@arbidocs/client 0.3.43 → 0.3.45

package/dist/index.d.cts

@@ -744,6 +744,35 @@ interface paths {
         patch?: never;
         trace?: never;
     };
+    '/v1/document/columns': {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        get?: never;
+        put?: never;
+        /**
+         * Fetch Document Columns Endpoint
+         * @description Fetch specific encrypted column groups for a batch of documents.
+         *
+         *     Companion to ``GET /v1/document/list?fields=lite``: the list endpoint
+         *     returns all rows fast with plain DB columns + decrypted ``file_name``,
+         *     and this endpoint fills in the heavy encrypted fields (``doc_metadata``
+         *     and ``doctags``) on demand when the frontend reveals the columns that
+         *     need them.
+         *
+         *     Only the column groups listed in ``fields`` trigger DB work and
+         *     decryption, so the cost scales with what's actually asked for.
+         */
+        post: operations['fetch_document_columns'];
+        delete?: never;
+        options?: never;
+        head?: never;
+        patch?: never;
+        trace?: never;
+    };
     '/v1/document/similar': {
         parameters: {
             query?: never;
@@ -2350,7 +2379,7 @@ interface components {
              * @default {
              *       "API_TYPE": "remote",
              *       "ENABLE_THINKING": true,
-             *       "MODEL_NAME": "GLM5@Fireworks",
+             *       "MODEL_NAME": "Premium",
              *       "SYSTEM_INSTRUCTION": "You are a code execution agent. You write and run code to accomplish tasks.\n\nYou have access to an execute_code tool that runs code in a sandboxed Docker container.\n\nThe sandbox has:\n- Python 3.12 with numpy, pandas, matplotlib, requests, beautifulsoup4,\n  httpx, pillow, sympy, pyyaml (plus uv for installing more)\n- Node.js 22 with TypeScript (ts-node) and npm\n- Bash with git, curl, wget, jq, sqlite3, build-essential\n- Network access (can pip/npm install additional packages)\n\nInstructions:\n1. Write clean, correct code to accomplish the user's task\n2. Always call execute_code to run your code — never guess the output\n3. If execution errors, analyze the traceback, fix the code, and retry\n4. Once you have the correct output, respond with ONLY the final result\n\nKeep code simple and direct. Prefer Python unless the task specifically requires another language.",
              *       "TEMPERATURE": 0.2,
              *       "MAX_TOKENS": 8000,
@@ -2386,7 +2415,7 @@ interface components {
              *       "MAX_NUMB_OF_CHUNKS": 30,
              *       "MAX_CONCURRENT_REQUESTS": 256,
              *       "MODEL_NAME": "Qwen/Qwen3-Reranker-0.6B",
-             *       "API_TYPE": "remote",
+             *       "API_TYPE": "local",
              *       "RETRIEVAL_INSTRUCTION": "Judge whether the passage contains the specific facts, figures, names, or references mentioned in the query. Only answer yes if the passage directly contains these details"
              *     }
              */
@@ -2408,7 +2437,7 @@ interface components {
             /**
              * @default {
              *       "MODEL_NAME": "Qwen/Qwen3-Embedding-0.6B",
-             *       "API_TYPE": "remote",
+             *       "API_TYPE": "local",
              *       "BATCH_SIZE": 128,
              *       "MAX_CONCURRENT_REQUESTS": 256,
              *       "DOCUMENT_PREFIX": "",
@@ -2820,7 +2849,7 @@ interface components {
             /**
              * Model Name
              * @description Model for code generation. Should be a strong coding model.
-             * @default GLM5@Fireworks
+             * @default Premium
              */
             MODEL_NAME: string;
             /**
@@ -3328,6 +3357,45 @@ interface components {
             /** External Ids */
             external_ids: string[];
         };
+        /**
+         * DocColumnData
+         * @description Per-document column data returned by POST /v1/document/columns.
+         *
+         *     Fields are populated only when requested in the ``fields`` list. Unrequested
+         *     fields are omitted from the response. Used by the frontend's on-demand
+         *     column loader to fill in encrypted/joined fields after the skeleton load.
+         */
+        DocColumnData: {
+            /** External Id */
+            external_id: string;
+            doc_metadata?: components['schemas']['DocMetadata'] | null;
+            /** Doctags */
+            doctags?: components['schemas']['DocTagResponse'][] | null;
+        };
+        /**
+         * DocColumnsRequest
+         * @description Request body for POST /v1/document/columns.
+         */
+        DocColumnsRequest: {
+            /**
+             * External Ids
+             * @description Document external IDs to fetch columns for (max 10k per request).
+             */
+            external_ids: string[];
+            /**
+             * Fields
+             * @description Column groups to decrypt and return for each document.
+             */
+            fields: ('doc_metadata' | 'doctags')[];
+        };
+        /**
+         * DocColumnsResponse
+         * @description Response for POST /v1/document/columns — batch column data for many docs.
