trs_dataframe/dataframe/column_store/
key_index.rs1use data_value::DataValue;
2use halfbrown::HashMap;
3use ndarray::ArrayView1;
4use serde::{Deserialize, Serialize};
5
6use crate::{error::Error, Key};
7
8#[derive(Debug, Clone, Default, Serialize, Deserialize, PartialEq)]
14pub struct KeyIndex {
15 pub keys: Vec<Key>,
16 indexes: HashMap<String, usize>,
17 pub alias: HashMap<String, String>,
18}
19
20impl KeyIndex {
21 pub fn new(keys: Vec<Key>) -> Self {
22 let mut indexes = HashMap::with_capacity(keys.len());
23 let mut removed = 0;
24 let mut actual_keys = Vec::with_capacity(keys.len());
25 for (idx, key) in keys.into_iter().enumerate() {
26 if indexes.contains_key(key.name()) {
27 removed += 1;
28 } else {
29 indexes.insert(key.name().to_string(), idx.saturating_sub(removed));
30 actual_keys.push(key)
31 }
32 }
33 Self {
34 keys: actual_keys,
35 indexes,
36 alias: HashMap::new(),
37 }
38 }
39
40 pub fn len(&self) -> usize {
41 self.keys.len()
42 }
43 pub fn is_empty(&self) -> bool {
44 self.keys.is_empty()
45 }
46
47 pub fn get_column_index(&self, key: &Key) -> Option<usize> {
48 self.get_column_index_by_name(key.name())
49 }
50
51 pub fn get_column_index_by_name(&self, key: &str) -> Option<usize> {
52 if let Some(f) = self.indexes.get(key) {
53 Some(*f)
54 } else {
55 self.alias
56 .get(key)
57 .and_then(|alias| self.indexes.get(alias).copied())
58 }
59 }
60
61 pub fn get_keys(&self) -> &[Key] {
62 &self.keys
63 }
64 pub fn get_keys_mut(&mut self) -> &mut [Key] {
65 &mut self.keys
66 }
67
68 pub fn get_key(&self, idx: usize) -> Option<Key> {
69 self.keys.get(idx).cloned()
70 }
71 pub fn get_complement_keys(&self, keys: &[Key]) -> Vec<Key> {
72 self.keys
73 .iter()
74 .filter(|key| !keys.contains(key))
75 .cloned()
76 .collect()
77 }
78
79 pub fn select(&self, keys: &[Key]) -> KeyIndex {
80 let mut new_keys = Vec::with_capacity(keys.len());
81 let mut new_indexes = HashMap::with_capacity(keys.len());
82
83 for key in keys.iter() {
84 if let Some(idx) = self.indexes.get(key.name()) {
85 new_indexes.insert(key.name().to_string(), *idx);
86 new_keys.push(key.to_owned());
87 } else if let Some(alias_key) = self.alias.get(key.name()) {
88 if let Some(idx) = self.indexes.get(alias_key) {
89 new_indexes.insert(key.name().to_string(), *idx);
90 new_keys.push(key.to_owned());
91 }
92 }
93 }
94 Self {
95 keys: new_keys,
96 indexes: new_indexes,
97 alias: HashMap::new(),
98 }
99 }
100
101 pub fn indexes(&self) -> Vec<usize> {
102 self.indexes.values().copied().collect()
103 }
104
105 pub fn store_key(&mut self, key: Key) {
106 if self.indexes.contains_key(key.name()) {
107 return;
108 }
109 self.keys.push(key.clone());
110 self.indexes
111 .insert(key.name().to_string(), self.keys.len() - 1);
112 }
113
114 pub fn remove_key(&mut self, key: &Key) -> Option<(Key, usize)> {
115 let idx = self.indexes.remove(key.name())?;
116 let current = self.keys.remove(idx);
117 Some((current, idx))
118 }
119
120 pub fn get_as_candidate(&self, row: ArrayView1<DataValue>) -> HashMap<Key, DataValue> {
121 let mut result = HashMap::with_capacity(self.keys.len());
122 for (key, idx) in self.indexes.iter() {
123 result.insert(key.into(), row[*idx].clone());
124 }
125 result
126 }
127
128 pub fn to_vec_row(&self, candidate: HashMap<Key, DataValue>) -> Vec<DataValue> {
129 self.keys
130 .iter()
131 .map(|key| candidate.get(key).cloned().unwrap_or_default())
132 .collect()
133 }
134
135 pub fn check_order_of_indexes(&self, other: &Self) -> Result<(), Error> {
136 for (self_key, other_key) in self.keys.iter().zip(other.keys.iter()) {
