multiwoven-integrations 0.40.0 → 0.41.0

data/lib/multiwoven/integrations/source/anthropic/config/spec.json

@@ -6,7 +6,13 @@
     "$schema": "http://json-schema.org/draft-07/schema#",
     "title": "Anthropic Endpoint",
     "type": "object",
-    "required": ["api_key", "request_format", "response_format"],
+    "x-dynamic-input-path": "messages.0.content",
+    "x-output-path": "content.0.text",
+    "required": [
+      "api_key",
+      "request_format",
+      "response_format"
+    ],
     "properties": {
       "api_key": {
         "type": "string",

data/lib/multiwoven/integrations/source/aws_bedrock_model/config/meta.json

@@ -1,16 +1,22 @@
 {
   "data": {
     "name": "AwsBedrockModel",
-    "title": "AWS Bedrock Model",
+    "title": "Amazon Bedrock",
     "connector_type": "source",
     "category": "AI Model",
     "sub_category": "LLM",
+    "provider_type": "cloud_provider",
+    "display_order": 40,
+    "description": "Access foundation models via AWS Bedrock. Supports Titan, Claude via AWS, Llama, Mistral, and more via your AWS credentials.",
     "documentation_url": "https://docs.squared.ai/activation/ai-ml-sources/aws_bedrock-model",
     "github_issue_label": "source-aws-bedrock-model",
     "icon": "icon.svg",
     "license": "MIT",
     "release_stage": "alpha",
     "support_level": "community",
-    "tags": ["language:ruby", "multiwoven"]
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
 }

data/lib/multiwoven/integrations/source/aws_bedrock_model/config/models.json

@@ -0,0 +1,96 @@
+{
+  "models": [
+    {
+      "id": "anthropic.claude-sonnet-4-5-20250929-v1:0",
+      "name": "Claude Sonnet 4.5 (Bedrock)",
+      "model_type": "llm",
+      "type": "completion",
+      "tasks": [
+        "Text Generation",
+        "Reasoning",
+        "Analysis"
+      ],
+      "context_window": 200000,
+      "max_output": 64000,
+      "pricing": {
+        "input": 3.0,
+        "output": 15.0,
+        "cached_read": 0.3,
+        "cached_write": 3.75,
+        "unit": "per_1m_tokens"
+      },
+      "capabilities": [
+        "Documents",
+        "Image analysis",
+        "Data analyst"
+      ]
+    },
+    {
+      "id": "amazon.titan-text-premier-v1:0",
+      "name": "Titan Text Premier",
+      "model_type": "llm",
+      "type": "completion",
+      "tasks": [
+        "Text Generation",
+        "Chat"
+      ],
+      "context_window": 32000,
+      "max_output": 3072,
+      "pricing": {
+        "input": 0.5,
+        "output": 1.5,
+        "cached_read": null,
+        "cached_write": null,
+        "unit": "per_1m_tokens"
+      },
+      "capabilities": [
+        "Documents"
+      ]
+    },
+    {
+      "id": "meta.llama3-70b-instruct-v1:0",
+      "name": "Llama 3 70B Instruct",
+      "model_type": "llm",
+      "type": "completion",
+      "tasks": [
+        "Text Generation",
+        "Reasoning"
+      ],
+      "context_window": 8192,
+      "max_output": 2048,
+      "pricing": {
+        "input": 2.65,
+        "output": 3.5,
+        "cached_read": null,
+        "cached_write": null,
+        "unit": "per_1m_tokens"
+      },
+      "capabilities": [
+        "Documents"
+      ]
+    },
+    {
+      "id": "amazon.titan-embed-text-v2:0",
+      "name": "Titan Text Embeddings V2",
+      "model_type": "embedding",
+      "type": "embedding",
+      "tasks": [
+        "Text Embeddings",
+        "Indexing",
+        "Retrieval"
+      ],
+      "context_window": 8192,
+      "max_output": null,
+      "pricing": {
+        "input": 0.02,
+        "output": null,
+        "cached_read": null,
+        "cached_write": null,
+        "unit": "per_1m_tokens"
+      },
+      "capabilities": [
+        "Documents"
+      ]
+    }
+  ]
+}

