata'
+                    },
+                    status_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Extraction lifecycle status field',
+                        default: 'extraction_status'
+                    }
+                }
+            },
+            // ── Chunking config (optional — creates embedding_chunks) ──────
+            chunks: {
+                type: 'object',
+                description: 'Chunking configuration. Creates an embedding_chunks record that drives ' +
+                    'automatic text splitting and per-chunk embedding. Only meaningful when ' +
+                    'extraction is also provided.',
+                properties: {
+                    content_field_name: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Name of the text content column in the chunks table',
+                        default: 'content'
+                    },
+                    chunk_size: {
+                        type: 'integer',
+                        description: 'Maximum number of characters per chunk',
+                        default: 1000
+                    },
+                    chunk_overlap: {
+                        type: 'integer',
+                        description: 'Number of overlapping characters between consecutive chunks',
+                        default: 200
+                    },
+                    chunk_strategy: {
+                        type: 'string',
+                        enum: ['fixed', 'sentence', 'paragraph', 'semantic'],
+                        description: 'Strategy for splitting text into chunks',
+                        default: 'paragraph'
+                    },
+                    metadata_fields: {
+                        type: 'object',
+                        description: 'Metadata fields from parent to copy into chunks'
+                    },
+                    enqueue_chunking_job: {
+                        type: 'boolean',
+                        description: 'Whether to auto-enqueue a chunking job on insert/update',
+                        default: true
+                    },
+                    chunking_task_name: {
+                        type: 'string',
+                        description: 'Task identifier for the chunking job queue',
+                        default: 'generate_chunks'
+                    }
+                }
+            },
+            // ── Stale tracking (meaningful in extract mode) ────────────────
+            stale_strategy: {
+                type: 'string',
+                enum: ['column', 'null', 'hash'],
+                description: 'Strategy for tracking embedding staleness when extraction is enabled. ' +
+                    'column: embedding_stale boolean. null: set embedding to NULL. hash: md5 hash.',
+                default: 'column'
+            },
+            include_stale_field: {
+                type: 'boolean',
+                description: 'Whether to include the embedding_stale boolean field (extract mode)',
+                default: true
+            }
+        }
+    },
+    tags: [
+        'embedding',
+        'vector',
+        'ai',
+        'composition',
+        'jobs',
+        'multimodal',
+        'files'
+    ]
+};

package/esm/data/data-image-embedding.d.ts

@@ -1,2 +1,14 @@
 import type { NodeTypeDefinition } from '../types';
+/**
+ * Image-specific preset of DataFileEmbedding.
+ *
+ * At the SQL layer, data_image_embedding delegates entirely to
+ * data_file_embedding, merging image-specific defaults before forwarding.
+ * The parameter schema here is intentionally identical to DataFileEmbedding;
+ * only the defaults differ (dimensions: 512, task: process_image_embedding,
+ * mime_patterns: ['image/%']).
+ *
+ * Kept as a separate node type for backward compatibility — existing
+ * blueprints that reference DataImageEmbedding continue to work unchanged.
+ */
 export declare const DataImageEmbedding: NodeTypeDefinition;

