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
use encode::Encode;
use stream::Write;
use {Bit, DataPoint};

// END_MARKER relies on the fact that when we encode the delta of delta for a number that requires
// more than 12 bits we write four control bits 1111 followed by the 32 bits of the value. Since
// encoding assumes the value is greater than 12 bits, we can store the value 0 to signal the end
// of the stream

/// END_MARKER is a special bit sequence used to indicate the end of the stream
pub const END_MARKER: u64 = 0b1111_0000_0000_0000_0000_0000_0000_0000_0000;

/// END_MARKER_LEN is the length, in bits, of END_MARKER
pub const END_MARKER_LEN: u32 = 36;

/// StdEncoder
///
/// StdEncoder is used to encode `DataPoint`s
#[derive(Debug)]
pub struct StdEncoder<T: Write> {
    time: u64,       // current time
    delta: u64,      // current time delta
    value_bits: u64, // current float value as bits

    // store the number of leading and trailing zeroes in the current xor as u32 so we
    // don't have to do any conversions after calling `leading_zeros` and `trailing_zeros`
    leading_zeroes: u32,
    trailing_zeroes: u32,

    first: bool, // will next DataPoint be the first DataPoint encoded

    w: T,
}

impl<T> StdEncoder<T>
where
    T: Write,
{
    /// new creates a new StdEncoder whose starting timestamp is `start` and writes its encoded
    /// bytes to `w`
    pub fn new(start: u64, w: T) -> Self {
        let mut e = StdEncoder {
            time: start,
            delta: 0,
            value_bits: 0,
            leading_zeroes: 64,  // 64 is an initial sentinel value
            trailing_zeroes: 64, // 64 is an intitial sentinel value
            first: true,
            w,
        };

        // write timestamp header
        e.w.write_bits(start, 64);

        e
    }

    fn write_first(&mut self, time: u64, value_bits: u64) {
        self.delta = time - self.time;
        self.time = time;
        self.value_bits = value_bits;

        // write one control bit so we can distinguish a stream which contains only an initial
        // timestamp, this assumes the first bit of the END_MARKER is 1
        self.w.write_bit(Bit::Zero);

        // store the first delta with 14 bits which is enough to span just over 4 hours
        // if one wanted to use a window larger than 4 hours this size would increase
        self.w.write_bits(self.delta, 14);

        // store the first value exactly
        self.w.write_bits(self.value_bits, 64);

        self.first = true
    }

    fn write_next_timestamp(&mut self, time: u64) {
        let delta = time - self.time; // current delta
        let dod = delta.wrapping_sub(self.delta) as i32; // delta of delta

        // store the delta of delta using variable length encoding
        #[cfg_attr(feature = "cargo-clippy", allow(clippy::match_overlapping_arm))]
        match dod {
            0 => {
                self.w.write_bit(Bit::Zero);
            }
            -63..=64 => {
                self.w.write_bits(0b10, 2);
                self.w.write_bits(dod as u64, 7);
            }
            -255..=256 => {
                self.w.write_bits(0b110, 3);
                self.w.write_bits(dod as u64, 9);
            }
            -2047..=2048 => {
                self.w.write_bits(0b1110, 4);
                self.w.write_bits(dod as u64, 12);
            }
            _ => {
                self.w.write_bits(0b1111, 4);
                self.w.write_bits(dod as u64, 32);
            }
        }

        self.delta = delta;
        self.time = time;
    }

    fn write_next_value(&mut self, value_bits: u64) {
        let xor = value_bits ^ self.value_bits;
        self.value_bits = value_bits;

        if xor == 0 {
            // if xor with previous value is zero just store single zero bit
            self.w.write_bit(Bit::Zero);
        } else {
            self.w.write_bit(Bit::One);

            let leading_zeroes = xor.leading_zeros();
            let trailing_zeroes = xor.trailing_zeros();

            if leading_zeroes >= self.leading_zeroes && trailing_zeroes >= self.trailing_zeroes {
                // if the number of leading and trailing zeroes in this xor are >= the leading and
                // trailing zeroes in the previous xor then we only need to store a control bit and
                // the significant digits of this xor
                self.w.write_bit(Bit::Zero);
                self.w.write_bits(
                    xor.wrapping_shr(self.trailing_zeroes),
                    64 - self.leading_zeroes - self.trailing_zeroes,
                );
            } else {
                // if the number of leading and trailing zeroes in this xor are not less than the
                // leading and trailing zeroes in the previous xor then we store a control bit and
                // use 6 bits to store the number of leading zeroes and 6 bits to store the number
                // of significant digits before storing the significant digits themselves

                self.w.write_bit(Bit::One);
                self.w.write_bits(u64::from(leading_zeroes), 6);