+         */
+        DocColumnsResponse: {
+            /** Docs */
+            docs: components['schemas']['DocColumnData'][];
+        };
         /**
          * DocMetadata
          * @description Structured model for document metadata stored in JSONB column.
@@ -3410,8 +3478,8 @@ interface components {
          * @description A pair of documents identified as near-duplicates by centroid similarity.
          */
         DocSimPair: {
-            doc_a: components['schemas']['DocResponse'];
-            doc_b: components['schemas']['DocResponse'];
+            doc_a: components['schemas']['DocSimSummary'];
+            doc_b: components['schemas']['DocSimSummary'];
             /** Similarity */
             similarity: number;
         };
@@ -3425,6 +3493,16 @@ interface components {
             /** Pairs */
             pairs: components['schemas']['DocSimPair'][];
         };
+        /**
+         * DocSimSummary
+         * @description Lightweight document summary for the similar-documents endpoint.
+         */
+        DocSimSummary: {
+            /** External Id */
+            external_id: string;
+            /** File Name */
+            file_name?: string | null;
+        };
         /**
          * DocTagResponse
          * @description Response for doctag operations - the link between a document and a tag.
@@ -3589,7 +3667,7 @@ interface components {
             /**
              * Api Type
              * @description The inference type (local or remote).
-             * @default remote
+             * @default local
              * @enum {string}
              */
             API_TYPE: 'local' | 'remote';
@@ -5278,7 +5356,7 @@ interface components {
             /**
              * Api Type
              * @description The inference type (local or remote).
-             * @default remote
+             * @default local
              * @enum {string}
              */
             API_TYPE: 'local' | 'remote';
@@ -8673,10 +8751,43 @@ interface operations {
             };
         };
     };
+    fetch_document_columns: {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        requestBody: {
+            content: {
+                'application/json': components['schemas']['DocColumnsRequest'];
+            };
+        };
+        responses: {
+            /** @description Successful Response */
+            200: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    'application/json': components['schemas']['DocColumnsResponse'];
+                };
+            };
+            /** @description Validation Error */
+            422: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    'application/json': components['schemas']['HTTPValidationError'];
+                };
+            };
+        };
+    };
     get_similar_documents: {
         parameters: {
             query?: {
-                /** @description Minimum similarity score (default 0.92 for near-duplicates) */
+                /** @description Minimum similarity score (default 0.95 for near-duplicates) */
                 threshold?: number;
                 /** @description Filter pairs involving a specific document */
                 doc_ext_id?: string | null;

package/dist/index.d.ts

@@ -744,6 +744,35 @@ interface paths {
         patch?: never;
         trace?: never;
     };
+    '/v1/document/columns': {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        get?: never;
+        put?: never;
+        /**
+         * Fetch Document Columns Endpoint
+         * @description Fetch specific encrypted column groups for a batch of documents.
+         *
+         *     Companion to ``GET /v1/document/list?fields=lite``: the list endpoint
+         *     returns all rows fast with plain DB columns + decrypted ``file_name``,
+         *     and this endpoint fills in the heavy encrypted fields (``doc_metadata``
+         *     and ``doctags``) on demand when the frontend reveals the columns that
+         *     need them.
+         *
+         *     Only the column groups listed in ``fields`` trigger DB work and
+         *     decryption, so the cost scales with what's actually asked for.
+         */
+        post: operations['fetch_document_columns'];
+        delete?: never;
+        options?: never;
+        head?: never;
+        patch?: never;
+        trace?: never;
+    };
     '/v1/document/similar': {
         parameters: {
             query?: never;
@@ -2350,7 +2379,7 @@ interface components {
              * @default {
              *       "API_TYPE": "remote",
              *       "ENABLE_THINKING": true,
-             *       "MODEL_NAME": "GLM5@Fireworks",
+             *       "MODEL_NAME": "Premium",
              *       "SYSTEM_INSTRUCTION": "You are a code execution agent. You write and run code to accomplish tasks.\n\nYou have access to an execute_code tool that runs code in a sandboxed Docker container.\n\nThe sandbox has:\n- Python 3.12 with numpy, pandas, matplotlib, requests, beautifulsoup4,\n  httpx, pillow, sympy, pyyaml (plus uv for installing more)\n- Node.js 22 with TypeScript (ts-node) and npm\n- Bash with git, curl, wget, jq, sqlite3, build-essential\n- Network access (can pip/npm install additional packages)\n\nInstructions:\n1. Write clean, correct code to accomplish the user's task\n2. Always call execute_code to run your code — never guess the output\n3. If execution errors, analyze the traceback, fix the code, and retry\n4. Once you have the correct output, respond with ONLY the final result\n\nKeep code simple and direct. Prefer Python unless the task specifically requires another language.",
              *       "TEMPERATURE": 0.2,
              *       "MAX_TOKENS": 8000,
@@ -2386,7 +2415,7 @@ interface components {
              *       "MAX_NUMB_OF_CHUNKS": 30,
              *       "MAX_CONCURRENT_REQUESTS": 256,
              *       "MODEL_NAME": "Qwen/Qwen3-Reranker-0.6B",
-             *       "API_TYPE": "remote",
+             *       "API_TYPE": "local",
              *       "RETRIEVAL_INSTRUCTION": "Judge whether the passage contains the specific facts, figures, names, or references mentioned in the query. Only answer yes if the passage directly contains these details"
              *     }
              */
@@ -2408,7 +2437,7 @@ interface components {
             /**
              * @default {
              *       "MODEL_NAME": "Qwen/Qwen3-Embedding-0.6B",
-             *       "API_TYPE": "remote",
+             *       "API_TYPE": "local",
              *       "BATCH_SIZE": 128,
              *       "MAX_CONCURRENT_REQUESTS": 256,
              *       "DOCUMENT_PREFIX": "",
@@ -2820,7 +2849,7 @@ interface components {
             /**
              * Model Name
              * @description Model for code generation. Should be a strong coding model.
