git_perf/
serialization.rs1use std::{
2 borrow::Borrow,
3 collections::{
4 hash_map::Entry::{Occupied, Vacant},
5 HashMap,
6 },
7};
8
9use itertools::Itertools;
10use log::warn;
11
12use crate::data::MeasurementData;
13
14pub const DELIMITER: &str = "";
15
16pub fn serialize_single<M>(measurement_data: &M, custom_delimiter: &str) -> String
17where
18 M: Borrow<MeasurementData>,
19{
20 let md: &MeasurementData = measurement_data.borrow();
21
22 let mut m = vec![
23 format!("{:?}", md.epoch),
24 md.name.clone(),
25 format!("{:?}", md.timestamp),
26 format!("{:?}", md.val),
27 ];
28
29 m.extend(md.key_values.iter().map(|(k, v)| format!("{k}={v}")));
30
31 m.join(custom_delimiter) + "\n"
32}
33
34pub fn serialize_multiple<M: Borrow<MeasurementData>>(measurement_data: &[M]) -> String {
35 measurement_data
36 .iter()
37 .map(|md| serialize_single(md, DELIMITER))
38 .join("")
39}
40
41fn deserialize_single(line: &str) -> Option<MeasurementData> {
42 let components = line
43 .split(DELIMITER)
44 .filter(|item| !item.is_empty())
45 .collect_vec();
46
47 let num_components = components.len();
48 if num_components < 4 {
49 warn!("Too few items with {num_components}, skipping record");
50 return None;
51 }
52
53 let epoch = components[0];
54 let epoch = match epoch.parse::<u32>() {
55 Ok(e) => e,
56 Err(err) => {
57 warn!("Cannot parse epoch '{epoch}': {err}, skipping record");
58 return None;
59 }
60 };
61
62 let name = components[1].to_string();
63
64 let timestamp = components[2];
65 let timestamp = match timestamp.parse::<f64>() {
66 Ok(ts) => ts,
67 Err(err) => {
68 warn!("Cannot parse timestamp '{timestamp}': {err}, skipping record");
69 return None;
70 }
71 };
72
73 let val = components[3];
74 let val = match val.parse::<f64>() {
75 Ok(val) => val,
76 Err(err) => {
77 warn!("Cannot parse value '{val}': {err}, skipping record");
78 return None;
79 }
80 };
81
82 let mut key_values = HashMap::new();
83
84 if components.len() > 4 {
85 for kv in components.iter().skip(4) {
86 if let Some((key, value)) = kv.split_once('=') {
87 let entry = key_values.entry(key.to_string());
88 let value = value.to_string();
89 match entry {
90 Occupied(mut e) => {
91 if e.get() == &value {
92 static DUPLICATE_KEY_SAME_VALUE: std::sync::Once =
93 std::sync::Once::new();
94 DUPLICATE_KEY_SAME_VALUE.call_once(|| {
95 warn!("Duplicate entries for key {key} with same value");
96 });
97 } else {
98 static DUPLICATE_KEY_CONFLICT: std::sync::Once = std::sync::Once::new();
99 DUPLICATE_KEY_CONFLICT.call_once(|| {
100 warn!(
101 "Conflicting values for key {key}: '{}' vs '{}'",
102 e.get(),
103 value
104 );
105 });
106 }
107 e.insert(value);
108 }
109 Vacant(e) => {
110 e.insert(value);
111 }
112 }
113 } else {
114 warn!("No equals sign in key value pair, skipping record");
115 return None;
116 }
117 }
118 }
119
120 Some(MeasurementData {
121 epoch,
122 name,
123 timestamp,
124 val,
125 key_values,
126 })
127}
128
129pub fn deserialize(lines: &str) -> Vec<MeasurementData> {
130 lines
131 .lines()
132 .filter(|l| !l.trim().is_empty())
133 .filter_map(deserialize_single)
134 .collect_vec()
135}
136
137#[cfg(test)]
138mod test {
139 use super::*;
140
141 #[test]
142 fn key_value_deserialization() {
143 let lines = "0test1234123key1=value1key2=value2";
144 let actual = deserialize(lines);
145 let expected = MeasurementData {
146 epoch: 0,
147 name: "test".to_string(),
148 timestamp: 1234.0,
149 val: 123.0,
150 key_values: [
151 ("key1".to_string(), "value1".to_string()),
152 ("key2".to_string(), "value2".to_string()),
153 ]
154 .into(),
155 };
156 assert_eq!(actual.len(), 1);
157 assert_eq!(actual[0], expected);
158 }
159
160 #[test]
161 fn key_value_invalid_pair() {
162 let lines = "0test1234123key1value1\n\
164 0test24567890key2=value2";
165
166 let expected = [MeasurementData {
167 epoch: 0,
168 name: "test2".to_string(),
169 timestamp: 4567.0,
170 val: 890.0,
171 key_values: [("key2".to_string(), "value2".to_string())].into(),
172 }];
173 let actual = deserialize(lines);
174 assert_eq!(actual, expected);
175 }
176
177 #[test]
178 fn additional_whitespace_deserialization() {
179 let lines = "0test1234123";
180 let actual = deserialize(lines);
181 assert_eq!(1, actual.len());
182 }
183
184 #[test]
185 fn test_serialize_single() {
186 let md = MeasurementData {
187 epoch: 3,
188 name: "Mymeasurement".into(),
189 timestamp: 1234567.0,
190 val: 42.0,
191 key_values: [("mykey".to_string(), "myvalue".to_string())].into(),
192 };
193 let serialized = serialize_single(&md, DELIMITER);
194 assert_eq!(serialized, "3Mymeasurement1234567.042.0mykey=myvalue\n");
195 }
196
197 #[test]
198 fn test_epoch_parsing() {
199 let valid_line = "42test1234123";
201 let result = deserialize_single(valid_line);
202 assert!(result.is_some());
203 assert_eq!(result.unwrap().epoch, 42);
204
205 let invalid_line = "not_a_numbertest1234123";
207 let result = deserialize_single(invalid_line);
208 assert!(result.is_none());
209
210 let out_of_range_line = "4294967296test1234123"; let result = deserialize_single(out_of_range_line);
213 assert!(result.is_none());
214 }
215}