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

use crate::internal::*;
use ndarray::*;

#[derive(Debug, Clone, new, Hash)]
pub struct Gather {
    pub axis: usize,
}
impl_dyn_hash!(Gather);

impl Op for Gather {
    fn name(&self) -> Cow<str> {
        "Gather".into()
    }

    op_as_typed_op!();
}

impl Gather {
    pub fn compute_output_shape<D: DimLike>(
        &self,
        input_shape: &[D],
        indices_shape: &[D],
    ) -> TractResult<TVec<D>> {
        let mut output_shape = tvec![];
        for (idx, dim) in input_shape.iter().enumerate() {
            if idx != self.axis {
                output_shape.push(dim.clone());
            } else {
                for idx2 in indices_shape {
                    output_shape.push(idx2.clone());
                }
            }
        }
        Ok(output_shape)
    }

    unsafe fn eval_t<T: Datum>(&self, data: TValue, indices: &TValue) -> TractResult<TValue> {
        let data_view = data.to_array_view_unchecked::<T>();
        let indices = indices.cast_to::<i64>()?;
        if indices.shape().len() == 0 {
            let mut index = *indices.to_scalar::<i64>()?;
            if index < 0 {
                index += data_view.shape()[0] as i64;
            }
            let mut tensor =
                data_view.index_axis(Axis(self.axis), index as usize).to_owned().into_tensor();
            tensor.set_datum_type(data.datum_type());
            return Ok(tensor.into_tvalue());
        }

        let mut output = Tensor::uninitialized_dt(
            data.datum_type(),
            &self.compute_output_shape(data.shape(), indices.shape())?,
        )?;
        let mut view = output.to_array_view_mut_unchecked::<T>();
        for (indices_coords, indices_value) in indices.to_array_view::<i64>()?.indexed_iter() {
            let mut to_update = view.index_axis_mut(Axis(self.axis), indices_coords[0]);
            for idx in 1..indices_coords.ndim() {
                to_update = to_update.index_axis_move(Axis(0), indices_coords[idx]);
            }
            let index_value = if *indices_value >= 0 {
                *indices_value
            } else {
                indices_value + data_view.shape()[self.axis] as i64
            } as usize;
            anyhow::ensure!(index_value < data_view.shape()[self.axis]);
            to_update.assign(&data_view.index_axis(Axis(self.axis), index_value));
        }
        Ok(output.into_tvalue())
    }
}

impl TypedOp for Gather {
    as_op!();

    fn output_facts(&self, inputs: &[&TypedFact]) -> TractResult<TVec<TypedFact>> {
        Ok(tvec!(inputs[0].datum_type.fact(
            &*self.compute_output_shape(&inputs[0].shape.to_tvec(), &inputs[1].shape.to_tvec())?
        )))
    }

    fn declutter(
        &self,
        model: &TypedModel,
        node: &TypedNode,
    ) -> TractResult<Option<TypedModelPatch>> {
        let indices_fact = model.outlet_fact(node.inputs[1])?;
        if let Some(indices) = indices_fact.konst.as_ref() {
            if indices.len() == 1 {
                let mut patch = TypedModelPatch::default();
                let mut wire = patch.tap_model(model, node.inputs[0])?;
                let index = indices.cast_to_scalar::<i64>()?;
                let index = if index < 0 {
                    let data_fact = model.outlet_fact(node.inputs[0])?;
                    data_fact.shape[self.axis].clone() + index.to_dim()
                } else {
                    index.to_dim()
                };
                wire = patch.wire_node(
                    format!("{}.slice", node.name),
                    crate::ops::array::Slice {
                        axis: self.axis,
                        start: index.clone(),
                        end: index + 1,
                    },
                    &[wire],
                )?[0];
                wire = patch.wire_node(
                    format!("{}.rm_axis", node.name),
                    crate::ops::change_axes::AxisOp::Rm(self.axis),
                    &[wire],
                )?[0];
                patch.shunt_outside(model, node.id.into(), wire)?;
                return Ok(Some(patch));
            }
        }
        Ok(None)
    }
}

impl EvalOp for Gather {
    fn is_stateless(&self) -> bool {
        true
    }

    fn eval(&self, mut inputs: TVec<TValue>) -> TractResult<TVec<TValue>> {
        let (data, indices) = args_2!(inputs);
        unsafe {
            Ok(tvec!(dispatch_datum_by_size!(Self::eval_t(data.datum_type())(
                self, data, &indices
            ))?))
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_should_gather_scalar_index() {
        let data = Tensor::from(arr1(&[1i64, 2, 3]));
        let gatherer = Gather::new(0);
        for idx in 2..3 {
            let index = Tensor::from(arr0(idx));
            let outputs =
                gatherer.eval(tvec![data.clone().into_tvalue(), index.into_tvalue()]).unwrap();
            let output = &outputs[0];
            assert_eq!(output.shape().len(), 0);
            assert_eq!(*output.to_scalar::<i64>().unwrap(), idx + 1);
        }
    }
}