Files
ddddocr-rs/ddddocr-ort/src/ocr/session.rs
CNWei a3c4614574 refactor(core): 提炼公共类型
- 将 AxisDim、TensorInfo 等公共类型下沉至 ddddocr_core::types
- 项目结构优化
2026-07-30 16:55:58 +08:00

151 lines
5.9 KiB
Rust
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

use crate::types::Session;
use ddddocr_core::ModelMetadata;
use ddddocr_core::OcrOutput;
use ddddocr_core::error::{DdddError, Result, TensorError};
use ddddocr_core::traits::{InferenceEngine, Info, OcrEngine};
use ddddocr_core::types::{AxisDim, ModelInfo, TensorInfo, TensorType};
use ddddocr_core::utils::normalize_ocr_logits;
use ort::inputs;
use ort::value::{TensorElementType, TensorRef};
use std::sync::Mutex;
// 引入核心层的统一错误类型
/// 明确命名为 AxisDim代表模型某一个轴的维度特征
// #[derive(Clone, PartialEq, Eq)]
// pub enum AxisDim {
// /// 静态固定维度(如通道数固定为 1高度固定为 64
// Static(usize),
// /// 动态符号维度(如宽度是动态的 "image_width"
// Dynamic(String),
// }
// impl AxisDim {
// /// 便捷方法:判断是否为动态维度
// pub fn is_dynamic(&self) -> bool {
// matches!(self, AxisDim::Dynamic(_))
// }
// }
// /// 自定义 Debug 格式化输出,彻底融化套娃外壳,保证日志干净漂亮
// impl std::fmt::Debug for AxisDim {
// fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
// match self {
// AxisDim::Static(size) => write!(f, "{}", size),
// AxisDim::Dynamic(expr) => write!(f, "Dynamic(\"{}\")", expr),
// }
// }
// }
/// 模拟 Python 的 input_info 和 output_info 结构
// #[derive(Debug, Clone)]
// pub struct TensorInfo {
// pub name: String,
// pub shape: Vec<AxisDim>, // 既包含 Fixed 静态维度,也包含 Dynamic 动态符号
// pub data_type: TensorElementType, // 对应 Python 的 type
// }
//
// /// 最终返回的模型完整信息
// #[derive(Debug, Clone)]
// pub struct ModelInfo {
// pub inputs: Vec<TensorInfo>,
// pub outputs: Vec<TensorInfo>,
// /// 硬件执行提供者(采用 Option 兼容不同底层的推理引擎)
// pub providers: Option<Vec<String>>,
// }
pub struct OcrRuntime {
pub session: Session,
pub metadata: ModelMetadata,
}
impl OcrRuntime {
pub fn new(session: Session, metadata: ModelMetadata) -> Self {
Self { session, metadata }
}
}
impl OcrEngine for OcrRuntime {
fn metadata(&self) -> &ModelMetadata {
&self.metadata
}
}
impl InferenceEngine for OcrRuntime {
type Output = OcrOutput;
/// 对应 Python 的 _inference
fn inference(&self, input_array: ndarray::Array4<f32>) -> Result<Self::Output, TensorError> {
// tract 的 run 会返回一个 Vec<TValue>,我们通常只需要第一个输出
// let result = self.ocr.run(tvec!(tensor.into()))?;
// let tensor = Tensor::from(input_array);
let mut session_guard = self
.session
.lock()
.map_err(|_| TensorError::Engine("获取 Session 锁失败 (Poisoned)".to_string()))?;
let result = session_guard
.run(inputs![TensorRef::from_array_view(&input_array).map_err(
|e| TensorError::Engine(format!("构建输入失败: {e}"))
)?])
.map_err(|e| TensorError::Engine(format!("执行模型推理失败: {e}")))?;
// .context("执行模型推理失败")?;
println!("模型输出原始数据: {:?}", result);
// Ok(result.swap_remove(0).into_tensor())
let raw_value = &result[0];
match raw_value.dtype().tensor_type().unwrap() {
TensorElementType::Int64 => {
let (array_d, slice) = raw_value
.try_extract_tensor::<i64>()
.map_err(|_| TensorError::Engine("Tract 无法获取 i64 内存视图".to_string()))?;
// .context("Tract 无法获取 i64 内存视图")?;
// 提前提取真实维度
let actual_shape = array_d
.to_vec()
.iter()
.map(|v| *v as usize)
.collect::<Vec<usize>>();
let view = ndarray::ArrayViewD::from_shape(actual_shape.as_slice(), slice)
.map_err(|_| TensorError::Engine("构建 ndarray ArrayViewD 失败".to_string()))?;
// 转成标准的 Array1 传给 core
let array1 = view
.to_owned()
.into_dimensionality::<ndarray::Ix1>()
.map_err(|_| TensorError::DimensionMismatch {
expected: "1D 字符索引静态矩阵".to_string(),
actual: actual_shape,
})?;
Ok(OcrOutput::Indices(array1))
}
TensorElementType::Float32 => {
let shape = raw_value.shape();
println!("模型输出shape数据: {:?}", shape);
// raw_tensor.to_plain_array_view()
let (shape_ref, slice) = raw_value
.try_extract_tensor::<f32>()
.map_err(|_| TensorError::Engine("Tract 无法获取 f32 内存视图".to_string()))?;
// 1. 极其纯粹的、无拷贝的多维 Shape 压扁清洗
let shape_vec: Vec<usize> =
shape_ref.to_vec().iter().map(|v| *v as usize).collect();
let shape_vec_slice = shape_vec.as_slice();
let view = ndarray::ArrayViewD::from_shape(shape_vec_slice, slice)
.map_err(|_| TensorError::Engine("构建 ndarray ArrayViewD 失败".to_string()))?;
normalize_ocr_logits(view, shape_vec_slice)
}
_ => Err(
// anyhow::anyhow!("不支持的模型输出数据类型: {:?}",raw_tensor.datum_type())
TensorError::UnknownOutputFormat,
),
}
}
}
impl Info for OcrRuntime {
fn input_info(&self) -> Result<Vec<TensorInfo>> {
todo!()
}
fn output_info(&self) -> Result<Vec<TensorInfo>> {
todo!()
}
fn model_info(&self) -> Result<ModelInfo> {
todo!()
}
}