codepro-ia 1.0.0

package/src/provider/transform.ts

@@ -0,0 +1,705 @@
+import type { APICallError, ModelMessage } from "ai"
+import { unique } from "remeda"
+import type { JSONSchema } from "zod/v4/core"
+import type { Provider } from "./provider"
+import type { ModelsDev } from "./models"
+import { iife } from "@/util/iife"
+
+type Modality = NonNullable<ModelsDev.Model["modalities"]>["input"][number]
+
+function mimeToModality(mime: string): Modality | undefined {
+  if (mime.startsWith("image/")) return "image"
+  if (mime.startsWith("audio/")) return "audio"
+  if (mime.startsWith("video/")) return "video"
+  if (mime === "application/pdf") return "pdf"
+  return undefined
+}
+
+export namespace ProviderTransform {
+  // Maps npm package to the key the AI SDK expects for providerOptions
+  function sdkKey(npm: string): string | undefined {
+    switch (npm) {
+      case "@ai-sdk/github-copilot":
+      case "@ai-sdk/openai":
+      case "@ai-sdk/azure":
+        return "openai"
+      case "@ai-sdk/amazon-bedrock":
+        return "bedrock"
+      case "@ai-sdk/anthropic":
+        return "anthropic"
+      case "@ai-sdk/google-vertex":
+      case "@ai-sdk/google":
+        return "google"
+      case "@ai-sdk/gateway":
+        return "gateway"
+      case "@openrouter/ai-sdk-provider":
+        return "openrouter"
+    }
+    return undefined
+  }
+
+  function normalizeMessages(
+    msgs: ModelMessage[],
+    model: Provider.Model,
+    options: Record<string, unknown>,
+  ): ModelMessage[] {
+    // Anthropic rejects messages with empty content - filter out empty string messages
+    // and remove empty text/reasoning parts from array content
+    if (model.api.npm === "@ai-sdk/anthropic") {
+      msgs = msgs
+        .map((msg) => {
+          if (typeof msg.content === "string") {
+            if (msg.content === "") return undefined
+            return msg
+          }
+          if (!Array.isArray(msg.content)) return msg
+          const filtered = msg.content.filter((part) => {
+            if (part.type === "text" || part.type === "reasoning") {
+              return part.text !== ""
+            }
+            return true
+          })
+          if (filtered.length === 0) return undefined
+          return { ...msg, content: filtered }
+        })
+        .filter((msg): msg is ModelMessage => msg !== undefined && msg.content !== "")
+    }
+
+    if (model.api.id.includes("claude")) {
+      return msgs.map((msg) => {
+        if ((msg.role === "assistant" || msg.role === "tool") && Array.isArray(msg.content)) {
+          msg.content = msg.content.map((part) => {
+            if ((part.type === "tool-call" || part.type === "tool-result") && "toolCallId" in part) {
+              return {
+                ...part,
+                toolCallId: part.toolCallId.replace(/[^a-zA-Z0-9_-]/g, "_"),
+              }
+            }
+            return part
+          })
+        }
+        return msg
+      })
+    }
+    if (model.providerID === "mistral" || model.api.id.toLowerCase().includes("mistral")) {
+      const result: ModelMessage[] = []
+      for (let i = 0; i < msgs.length; i++) {
+        const msg = msgs[i]
+        const nextMsg = msgs[i + 1]
+
+        if ((msg.role === "assistant" || msg.role === "tool") && Array.isArray(msg.content)) {
+          msg.content = msg.content.map((part) => {
+            if ((part.type === "tool-call" || part.type === "tool-result") && "toolCallId" in part) {
+              // Mistral requires alphanumeric tool call IDs with exactly 9 characters
+              const normalizedId = part.toolCallId
+                .replace(/[^a-zA-Z0-9]/g, "") // Remove non-alphanumeric characters
+                .substring(0, 9) // Take first 9 characters
+                .padEnd(9, "0") // Pad with zeros if less than 9 characters
+
+              return {
+                ...part,