package/esm/data/data-image-embedding.js

@@ -1,12 +1,28 @@
+/**
+ * Image-specific preset of DataFileEmbedding.
+ *
+ * At the SQL layer, data_image_embedding delegates entirely to
+ * data_file_embedding, merging image-specific defaults before forwarding.
+ * The parameter schema here is intentionally identical to DataFileEmbedding;
+ * only the defaults differ (dimensions: 512, task: process_image_embedding,
+ * mime_patterns: ['image/%']).
+ *
+ * Kept as a separate node type for backward compatibility — existing
+ * blueprints that reference DataImageEmbedding continue to work unchanged.
+ */
 export const DataImageEmbedding = {
     name: 'DataImageEmbedding',
     slug: 'data_image_embedding',
     category: 'data',
     display_name: 'Image Embedding',
-    description: 'Composition wrapper that creates a vector embedding field with HNSW/IVFFlat index (via SearchVector) and a job trigger with compound conditions (via DataJobTrigger) that fires on INSERT for image files matching mime_type patterns. Designed for storage file tables.',
+    description: 'Image-specific preset of DataFileEmbedding. Delegates to DataFileEmbedding ' +
+        'with image-oriented defaults: dimensions=512 (CLIP), mime_patterns=[\'image/%\'], ' +
+        'task_identifier=\'process_image_embedding\', direct mode (no extraction). ' +
+        'Accepts all DataFileEmbedding parameters — any overrides are forwarded through.',
     parameter_schema: {
         type: 'object',
         properties: {
+            // ── Vector config (passed through to DataFileEmbedding) ──────────
             field_name: {
                 type: 'string',
                 format: 'column-ref',
@@ -15,47 +31,48 @@ export const DataImageEmbedding = {
             },
             dimensions: {
                 type: 'integer',
-                description: 'Vector dimensions',
+                description: 'Vector dimensions (default 512 for CLIP-style image embeddings)',
                 default: 512
             },
             index_method: {
                 type: 'string',
-                enum: [
-                    'hnsw',
-                    'ivfflat'
-                ],
+                enum: ['hnsw', 'ivfflat'],
                 description: 'Index type for similarity search',
                 default: 'hnsw'
             },
             metric: {
                 type: 'string',
-                enum: [
-                    'cosine',
-                    'l2',
-                    'ip'
-                ],
+                enum: ['cosine', 'l2', 'ip'],
                 description: 'Distance metric',
                 default: 'cosine'
             },
+            index_options: {
+                type: 'object',
+                description: 'Index-specific options. HNSW: {m, ef_construction}. IVFFlat: {lists}.',
+                default: {}
+            },
+            // ── MIME scoping ───────────────────────────────────────────────
+            mime_patterns: {
+                type: 'array',
+                items: { type: 'string' },
+                description: 'MIME type LIKE patterns to match. Multiple patterns are OR\'d together.',
+                default: ['image/%']
+            },
+            // ── Job routing ────────────────────────────────────────────────
             task_identifier: {
                 type: 'string',
-                description: 'Job task identifier for the embedding worker',
+                description: 'Job task identifier for the image embedding worker',
                 default: 'process_image_embedding'
             },
-            mime_patterns: {
+            events: {
                 type: 'array',
-                items: {
-                    type: 'string'
-                },
-                description: 'MIME type LIKE patterns to match (e.g., image/%, video/%). Multiple patterns are OR\'d together.',
-                default: ['image/%']
+                items: { type: 'string', enum: ['INSERT', 'UPDATE'] },
+                description: 'Trigger events that fire the job',
+                default: ['INSERT']
             },
             payload_custom: {
                 type: 'object',
-                additionalProperties: {
-                    type: 'string',
-                    format: 'column-ref'
-                },
+                additionalProperties: { type: 'string', format: 'column-ref' },
                 description: 'Custom payload key-to-column mapping for the job trigger',
                 default: {
                     file_id: 'id',
@@ -63,6 +80,76 @@ export const DataImageEmbedding = {
                     mime_type: 'mime_type',
                     bucket_id: 'bucket_id'
                 }
+            },
+            trigger_conditions: {
+                description: 'Additional compound conditions beyond MIME filtering. ' +
+                    'Merged with the auto-generated MIME conditions via AND.',
+                'x-codegen-type': 'TriggerCondition | TriggerCondition[]',
+                oneOf: [
+                    { $ref: '#/$defs/triggerCondition' },
+                    { type: 'array', items: { $ref: '#/$defs/triggerCondition' } }
+                ]
+            },
+            // ── Extraction config (optional — enables extract mode) ────────
+            extraction: {
+                type: 'object',
+                description: 'Text extraction configuration. Forwarded to DataFileEmbedding. ' +
+                    'When present, enables extract mode (e.g., OCR for images).',
+                properties: {
+                    text_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Field to store extracted text',
+                        default: 'extracted_text'
+                    },
+                    metadata_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'JSONB field for extraction metadata',
+                        default: 'extracted_metadata'
+                    },
+                    status_field: {
+                        type: 'string',
+                        format: 'column-ref',
+                        description: 'Extraction lifecycle status field',
+                        default: 'extraction_status'
+                    }
+                }
+            },
+            // ── Chunking config (optional — forwarded to DataFileEmbedding) ─
+            chunks: {
+                type: 'object',
+                description: 'Chunking configuration. Forwarded to DataFileEmbedding. ' +
+                    'Only meaningful when extraction is also provided.',
+                properties: {
+                    content_field_name: {
+                        type: 'string',
+                        format: 'column-ref',
+                        default: 'content'
+                    },
+                    chunk_size: { type: 'integer', default: 1000 },
+                    chunk_overlap: { type: 'integer', default: 200 },
+                    chunk_strategy: {
+                        type: 'string',
+                        enum: ['fixed', 'sentence', 'paragraph', 'semantic'],
+                        default: 'paragraph'
+                    },
+                    metadata_fields: { type: 'object' },
+                    enqueue_chunking_job: { type: 'boolean', default: true },
+                    chunking_task_name: { type: 'string', default: 'generate_chunks' }
+                }
+            },
+            // ── Stale tracking (forwarded to DataFileEmbedding) ────────────
+            stale_strategy: {
+                type: 'string',
+                enum: ['column', 'null', 'hash'],
+                description: 'Strategy for tracking embedding staleness in extract mode',
+                default: 'column'
+            },
+            include_stale_field: {
+                type: 'boolean',
+                description: 'Whether to include the embedding_stale boolean field',
+                default: true
             }
         }
     },

package/esm/data/index.d.ts

@@ -1,6 +1,7 @@
 export { DataCompositeField } from './data-composite-field';
 export { DataDirectOwner } from './data-direct-owner';
 export { DataEntityMembership } from './data-entity-membership';
+export { DataFileEmbedding } from './data-file-embedding';
 export { DataFeatureFlag } from './data-feature-flag';
 export { DataForceCurrentUser } from './data-force-current-user';
 export { DataId } from './data-id';

package/esm/data/index.js

@@ -1,6 +1,7 @@
 export { DataCompositeField } from './data-composite-field';
 export { DataDirectOwner } from './data-direct-owner';
 export { DataEntityMembership } from './data-entity-membership';
+export { DataFileEmbedding } from './data-file-embedding';
 export { DataFeatureFlag } from './data-feature-flag';
 export { DataForceCurrentUser } from './data-force-current-user';
 export { DataId } from './data-id';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "node-type-registry",
-  "version": "0.24.0",
+  "version": "0.26.0",
   "description": "Node type definitions for the Constructive blueprint system. Single source of truth for all Authz*, Data*, Relation*, and View* node types.",
   "author": "Constructive <developers@constructive.io>",
   "main": "index.js",
@@ -47,5 +47,5 @@
     "registry",
     "graphile"
   ],
-  "gitHead": "04633dc47399ffc9a5cefacfb0a90451acee988d"
+  "gitHead": "52d3f163860f219b34e7e918b2ddbf2bded7c31a"
 }