1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
//! Anthropic request building
use crate::llm::{ContentPart, LlmError, Message, ToolChoice, ToolDefinition};
use super::types::*;
impl super::ChatAnthropic {
/// Build an Anthropic API request
pub(super) fn build_request(
&self,
messages: Vec<Message>,
tools: Option<Vec<ToolDefinition>>,
tool_choice: Option<ToolChoice>,
stream: bool,
) -> Result<AnthropicRequest, LlmError> {
// Separate system message from other messages
let mut system: Option<String> = None;
let mut anthropic_messages: Vec<AnthropicMessage> = Vec::new();
for message in messages {
match message {
Message::System(s) => {
system = Some(s.content);
}
Message::User(u) => {
let content: Vec<AnthropicContent> = u
.content
.into_iter()
.map(|c| match c {
ContentPart::Text(t) => AnthropicContent::Text {
text: t.text,
content_type: "text".to_string(),
},
ContentPart::Image(img) => {
let (media_type, data) = if img.image_url.url.starts_with("data:") {
// Parse data URL
let parts: Vec<&str> =
img.image_url.url.splitn(2, ',').collect();
let mime = parts[0]
.strip_prefix("data:")
.and_then(|s| s.strip_suffix(";base64"))
.unwrap_or("image/png");
(mime.to_string(), parts.get(1).unwrap_or(&"").to_string())
} else {
("image/png".to_string(), img.image_url.url.clone())
};
AnthropicContent::Image {
source: AnthropicImageSource {
source_type: "base64".to_string(),
media_type,
data,
},
content_type: "image".to_string(),
}
}
_ => AnthropicContent::Text {
text: "[Unsupported content type]".to_string(),
content_type: "text".to_string(),
},
})
.collect();
anthropic_messages.push(AnthropicMessage {
role: "user".to_string(),
content,
});
}
Message::Assistant(a) => {
let mut content = Vec::new();
if let Some(t) = a.thinking {
content.push(AnthropicContent::Text {
text: t,
content_type: "thinking".to_string(),
});
}
if let Some(c) = a.content {
content.push(AnthropicContent::Text {
text: c,
content_type: "text".to_string(),
});
}
for tc in a.tool_calls {
let input: serde_json::Value = serde_json::from_str(&tc.function.arguments)
.unwrap_or(serde_json::json!({}));
content.push(AnthropicContent::ToolUse {
id: tc.id,
name: tc.function.name,
input,
content_type: "tool_use".to_string(),
});
}
anthropic_messages.push(AnthropicMessage {
role: "assistant".to_string(),
content,
});
}
Message::Tool(t) => {
anthropic_messages.push(AnthropicMessage {
role: "user".to_string(),
content: vec![AnthropicContent::ToolResult {
tool_use_id: t.tool_call_id,
content: t.content,
content_type: "tool_result".to_string(),
}],
});
}
Message::Developer(d) => {
system = Some(d.content);
}
}
}
let anthropic_tools = tools.map(|ts| {
ts.into_iter()
.map(|t| AnthropicTool {
name: t.name,
description: t.description,
input_schema: t.parameters,
})
.collect()
});
let tool_choice_value = tool_choice.map(|tc| match tc {
ToolChoice::Auto => serde_json::json!({"type": "auto"}),
ToolChoice::Required => serde_json::json!({"type": "any"}),
ToolChoice::None => serde_json::json!({"type": "none"}),
ToolChoice::Named(name) => serde_json::json!({"type": "tool", "name": name}),
});
let thinking_config = if self.thinking && self.supports_thinking() {
Some(ThinkingConfig {
thinking_type: "enabled".to_string(),
budget_tokens: self.thinking_budget,
})
} else {
None
};
Ok(AnthropicRequest {
model: self.model.clone(),
max_tokens: self.max_tokens,
messages: anthropic_messages,
system,
tools: anthropic_tools,
tool_choice: tool_choice_value,
temperature: Some(self.temperature),
stream: if stream { Some(true) } else { None },
thinking: thinking_config,
})
}
}