sm-golem-plugin/classes/providers/anthropic.go

package providers

import (
	"encoding/json"
	"net/http"
	"regexp"
	"strings"

	"git.golem15.com/golem15/sm-golem-plugin/classes/valueobjects"
	"git.golem15.com/golem15/summercms/modules/fetchguard"
)

// Anthropic API headers PHP sends on every request.
const (
	AnthropicVersion = "2023-06-01"
	AnthropicBeta    = "files-api-2025-04-14"
)

// AnthropicAdapter is the Anthropic Messages API mapping (PHP
// AnthropicAdapter).
type AnthropicAdapter struct{}

var _ Adapter = AnthropicAdapter{}

var dataURL = regexp.MustCompile(`^data:([^;]+);base64,(.+)$`)

func anthropicHeaders(apiKey string) http.Header {
	return http.Header{
		"X-Api-Key":         {apiKey},
		"Anthropic-Version": {AnthropicVersion},
		"Anthropic-Beta":    {AnthropicBeta},
	}
}

// Headers are x-api-key, anthropic-version and anthropic-beta.
func (AnthropicAdapter) Headers(apiKey string) http.Header { return anthropicHeaders(apiKey) }

// ChatEndpoint is <base>/messages.
func (AnthropicAdapter) ChatEndpoint(baseURL string) string {
	return endpoint(baseURL, AnthropicBaseURL, "/messages")
}

// FilesEndpoint is <base>/files.
func (AnthropicAdapter) FilesEndpoint(baseURL string) string {
	return endpoint(baseURL, AnthropicBaseURL, "/files")
}

// ImageEndpoint is "": Anthropic generates no images.
func (AnthropicAdapter) ImageEndpoint(string) string { return "" }

// BuildChatPayload is the Messages body: model (the prompt's override or
// modelID), the messages with image_url blocks rewritten to image blocks
// (base64 for a data: URL, url otherwise) and file blocks to document
// blocks, the system prompt when set, max_tokens (the max_tokens option,
// else max_completion_tokens, else 4096) and every other option except
// response_format and max_completion_tokens.
func (AnthropicAdapter) BuildChatPayload(prompt valueobjects.Prompt, modelID string) *valueobjects.Payload {
	msgs := prompt.Messages()
	out := make([]map[string]any, 0, len(msgs))
	for _, m := range msgs {
		out = append(out, map[string]any{"role": m.Role, "content": anthropicContent(m)})
	}
	payload := valueobjects.NewPayload()
	model := modelID
	if p := prompt.Model(); p != nil {
		model = *p
	}
	payload.Set("model", model)
	payload.Set("messages", out)
	if s := prompt.SystemPrompt(); s != "" {
		payload.Set("system", s)
	}
	opts := prompt.Options()
	var maxTokens any = 4096
	if opts.IsSet("max_tokens") {
		maxTokens, _ = opts.Get("max_tokens")
	} else if opts.IsSet("max_completion_tokens") {
		maxTokens, _ = opts.Get("max_completion_tokens")
	}
	payload.Set("max_tokens", maxTokens)
	for _, k := range opts.Keys() {
		if k == "response_format" || k == "max_completion_tokens" {
			continue
		}
		v, _ := opts.Get(k)
		payload.Set(k, v)
	}
	return payload
}

func anthropicContent(m valueobjects.Message) any {
	if m.Parts == nil {
		return m.Text
	}
	blocks := make([]any, 0, len(m.Parts))
	for _, part := range m.Parts {
		switch part.Type {
		case "image_url":
			u := ""
			if part.ImageURL != nil {
				u = part.ImageURL.URL
			}
			if strings.HasPrefix(u, "data:") {
				media, data := "image/jpeg", ""
				if mm := dataURL.FindStringSubmatch(u); mm != nil {
					media, data = mm[1], mm[2]
				}
				blocks = append(blocks, map[string]any{"type": "image", "source": map[string]any{"type": "base64", "media_type": media, "data": data}})
			} else {
				blocks = append(blocks, map[string]any{"type": "image", "source": map[string]any{"type": "url", "url": u}})
			}
		case "file":
			id := ""
			if part.File != nil {
				id = part.File.FileID
			}
			blocks = append(blocks, map[string]any{"type": "document", "source": map[string]any{"type": "file", "file_id": id}})
		default:
			blocks = append(blocks, part)
		}
	}
	return blocks
}

// FileUploadFields is empty: the Anthropic files API takes the file only.
func (AnthropicAdapter) FileUploadFields(string) []fetchguard.FormField { return nil }

// FileUploadHeaders are the Messages headers.
func (AnthropicAdapter) FileUploadHeaders(apiKey string) http.Header {
	return anthropicHeaders(apiKey)
}

// ParseChatResponse reads content[0].text; usage gains PHP's
// prompt_tokens, completion_tokens and total_tokens next to Anthropic's
// input_tokens and output_tokens; Raw keeps the whole answer, stop_reason
// included.
func (AnthropicAdapter) ParseChatResponse(data map[string]any) valueobjects.AIResponse {
	if msg, ok := errorMessage(data); ok {
		return valueobjects.Failure(msg, data)
	}
	content := dig(data, "content", 0, "text")
	if content == nil {
		return valueobjects.Failure("No content in AI response.", data)
	}
	usage := map[string]any{}
	if u, ok := data["usage"].(map[string]any); ok {
		for k, v := range u {
			usage[k] = v
		}
	}
	if in, ok := usage["input_tokens"]; ok && in != nil {
		prompt := number(in)
		completion := number(usage["output_tokens"])
		usage["prompt_tokens"] = in
		if usage["output_tokens"] == nil {
			usage["completion_tokens"] = 0
		} else {
			usage["completion_tokens"] = usage["output_tokens"]
		}
		usage["total_tokens"] = prompt + completion
	}
	return valueobjects.Success(phpString(content), usage, data)
}

func number(v any) float64 {
	if f, ok := v.(float64); ok {
		return f
	}
	return 0
}

// ParseStreamChunk reads Anthropic's event stream: `event: message_stop`
// ends it, a `data:` line of type content_block_delta carries the text
// delta, anything else is skipped.
func (AnthropicAdapter) ParseStreamChunk(line string) (string, bool) {
	if line == "" {
		return "", false
	}
	if event, ok := strings.CutPrefix(line, "event: "); ok {
		return "", event == "message_stop"
	}
	payload, ok := strings.CutPrefix(line, "data: ")
	if !ok {
		return "", false
	}
	var v map[string]any
	if json.Unmarshal([]byte(payload), &v) != nil || len(v) == 0 {
		return "", false
	}
	if v["type"] != "content_block_delta" {
		return "", false
	}
	text, _ := dig(v, "delta", "text").(string)
	return text, false
}