> ## Documentation Index > Fetch the complete documentation index at: https://api-docs.datagol.ai/llms.txt > Use this file to discover all available pages before exploring further. # Generate Schema for Advanced Extraction > ## 1️⃣ Overview **Purpose:** Generates a schema based on provided file content or file identifier. The schema generation is driven by an instruction prompt. --- ## 2️⃣ Endpoint ``` bash POST /noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate ``` --- ## 3️⃣ Path Parameters | Parameter | Type | Required | Description | | --- | --- | --- | --- | | workspaceId | UUID | ✅ Yes | Identifier of the workspace | --- ## 4️⃣ Authentication Requires authentication using: ``` id="c3yq8n" Authorization: Bearer ``` --- ## 5️⃣ Request Headers | Header | Required | | --- | --- | | Authorization | ✅ Yes | | Content-Type: application/json | ✅ Yes | --- ## 6️⃣ Request Body Schema ``` json { "prompt": "string", "content": "string", "fileId": "string" } ``` --- ## 7️⃣ Field Descriptions ### Schema Generation Fields | Field | Type | Required | Description | | --- | --- | --- | --- | | prompt | string | ✅ Yes | Instruction for schema generation | | content | string | ❌ Optional | File content (if not using fileId) | | fileId | string | ❌ Optional | File ID (if not using content) | --- ## 8️⃣ Example Request ``` bash curl -X POST https://be.datagol.ai/noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{ "prompt": "Extract schema from invoice file", "content": "file raw text content here" }' ``` --- ## 9️⃣ Behavior Summary Generates a schema using either file content or file ID based on the provided prompt instruction. ## OpenAPI ````yaml /openapi/openapi.yaml post /noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate openapi: 3.0.0 info: title: DataGOL APIs version: 1.0.0 description: '' servers: - url: https://be.datagol.ai - url: https://be.datagol.ai - url: https://kg.datagol.ai - url: https://ai.datagol.ai security: [] tags: - name: Links & Joins description: > ## 🔗 Links & Joins LINK and LOOKUP are special workbook column types that enable relational data modeling across tables. | Column Type | Purpose | |---|---| | `LINK` | Defines a relationship between two tables | | `LOOKUP` | Projects columns from an associated table via a link | Both are configured through `ColumnDTO.colOptions`. A table can have multiple link columns, and each link can have multiple lookup columns. --- ## 📐 UI Data Types - **LINK** — Special column used to represent inter-table relations - **LOOKUP** — Special column used to project associated table fields via a link --- ## 🔗 LINK Semantics A LINK column stores metadata for: - `tableId` — Current table - `associatedTableId` — Associated (linked) table - `relationType` — Type of relationship - `linkColumnId` — Link column identifier **Supported relation patterns:** | Pattern | Description | |---|---| | `ONE_TO_ONE` | Each record in the source maps to one record in the destination | | `ONE_TO_MANY` | One source record maps to many destination records | | `MANY_TO_MANY` | Many source records map to many destination records | --- ## 🔍 LOOKUP Semantics - A LOOKUP column is tightly coupled with a specific LINK column - References a link using `linkColumnId` - Represents a selected column from the associated table - Multiple lookup columns can be defined for the same link --- ## 🗂️ Metadata in `ColumnDTO.colOptions` Use `colOptions` to persist link/lookup metadata. | Field | Description | |---|---| | `columnId` | Current column ID | | `tableId` | Current table ID | | `associatedTableId` | Associated table ID | | `linkColumnId` | Link column ID — used by LOOKUP; references the LINK column's `columnId` | | `relationType` | Relation semantics for LINK | --- ## 🔄 Manual Linking Record Flow To create manual record-level links: 1. Fetch associated table schema/data 2. User selects associated record(s) to link 3. Send linking payload with: - `sourceRecordId` — PK of the current table row - `destinationRecordId` — PK of the associated table row - `linkColumnId` — LINK column identifier 4. Backend resolves link metadata from `linkColumnId` (relation type, associated table, direction/constraints) 5. Backend applies relation rules and stores the link relation --- ## 📋 Record Linking Contract | Field | Description | |---|---| | `sourceRecordId` | PK of the current table row | | `destinationRecordId` | PK of