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
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
//! Floyd-Warshall algorithm.

use crate::result::FloydWarshallResult;
use safe_graph::{Graph, NodeTrait};
use num_traits::Num;
use std::clone::Clone;
use std::cmp::Ord;
use std::cmp::Ordering::{Greater, Less};
use std::cmp::PartialOrd;

/// A trait group for `FloydWarshall`'s weighted edges.
pub trait FloydWarshallTrait: Copy + Num + PartialOrd {}

/// Implement the `FloydWarshallTrait` for all types satisfying bounds.
impl<F> FloydWarshallTrait for F where F: Copy + Num + PartialOrd {}

/// Floyd-Warshall algorithm structure.
///
/// # `FloydWarshall` algorithm is parameterized over:
///
/// - Number type `F` giving a weights to edges.
pub struct FloydWarshall<F: FloydWarshallTrait> {
    /// Operator to be used for weighted edges.
    op: Box<Fn(F, F) -> F>,
    /// Comparison to be used for weighted paths to determine if a newly tested path through `k`
    /// should be added or not.
    ///
    /// # Examples
    ///
    /// ```
    /// use std::cmp::Ordering::{Greater, Less};
    /// use std::cmp::PartialOrd;
    ///
    /// let cmp: Box<Fn(f32, f32) -> bool> = Box::new(|x: f32, y: f32| x.partial_cmp(&y)
    ///   .unwrap_or(Greater) == Less);
    ///
    /// let new_weight = 4.2;
    /// let old_weight = 10.1;
    ///
    /// if cmp(new_weight, old_weight) {
    ///   // Use the new weight and path through `k`.
    /// }
    /// ```
    cmp: Box<Fn(F, F) -> bool>,
    /// Discard loops (e.g. edges starting and ending in the same node) from calculation.
    discard_loops: bool,
}

impl<F: FloydWarshallTrait> FloydWarshall<F> {
    /// Create a new instance of FloydWarshall structure with default settings.
    ///
    /// # Examples
    ///
    /// ```
    /// use floyd_warshall_alg::FloydWarshall;
    ///
    /// let alg: FloydWarshall<f32> = FloydWarshall::new();
    /// ```
    pub fn new() -> Self {
        // Initialize defaults.
        let add = Box::new(|x: F, y: F| x + y);
        let sharp_less = Box::new(|x: F, y: F| x.partial_cmp(&y).unwrap_or(Greater) == Less);

        Self {
            op: add,
            cmp: sharp_less,
            discard_loops: true,
        }
    }

    /// Create a new instance of FloydWarshall structure with customized settings.
    ///
    /// You can set:
    /// - the `op` (operator) to be used for weighted edges
    /// - the `cmp` (comparison) to be used for weighted paths
    ///
    /// # Examples
    ///
    /// ```
    /// use floyd_warshall_alg::FloydWarshall;
    /// use std::cmp::Ordering::{Greater, Less};
    ///
    /// let mul = Box::new(|x: f32, y: f32| x * y);
    /// let sharp_greater = Box::new(|x: f32, y: f32| x.partial_cmp(&y).unwrap_or(Less) == Greater);
    ///
    /// let alg: FloydWarshall<f32> = FloydWarshall::new_customized(mul, sharp_greater);
    /// ```
    pub fn new_customized(op: Box<Fn(F, F) -> F>, cmp: Box<Fn(F, F) -> bool>) -> Self {
        Self::new_fully_customized(op, cmp, true)
    }

    /// Create a new instance of FloydWarshall structure with customized settings.
    ///
    /// You can set:
    /// - the `op` (operator) to be used for weighted edges
    /// - the `cmp` (comparison) to be used for weighted paths
    /// - the `discard_loops` to discard loops (e.g. edges starting and ending in the same node)
    ///   from calculation.
    ///
    /// # Examples
    ///
    /// ```
    /// use floyd_warshall_alg::FloydWarshall;
    /// use std::cmp::Ordering::{Greater, Less};
    ///
    /// let mul = Box::new(|x: f32, y: f32| x * y);
    /// let sharp_greater = Box::new(|x: f32, y: f32| x.partial_cmp(&y).unwrap_or(Less) == Greater);
    /// let discard_loops = false;
    ///
    /// let alg: FloydWarshall<f32> = FloydWarshall::new_customized(mul, sharp_greater);
    pub fn new_fully_customized(
        op: Box<Fn(F, F) -> F>,
        cmp: Box<Fn(F, F) -> bool>,
        discard_loops: bool,
    ) -> Self {
        Self {
            op,
            cmp,
            discard_loops,
        }
    }

    /// Find all the shortest paths (or best rated paths based on algorithm customized settings).
    ///
    /// The result of type FloydWarshallResult holds both:
    /// - best rates for all possible paths
    /// - next node on the best rated (shortest) path for each possible path
    ///
    /// # Method is parameterized over:
    ///
    /// - Node index / label `N`.
    /// - Number type `E` giving a weight to edges.
    pub fn find_paths<N>(&self, graph: &Graph<N, F>) -> FloydWarshallResult<N, F>
    where
        N: NodeTrait,
    {
        let mut path: Graph<N, F> = graph.clone();
        let mut next: Graph<N, N> = Graph::with_capacity(graph.node_count(), graph.edge_count());

        // Initialize next steps of each edge existing in `graph` with its end node.
        for (a, b, _) in graph.all_edges() {
            next.add_edge(a, b, b);
        }

