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
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
use crate::operation::request::RequestTrait;
use crate::{
error::{DashScopeError, Result},
operation::{common::Parameters, embeddings::EmbeddingsParameters},
};
/// Defines the validation strategy for a given model.
///
/// This enum replaces the previous trait-based approach to resolve `dyn` compatibility issues.
/// It allows for static dispatch, which is more performant and avoids the complexities of
/// object safety.
pub enum ModelValidator {
/// The default validation, which performs no special checks.
Default,
/// Validation specific to the `deepseek-r1` model.
NotSupportResultFormatText,
/// validation enable_thinking
NotSupportEnableThinking,
NotSupportToolCall,
NotSupportJsonOutput,
// dimensions 不匹配
DimensionNotMatch,
OnlyStreaming,
}
pub trait Validator<T> {
/// 验证请求的参数
#[allow(clippy::result_large_err)]
fn validate<R: RequestTrait<P = T> + ?Sized>(&self, params: &R) -> Result<()>;
}
impl Validator<EmbeddingsParameters> for ModelValidator {
fn validate<R: RequestTrait<P = EmbeddingsParameters> + ?Sized>(
&self,
params: &R,
) -> Result<()> {
match self {
ModelValidator::DimensionNotMatch => {
// text-embedding-v4 向量维度 只能是以下值: 2,048、1,536、1,024（默认）、768、512、256、128、64
// text-embedding-v3 向量维度 只能是以下值: 1,024（默认）、768、512、256、128或64
// text-embedding-v2 向量维度 只能是1,536
// text-embedding-v1 向量维度 只能是1,536
if let Some(p) = params.parameters() {
let valid_dimensions = match params.model() {
"text-embedding-v1" | "text-embedding-v2" => {
vec![1536]
}
"text-embedding-v3" => {
vec![1024, 768, 512, 256, 128, 64]
}
"text-embedding-v4" => {
vec![2048, 1536, 1024, 768, 512, 256, 128, 64]
}
_ => vec![], // 未知模型不验证
};
if let Some(dimension) = p.dimension {
if !valid_dimensions.contains(&dimension) {
return Err(DashScopeError::InvalidArgument(format!(
"Invalid dimension: {} for model: {}",
dimension,
params.model()
)));
}
}
}
Ok(())
}
_ => Ok(()),
}
}
}
impl Validator<Parameters> for ModelValidator {
fn validate<R: RequestTrait<P = Parameters> + ?Sized>(&self, params: &R) -> Result<()> {
match self {
ModelValidator::Default => {
// No specific validation rules for the default case.
Ok(())
}
ModelValidator::NotSupportResultFormatText => {
// The deepseek-r1 model does not support `result_format: "text"`.
if let Some(p) = params.parameters() {
if let Some(format) = &p.result_format {
if format == "text" {
return Err(DashScopeError::InvalidArgument(
"deepseek-r1 does not support result_format = text".into(),
));
}
}
}
Ok(())
}
ModelValidator::NotSupportEnableThinking => {
if let Some(p) = params.parameters() {
if let Some(thinking) = p.enable_thinking {
if thinking {
return Err(DashScopeError::InvalidArgument(
"The model does not support enable_thinking = true".into(),
));
}
}
}
Ok(())
}
ModelValidator::NotSupportJsonOutput => {
if let Some(p) = params.parameters() {
if let Some(response_format) = p.response_format.as_ref() {
if response_format.type_ == "json_object" {
return Err(DashScopeError::InvalidArgument(
"The model does not support response_format=json_object".into(),
));
}
}
}
Ok(())
}
ModelValidator::NotSupportToolCall => {
if let Some(p) = params.parameters() {
if p.tools.is_some() {
return Err(DashScopeError::InvalidArgument(
"The model does not support tool call".into(),
));
}
}
Ok(())
}
ModelValidator::OnlyStreaming => {
if let Some(p) = params.parameters() {
#[allow(deprecated)]
if p.incremental_output == Some(false) {
return Err(DashScopeError::InvalidArgument(
"The model does not support streaming".into(),
));
}
}
Ok(())
}
_ => Ok(()),
}
}
}
/// Selects the appropriate validator for the given model name.
///
/// # Arguments
///
/// * `model` - The name of the model as a string slice.
///
/// # Returns
///
/// A `ModelValidator` enum variant corresponding to the required validation strategy.
pub(crate) fn check_model_parameters(model: &str) -> Vec<ModelValidator> {
match model {
"deepseek-r1" => vec![
ModelValidator::NotSupportResultFormatText,
ModelValidator::NotSupportJsonOutput,
],
"qwen-vl" | "qwen-audio" => vec![ModelValidator::NotSupportToolCall],
"Moonshot-Kimi-K2-Instruct" => vec![
ModelValidator::NotSupportEnableThinking,
ModelValidator::NotSupportResultFormatText,
ModelValidator::NotSupportJsonOutput,
ModelValidator::NotSupportToolCall,
],
"text-embedding-v4" | "text-embedding-v3" | "text-embedding-v2" | "text-embedding-v1" => {
vec![ModelValidator::DimensionNotMatch]
}
"qwen-mt-image" => {
vec![ModelValidator::Default]
}
"glm-4.6" | "glm-4.5" | "glm-4.5-air" => {
vec![ModelValidator::OnlyStreaming]
}
_ => vec![ModelValidator::Default],
}
}