the associated table row | | `linkColumnId` | LINK column identifier | | `tableId` | Current table ID | > Backend derives all remaining relation metadata from the link configuration. --- ## ⚠️ Important Notes > - LINK and LOOKUP are **column-level constructs** first; record linking is a data-level operation built on top > - LOOKUP behavior depends on LINK existence and metadata integrity > - With multiple links to the same associated table, each LOOKUP must bind to the correct `linkColumnId` > - Always treat the associated row ID as the associated table PK for linking UI selection paths: /noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate: parameters: - name: workspaceId in: path required: true schema: type: string example: '{{workspaceId}}' post: tags: - AI Extraction summary: Generate Schema for Advanced Extraction description: >- ## 1️⃣ Overview **Purpose:** Generates a schema based on provided file content or file identifier. The schema generation is driven by an instruction prompt. --- ## 2️⃣ Endpoint ``` bash POST /noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate ``` --- ## 3️⃣ Path Parameters | Parameter | Type | Required | Description | | --- | --- | --- | --- | | workspaceId | UUID | ✅ Yes | Identifier of the workspace | --- ## 4️⃣ Authentication Requires authentication using: ``` id="c3yq8n" Authorization: Bearer ``` --- ## 5️⃣ Request Headers | Header | Required | | --- | --- | | Authorization | ✅ Yes | | Content-Type: application/json | ✅ Yes | --- ## 6️⃣ Request Body Schema ``` json { "prompt": "string", "content": "string", "fileId": "string" } ``` --- ## 7️⃣ Field Descriptions ### Schema Generation Fields | Field | Type | Required | Description | | --- | --- | --- | --- | | prompt | string | ✅ Yes | Instruction for schema generation | | content | string | ❌ Optional | File content (if not using fileId) | | fileId | string | ❌ Optional | File ID (if not using content) | --- ## 8️⃣ Example Request ``` bash curl -X POST https://be.datagol.ai/noCo/api/v2/workspaces/{workspaceId}/extraction/schema/generate \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{ "prompt": "Extract schema from invoice file", "content": "file raw text content here" }' ``` --- ## 9️⃣ Behavior Summary Generates a schema using either file content or file ID based on the provided prompt instruction. requestBody: content: application/json: schema: type: object properties: prompt: type: string content: type: string example: prompt: extract detailed transcript content: >- [00:00:02] Sarah Miller: Yeah, great to connect.[00:00:04] Mark Thompson: Likewise. So just for context — Alex and I started working together recently. We’re very early, mostly speaking with potential users and understanding their pain points.[00:00:18] Sarah Miller: Quick intro from my side — I’ve spent the last 10 years in startups. Started in fintech, then built a SaaS company focused on workflow tools. Exited last year and now exploring the next venture.[00:00:32] Alex Rivera: I’m Alex, engineering background. Around 15 years building developer tools and infrastructure products. This is my second startup.[00:00:44] John Doe: Nice to meet you both. I run a small fund investing in early-stage AI and infrastructure companies. Also previously built and bootstrapped a company for nearly a decade, so I understand the founder journey well.[00:01:00] Sarah Miller: Amazing. So our thesis is simple — AI isn’t just speeding up software development, it’s changing how software should be built entirely. Instead of static apps, we’re exploring the idea of “software you can summon” — lightweight, adaptive tools that exist only when needed. [00:01:20] John Doe: Interesting. Who’s the target user?[00:01:23] Alex Rivera: Product managers, designers, and technical founders who are building tools themselves but feel underserved by existing infrastructure. We want to rethink the layer beneath the coding agents.[00:01:38] John Doe: Makes sense. The key will be articulating why you, why now, and what gives you the right to win.[00:01:46] Sarah Miller: Exactly. That’s what we’re refining — and conversations like this really help. responses: '200': description: Generate Schema for Advanced Extraction security: - bearerAuth: [] components: securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: JWT description: | This API uses OAuth 2.0 with the authorization code grant flow. ````