        // `k` is the "intermediate" node which is currently considered.
        for k in graph.nodes() {
            // `i` is a starting node of the path we try to optimize.
            for i in graph.nodes() {
                // `j` is an end node of the path we try to optimize.
                for j in graph.nodes() {
                    // Skip calculation for loops if requested.
                    if self.discard_loops && !Self::unique(vec![k, i, j]) {
                        continue;
                    }

                    // Calculation of a new weight of the path from `i` to `j` through `k`.
                    let left_operand = path.edge_weight(i, k);
                    let right_operand = path.edge_weight(k, j);

                    // There's nothing to calculate if the left `(i, k)` or right `(k, j)` path misses.
                    if left_operand.is_none() || right_operand.is_none() {
                        continue;
                    }

                    // It is safe to unwrap the operands now.
                    let left_operand = left_operand.unwrap();
                    let right_operand = right_operand.unwrap();

                    // Use the weight operator. It can be customized.
                    let new_weight = (self.op)(*left_operand, *right_operand);

                    // The `(i, j)` path already exists.
                    if let Some(&old_weight) = path.edge_weight(i, j) {
                        // Use the comparison. It can be customized.
                        if (self.cmp)(new_weight, old_weight) {
                            path.add_edge(i, j, new_weight);

                            // The algorithm invariants guarantee the edge exists.
                            let direction = next.edge_weight(i, k).unwrap();
                            next.add_edge(i, j, *direction);
                        }
                    } else {
                        // The path was missing so add a new one.
                        path.add_edge(i, j, new_weight);

                        // The algorithm invariants guarantee the edge exists.
                        let direction = next.edge_weight(i, k).unwrap();
                        next.add_edge(i, j, *direction);
                    }
                }
            }
        }

        FloydWarshallResult::new(path, next)
    }

    /// Are elements unique (no duplicates present).
    fn unique<T: Ord>(mut elements: Vec<T>) -> bool {
        let length = elements.len();

        Self::dedup(&mut elements);

        elements.len() == length
    }

    /// De-duplicate vector elements.
    fn dedup<T: Ord>(elements: &mut Vec<T>) {
        // `sort_unstable` may not preserve the order of equal elements, but it is faster and less
        // memory consuming algorithm.
        elements.sort();
        elements.dedup();
    }
}

#[cfg(test)]
mod tests {
    use crate::floyd_warshall::FloydWarshall;
    use safe_graph::Graph;
    use std::cmp::Ordering::{Greater, Less};

    #[test]
    fn new() {
        let _alg: FloydWarshall<f32> = FloydWarshall::new();
    }

    #[test]
    fn new_customized() {
        let mul = Box::new(|x: f32, y: f32| x * y);
        let sharp_greater = Box::new(|x: f32, y: f32| x.partial_cmp(&y).unwrap_or(Less) == Greater);

        let _alg: FloydWarshall<f32> = FloydWarshall::new_customized(mul, sharp_greater);
    }

    #[test]
    fn new_fully_customized() {
        let mul = Box::new(|x: f32, y: f32| x * y);
        let sharp_less = Box::new(|x: f32, y: f32| x.partial_cmp(&y).unwrap_or(Greater) == Less);
        let discard_loops = false;

        let _alg: FloydWarshall<f32> =
            FloydWarshall::new_fully_customized(mul, sharp_less, discard_loops);
    }

    #[test]
    fn find_paths() {
        let alg: FloydWarshall<f32> = FloydWarshall::new();

        let w_a_b = 0.12;
        let w_a_c = 1.99;
        let w_b_c = 3.0;
        let w_a_d = 2.1;
        let w_a_e = 0.9;
        let w_a_f = 4.44;
        let w_a_g = 0.8;
        let w_g_f = 0.6;
        let w_a_h = 8.8;
        let w_f_h = 1.0;

        let graph = Graph::<_, _>::from_edges(&[
            ("a", "b", w_a_b),
            ("a", "c", w_a_c),
            ("b", "c", w_b_c),
            ("a", "d", w_a_d),
            ("a", "e", w_a_e),
            ("a", "f", w_a_f),
            ("a", "g", w_a_g),
            ("g", "f", w_g_f),
            ("a", "h", w_a_h),
            ("f", "h", w_f_h),
        ]);

        let result = alg.find_paths(&graph);

        let path = result.path;
        let next = result.next;

        // Test that the initial `(a, b)` edge is still the shortest path.
        assert_eq!(path.edge_weight("a", "b"), Some(&w_a_b));
        assert_eq!(next.edge_weight("a", "b"), Some(&"b"));

        // Test that the initial `(a, c)` edge is still the shortest path.
        assert_eq!(path.edge_weight("a", "c"), Some(&w_a_c));
        assert_eq!(next.edge_weight("a", "c"), Some(&"c"));

        // Test that a shorter path was found for the `(a, f)` and it is through `g`.
        assert_eq!(path.edge_weight("a", "f"), Some(&(w_a_g + w_g_f)));
        assert_eq!(next.edge_weight("a", "f"), Some(&"g"));
        assert_eq!(next.edge_weight("g", "f"), Some(&"f"));

        // Test that a shorter path was found for the `(a, f)` and it is through `g` and `f`.
        assert_eq!(path.edge_weight("a", "h"), Some(&(w_a_g + w_g_f + w_f_h)));
        assert_eq!(next.edge_weight("a", "h"), Some(&"g"));
        assert_eq!(next.edge_weight("g", "h"), Some(&"f"));
        assert_eq!(next.edge_weight("f", "h"), Some(&"h"));
    }
}