+                toolCallId: normalizedId,
+              }
+            }
+            return part
+          })
+        }
+
+        result.push(msg)
+
+        // Fix message sequence: tool messages cannot be followed by user messages
+        if (msg.role === "tool" && nextMsg?.role === "user") {
+          result.push({
+            role: "assistant",
+            content: [
+              {
+                type: "text",
+                text: "Done.",
+              },
+            ],
+          })
+        }
+      }
+      return result
+    }
+
+    if (
+      model.capabilities.interleaved &&
+      typeof model.capabilities.interleaved === "object" &&
+      model.capabilities.interleaved.field === "reasoning_content"
+    ) {
+      return msgs.map((msg) => {
+        if (msg.role === "assistant" && Array.isArray(msg.content)) {
+          const reasoningParts = msg.content.filter((part: any) => part.type === "reasoning")
+          const reasoningText = reasoningParts.map((part: any) => part.text).join("")
+
+          // Filter out reasoning parts from content
+          const filteredContent = msg.content.filter((part: any) => part.type !== "reasoning")
+
+          // Include reasoning_content directly on the message for all assistant messages
+          if (reasoningText) {
+            return {
+              ...msg,
+              content: filteredContent,
+              providerOptions: {
+                ...msg.providerOptions,
+                openaiCompatible: {
+                  ...(msg.providerOptions as any)?.openaiCompatible,
+                  reasoning_content: reasoningText,
+                },
+              },
+            }
+          }
+
+          return {
+            ...msg,
+            content: filteredContent,
+          }
+        }
+
+        return msg
+      })
+    }
+
+    return msgs
+  }
+
+  function applyCaching(msgs: ModelMessage[], providerID: string): ModelMessage[] {
+    const system = msgs.filter((msg) => msg.role === "system").slice(0, 2)
+    const final = msgs.filter((msg) => msg.role !== "system").slice(-2)
+
+    const providerOptions = {
+      anthropic: {
+        cacheControl: { type: "ephemeral" },
+      },
+      openrouter: {
+        cacheControl: { type: "ephemeral" },
+      },
+      bedrock: {
+        cachePoint: { type: "ephemeral" },
+      },
+      openaiCompatible: {
+        cache_control: { type: "ephemeral" },
+      },
+    }
+
+    for (const msg of unique([...system, ...final])) {
+      const shouldUseContentOptions = providerID !== "anthropic" && Array.isArray(msg.content) && msg.content.length > 0
+
+      if (shouldUseContentOptions) {
+        const lastContent = msg.content[msg.content.length - 1]
+        if (lastContent && typeof lastContent === "object") {
+          lastContent.providerOptions = {
+            ...lastContent.providerOptions,
+            ...providerOptions,
+          }
+          continue
+        }
+      }
+
+      msg.providerOptions = {
+        ...msg.providerOptions,
+        ...providerOptions,
+      }
+    }
+
+    return msgs
+  }
+
+  function unsupportedParts(msgs: ModelMessage[], model: Provider.Model): ModelMessage[] {
+    return msgs.map((msg) => {
+      if (msg.role !== "user" || !Array.isArray(msg.content)) return msg
+
+      const filtered = msg.content.map((part) => {
+        if (part.type !== "file" && part.type !== "image") return part
+
+        // Check for empty base64 image data
+        if (part.type === "image") {
+          const imageStr = part.image.toString()
+          if (imageStr.startsWith("data:")) {
+            const match = imageStr.match(/^data:([^;]+);base64,(.*)$/)
+            if (match && (!match[2] || match[2].length === 0)) {
+              return {
+                type: "text" as const,
+                text: "ERROR: Image file is empty or corrupted. Please provide a valid image.",
