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

use std::{
    collections::{hash_map::RandomState, HashSet},
    hash::{BuildHasher, Hash},
};

use crate::HandlerDescription;

/// Description for a handler that describes what event kinds are interesting to
/// the handler.
///
/// This can be useful if you can filter events before they are passed to
/// `dptree`. In this case you should keep updates that are in the `observed`
/// set.
#[derive(Debug, Clone)]
pub struct InterestSet<K, S = RandomState> {
    /// Event kinds that are of interested for a given handler.
    ///
    /// I.e. the ones that can cause meaningful side-effects.
    pub observed: HashSet<K, S>,

    /// Event kinds that can be observed by handlers chained to this one.
    pub filtered: HashSet<K, S>,
}

/// An event kind that can be used with [`InterestSet`].
///
/// Usually this would be implemented by a field-less enumeration of all update
/// kinds.
pub trait EventKind<S = RandomState>: Sized {
    /// Set of all event kinds.
    fn full_set() -> HashSet<Self, S>;

    /// An empty set.
    fn empty_set() -> HashSet<Self, S>;
}

impl<K: EventKind<S>, S> InterestSet<K, S> {
    /// Constructs an [`InterestSet`] for a filter that allows to pass through
    /// it only updates with kinds in the `filtered` set.
    ///
    /// Note that the filter should not have observable side-effects, for
    /// example:
    /// ```
    /// use dptree::{description::{InterestSet, EventKind}, filter_with_description};
    /// use maplit::hashset;
    ///
    /// # enum K {} impl EventKind for K { fn full_set() -> std::collections::HashSet<Self> { hashset!{} } fn empty_set() -> std::collections::HashSet<Self> { hashset!{} } }
    /// # let _: dptree::Handler<(), (), InterestSet<K>> =
    /// filter_with_description(InterestSet::new_filter(hashset! {}), || {
    ///     println!("Filter called!"); // <-- bad
    ///
    ///     false
    /// });
    ///
    /// # #[derive(Clone)] struct Db; impl Db { fn fetch_enabled(&self) -> bool { false } }
    /// # let _: dptree::Handler<dptree::di::DependencyMap, (), InterestSet<K>> =
    /// filter_with_description(InterestSet::new_filter(hashset! {}), |db: Db| {
    ///     let pass = db.fetch_enabled(); // <-- fine
    ///
    ///     pass
    /// });
    /// ```
    pub fn new_filter(filtered: HashSet<K, S>) -> Self {
        // We assume that well behaved filters don't observe anything (filters should
        // only filter!).
        Self { observed: K::empty_set(), filtered }
    }
}

impl<T, S> HandlerDescription for InterestSet<T, S>
where
    T: EventKind<S> + Eq + Hash + Clone,
    S: BuildHasher + Clone,
    T: Send + Sync + 'static,
    S: Send + Sync + 'static,
{
    fn entry() -> Self {
        // Entry does not observe anything and allows everything.
        Self { observed: T::empty_set(), filtered: T::full_set() }
    }

    fn user_defined() -> Self {
        // We don't know what user defined code does, so we assume the most forgiving
        // case: it observes everything and allows everything.
        //
        // Were we to choose anything else, there would be user code that would be
        // broken.
        Self { observed: T::full_set(), filtered: T::full_set() }
    }

    fn endpoint() -> Self {
        // Endpoint observes everything (again, as per the same reasoning as
        // `user_defined`), but does not allow anything to pass through (it's
        // the last handler).
        Self { observed: T::full_set(), filtered: T::empty_set() }
    }

    fn merge_chain(&self, other: &Self) -> Self {
        let Self { observed: l_obs, filtered: l_flt } = self;
        let Self { observed: r_obs, filtered: r_flt } = other;

        // Second handler can only observe things that were passed through by the first
        // one.
        let observed = {
            let hasher = l_obs.hasher().clone();
            let mut tmp = HashSet::with_hasher(hasher);

            tmp.extend(l_obs.iter().cloned());
            tmp.extend(l_flt.intersection(r_obs).cloned());
            tmp
        };

        // If we chain two filters together, we are only interested in events that can
        // pass both of them.
        let filtered = {
            let hasher = l_flt.hasher().clone();
            let mut tmp = HashSet::with_hasher(hasher);

            tmp.extend(l_flt.intersection(r_flt).cloned());
            tmp
        };

        Self { filtered, observed }
    }

    fn merge_branch(&self, other: &Self) -> Self {
        let Self { observed: l_obs, filtered: l_flt } = self;
        let Self { observed: r_obs, filtered: _ } = other;

        // Second handler can only observe things that were passed through by the first
        // one.
        let observed = {
            let hasher = l_obs.hasher().clone();
            let mut tmp = HashSet::with_hasher(hasher);

            tmp.extend(l_obs.iter().cloned());
            tmp.extend(l_flt.intersection(r_obs).cloned());
            tmp
        };

        // Even if second filter did not pass something through, the execution still
        // continues.
        let filtered = l_flt.clone();

        Self { observed, filtered }
    }
}

impl<K: Hash + Eq, S: BuildHasher> Eq for InterestSet<K, S> {}

impl<K: Hash + Eq, S: BuildHasher> PartialEq for InterestSet<K, S> {
    fn eq(&self, other: &Self) -> bool {
        let Self { observed: l_obs, filtered: l_flt } = self;
        let Self { observed: r_obs, filtered: r_flt } = other;

        l_obs == r_obs && l_flt == r_flt
    }
}