137 if self_key != other_key {
138 return Err(Error::IndexOutOfOrder(
139 self.keys.clone(),
140 other.keys.clone(),
141 ));
142 }
143 }
144 Ok(())
145 }
146
147 pub fn rename_key(&mut self, key: &str, new_key: Key) -> Result<(), Error> {
148 if let Some(idx) = self.indexes.remove(key) {
149 self.indexes.insert(new_key.to_string(), idx);
150 self.keys[idx] = new_key;
151 Ok(())
152 } else {
153 Err(Error::NotFound(key.into()))
154 }
155 }
156
157 pub fn add_alias(&mut self, key: &str, alias: &str) -> Result<(), Error> {
158 if !self.indexes.contains_key(key) {
159 return Err(Error::NotFound(key.into()));
160 }
161 self.alias.insert(alias.to_string(), key.to_string());
162 Ok(())
163 }
164}
165
166impl From<Vec<Key>> for KeyIndex {
167 fn from(keys: Vec<Key>) -> Self {
168 Self::new(keys)
169 }
170}
171
172#[cfg(test)]
173mod test {
174 use crate::DataType;
175
176 use super::*;
177 use rstest::*;
178
179 #[rstest]
180 fn test_alias() {
181 let key = Key::new("a", DataType::U32);
182 let mut key_index = KeyIndex::new(vec![key.clone()]);
183 assert_eq!(key_index.add_alias(key.name(), "alias"), Ok(()));
184 assert!(key_index.add_alias("c", "alias").is_err());
185 assert_eq!(key_index.alias.get("alias"), Some(&key.name().to_string()));
186 assert_eq!(key_index.get_column_index(&key), Some(0));
187 assert_eq!(
188 key_index.get_column_index(&Key::new("alias", DataType::U32)),
189 Some(0)
190 );
191 }
192
193 #[rstest]
194 fn test_rename() {
195 let key = Key::new("a", DataType::U32);
196 let mut key_index = KeyIndex::new(vec![key.clone(), Key::new("b", DataType::U32)]);
197 assert_eq!(key_index.rename_key("a", "new_key".into()), Ok(()));
198 assert_eq!(
199 key_index.get_column_index(&Key::new("new_key", DataType::U32)),
200 Some(0)
201 );
202 assert!(key_index.rename_key("c", "alias".into()).is_err());
203 }
204
205 #[rstest]
206 #[case(
207 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
208 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]
209 )]
210 fn test_key_index_new(#[case] keys: Vec<Key>, #[case] expected: Vec<Key>) {
211 let key_index = KeyIndex::new(keys);
212 assert_eq!(key_index.keys, expected);
213 assert_eq!(key_index.get_keys(), expected);
214 }
215
216 #[rstest]
217 #[case(
218 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
219 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
220 Ok(())
221 )]
222 #[case(
223 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
224 vec![Key::new("b", DataType::U32), Key::new("a", DataType::U32)],
225 Err(Error::IndexOutOfOrder(
226 vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
227 vec![Key::new("b", DataType::U32), Key::new("a", DataType::U32)]
228 ))
229 )]
230 fn test_key_index_check_order_of_indexes(
231 #[case] keys: Vec<Key>,
232 #[case] other_keys: Vec<Key>,
233 #[case] expected: Result<(), Error>,
234 ) {
235 let key_index: KeyIndex = keys.into();
236 let other_key_index = KeyIndex::new(other_keys);
237 assert_eq!(key_index.check_order_of_indexes(&other_key_index), expected);
238 }
239
240 #[rstest]
241 #[case(
242 KeyIndex::new(vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]),
243 (Key::new("a", DataType::U32), 0)
244 )]
245 #[case(
246 KeyIndex::new(vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]),
247 (Key::new("b", DataType::U32), 1)
248 )]
249 fn test_key_index_remove_key(#[case] mut key_index: KeyIndex, #[case] expected: (Key, usize)) {
250 let key = expected.0.clone();
251 assert_eq!(key_index.remove_key(&key), Some(expected));
252 }
253}