+              }
+            }
+          }
+        }
+
+        const mime = part.type === "image" ? part.image.toString().split(";")[0].replace("data:", "") : part.mediaType
+        const filename = part.type === "file" ? part.filename : undefined
+        const modality = mimeToModality(mime)
+        if (!modality) return part
+        if (model.capabilities.input[modality]) return part
+
+        const name = filename ? `"${filename}"` : modality
+        return {
+          type: "text" as const,
+          text: `ERROR: Cannot read ${name} (this model does not support ${modality} input). Inform the user.`,
+        }
+      })
+
+      return { ...msg, content: filtered }
+    })
+  }
+
+  export function message(msgs: ModelMessage[], model: Provider.Model, options: Record<string, unknown>) {
+    msgs = unsupportedParts(msgs, model)
+    msgs = normalizeMessages(msgs, model, options)
+    if (
+      model.providerID === "anthropic" ||
+      model.api.id.includes("anthropic") ||
+      model.api.id.includes("claude") ||
+      model.api.npm === "@ai-sdk/anthropic"
+    ) {
+      msgs = applyCaching(msgs, model.providerID)
+    }
+
+    // Remap providerOptions keys from stored providerID to expected SDK key
+    const key = sdkKey(model.api.npm)
+    if (key && key !== model.providerID && model.api.npm !== "@ai-sdk/azure") {
+      const remap = (opts: Record<string, any> | undefined) => {
+        if (!opts) return opts
+        if (!(model.providerID in opts)) return opts
+        const result = { ...opts }
+        result[key] = result[model.providerID]
+        delete result[model.providerID]
+        return result
+      }
+
+      msgs = msgs.map((msg) => {
+        if (!Array.isArray(msg.content)) return { ...msg, providerOptions: remap(msg.providerOptions) }
+        return {
+          ...msg,
+          providerOptions: remap(msg.providerOptions),
+          content: msg.content.map((part) => ({ ...part, providerOptions: remap(part.providerOptions) })),
+        } as typeof msg
+      })
+    }
+
+    return msgs
+  }
+
+  export function temperature(model: Provider.Model) {
+    const id = model.id.toLowerCase()
+    if (id.includes("qwen")) return 0.55
+    if (id.includes("claude")) return undefined
+    if (id.includes("gemini")) return 1.0
+    if (id.includes("glm-4.6")) return 1.0
+    if (id.includes("glm-4.7")) return 1.0
+    if (id.includes("minimax-m2")) return 1.0
+    if (id.includes("kimi-k2")) {
+      if (id.includes("thinking")) return 1.0
+      return 0.6
+    }
+    return undefined
+  }
+
+  export function topP(model: Provider.Model) {
+    const id = model.id.toLowerCase()
+    if (id.includes("qwen")) return 1
+    if (id.includes("minimax-m2")) {
+      return 0.95
+    }
+    if (id.includes("gemini")) return 0.95
+    return undefined
+  }
+
+  export function topK(model: Provider.Model) {
+    const id = model.id.toLowerCase()
+    if (id.includes("minimax-m2")) {
+      if (id.includes("m2.1")) return 40
+      return 20
+    }
+    if (id.includes("gemini")) return 64
+    return undefined
+  }
+
+  const WIDELY_SUPPORTED_EFFORTS = ["low", "medium", "high"]
+  const OPENAI_EFFORTS = ["none", "minimal", ...WIDELY_SUPPORTED_EFFORTS, "xhigh"]
+
+  export function variants(model: Provider.Model): Record<string, Record<string, any>> {
+    if (!model.capabilities.reasoning) return {}
+
+    const id = model.id.toLowerCase()
+    if (id.includes("deepseek") || id.includes("minimax") || id.includes("glm") || id.includes("mistral")) return {}
+
+    switch (model.api.npm) {
+      case "@openrouter/ai-sdk-provider":
+        if (!model.id.includes("gpt") && !model.id.includes("gemini-3") && !model.id.includes("grok-4")) return {}
+        return Object.fromEntries(OPENAI_EFFORTS.map((effort) => [effort, { reasoning: { effort } }]))
+
+      // TODO: YOU CANNOT SET max_tokens if this is set!!!