data/lib/multiwoven/integrations/source/aws_sagemaker_model/config/meta.json

@@ -1,16 +1,22 @@
 {
   "data": {
     "name": "AwsSagemakerModel",
-    "title": "AWS Sagemaker Model",
+    "title": "AWS SageMaker",
     "connector_type": "source",
     "category": "AI Model",
     "sub_category": "AI_ML Service",
+    "provider_type": "cloud_provider",
+    "display_order": 80,
+    "description": "Deploy and serve custom models via AWS SageMaker endpoints. Supports any model you have hosted on SageMaker.",
     "documentation_url": "https://docs.squared.ai/guides/sources/data-sources/aws_sagemaker-model",
     "github_issue_label": "source-aws-sagemaker-model",
     "icon": "icon.svg",
     "license": "MIT",
     "release_stage": "alpha",
     "support_level": "community",
-    "tags": ["language:ruby", "multiwoven"]
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
 }

data/lib/multiwoven/integrations/source/databrics_model/config/meta.json

@@ -1,17 +1,22 @@
 {
-    "data": {
-      "name": "DatabricksModel",
-      "title": "Databricks Model",
-      "connector_type": "source",
-      "category": "AI Model",
-      "sub_category": "AI_ML Service",
-      "documentation_url": "https://docs.squared.ai/guides/sources/data-sources/databricks-model",
-      "github_issue_label": "source-databricks-foundation",
-      "icon": "icon.svg",
-      "license": "MIT",
-      "release_stage": "alpha",
-      "support_level": "community",
-      "tags": ["language:ruby", "multiwoven"]
-    }
+  "data": {
+    "name": "DatabricksModel",
+    "title": "Databricks",
+    "connector_type": "source",
+    "category": "AI Model",
+    "sub_category": "AI_ML Service",
+    "provider_type": "cloud_provider",
+    "display_order": 100,
+    "description": "Serve models from Databricks Model Serving endpoints using your workspace credentials.",
+    "documentation_url": "https://docs.squared.ai/guides/sources/data-sources/databricks-model",
+    "github_issue_label": "source-databricks-foundation",
+    "icon": "icon.svg",
+    "license": "MIT",
+    "release_stage": "alpha",
+    "support_level": "community",
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
-  
+}

data/lib/multiwoven/integrations/source/epic_fhir/client.rb

@@ -8,6 +8,7 @@ module Multiwoven::Integrations::Source
     NDJSON_ACCEPT = "application/fhir+ndjson"
     DEFAULT_EXPORT_POLL_INTERVAL = 5
     DEFAULT_EXPORT_TIMEOUT = 120
+    CONCURRENT_EXPORT_MESSAGE = "Another request for this same Client and Group is in progress."
     REQUIRED_CONFIG_KEYS = %w[fhir_base_url token_url client_id private_key kid export_group_id].freeze
     DEFAULT_RESOURCES = %w[
       AllergyIntolerance
@@ -203,36 +204,68 @@ module Multiwoven::Integrations::Source
       end
 
       def run_bulk_export(connection_config, resource_type)
-        status_url = kickoff_bulk_export(connection_config, resource_type)
-        poll_bulk_export(connection_config, status_url)
+        deadline = Time.now + export_timeout(connection_config)
+        status_url = kickoff_bulk_export(connection_config, resource_type, deadline: deadline)
+        poll_bulk_export(connection_config, status_url, deadline: deadline)
       end
 
-      def kickoff_bulk_export(connection_config, resource_type)
+      def kickoff_bulk_export(connection_config, resource_type, deadline:)
         url = bulk_export_url(connection_config, resource_type)
-        response = fhir_get(connection_config, url, extra_headers: { "Prefer" => "respond-async" })
-        raise fhir_api_error(response, url) unless response.code.to_s == "202"
 
-        status_url = response["content-location"].presence
-        raise StandardError, "Epic FHIR 202 #{url} missing Content-Location header" if status_url.nil?
+        loop do
+          response = fhir_get(connection_config, url, extra_headers: { "Prefer" => "respond-async" })
+          if response.code.to_s == "202"
+            status_url = response["content-location"].presence
+            raise StandardError, "Epic FHIR 202 #{url} missing Content-Location header" if status_url.nil?
 
