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
//! # Library to perform operations over dependency graphs.
//!
//! This library allow running iterative operations over a dependency graph in
//! the correct evaluation order, or will fail if there are a circular
//! dependencies.
//!
//! To use this library, you create a list of [Nodes](trait.Node.html)
//! containing dependency information (which node depends on which). You then
//! create a [DepGraph](struct.DepGraph.html) which will allow you to traverse
//! the graph so that you will always get an item for which all dependencies
//! have been processed.
//!
//! ## Processing order
//!
//! DepGraphs have two methods: one for sequential operations and one for
//! parallel (multi-threaded) operations. In the first case, it's easy to know
//! in which order nodes can be processed, as only one node will be processed
//! at a time. However, in parallel operations, we need to know if a given node
//! is done processing.
//!
//! This leads to a situation where a given worker thread might not be able to
//! pull a node temporarily, as it needs to wait for another worker to finish
//! processing another node.
//!
//! Let's look at the following case:
//!
//! ```text,no_run
//! (A) <-|
//!       |-- [E] <-|-- [G]
//! (B) <-|         |
//!       |-- [F] <-|-- [H]
//! [C] <-|
//! ```
//!
//! In this case, the nodes __E__ and __F__ are dependent on __A__, __B__ and
//! __C__ and __G__ and __H__ are dependent on both __E__ and __F__. If we
//! process the nodes with two workers, they might pick up nodes A and B first.
//! Since these nodes don't have any dependencies, there is no problem right
//! now.
//!
//! ```text,no_run
//! [ ] <-|
//!       |-- [E] <-|-- [G]
//! [ ] <-|         |
//!       |-- [F] <-|-- [H]
//! (C) <-|
//! ```
//!
//! When one of the worker is done, it can immediately start working on node
//! __C__, as it does not have any dependencies. However, when the second
//! worker is done, there are no available nodes for processing: we need to
//! wait until __C__ is processed before we can start working on __E__ or
//! __F__. One of the worker will then stay idle until the other one is done.
//!
//! ```text,no_run
//! [ ] <-|
//!       |-- (E) <-|-- [G]
//! [ ] <-|         |
//!       |-- (F) <-|-- [H]
//! [ ] <-|
//! ```
//!
//! Once that is done, both workers can work on __E__ and __F__. However, if
//! __E__ takes only a fraction of the time compared to __F__, we will end up
//! in the same situation, as there are no nodes without un-processed
//! dependencies.
//!
//! ## Basic usage
//!
//! ```rust
//! use dep_graph::{Node, DepGraph};
//! #[cfg(feature = "rayon")]
//! use rayon::prelude::*;
//!
//! // Create a list of nodes
//! let mut root = Node::new("root");
//! let mut dep1 = Node::new("dep1");
//! let mut dep2 = Node::new("dep2");
//! let leaf = Node::new("leaf");
//!
//! // Map their connections
//! root.add_dep(dep1.id());
//! root.add_dep(dep2.id());
//! dep1.add_dep(leaf.id());
//! dep2.add_dep(leaf.id());
//!
//! // Create a graph
//! let nodes = vec![root, dep1, dep2, leaf];
//!
//! // Print the name of all nodes in the dependency graph.
//! // This will parse the dependency graph sequentially
//! {
//!     let graph = DepGraph::new(&nodes);
//!     graph
//!         .into_iter()
//!         .for_each(|node| {
//!             println!("{:?}", node)
//!         });
//! }
//!
//! // This is the same as the previous command, excepts it leverages rayon
//! // to process them in parallel as much as possible.
//! #[cfg(feature = "rayon")]
//! {
//!     let graph = DepGraph::new(&nodes);
//!     graph
//!         .into_par_iter()
//!         .for_each(|node| {
//!             // The node is a depgraph::Wrapper object, not a String.
//!             // We need to use `*node` to get its value.
//!             println!("{:?}", *node)
//!         });
//! }
//! ```
//!
//! ## Create your own node type
//!
//! This library provides a node for string types
//! [`Node`](struct.Node.html).
//!
//! However, you may want to implement your own node type to hold another type
//! of identity information. For this purpose, you can implement the
//! [`Node trait`](trait.Node.html).

pub mod error;
mod graph;
#[cfg(feature = "rayon")]
mod graph_par;
mod node;

pub use graph::DepGraph;
#[cfg(feature = "rayon")]
pub use graph_par::Wrapper;
pub use node::Node;