+      case "@ai-sdk/gateway":
+        return Object.fromEntries(OPENAI_EFFORTS.map((effort) => [effort, { reasoningEffort: effort }]))
+
+      case "@ai-sdk/cerebras":
+      // https://v5.ai-sdk.dev/providers/ai-sdk-providers/cerebras
+      case "@ai-sdk/togetherai":
+      // https://v5.ai-sdk.dev/providers/ai-sdk-providers/togetherai
+      case "@ai-sdk/xai":
+      // https://v5.ai-sdk.dev/providers/ai-sdk-providers/xai
+      case "@ai-sdk/deepinfra":
+      // https://v5.ai-sdk.dev/providers/ai-sdk-providers/deepinfra
+      case "@ai-sdk/openai-compatible":
+        return Object.fromEntries(WIDELY_SUPPORTED_EFFORTS.map((effort) => [effort, { reasoningEffort: effort }]))
+
+      case "@ai-sdk/azure":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/azure
+        if (id === "o1-mini") return {}
+        const azureEfforts = ["low", "medium", "high"]
+        if (id.includes("gpt-5-") || id === "gpt-5") {
+          azureEfforts.unshift("minimal")
+        }
+        return Object.fromEntries(
+          azureEfforts.map((effort) => [
+            effort,
+            {
+              reasoningEffort: effort,
+              reasoningSummary: "auto",
+              include: ["reasoning.encrypted_content"],
+            },
+          ]),
+        )
+      case "@ai-sdk/openai":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/openai
+        if (id === "gpt-5-pro") return {}
+        const openaiEfforts = iife(() => {
+          if (id.includes("codex")) {
+            if (id.includes("5.2")) return [...WIDELY_SUPPORTED_EFFORTS, "xhigh"]
+            return WIDELY_SUPPORTED_EFFORTS
+          }
+          const arr = [...WIDELY_SUPPORTED_EFFORTS]
+          if (id.includes("gpt-5-") || id === "gpt-5") {
+            arr.unshift("minimal")
+          }
+          if (model.release_date >= "2025-11-13") {
+            arr.unshift("none")
+          }
+          if (model.release_date >= "2025-12-04") {
+            arr.push("xhigh")
+          }
+          return arr
+        })
+        return Object.fromEntries(
+          openaiEfforts.map((effort) => [
+            effort,
+            {
+              reasoningEffort: effort,
+              reasoningSummary: "auto",
+              include: ["reasoning.encrypted_content"],
+            },
+          ]),
+        )
+
+      case "@ai-sdk/anthropic":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/anthropic
+        return {
+          high: {
+            thinking: {
+              type: "enabled",
+              budgetTokens: 16000,
+            },
+          },
+          max: {
+            thinking: {
+              type: "enabled",
+              budgetTokens: 31999,
+            },
+          },
+        }
+
+      case "@ai-sdk/amazon-bedrock":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/amazon-bedrock
+        // For Anthropic models on Bedrock, use reasoningConfig with budgetTokens
+        if (model.api.id.includes("anthropic")) {
+          return {
+            high: {
+              reasoningConfig: {
+                type: "enabled",
+                budgetTokens: 16000,
+              },
+            },
+            max: {
+              reasoningConfig: {
+                type: "enabled",
+                budgetTokens: 31999,
+              },
+            },
+          }
+        }
+
+        // For Amazon Nova models, use reasoningConfig with maxReasoningEffort
+        return Object.fromEntries(
+          WIDELY_SUPPORTED_EFFORTS.map((effort) => [
+            effort,
+            {
+              reasoningConfig: {
+                type: "enabled",
+                maxReasoningEffort: effort,
+              },
+            },
+          ]),
+        )
+
+      case "@ai-sdk/google-vertex":
+      // https://v5.ai-sdk.dev/providers/ai-sdk-providers/google-vertex
+      case "@ai-sdk/google":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/google-generative-ai
+        if (id.includes("2.5")) {
+          return {
+            high: {
+              thinkingConfig: {
+                includeThoughts: true,