-             * @default GLM5@Fireworks
+             * @default Premium
              */
             MODEL_NAME: string;
             /**
@@ -3328,6 +3357,45 @@ interface components {
             /** External Ids */
             external_ids: string[];
         };
+        /**
+         * DocColumnData
+         * @description Per-document column data returned by POST /v1/document/columns.
+         *
+         *     Fields are populated only when requested in the ``fields`` list. Unrequested
+         *     fields are omitted from the response. Used by the frontend's on-demand
+         *     column loader to fill in encrypted/joined fields after the skeleton load.
+         */
+        DocColumnData: {
+            /** External Id */
+            external_id: string;
+            doc_metadata?: components['schemas']['DocMetadata'] | null;
+            /** Doctags */
+            doctags?: components['schemas']['DocTagResponse'][] | null;
+        };
+        /**
+         * DocColumnsRequest
+         * @description Request body for POST /v1/document/columns.
+         */
+        DocColumnsRequest: {
+            /**
+             * External Ids
+             * @description Document external IDs to fetch columns for (max 10k per request).
+             */
+            external_ids: string[];
+            /**
+             * Fields
+             * @description Column groups to decrypt and return for each document.
+             */
+            fields: ('doc_metadata' | 'doctags')[];
+        };
+        /**
+         * DocColumnsResponse
+         * @description Response for POST /v1/document/columns — batch column data for many docs.
+         */
+        DocColumnsResponse: {
+            /** Docs */
+            docs: components['schemas']['DocColumnData'][];
+        };
         /**
          * DocMetadata
          * @description Structured model for document metadata stored in JSONB column.
@@ -3410,8 +3478,8 @@ interface components {
          * @description A pair of documents identified as near-duplicates by centroid similarity.
          */
         DocSimPair: {
-            doc_a: components['schemas']['DocResponse'];
-            doc_b: components['schemas']['DocResponse'];
+            doc_a: components['schemas']['DocSimSummary'];
+            doc_b: components['schemas']['DocSimSummary'];
             /** Similarity */
             similarity: number;
         };
@@ -3425,6 +3493,16 @@ interface components {
             /** Pairs */
             pairs: components['schemas']['DocSimPair'][];
         };
+        /**
+         * DocSimSummary
+         * @description Lightweight document summary for the similar-documents endpoint.
+         */
+        DocSimSummary: {
+            /** External Id */
+            external_id: string;
+            /** File Name */
+            file_name?: string | null;
+        };
         /**
          * DocTagResponse
          * @description Response for doctag operations - the link between a document and a tag.
@@ -3589,7 +3667,7 @@ interface components {
             /**
              * Api Type
              * @description The inference type (local or remote).
-             * @default remote
+             * @default local
              * @enum {string}
              */
             API_TYPE: 'local' | 'remote';
@@ -5278,7 +5356,7 @@ interface components {
             /**
              * Api Type
              * @description The inference type (local or remote).
-             * @default remote
+             * @default local
              * @enum {string}
              */
             API_TYPE: 'local' | 'remote';
@@ -8673,10 +8751,43 @@ interface operations {
             };
         };
     };
+    fetch_document_columns: {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        requestBody: {
+            content: {
+                'application/json': components['schemas']['DocColumnsRequest'];
+            };
+        };
+        responses: {
+            /** @description Successful Response */
+            200: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    'application/json': components['schemas']['DocColumnsResponse'];
+                };
+            };
+            /** @description Validation Error */
+            422: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    'application/json': components['schemas']['HTTPValidationError'];
+                };
+            };
+        };
+    };
     get_similar_documents: {
         parameters: {
             query?: {
-                /** @description Minimum similarity score (default 0.92 for near-duplicates) */
+                /** @description Minimum similarity score (default 0.95 for near-duplicates) */
                 threshold?: number;
                 /** @description Filter pairs involving a specific document */
                 doc_ext_id?: string | null;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arbidocs/client",
-  "version": "0.3.43",
+  "version": "0.3.45",
   "description": "TypeScript SDK for the ARBI API — zero-knowledge auth, E2E encryption, and type-safe REST client",
   "type": "module",
   "main": "dist/index.cjs",