                // if significant_digits is 64 we cannot encode it using 6 bits, however since
                // significant_digits is guaranteed to be at least 1 we can subtract 1 to ensure
                // significant_digits can always be expressed with 6 bits or less
                let significant_digits = 64 - leading_zeroes - trailing_zeroes;
                self.w.write_bits(u64::from(significant_digits - 1), 6);
                self.w
                    .write_bits(xor.wrapping_shr(trailing_zeroes), significant_digits);

                // finally we need to update the number of leading and trailing zeroes
                self.leading_zeroes = leading_zeroes;
                self.trailing_zeroes = trailing_zeroes;
            }
        }
    }
}

impl<T> Encode for StdEncoder<T>
where
    T: Write,
{
    fn encode(&mut self, dp: DataPoint) {
        let value_bits = dp.value.to_bits();

        if self.first {
            self.write_first(dp.time, value_bits);
            self.first = false;
            return;
        }

        self.write_next_timestamp(dp.time);
        self.write_next_value(value_bits)
    }

    fn close(mut self) -> Box<[u8]> {
        self.w.write_bits(END_MARKER, 36);
        self.w.close()
    }
}

#[cfg(test)]
mod tests {
    use super::StdEncoder;
    use encode::Encode;
    use stream::BufferedWriter;
    use DataPoint;

    #[test]
    fn create_new_encoder() {
        let w = BufferedWriter::new();
        let start_time = 1482268055; // 2016-12-20T21:07:35+00:00
        let e = StdEncoder::new(start_time, w);

        let bytes = e.close();
        let expected_bytes: [u8; 13] = [0, 0, 0, 0, 88, 89, 157, 151, 240, 0, 0, 0, 0];

        assert_eq!(bytes[..], expected_bytes[..]);
    }

    #[test]
    fn encode_datapoint() {
        let w = BufferedWriter::new();
        let start_time = 1482268055; // 2016-12-20T21:07:35+00:00
        let mut e = StdEncoder::new(start_time, w);

        let d1 = DataPoint::new(1482268055 + 10, 1.24);

        e.encode(d1);

        let bytes = e.close();
        let expected_bytes: [u8; 23] = [
            0, 0, 0, 0, 88, 89, 157, 151, 0, 20, 127, 231, 174, 20, 122, 225, 71, 175, 224, 0, 0,
            0, 0,
        ];

        assert_eq!(bytes[..], expected_bytes[..]);
    }

    #[test]
    fn encode_multiple_datapoints() {
        let w = BufferedWriter::new();
        let start_time = 1482268055; // 2016-12-20T21:07:35+00:00
        let mut e = StdEncoder::new(start_time, w);

        let d1 = DataPoint::new(1482268055 + 10, 1.24);

        e.encode(d1);

        let d2 = DataPoint::new(1482268055 + 20, 1.98);

        let d3 = DataPoint::new(1482268055 + 32, 2.37);
        let d4 = DataPoint::new(1482268055 + 44, -7.41);
        let d5 = DataPoint::new(1482268055 + 52, 103.50);

        e.encode(d2);
        e.encode(d3);
        e.encode(d4);
        e.encode(d5);

        let bytes = e.close();
        let expected_bytes: [u8; 61] = [
            0, 0, 0, 0, 88, 89, 157, 151, 0, 20, 127, 231, 174, 20, 122, 225, 71, 174, 204, 207,
            30, 71, 145, 228, 121, 30, 96, 88, 61, 255, 253, 91, 214, 245, 189, 111, 91, 3, 232, 1,
            245, 97, 88, 86, 21, 133, 55, 202, 1, 17, 15, 92, 40, 245, 194, 151, 128, 0, 0, 0, 0,
        ];

        assert_eq!(bytes[..], expected_bytes[..]);
    }
}