Add lingua support for system message with user content arrays

crates/coverage-report/src/requests_expected_differences.json

@@ -218,6 +218,20 @@
       "skip": true,
       "reason": "Anthropic extracts system messages to separate system parameter"
     },
+    {
+      "testCase": "systemMessageArrayContent",
+      "source": "ChatCompletions",
+      "target": "Anthropic",
+      "skip": true,
+      "reason": "Anthropic extracts system messages to separate system parameter"
+    },
+    {
+      "testCase": "systemMessageArrayContent",
+      "source": "Responses",
+      "target": "Anthropic",
+      "skip": true,
+      "reason": "Anthropic extracts system messages to separate system parameter"
+    },
     {
       "testCase": "multimodalRequest",
       "source": "Responses",

crates/lingua/src/providers/anthropic/adapter.rs

@@ -17,7 +17,7 @@ use crate::providers::anthropic::params::AnthropicParams;
 use crate::providers::anthropic::try_parse_anthropic;
 use crate::serde_json::{self, Map, Value};
 use crate::universal::convert::TryFromLLM;
-use crate::universal::message::{Message, UserContent};
+use crate::universal::message::{Message, UserContent, UserContentPart};
 use crate::universal::tools::{tools_to_anthropic_value, UniversalTool};
 use crate::universal::transform::extract_system_messages;
 use crate::universal::{
@@ -147,10 +147,20 @@ impl ProviderAdapter for AnthropicAdapter {
         // Add system message if present
         if !system_contents.is_empty() {
             let system_text: String = system_contents
-                .iter()
-                .filter_map(|c| match c {
-                    UserContent::String(s) => Some(s.as_str()),
-                    _ => None,
+                .into_iter()
+                .map(|c| match c {
+                    UserContent::String(s) => s,
+                    UserContent::Array(parts) => parts
+                        .into_iter()
+                        .filter_map(|p| {
+                            if let UserContentPart::Text(t) = p {
+                                Some(t.text)
+                            } else {
+                                None
+                            }
+                        })
+                        .collect::<Vec<_>>()
+                        .join("\n"),
                 })
                 .collect::<Vec<_>>()
                 .join("\n\n");

payloads/cases/advanced.ts

@@ -395,4 +395,73 @@ export const advancedCases: TestCaseCollection = {
       },
     },
   },
+
+  systemMessageArrayContent: {
+    "chat-completions": {
+      model: OPENAI_CHAT_COMPLETIONS_MODEL,
+      messages: [
+        {
+          role: "system",
+          content: [
+            {
+              type: "text",
+              text: "You are a helpful data analyst. The default data source is project_logs with id abc-123.",
+            },
+          ],
+        },
+        {
+          role: "user",
+          content: "What errors occurred recently?",
+        },
+      ],
+      max_completion_tokens: 300,
+    },
+    responses: {
+      model: OPENAI_RESPONSES_MODEL,
+      instructions:
+        "You are a helpful data analyst. The default data source is project_logs with id abc-123.",
+      input: [
+        {
+          role: "user",
+          content: "What errors occurred recently?",
+        },
+      ],
+      max_output_tokens: 300,
+    },
+    anthropic: {
+      model: ANTHROPIC_MODEL,
+      max_tokens: 300,
+      system: [
+        {
+          type: "text",
+          text: "You are a helpful data analyst. The default data source is project_logs with id abc-123.",
+        },
+      ],
+      messages: [
+        {
+          role: "user",
+          content: "What errors occurred recently?",
+        },
+      ],
+    },
+    google: {
+      systemInstruction: {
+        parts: [
+          {
+            text: "You are a helpful data analyst. The default data source is project_logs with id abc-123.",
+          },
+        ],
+      },
+      contents: [
+        {
+          role: "user",
+          parts: [{ text: "What errors occurred recently?" }],
+        },
+      ],
+      config: {
+        maxOutputTokens: 300,
+      },
+    },
+    bedrock: null,
+  },
 };

payloads/snapshots/systemMessageArrayContent/anthropic/followup-request.json

@@ -0,0 +1,29 @@
+{
+  "model": "claude-sonnet-4-20250514",
+  "max_tokens": 300,
+  "system": [
+    {
+      "type": "text",
+      "text": "You are a helpful data analyst. The default data source is project_logs with id abc-123."
+    }
+  ],
+  "messages": [
+    {
+      "role": "user",
+      "content": "What errors occurred recently?"
+    },
+    {
+      "role": "assistant",
+      "content": [
+        {
+          "type": "text",
+          "text": "I'll help you find recent errors from the project logs. Let me query the data to identify error events.\n\n```sql\nSELECT \n    timestamp,\n    level,\n    message,\n    source,\n    error_code\nFROM project_logs \nWHERE level IN ('ERROR', 'CRITICAL', 'FATAL')\n    AND timestamp >= DATETIME_SUB(CURRENT_TIMESTAMP(), INTERVAL 24 HOURS)\nORDER BY timestamp DESC\nLIMIT 20;\n```\n\nThis query will show:\n- **Recent errors** from the last 24 hours\n- **Error details** including timestamp, severity level, and error messages\n- **Source information** to help identify where errors originated\n- **Error codes** if available\n\nWould you like me to:\n1. Extend the time range (e.g., last week)?\n2. Filter by specific error types or sources?\n3. Group errors by frequency to show the most common issues?\n4. Include additional context like user IDs or request IDs if available?\n\nPlease let me know if you need the query adjusted or if you'd like me to analyze the results in a specific way!"
+        }
+      ]
+    },
+    {
+      "role": "user",
+      "content": "What should I do next?"
+    }
+  ]
+}