use crate::samplerplustree::boundingbox::BoundingBox;

#[repr(C)]
#[derive(Clone)]
pub struct DiVector {
    pub high: Vec<f64>,
    pub low: Vec<f64>,
}

impl DiVector {
    pub fn empty(dimension: usize) -> Self {
        DiVector {
            high: vec![0.0; dimension],
            low: vec![0.0; dimension],
        }
    }

    pub fn new(high: &[f64], low: &[f64]) -> Self {
        assert!(high.len() == low.len(), " incorrect lengths");
        DiVector {
            high: Vec::from(high),
            low: Vec::from(low),
        }
    }

    pub fn assign_as_probability_of_cut(&mut self, bounding_box: &BoundingBox, point: &[f32]) {
        let minsum: f64 = self
            .low
            .iter_mut()
            .zip(bounding_box.get_min_values())
            .zip(point)
            .map(|((x, &y), &z)| {
                if y - z > 0.0 {
                    *x = (y - z) as f64;
                    *x
                } else {
                    *x = 0.0;
                    *x
                }
            })
            .sum();
        let maxsum: f64 = self
            .high
            .iter_mut()
            .zip(point)
            .zip(bounding_box.get_max_values())
            .map(|((x, &y), &z)| {
                if y - z > 0.0 {
                    *x = (y - z) as f64;
                    *x
                } else {
                    *x = 0.0;
                    *x
                }
            })
            .sum();

        let sum = minsum + maxsum;
        if sum != 0.0 {
            self.scale(1.0 / (bounding_box.get_range_sum() + sum));
        }
    }

    pub fn assign_as_probability_of_cut_with_missing_coordinates(
        &mut self,
        bounding_box: &BoundingBox,
        point: &[f32],
        missing_coordinates: &[bool],
    ) {
        let minsum: f64 = self
            .low
            .iter_mut()
            .zip(bounding_box.get_min_values())
            .zip(point)
            .zip(missing_coordinates)
            .map(|(((x, &y), &z), &b)| {
                if !b && y - z > 0.0 {
                    *x = (y - z) as f64;
                    *x
                } else {
                    *x = 0.0;
                    *x
                }
            })
            .sum();
        let maxsum: f64 = self
            .high
            .iter_mut()
            .zip(point)
            .zip(bounding_box.get_max_values())
            .zip(missing_coordinates)
            .map(|(((x, &y), &z), &b)| {
                if !b && y - z > 0.0 {
                    *x = (y - z) as f64;
                    *x
                } else {
                    *x = 0.0;
                    *x
                }
            })
            .sum();

        let sum = minsum + maxsum;
        if sum != 0.0 {
            self.scale(1.0 / (bounding_box.get_range_sum() + sum));
        }
    }

    pub fn assign(&mut self, other: &DiVector) {
        for (x, &y) in self.high.iter_mut().zip(&other.high) {
            *x = y;
        }
        for (x, &y) in self.low.iter_mut().zip(&other.low) {
            *x = y;
        }
    }

    pub fn add_from(&mut self, other: &DiVector, factor: f64) {
        other.add_to_scaled(self, factor);
    }

    pub fn add_to(&self, other: &mut DiVector) {
        for (x, &y) in other.high.iter_mut().zip(&self.high) {
            *x += y;
        }
        for (x, &y) in other.low.iter_mut().zip(&self.low) {
            *x += y;
        }
    }

    pub fn add_to_scaled(&self, other: &mut DiVector, factor: f64) {
        for (x, &y) in other.high.iter_mut().zip(&self.high) {
            *x += y * factor;
        }
        for (x, &y) in other.low.iter_mut().zip(&self.low) {
            *x += y * factor;
        }
    }

    pub fn divide(&mut self, num: usize) {
        self.scale(1.0 / num as f64)
    }

    pub fn scale(&mut self, factor: f64) {
        for x in self.high.iter_mut() {
            *x *= factor;
        }
        for x in self.low.iter_mut() {
            *x *= factor;
        }
    }

    pub fn total(&self) -> f64 {
        self.high.iter().sum::<f64>() + self.low.iter().sum::<f64>()
    }

    pub fn normalize(&mut self, value: f64) {
        let current = self.total();
        if current <= 0.0 {
            let v = value / (2.0 * self.high.len() as f64);
            for x in self.high.iter_mut() {
                *x = v;
            }
            for x in self.low.iter_mut() {
                *x = v;
            }
        } else {
            self.scale(value / current);
        }
    }

    pub fn dimensions(&self) -> usize {
        self.high.len()
    }

    pub fn high_low_sum(&self, index: usize) -> f64 {
        self.high[index] + self.low[index]
    }

    pub fn max_contribution(&self, base_dimension: usize) -> usize {
        let mut val = 0.0;
        let mut index = 0;
        for i in 0..base_dimension {
            val += self.high_low_sum(i);
        }
        for j in 1..(self.dimensions() / base_dimension) {
            let mut sum = 0.0;
            for i in 0..base_dimension {
                sum += self.high_low_sum(j * base_dimension + i);
            }
            if sum > val {
                val = sum;
                index = j;
            }
        }
        index
    }
}