Move AI SDK to opensource (#483)

## Summary
Copies over the code for AI SDK to our open source repo. Once this is
in, I will delete the AI SDK from Axiom so that it can start using the
open source version.

## How was it tested?
Ran the unit tests.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

Author: loreto

.golangci.yml

@@ -42,6 +42,12 @@ issues:
       linters:
         - dupl
 
+    # Disable only cognitive-complexity check in test files
+    - path: _test\.go
+      linters:
+        - revive
+      text: "cognitive-complexity:"
+
     # TODO(gcurtis): temporary until this file is used.
     - path: terminal.*\.go
       linters:
@@ -60,7 +66,7 @@ linters-settings:
       - name: bool-literal-in-expr
       - name: cognitive-complexity
         arguments:
-          - 27
+          - 30
       - name: datarace
       - name: duplicated-imports
       - name: early-return

aisdk/ai/aisdk.go

@@ -0,0 +1,199 @@
+package aisdk
+
+import (
+	"context"
+
+	"go.jetify.com/ai/api"
+)
+
+// GenerateText generates a text response for a given prompt using a language model.
+// This function does not stream its output.
+//
+// It returns a [api.Response] containing the generated text, the results of
+// any tool calls, and additional information.
+//
+// It supports either a string argument, which will be converted to a
+// [api.UserMessage] with
+//
+//	GenerateText(ctx, "Hello, world!")
+//
+// Or a series of [api.Message] arguments:
+//
+//	GenerateText(ctx,
+//	  UserMessage("Show me a picture of a cat"),
+//	  AssistantMessage(
+//	    "Here is a picture of a cat",
+//	    ImageBlock{URL: "https://example.com/cat.png"},
+//	  ),
+//	)
+//
+// The last argument can optionally be a series of [api.CallOption] arguments:
+//
+//	GenerateText(ctx, "Hello, world!", WithMaxTokens(100))
+func GenerateText(ctx context.Context, args ...any) (api.Response, error) {
+	llmArgs, err := toLLMArgs(args...)
+	if err != nil {
+		return api.Response{}, err
+	}
+
+	return generate(ctx, llmArgs.Prompt, llmArgs.Config)
+}
+
+func generate(ctx context.Context, prompt []api.Message, config GenerateTextConfig) (api.Response, error) {
+	return config.Model.Generate(ctx, prompt, config.CallOptions)
+}
+
+type GenerateTextConfig struct {
+	CallOptions api.CallOptions
+	Model       api.LanguageModel
+}
+
+// GenerateOption is a function that modifies GenerateConfig.
+type GenerateOption func(*GenerateTextConfig)
+
+// WithModel sets the language model to use for generation
+func WithModel(model api.LanguageModel) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.Model = model
+	}
+}
+
+// WithMaxTokens specifies the maximum number of tokens to generate
+func WithMaxTokens(maxTokens int) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.MaxTokens = maxTokens
+	}
+}
+
+// WithTemperature controls randomness in the model's output.
+// It is recommended to set either Temperature or TopP, but not both.
+func WithTemperature(temperature float64) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.Temperature = &temperature
+	}
+}
+
+// WithStopSequences specifies sequences that will stop generation when produced.
+// Providers may have limits on the number of stop sequences.
+func WithStopSequences(stopSequences ...string) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.StopSequences = stopSequences
+	}
+}
+
+// WithTopP controls nucleus sampling.
+// It is recommended to set either Temperature or TopP, but not both.
+func WithTopP(topP float64) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.TopP = topP
+	}
+}
+
+// WithTopK limits sampling to the top K options for each token.
+// Used to remove "long tail" low probability responses.
+// Recommended for advanced use cases only.
+func WithTopK(topK int) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.TopK = topK
+	}
+}
+
+// WithPresencePenalty affects the likelihood of the model repeating
+// information that is already in the prompt
+func WithPresencePenalty(penalty float64) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.PresencePenalty = penalty
+	}
+}
+
+// WithFrequencyPenalty affects the likelihood of the model
+// repeatedly using the same words or phrases
+func WithFrequencyPenalty(penalty float64) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.FrequencyPenalty = penalty
+	}
+}
+
+// WithResponseFormat specifies whether the output should be text or JSON.
+// For JSON output, a schema can optionally guide the model.
+func WithResponseFormat(format *api.ResponseFormat) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.ResponseFormat = format
+	}
+}
+
+// WithSeed provides an integer seed for random sampling.
+// If supported by the model, calls will generate deterministic results.
+func WithSeed(seed int) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.Seed = seed
+	}
+}
+
+// WithHeaders specifies additional HTTP headers to send with the request.
+// Only applicable for HTTP-based providers.
+func WithHeaders(headers map[string]string) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.Headers = headers
+	}
+}
+
+// WithInputFormat specifies whether the user provided the input as messages or as a prompt.
+// This can help guide non-chat models in the expansion, as different expansions
+// may be needed for chat vs non-chat use cases.
+func WithInputFormat(format api.InputFormat) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.InputFormat = format
+	}
+}
+
+// WithMode affects the behavior of the language model. It is required to
+// support provider-independent streaming and generation of structured objects.
+// The model can take this information and e.g. configure json mode, the correct
+// low level grammar, etc. It can also be used to optimize the efficiency of the
+// streaming, e.g. tool-delta stream parts are only needed in the
+// object-tool mode.
+//
+// Mode will be removed in v2, and at that point it will be deprecated.
+// All necessary settings will be directly supported through the call settings,
+// in particular responseFormat, toolChoice, and tools.
+func WithMode(mode api.ModeConfig) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.Mode = mode
+	}
+}
+
+// WithProviderMetadata sets additional provider-specific metadata.
+// The metadata is passed through to the provider from the AI SDK and enables
+// provider-specific functionality that can be fully encapsulated in the provider.
+func WithProviderMetadata(providerName string, metadata any) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		if o.CallOptions.ProviderMetadata == nil {
+			o.CallOptions.ProviderMetadata = api.NewProviderMetadata(map[string]any{})
+		}
+		o.CallOptions.ProviderMetadata.Set(providerName, metadata)
+	}
+}
+
+// WithTools specifies the tools available to the model during generation.
+func WithTools(tools ...api.ToolDefinition) GenerateOption {
+	return func(o *GenerateTextConfig) {
+		o.CallOptions.Mode = api.RegularMode{Tools: tools}
+	}
+}
+
+// buildGenerateConfig combines multiple generate options into a single GenerateConfig struct.
+func buildGenerateConfig(opts []GenerateOption) GenerateTextConfig {
+	config := GenerateTextConfig{
+		CallOptions: api.CallOptions{
+			InputFormat:      "prompt",          // default
+			Mode:             api.RegularMode{}, // default
+			ProviderMetadata: api.NewProviderMetadata(map[string]any{}),
+		},
+		Model: DefaultLanguageModel(),
+	}
+	for _, opt := range opts {
+		opt(&config)
+	}
+	return config
+}