Feature/environment var creds (#116)

- Change templates to interpolate environment variables in docker compose
- Change templates to invoke secrets for environment variable credentials in K8s configuration
- Update LLMs to pull in credentials from environment variables if not specified

trustgraph-flow/trustgraph/model/text_completion/azure/llm.py

@@ -7,6 +7,7 @@ serverless endpoint service.  Input is prompt, output is response.
 import requests
 import json
 from prometheus_client import Histogram
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -23,6 +24,8 @@ default_subscriber = module
 default_temperature = 0.0
 default_max_output = 4192
 default_model = "AzureAI"
+default_endpoint = os.getenv("AZURE_ENDPOINT")
+default_token = os.getenv("AZURE_TOKEN")
 
 class Processor(ConsumerProducer):
 
@@ -31,12 +34,18 @@ class Processor(ConsumerProducer):
         input_queue = params.get("input_queue", default_input_queue)
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
-        endpoint = params.get("endpoint")
-        token = params.get("token")
+        endpoint = params.get("endpoint", default_endpoint)
+        token = params.get("token", default_token)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
         model = default_model
 
+        if endpoint is None:
+            raise RuntimeError("Azure endpoint not specified")
+
+        if token is None:
+            raise RuntimeError("Azure token not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,

trustgraph-flow/trustgraph/model/text_completion/azure_openai/llm.py

@@ -8,6 +8,7 @@ import requests
 import json
 from prometheus_client import Histogram
 from openai import AzureOpenAI
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -24,6 +25,8 @@ default_subscriber = module
 default_temperature = 0.0
 default_max_output = 4192
 default_api = "2024-02-15-preview"
+default_endpoint = os.getenv("AZURE_ENDPOINT")
+default_token = os.getenv("AZURE_TOKEN")
 
 class Processor(ConsumerProducer):
 
@@ -32,13 +35,19 @@ class Processor(ConsumerProducer):
         input_queue = params.get("input_queue", default_input_queue)
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
-        endpoint = params.get("endpoint")
-        token = params.get("token")
+        endpoint = params.get("endpoint", default_endpoint)
+        token = params.get("token", default_token)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
         model = params.get("model")
         api = params.get("api_version", default_api)
 
+        if endpoint is None:
+            raise RuntimeError("Azure endpoint not specified")
+
+        if token is None:
+            raise RuntimeError("Azure token not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,

trustgraph-flow/trustgraph/model/text_completion/claude/llm.py

@@ -6,6 +6,7 @@ Input is prompt, output is response.
 
 import anthropic
 from prometheus_client import Histogram
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -22,6 +23,7 @@ default_subscriber = module
 default_model = 'claude-3-5-sonnet-20240620'
 default_temperature = 0.0
 default_max_output = 8192
+default_api_key = os.getenv("CLAUDE_KEY")
 
 class Processor(ConsumerProducer):
 
@@ -31,10 +33,13 @@ class Processor(ConsumerProducer):
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
         model = params.get("model", default_model)
-        api_key = params.get("api_key")
+        api_key = params.get("api_key", default_api_key)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
 
+        if api_key is None:
+            raise RuntimeError("Claude API key not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,

trustgraph-flow/trustgraph/model/text_completion/cohere/llm.py

@@ -6,6 +6,7 @@ Input is prompt, output is response.
 
 import cohere
 from prometheus_client import Histogram
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -21,6 +22,7 @@ default_output_queue = text_completion_response_queue
 default_subscriber = module
 default_model = 'c4ai-aya-23-8b'
 default_temperature = 0.0
+default_api_key = os.getenv("COHERE_KEY")
 
 class Processor(ConsumerProducer):
 
@@ -30,9 +32,12 @@ class Processor(ConsumerProducer):
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
         model = params.get("model", default_model)
-        api_key = params.get("api_key")
+        api_key = params.get("api_key", default_api_key)
         temperature = params.get("temperature", default_temperature)
 
+        if api_key is None:
+            raise RuntimeError("Cohere API key not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,

trustgraph-flow/trustgraph/model/text_completion/googleaistudio/llm.py

@@ -7,6 +7,7 @@ Input is prompt, output is response.
 import google.generativeai as genai
 from google.generativeai.types import HarmCategory, HarmBlockThreshold
 from prometheus_client import Histogram
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -23,6 +24,7 @@ default_subscriber = module
 default_model = 'gemini-1.5-flash-002'
 default_temperature = 0.0
 default_max_output = 8192
+default_api_key = os.getenv("GOOGLE_AI_STUDIO_KEY")
 
 class Processor(ConsumerProducer):
 
@@ -32,10 +34,13 @@ class Processor(ConsumerProducer):
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
         model = params.get("model", default_model)
-        api_key = params.get("api_key")
+        api_key = params.get("api_key", default_api_key)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
 
+        if api_key is None:
+            raise RuntimeError("Google AI Studio API key not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,
@@ -211,4 +216,4 @@ def run():
 
     Processor.start(module, __doc__)
 
-    
+    

trustgraph-flow/trustgraph/model/text_completion/llamafile/llm.py

@@ -20,7 +20,7 @@ default_input_queue = text_completion_request_queue
 default_output_queue = text_completion_response_queue
 default_subscriber = module
 default_model = 'LLaMA_CPP'
-default_llamafile = 'http://localhost:8080/v1'
+default_llamafile = os.getenv("LLAMAFILE_URL", "http://localhost:8080/v1")
 default_temperature = 0.0
 default_max_output = 4096
 

trustgraph-flow/trustgraph/model/text_completion/ollama/llm.py

@@ -6,6 +6,7 @@ Input is prompt, output is response.
 
 from ollama import Client
 from prometheus_client import Histogram, Info
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -19,8 +20,8 @@ module = ".".join(__name__.split(".")[1:-1])
 default_input_queue = text_completion_request_queue
 default_output_queue = text_completion_response_queue
 default_subscriber = module
-default_model = 'gemma2'
-default_ollama = 'http://localhost:11434'
+default_model = 'gemma2:9b'
+default_ollama = os.getenv("OLLAMA_HOST", 'http://localhost:11434')
 
 class Processor(ConsumerProducer):
 
@@ -152,7 +153,7 @@ class Processor(ConsumerProducer):
         parser.add_argument(
             '-m', '--model',
             default="gemma2",
-            help=f'LLM model (default: gemma2)'
+            help=f'LLM model (default: {default_model})'
         )
 
         parser.add_argument(

trustgraph-flow/trustgraph/model/text_completion/openai/llm.py

@@ -6,6 +6,7 @@ Input is prompt, output is response.
 
 from openai import OpenAI
 from prometheus_client import Histogram
+import os
 
 from .... schema import TextCompletionRequest, TextCompletionResponse, Error
 from .... schema import text_completion_request_queue
@@ -22,6 +23,7 @@ default_subscriber = module
 default_model = 'gpt-3.5-turbo'
 default_temperature = 0.0
 default_max_output = 4096
+default_api_key = os.getenv("OPENAI_KEY")
 
 class Processor(ConsumerProducer):
 
@@ -31,10 +33,13 @@ class Processor(ConsumerProducer):
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
         model = params.get("model", default_model)
-        api_key = params.get("api_key")
+        api_key = params.get("api_key", default_api_key)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
 
+        if api_key is None:
+            raise RuntimeError("OpenAI API key not specified")
+
         super(Processor, self).__init__(
             **params | {
                 "input_queue": input_queue,