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
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
//! Edit operations.

use std::cmp;
use std::fmt::Debug;
use std::hash::Hash;

use {Measure, SeqPair};

pub mod archetype;

/// Trait for sequence edit operations.
pub trait Operation<T>: Clone + Debug + Eq + Hash {
    /// Return the cell after backtracking from the given cell with this operation.
    ///
    /// Must return `None` if backtracking is not possible (e.g. would lead
    /// to an invalid cell). This method is used for the construction of
    /// traces and edit scripts.
    fn backtrack(
        &self,
        seq_pair: &SeqPair<T>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<(usize, usize)>;

    /// Compute the cost after applying the operation.
    ///
    /// Returns `None` if the operation cannot be applied. Otherwise, it
    /// returns the cost for the alignment at `source_idx`, `target_idx`
    /// using this operation.
    fn cost(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<usize>
    where
        T: Eq;
}

///An indexed edit operation.
///
/// Indexed edit operations are a pairing of an edit operation and the
/// sequence positions when/where the operation was applied. The indexes
/// can be used to simplify external use of the operations. For example,
/// if we are interested in which elements were aligned, then a
/// subsequence of matches
///
/// * *match 1 2*
/// * *match 3 7*
/// * *match 4 10*
///
/// Tells us that indices 1/2, 3/7, and 4/10 of the source/target sequence
/// were aligned.
#[derive(Clone, Debug, Eq, Hash, PartialEq)]
pub struct IndexedOperation<O>
where
    O: Debug,
{
    operation: O,
    source_idx: usize,
    target_idx: usize,
}

impl<'a, O> IndexedOperation<O>
where
    O: Debug,
{
    pub fn new(operation: O, source_idx: usize, target_idx: usize) -> Self {
        IndexedOperation {
            operation,
            source_idx,
            target_idx,
        }
    }

    pub fn operation(&self) -> &O {
        &self.operation
    }

    pub fn source_idx(&self) -> usize {
        self.source_idx
    }

    pub fn target_idx(&self) -> usize {
        self.target_idx
    }
}

pub(crate) trait Backtrack<T> {
    type Operation: Operation<T>;

    fn backtrack(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<Self::Operation>
    where
        T: Eq;

    fn backtracks(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Vec<Self::Operation>
    where
        T: Eq;
}

impl<M, T> Backtrack<T> for M
where
    M: Measure<T>,
{
    type Operation = M::Operation;

    /// Give the operation that was used to construct the cost matrix cell
    /// at (`source_idx`, `target_idx`).
    fn backtrack(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<Self::Operation>
    where
        T: Eq,
    {
        for op in self.operations() {
            if let Some(cost) = op.cost(seq_pair, cost_matrix, source_idx, target_idx) {
                if cost == cost_matrix[source_idx][target_idx] {
                    return Some(op.clone());
                }
            }
        }

        None
    }

    /// Give the operations that were used to construct the cost matrix cell
    /// at (`source_idx`, `taget_idx`).
    fn backtracks(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Vec<Self::Operation>
    where
        T: Eq,
    {
        let mut ops = Vec::new();

        for op in self.operations() {
            if let Some(cost) = op.cost(seq_pair, cost_matrix, source_idx, target_idx) {
                if cost == cost_matrix[source_idx][target_idx] {
                    ops.push(op.clone());
                }
            }
        }

        ops
    }
}

pub(crate) trait BestCost<T> {
    type Operation: Operation<T>;

    fn best_cost(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<usize>
    where
        T: Eq;
}

impl<M, T> BestCost<T> for M
where
    M: Measure<T>,
{
    type Operation = M::Operation;

    /// Compute the cost of the best operation.
    ///
    /// Returns `None` if the operation cannot be applied. Otherwise, it
    /// returns the cost for the alignment at `source_idx`, `target_idx`
    /// using this operation.
    fn best_cost(
        &self,
        seq_pair: &SeqPair<T>,
        cost_matrix: &Vec<Vec<usize>>,
        source_idx: usize,
        target_idx: usize,
    ) -> Option<usize>
    where
        T: Eq,
    {
        // Used filter_map + min_by_key before, but the compiler is not able
        // to optimize to a better loop.
        let mut best = None;
        for op in self.operations() {
            if let Some(cost) = op.cost(seq_pair, cost_matrix, source_idx, target_idx) {
                best = best.map(|best_cost| cmp::min(cost, best_cost))
                    .or(Some(cost))
            }
        }

        best
    }
}