-        status_url
-      end
+            return status_url
+          end
 
-      def poll_bulk_export(connection_config, status_url)
-        timeout = export_timeout(connection_config)
-        deadline = Time.now + timeout
+          raise fhir_api_error(response, url) unless concurrent_export?(response)
+          raise concurrent_export_timeout_error unless wait_for_retry?(connection_config, response, deadline)
+        end
+      end
 
+      def poll_bulk_export(connection_config, status_url, deadline:)
         loop do
           response = fhir_get(connection_config, status_url)
           code = response.code.to_s
           return JSON.parse(response.body) if code == "200"
           raise fhir_api_error(response, status_url) unless code == "202"
-          raise StandardError, "Epic FHIR 202 #{status_url} still in progress after #{timeout}s" if Time.now >= deadline
 
-          sleep(retry_after(response) || export_poll_interval(connection_config))
+          unless wait_for_retry?(connection_config, response, deadline)
+            raise StandardError,
+                  "Epic FHIR 202 #{status_url} still in progress after #{export_timeout(connection_config)}s"
+          end
         end
       end
 
+      # Waits between attempts without ever waiting past the shared export
+      # deadline: sleeping longer and then retrying would issue a request the
+      # caller has already given up on, and Epic counts it as one more
+      # concurrent export for this client and group.
+      def wait_for_retry?(connection_config, response, deadline)
+        remaining = deadline - Time.now
+        return false unless remaining.positive?
+
+        delay = retry_after(response) || export_poll_interval(connection_config)
+        return false if delay > remaining
+
+        sleep(delay)
+        true
+      end
+
+      def concurrent_export?(response)
+        response.body.to_s.downcase.include?(CONCURRENT_EXPORT_MESSAGE.downcase)
+      end
+
+      def concurrent_export_timeout_error
+        StandardError.new(
+          "Epic is still preparing another bulk export for this Client and Group. " \
+          "Retry after the existing export completes."
+        )
+      end
+
       def download_bulk_output(connection_config, manifest, resource_type, limit:, offset:)
         return [] if limit&.zero?
 

data/lib/multiwoven/integrations/source/generic_open_ai/client.rb

@@ -15,7 +15,7 @@ module Multiwoven::Integrations::Source
         )
         success?(response) ? success_status : failure_status(nil)
       rescue StandardError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:CHECK_CONNECTION:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:CHECK_CONNECTION:EXCEPTION", type: "error" })
         failure_status(e)
       end
 
@@ -24,7 +24,7 @@ module Multiwoven::Integrations::Source
         catalog = build_catalog(catalog_json)
         catalog.to_multiwoven_message
       rescue StandardError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:DISCOVER:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:DISCOVER:EXCEPTION", type: "error" })
       end
 
       def read(sync_config)
@@ -41,11 +41,18 @@ module Multiwoven::Integrations::Source
           run_model(connection_config, payload)
         end
       rescue StandardError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:READ:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:READ:EXCEPTION", type: "error" })
       end
 
       private
 
+      # Prefixes this connector's log lines. Subclasses that reuse this client
+      # for another OpenAI-compatible provider override it so their exceptions
+      # are attributable.
+      def log_context
+        "GENERIC OPEN AI"
+      end
+
       def prepare_config(config)
         config.with_indifferent_access.tap do |conf|
           conf[:config][:timeout] ||= 30
@@ -55,7 +62,7 @@ module Multiwoven::Integrations::Source
       def parse_json(json_string)
         JSON.parse(json_string)
       rescue JSON::ParserError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:PARSE_JSON:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:PARSE_JSON:EXCEPTION", type: "error" })
         {}
       end
 