+                thinkingBudget: 16000,
+              },
+            },
+            max: {
+              thinkingConfig: {
+                includeThoughts: true,
+                thinkingBudget: 24576,
+              },
+            },
+          }
+        }
+        return Object.fromEntries(
+          ["low", "high"].map((effort) => [
+            effort,
+            {
+              includeThoughts: true,
+              thinkingLevel: effort,
+            },
+          ]),
+        )
+
+      case "@ai-sdk/mistral":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/mistral
+        return {}
+
+      case "@ai-sdk/cohere":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/cohere
+        return {}
+
+      case "@ai-sdk/groq":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/groq
+        const groqEffort = ["none", ...WIDELY_SUPPORTED_EFFORTS]
+        return Object.fromEntries(
+          groqEffort.map((effort) => [
+            effort,
+            {
+              includeThoughts: true,
+              thinkingLevel: effort,
+            },
+          ]),
+        )
+
+      case "@ai-sdk/perplexity":
+        // https://v5.ai-sdk.dev/providers/ai-sdk-providers/perplexity
+        return {}
+    }
+    return {}
+  }
+
+  export function options(input: {
+    model: Provider.Model
+    sessionID: string
+    providerOptions?: Record<string, any>
+  }): Record<string, any> {
+    const result: Record<string, any> = {}
+
+    // openai and providers using openai package should set store to false by default.
+    if (input.model.providerID === "openai" || input.model.api.npm === "@ai-sdk/openai") {
+      result["store"] = false
+    }
+
+    if (input.model.api.npm === "@openrouter/ai-sdk-provider") {
+      result["usage"] = {
+        include: true,
+      }
+      if (input.model.api.id.includes("gemini-3")) {
+        result["reasoning"] = { effort: "high" }
+      }
+    }
+
+    if (
+      input.model.providerID === "baseten" ||
+      (input.model.providerID === "opencode" && ["kimi-k2-thinking", "glm-4.6"].includes(input.model.api.id))
+    ) {
+      result["chat_template_args"] = { enable_thinking: true }
+    }
+
+    if (["zai", "zhipuai"].includes(input.model.providerID) && input.model.api.npm === "@ai-sdk/openai-compatible") {
+      result["thinking"] = {
+        type: "enabled",
+        clear_thinking: false,
+      }
+    }
+
+    if (input.model.providerID === "openai" || input.providerOptions?.setCacheKey) {
+      result["promptCacheKey"] = input.sessionID
+    }
+
+    if (input.model.api.npm === "@ai-sdk/google" || input.model.api.npm === "@ai-sdk/google-vertex") {
+      result["thinkingConfig"] = {
+        includeThoughts: true,
+      }
+      if (input.model.api.id.includes("gemini-3")) {
+        result["thinkingConfig"]["thinkingLevel"] = "high"
+      }
+    }
+
+    if (input.model.api.id.includes("gpt-5") && !input.model.api.id.includes("gpt-5-chat")) {
+      if (input.model.providerID.includes("codex")) {
+        result["store"] = false
+      }
+
+      if (!input.model.api.id.includes("codex") && !input.model.api.id.includes("gpt-5-pro")) {
+        result["reasoningEffort"] = "medium"
+      }
+
+      if (input.model.api.id.endsWith("gpt-5.") && input.model.providerID !== "azure") {
+        result["textVerbosity"] = "low"
+      }
+
+      if (input.model.providerID.startsWith("opencode")) {
+        result["promptCacheKey"] = input.sessionID
+        result["include"] = ["reasoning.encrypted_content"]
+        result["reasoningSummary"] = "auto"
+      }
+    }
+    return result
+  }
+
+  export function smallOptions(model: Provider.Model) {
+    if (model.providerID === "openai" || model.api.id.includes("gpt-5")) {
+      if (model.api.id.includes("5.")) {
+        return { reasoningEffort: "low" }
+      }
+      return { reasoningEffort: "minimal" }
+    }
+    if (model.providerID === "google") {
+      // gemini-3 uses thinkingLevel, gemini-2.5 uses thinkingBudget
+      if (model.api.id.includes("gemini-3")) {