#[cfg(test)]
mod tests {
    use super::*;
    #[cfg(feature = "rayon")]
    use rayon::prelude::*;
    use std::time::Duration;

    /// Run against a diamond graph
    ///
    /// ```no_run
    ///   1
    ///  / \
    /// 2   3
    ///  \ /
    ///   4
    /// ```
    #[cfg(feature = "rayon")]
    #[test]
    fn par_diamond_graph() {
        let mut n1 = Node::new("1");
        let mut n2 = Node::new("2");
        let mut n3 = Node::new("3");
        let n4 = Node::new("4");

        n1.add_dep(n2.id());
        n1.add_dep(n3.id());
        n2.add_dep(n4.id());
        n3.add_dep(n4.id());

        let deps = vec![n1, n2, n3, n4];

        let r = DepGraph::new(&deps);
        let result = r.into_par_iter().map(|_| true).collect::<Vec<bool>>();

        assert_eq!(result.len(), deps.len());
    }

    #[cfg(feature = "rayon")]
    #[test]
    fn par_diamond_graph_steps() {
        let mut n1 = Node::new("1");
        let mut n2 = Node::new("2");
        let mut n3 = Node::new("3");
        let n4 = Node::new("4");

        n1.add_dep(n2.id());
        n1.add_dep(n3.id());
        n2.add_dep(n4.id());
        n3.add_dep(n4.id());

        let deps = vec![n1, n2, n3, n4];

        let r = DepGraph::new(&deps);
        let result = r
            .into_par_iter()
            .map(|node_id| (*node_id).parse::<u64>().unwrap())
            .reduce(|| 0, |acc, x| acc + x);

        assert_eq!(result, 10);
    }

    #[cfg(feature = "rayon")]
    #[test]
    fn par_diamond_graph_with_timeout() {
        let mut n1 = Node::new("1");
        let mut n2 = Node::new("2");
        let mut n3 = Node::new("3");
        let n4 = Node::new("4");

        n1.add_dep(n2.id());
        n1.add_dep(n3.id());
        n2.add_dep(n4.id());
        n3.add_dep(n4.id());

        let deps = vec![n1, n2, n3, n4];

        let r = DepGraph::new(&deps);
        let result = r
            .into_par_iter()
            .with_timeout(Duration::from_secs(2))
            .map(|_| true)
            .collect::<Vec<bool>>();

        assert_eq!(result.len(), deps.len());
    }

    #[test]
    fn iter_diamond_graph() {
        let mut n1 = Node::new("1");
        let mut n2 = Node::new("2");
        let mut n3 = Node::new("3");
        let n4 = Node::new("4");

        n1.add_dep(n2.id());
        n1.add_dep(n3.id());
        n2.add_dep(n4.id());
        n3.add_dep(n4.id());

        let deps = vec![n1, n2, n3, n4];

        let r = DepGraph::new(&deps);
        let result = r.into_iter().map(|_| true).collect::<Vec<bool>>();

        assert_eq!(result.len(), deps.len());
    }

    /// 1 000 nodes with 999 depending on one
    #[cfg(feature = "rayon")]
    #[test]
    fn par_thousand_graph() {
        let mut nodes: Vec<Node<_>> = (0..1000).map(|i| Node::new(format!("{}", i))).collect();
        for i in 1..1000 {
            nodes[i].add_dep("0".to_string());
        }

        let r = DepGraph::new(&nodes);
        let result = r.into_par_iter().map(|_| true).collect::<Vec<bool>>();

        assert_eq!(result.len(), nodes.len());
    }

    #[test]
    fn iter_thousand_graph() {
        let mut nodes: Vec<Node<_>> = (0..1000).map(|i| Node::new(format!("{}", i))).collect();
        for i in 1..1000 {
            nodes[i].add_dep("0".to_string());
        }

        let r = DepGraph::new(&nodes);
        let result = r.into_iter().map(|_| true).collect::<Vec<bool>>();

        assert_eq!(result.len(), nodes.len());
    }

    // #[test]
    // #[should_panic]
    // fn par_circular_graph() {
    //     let mut n1 = Node::new("1");
    //     let mut n2 = Node::new("2");
    //     let mut n3 = Node::new("3");

    //     n1.add_dep(n2.id());
    //     n2.add_dep(n3.id());
    //     n3.add_dep(n1.id());

    //     let deps = vec![n1, n2, n3];

    //     // This should return an exception
    //     let r = DepGraph::new(&deps);
    //     r.into_par_iter().for_each(|_node_id| {});
    // }
}