@@ -69,7 +76,7 @@ module Multiwoven::Integrations::Source
         )
         process_response(response)
       rescue StandardError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:RUN_MODEL:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:RUN_MODEL:EXCEPTION", type: "error" })
       end
 
       def run_model_stream(connection_config, payload)
@@ -83,7 +90,7 @@ module Multiwoven::Integrations::Source
           process_streaming_response(chunk) { |message| yield message if block_given? }
         end
       rescue StandardError => e
-        handle_exception(e, { context: "GENERIC OPEN AI:RUN_STREAM_MODEL:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:RUN_STREAM_MODEL:EXCEPTION", type: "error" })
       end
 
       def process_response(response)
@@ -91,10 +98,10 @@ module Multiwoven::Integrations::Source
           data = JSON.parse(response.body)
           [RecordMessage.new(data: data, emitted_at: Time.now.to_i).to_multiwoven_message]
         else
-          create_log_message("GENERIC OPEN AI:RUN_MODEL", "error", "request failed: #{response.body}")
+          create_log_message("#{log_context}:RUN_MODEL", "error", "request failed: #{response.body}")
         end
       rescue StandardError => e
-        handle_exception(e, { context: "OPEN AI:PROCESS_RESPONSE:EXCEPTION", type: "error" })
+        handle_exception(e, { context: "#{log_context}:PROCESS_RESPONSE:EXCEPTION", type: "error" })
       end
 
       def extract_data_entries(chunk)

data/lib/multiwoven/integrations/source/generic_open_ai/config/meta.json

@@ -5,12 +5,19 @@
     "connector_type": "source",
     "category": "AI Model",
     "sub_category": "LLM",
+    "provider_type": "third_party",
+    "display_order": 120,
+    "description": "Connect any endpoint that implements the OpenAI API specification.",
+    "show_payload_formats": true,
     "documentation_url": "https://docs.squared.ai/activation/ai-ml-sources/generic_open_ai-endpoint",
     "github_issue_label": "source-generic-open-ai-model",
     "icon": "icon.svg",
     "license": "MIT",
     "release_stage": "alpha",
     "support_level": "community",
-    "tags": ["language:ruby", "multiwoven"]
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
 }

data/lib/multiwoven/integrations/source/google_gemini/client.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Multiwoven::Integrations::Source
+  module GoogleGemini
+    include Multiwoven::Integrations::Core
+    # Gemini exposes an OpenAI-compatible chat completions API, so the request,
+    # streaming and response handling is identical to GenericOpenAI. Only the
+    # config/ directory differs, and BaseConnector#read_json resolves that from
+    # this class's own source location.
+    class Client < GenericOpenAI::Client
+      private
+
+      def log_context
+        "GOOGLE GEMINI"
+      end
+    end
+  end
+end

data/lib/multiwoven/integrations/source/google_gemini/config/catalog.json

@@ -0,0 +1,6 @@
+{
+  "request_rate_limit": 600,
+  "request_rate_limit_unit": "minute",
+  "request_rate_concurrency": 10,
+  "streams": []
+}

data/lib/multiwoven/integrations/source/google_gemini/config/meta.json

@@ -0,0 +1,23 @@
+{
+  "data": {
+    "name": "GoogleGemini",
+    "title": "Google Gemini",
+    "connector_type": "source",
+    "category": "AI Model",
+    "sub_category": "LLM",
+    "provider_type": "third_party",
+    "display_order": 70,
+    "openrouter_slug": "google",
+    "description": "Gemini model family from Google. Multimodal capabilities with strong performance on reasoning, code generation, and long-context tasks.",
+    "documentation_url": "https://ai.google.dev/gemini-api/docs/openai",
+    "github_issue_label": "source-google-gemini",
+    "icon": "icon.svg",
+    "license": "MIT",
+    "release_stage": "alpha",
+    "support_level": "community",
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
+  }
+}

