add LLM model & manager, get-all, get and completion API endpoints

apis/paios/openapi.yaml

@@ -700,6 +700,8 @@ paths:
       - $ref: '#/components/parameters/id'  
   /shares:
     get:
+      security:
+        - jwt: []
       tags:
         - Share Management
       summary: Retrieve all share links
@@ -721,6 +723,8 @@ paths:
             X-Total-Count:
               $ref: '#/components/headers/X-Total-Count'
     post:
+      security:
+        - jwt: []
       summary: Create new share link
       tags:
         - Share Management
@@ -737,6 +741,8 @@ paths:
               $ref: '#/components/schemas/ShareCreate'
   '/shares/{id}':
     get:
+      security:
+        - jwt: []
       tags:
         - Share Management
       summary: Retrieve share link by id
@@ -751,6 +757,8 @@ paths:
               schema:
                 $ref: '#/components/schemas/Share'
     put:
+      security:
+        - jwt: []
       tags:
         - Share Management
       summary: Update share link by id
@@ -766,6 +774,8 @@ paths:
         '200':
           description: OK
     delete:
+      security:
+        - jwt: []
       tags:
         - Share Management
       summary: Delete share link by id
@@ -777,6 +787,74 @@ paths:
           description: No Content
         '404':
           description: Not Found
+  /llms:
+    get:
+      tags:
+        - LLM Management
+      summary: Retrieve all available LLMs
+      description: Get all installed / available LLMs.
+      parameters:
+        - $ref: '#/components/parameters/sort'
+        - $ref: '#/components/parameters/range'
+        - $ref: '#/components/parameters/filter'
+      responses:
+        '200':
+          description: OK
+          content:
+            application/json:
+              schema:
+                type: array
+                items:
+                  $ref: '#/components/schemas/Llm'
+          headers:
+            X-Total-Count:
+              $ref: '#/components/headers/X-Total-Count'
+  '/llms/{id}':
+    get:
+      tags:
+        - LLM Management
+      summary: Retrieve LLM by id
+      description: Retrieve the LLM with the specified id.
+      parameters:
+        - $ref: '#/components/parameters/kebab-dot_id'
+      responses:
+        '200':
+          description: OK
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/Llm'
+        '404':
+          description: LLM not found
+  '/llms/{id}/completion':
+    post:
+      tags:
+        - LLM Management
+      summary: Invoke Completion on LLM
+      description: Invoke Completion on the LLM with the specified id.
+      operationId: backend.api.LlmsView.completion
+      parameters:
+        - $ref: '#/components/parameters/kebab-dot_id'
+      requestBody:
+        description: Messages to input to Completion function.
+        content:
+          application/json:
+            schema:
+              type: object
+              properties:
+                messages:
+                  $ref: '#/components/schemas/messagesList'
+      responses:
+        '200':
+          description: Completion succeeded
+          content:
+            application/json:
+              schema:
+                type: object
+        '400':
+          description: Completion failed
+        '404':
+          description: LLM not found
   /auth/webauthn/register-options:
     post:
       summary: Generate WebAuthn registration options
@@ -936,6 +1014,13 @@ components:
       required: true
       schema:
         $ref: '#/components/schemas/kebab-snake_id'
+    kebab-dot_id:
+      name: id
+      in: path
+      description: id of the object
+      required: true
+      schema:
+        $ref: '#/components/schemas/kebab-dot_id'
     key:
       name: key
       in: path
@@ -1003,6 +1088,12 @@ components:
       maxLength: 100
       example: langchain-core
       pattern: ^[a-z0-9]+([_-][a-z0-9]+)*$
+    kebab-dot_id:
+      type: string
+      minLength: 2
+      maxLength: 100
+      example: ollama-llama3.2
+      pattern: ^[a-z0-9]+([.-][a-z0-9]+)*$
     semVer:
       type: string
       example: '1.1.0'
@@ -1068,6 +1159,16 @@ components:
       readOnly: true
       example: abcd-efgh-ijkl
       pattern: ^[a-z]{4}-[a-z]{4}-[a-z]{4}$
+    messagesList:
+      type: array
+      example: [{"role": "user", "content": "What is Kwaai.ai?"}]
+      items:
+        type: object
+        properties:
+          role:
+            type: string
+          content:
+            type: string
     download:
       type: object
       properties:
@@ -1390,6 +1491,28 @@ components:
           example: false
       required:        
         - resource_id
+    Llm:
+      type: object
+      title: Llm
+      properties:
+        id:
+          type: string
+        name:
+          type: string
+        llm_name:
+          type: string
+        provider:
+          type: string
+        api_base:
+          type: string
+          nullable: true
+        is_active:
+          type: boolean
+      required:
+        - id
+        - name
+        - provider
+        - is_active
     RegistrationOptions:
       type: object
       properties:

backend/api/LlmsView.py

@@ -0,0 +1,43 @@
+from starlette.responses import JSONResponse
+from backend.managers.LlmsManager import LlmsManager
+from backend.pagination import parse_pagination_params
+
+class LlmsView:
+    def __init__(self):
+        self.llmm = LlmsManager()
+
+    async def get(self, id: str):
+        llm = await self.llmm.get_llm(id)
+        if llm is None:
+            return JSONResponse(headers={"error": "LLM not found"}, status_code=404)
+        return JSONResponse(llm.model_dump(), status_code=200)
+
+    async def search(self, filter: str = None, range: str = None, sort: str = None):
+        result = parse_pagination_params(filter, range, sort)
+        if isinstance(result, JSONResponse):
+            return result
+
+        offset, limit, sort_by, sort_order, filters = result
+
+        llms, total_count = await self.llmm.retrieve_llms(limit=limit, offset=offset, sort_by=sort_by, sort_order=sort_order, filters=filters)
+        headers = {
+            'X-Total-Count': str(total_count),
+            'Content-Range': f'shares {offset}-{offset + len(llms) - 1}/{total_count}'
+        }
+        return JSONResponse([llm.model_dump() for llm in llms], status_code=200, headers=headers)
+
+    async def completion(self, id: str, body: dict):
+        print("completion.  body: {}".format(body))
+        messages = []
+        if 'messages' in body and body['messages']:
+            messages = body['messages']
+        llm = await self.llmm.get_llm(id)
+        if llm:
+            print("Selected LLM is {}".format(llm.llm_name))
+            response = self.llmm.completion(llm, messages)
+            if response:
+                return JSONResponse(response.model_dump(), status_code=200)
+            else:
+                return JSONResponse(status_code=400, content={"message": "Completion failed"})
+        else:
+            return JSONResponse(status_code=404, content={"message": "LLM not found"})

backend/api/__init__.py

@@ -9,3 +9,4 @@
 from .PersonasView import PersonasView
 from .SharesView import SharesView
 from .AuthView import AuthView
+from .LlmsView import LlmsView