+        return { thinkingConfig: { thinkingLevel: "minimal" } }
+      }
+      return { thinkingConfig: { thinkingBudget: 0 } }
+    }
+    if (model.providerID === "openrouter") {
+      if (model.api.id.includes("google")) {
+        return { reasoning: { enabled: false } }
+      }
+      return { reasoningEffort: "minimal" }
+    }
+    return {}
+  }
+
+  export function providerOptions(model: Provider.Model, options: { [x: string]: any }) {
+    const key = sdkKey(model.api.npm) ?? model.providerID
+    return { [key]: options }
+  }
+
+  export function maxOutputTokens(
+    npm: string,
+    options: Record<string, any>,
+    modelLimit: number,
+    globalLimit: number,
+  ): number {
+    const modelCap = modelLimit || globalLimit
+    const standardLimit = Math.min(modelCap, globalLimit)
+
+    if (npm === "@ai-sdk/anthropic") {
+      const thinking = options?.["thinking"]
+      const budgetTokens = typeof thinking?.["budgetTokens"] === "number" ? thinking["budgetTokens"] : 0
+      const enabled = thinking?.["type"] === "enabled"
+      if (enabled && budgetTokens > 0) {
+        // Return text tokens so that text + thinking <= model cap, preferring 32k text when possible.
+        if (budgetTokens + standardLimit <= modelCap) {
+          return standardLimit
+        }
+        return modelCap - budgetTokens
+      }
+    }
+
+    return standardLimit
+  }
+
+  export function schema(model: Provider.Model, schema: JSONSchema.BaseSchema) {
+    /*
+    if (["openai", "azure"].includes(providerID)) {
+      if (schema.type === "object" && schema.properties) {
+        for (const [key, value] of Object.entries(schema.properties)) {
+          if (schema.required?.includes(key)) continue
+          schema.properties[key] = {
+            anyOf: [
+              value as JSONSchema.JSONSchema,
+              {
+                type: "null",
+              },
+            ],
+          }
+        }
+      }
+    }
+    */
+
+    // Convert integer enums to string enums for Google/Gemini
+    if (model.providerID === "google" || model.api.id.includes("gemini")) {
+      const sanitizeGemini = (obj: any): any => {
+        if (obj === null || typeof obj !== "object") {
+          return obj
+        }
+
+        if (Array.isArray(obj)) {
+          return obj.map(sanitizeGemini)
+        }
+
+        const result: any = {}
+        for (const [key, value] of Object.entries(obj)) {
+          if (key === "enum" && Array.isArray(value)) {
+            // Convert all enum values to strings
+            result[key] = value.map((v) => String(v))
+            // If we have integer type with enum, change type to string
+            if (result.type === "integer" || result.type === "number") {
+              result.type = "string"
+            }
+          } else if (typeof value === "object" && value !== null) {
+            result[key] = sanitizeGemini(value)
+          } else {
+            result[key] = value
+          }
+        }
+
+        // Filter required array to only include fields that exist in properties
+        if (result.type === "object" && result.properties && Array.isArray(result.required)) {
+          result.required = result.required.filter((field: any) => field in result.properties)
+        }
+
+        if (result.type === "array" && result.items == null) {
+          result.items = {}
+        }
+
+        return result
+      }
+
+      schema = sanitizeGemini(schema)
+    }
+
+    return schema
+  }
+
+  export function error(providerID: string, error: APICallError) {
+    let message = error.message
+    if (providerID.includes("github-copilot") && error.statusCode === 403) {
+      return "Please reauthenticate with the copilot provider to ensure your credentials work properly with OpenCode."
+    }
+    if (providerID.includes("github-copilot") && message.includes("The requested model is not supported")) {
+      return (
+        message +
+        "\n\nMake sure the model is enabled in your copilot settings: https://github.com/settings/copilot/features"
+      )
+    }
+
+    return message
+  }
+}