data/lib/multiwoven/integrations/source/google_gemini/config/models.json

@@ -0,0 +1,26 @@
+{
+  "//models": "Sourced live from OpenRouter. Embedding models are curated here because OpenRouter lists none.",
+  "models": [
+    {
+      "id": "text-embedding-004",
+      "name": "Text Embedding 004",
+      "model_type": "embedding",
+      "type": "embedding",
+      "tasks": [
+        "Text Embeddings"
+      ],
+      "context_window": 2048,
+      "max_output": null,
+      "pricing": {
+        "input": 0.0,
+        "output": null,
+        "cached_read": null,
+        "cached_write": null,
+        "unit": "per_1m_tokens"
+      },
+      "capabilities": [
+        "Documents"
+      ]
+    }
+  ]
+}

data/lib/multiwoven/integrations/source/google_gemini/config/spec.json

@@ -0,0 +1,71 @@
+{
+  "documentation_url": "https://ai.google.dev/gemini-api/docs/openai",
+  "stream_type": "user_defined",
+  "connector_query_type": "ai_ml",
+  "connection_specification": {
+    "$schema": "http://json-schema.org/draft-07/schema#",
+    "title": "Google Gemini",
+    "type": "object",
+    "x-dynamic-input-path": "messages.0.content",
+    "x-output-path": "choices.0.message.content",
+    "required": [
+      "url",
+      "api_key",
+      "request_format",
+      "response_format"
+    ],
+    "properties": {
+      "url": {
+        "type": "string",
+        "title": "Endpoint URL",
+        "description": "The Gemini OpenAI-compatible chat completions endpoint.",
+        "default": "https://generativelanguage.googleapis.com/v1beta/openai/chat/completions",
+        "order": 0
+      },
+      "api_key": {
+        "type": "string",
+        "multiwoven_secret": true,
+        "title": "API Key",
+        "description": "Your Google AI Studio API key.",
+        "order": 1
+      },
+      "is_stream": {
+        "type": "boolean",
+        "title": "Enable streaming",
+        "description": "Enables data streaming when supported by the model. When true, responses are processed in chunks for immediate delivery. Default is false.",
+        "default": false,
+        "order": 2
+      },
+      "config": {
+        "title": "",
+        "type": "object",
+        "properties": {
+          "timeout": {
+            "type": "string",
+            "default": "30",
+            "title": "HTTP Timeout",
+            "description": "The maximum time, in seconds, to wait for a response from the server before the request is canceled.",
+            "order": 0
+          }
+        },
+        "order": 3
+      },
+      "request_format": {
+        "title": "Request Format",
+        "description": "Sample Request Format",
+        "type": "string",
+        "default": "{\"model\":\"gemini-2.0-flash\", \"messages\":[{\"role\": \"user\", \"content\": \"What do you know about APIs?\"}], \"stream\": false}",
+        "x-request-format": true,
+        "order": 4
+      },
+      "response_format": {
+        "title": "Response Format",
+        "description": "Sample Response Format",
+        "type": "string",
+        "default": "{\"choices\":[{\"index\": 0, \"message\":{\"role\": \"assistant\", \"content\": \"APIs let software applications communicate with each other.\"}, \"finish_reason\": \"stop\"}]}",
+        "x-response-format": true,
+        "order": 5
+      }
+    }
+  }
+}

data/lib/multiwoven/integrations/source/google_gemini/icon.svg

@@ -0,0 +1,10 @@
+<svg width="40" height="40" viewBox="0 0 40 40" fill="none" xmlns="http://www.w3.org/2000/svg">
+  <defs>
+    <linearGradient id="geminiSparkle" x1="6" y1="34" x2="34" y2="6" gradientUnits="userSpaceOnUse">
+      <stop offset="0" stop-color="#1C7DFF"/>
+      <stop offset="0.52" stop-color="#1C69FF"/>
+      <stop offset="1" stop-color="#F0DCD6"/>
+    </linearGradient>
+  </defs>
+  <path d="M20 4c1.1 6.2 5.8 10.9 12 12-6.2 1.1-10.9 5.8-12 12-1.1-6.2-5.8-10.9-12-12 6.2-1.1 10.9-5.8 12-12Z" fill="url(#geminiSparkle)"/>
+</svg>

data/lib/multiwoven/integrations/source/google_vertex_model/config/meta.json

@@ -5,13 +5,18 @@
     "connector_type": "source",
     "category": "AI Model",
     "sub_category": "AI_ML Service",
+    "provider_type": "cloud_provider",
+    "display_order": 90,
+    "description": "Access Gemini and other models via Google Vertex AI. Enterprise security, VPC deployment, and data residency controls.",
     "documentation_url": "https://docs.squared.ai/activation/ai-ml-sources/google_vertex-model",
     "github_issue_label": "source-vertex",
     "icon": "icon.svg",
     "license": "MIT",
     "release_stage": "alpha",
     "support_level": "community",
-    "tags": ["language:ruby", "multiwoven"]
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
 }
-  

data/lib/multiwoven/integrations/source/http_model/config/meta.json

@@ -5,12 +5,18 @@
     "connector_type": "source",
     "category": "AI Model",
     "sub_category": "AI_ML Service",
+    "provider_type": "cloud_provider",
+    "display_order": 110,
+    "description": "Connect any model exposed over HTTP by describing its request and response format.",
     "documentation_url": "https://docs.squared.ai/activation/ai-ml-sources/http-model-endpoint",
     "github_issue_label": "source-http-model",
     "icon": "icon.svg",
     "license": "MIT",
     "release_stage": "alpha",
     "support_level": "community",
-    "tags": ["language:ruby", "multiwoven"]
+    "tags": [
+      "language:ruby",
+      "multiwoven"
+    ]
   }
 }

data/lib/multiwoven/integrations/source/john_snow_labs/client.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Multiwoven::Integrations::Source
+  module JohnSnowLabs
+    include Multiwoven::Integrations::Core
+    # John Snow Labs exposes an OpenAI-compatible chat completions API, so the request,
+    # streaming and response handling is identical to GenericOpenAI. Only the
+    # config/ directory differs, and BaseConnector#read_json resolves that from
+    # this class's own source location.
+    class Client < GenericOpenAI::Client
+      private
+
+      def log_context
+        "JOHN SNOW LABS"
+      end
+    end
+  end
+end

data/lib/multiwoven/integrations/source/john_snow_labs/config/catalog.json

@@ -0,0 +1,6 @@
+{
+  "request_rate_limit": 600,
+  "request_rate_limit_unit": "minute",
+  "request_rate_concurrency": 10,
+  "streams": []
+}

data/lib/multiwoven/integrations/source/john_snow_labs/config/meta.json

@@ -0,0 +1,24 @@
+{
+  "data": {
+    "name": "JohnSnowLabs",
+    "title": "John Snow Labs",
+    "connector_type": "source",
+    "category": "AI Model",
+    "sub_category": "LLM",
+    "provider_type": "partner_provider",
+    "display_order": 50,
+    "description": "Spark NLP and healthcare AI models deployed and managed by AISquared. Clinical NLP, NER, and domain-specific models for life sciences and healthcare verticals.",
+    "managed_by_ais": true,
+    "documentation_url": "https://nlp.johnsnowlabs.com/",
+    "github_issue_label": "source-john-snow-labs",
+    "icon": "icon.svg",
+    "license": "MIT",
+    "release_stage": "alpha",
+    "support_level": "partner",
+    "tags": [
+      "language:ruby",
+      "multiwoven",
+      "partner:john-snow-labs"
+    ]
+  }
+}