trs_dataframe/dataframe/column_store/
key_index.rs

1use data_value::DataValue;
2use halfbrown::HashMap;
3use ndarray::ArrayView1;
4use serde::{Deserialize, Serialize};
5
6use crate::{error::Error, Key};
7
8/// [`KeyIndex`] is used to store the keys for the [`super::ColumnFrame`]
9/// The keys are stored in the order they are added - the order is preserved
10/// The keys are stored in the [`Vec`] and the indexes are stored in the [`HashMap`]
11/// The indexes are used to access the data in the [`super::ColumnFrame`] by the column [`Key`]
12/// NOTE: The keys are unique - if the key is already present, it will be removed
13#[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
65    pub fn get_key(&self, idx: usize) -> Option<Key> {
66        self.keys.get(idx).cloned()
67    }
68    pub fn get_complement_keys(&self, keys: &[Key]) -> Vec<Key> {
69        self.keys
70            .iter()
71            .filter(|key| !keys.contains(key))
72            .cloned()
73            .collect()
74    }
75
76    pub fn select(&self, keys: &[Key]) -> KeyIndex {
77        let mut new_keys = Vec::with_capacity(keys.len());
78        let mut new_indexes = HashMap::with_capacity(keys.len());
79
80        for key in keys.iter() {
81            if let Some(idx) = self.indexes.get(key.name()) {
82                new_indexes.insert(key.name().to_string(), *idx);
83                new_keys.push(key.to_owned());
84            } else if let Some(alias_key) = self.alias.get(key.name()) {
85                if let Some(idx) = self.indexes.get(alias_key) {
86                    new_indexes.insert(key.name().to_string(), *idx);
87                    new_keys.push(key.to_owned());
88                }
89            }
90        }
91        Self {
92            keys: new_keys,
93            indexes: new_indexes,
94            alias: HashMap::new(),
95        }
96    }
97
98    pub fn indexes(&self) -> Vec<usize> {
99        self.indexes.values().copied().collect()
100    }
101
102    pub fn store_key(&mut self, key: Key) {
103        if self.indexes.contains_key(key.name()) {
104            return;
105        }
106        self.keys.push(key.clone());
107        self.indexes
108            .insert(key.name().to_string(), self.keys.len() - 1);
109    }
110
111    pub fn remove_key(&mut self, key: &Key) -> Option<(Key, usize)> {
112        let idx = self.indexes.remove(key.name())?;
113        let current = self.keys.remove(idx);
114        Some((current, idx))
115    }
116
117    pub fn get_as_candidate(&self, row: ArrayView1<DataValue>) -> HashMap<Key, DataValue> {
118        let mut result = HashMap::with_capacity(self.keys.len());
119        for (key, idx) in self.indexes.iter() {
120            result.insert(key.into(), row[*idx].clone());
121        }
122        result
123    }
124
125    pub fn to_vec_row(&self, candidate: HashMap<Key, DataValue>) -> Vec<DataValue> {
126        self.keys
127            .iter()
128            .map(|key| candidate.get(key).cloned().unwrap_or_default())
129            .collect()
130    }
131
132    pub fn check_order_of_indexes(&self, other: &Self) -> Result<(), Error> {
133        for (self_key, other_key) in self.keys.iter().zip(other.keys.iter()) {
134            if self_key != other_key {
135                return Err(Error::IndexOutOfOrder(
136                    self.keys.clone(),
137                    other.keys.clone(),
138                ));
139            }
140        }
141        Ok(())
142    }
143
144    pub fn rename_key(&mut self, key: &str, new_key: Key) -> Result<(), Error> {
145        if let Some(idx) = self.indexes.remove(key) {
146            self.indexes.insert(new_key.to_string(), idx);
147            self.keys[idx] = new_key;
148            Ok(())
149        } else {
150            Err(Error::NotFound(key.into()))
151        }
152    }
153
154    pub fn add_alias(&mut self, key: &str, alias: &str) -> Result<(), Error> {
155        if !self.indexes.contains_key(key) {
156            return Err(Error::NotFound(key.into()));
157        }
158        self.alias.insert(alias.to_string(), key.to_string());
159        Ok(())
160    }
161}
162
163impl From<Vec<Key>> for KeyIndex {
164    fn from(keys: Vec<Key>) -> Self {
165        Self::new(keys)
166    }
167}
168
169#[cfg(test)]
170mod test {
171    use crate::DataType;
172
173    use super::*;
174    use rstest::*;
175
176    #[rstest]
177    fn test_alias() {
178        let key = Key::new("a", DataType::U32);
179        let mut key_index = KeyIndex::new(vec![key.clone()]);
180        assert_eq!(key_index.add_alias(key.name(), "alias"), Ok(()));
181        assert!(key_index.add_alias("c", "alias").is_err());
182        assert_eq!(key_index.alias.get("alias"), Some(&key.name().to_string()));
183        assert_eq!(key_index.get_column_index(&key), Some(0));
184        assert_eq!(
185            key_index.get_column_index(&Key::new("alias", DataType::U32)),
186            Some(0)
187        );
188    }
189
190    #[rstest]
191    fn test_rename() {
192        let key = Key::new("a", DataType::U32);
193        let mut key_index = KeyIndex::new(vec![key.clone(), Key::new("b", DataType::U32)]);
194        assert_eq!(key_index.rename_key("a", "new_key".into()), Ok(()));
195        assert_eq!(
196            key_index.get_column_index(&Key::new("new_key", DataType::U32)),
197            Some(0)
198        );
199        assert!(key_index.rename_key("c", "alias".into()).is_err());
200    }
201
202    #[rstest]
203    #[case(
204        vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
205        vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]
206    )]
207    fn test_key_index_new(#[case] keys: Vec<Key>, #[case] expected: Vec<Key>) {
208        let key_index = KeyIndex::new(keys);
209        assert_eq!(key_index.keys, expected);
210        assert_eq!(key_index.get_keys(), expected);
211    }
212
213    #[rstest]
214    #[case(
215        vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
216        vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
217        Ok(())
218    )]
219    #[case(
220        vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)],
221        vec![Key::new("b", DataType::U32), Key::new("a", DataType::U32)],
222        Err(Error::IndexOutOfOrder(
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        ))
226    )]
227    fn test_key_index_check_order_of_indexes(
228        #[case] keys: Vec<Key>,
229        #[case] other_keys: Vec<Key>,
230        #[case] expected: Result<(), Error>,
231    ) {
232        let key_index: KeyIndex = keys.into();
233        let other_key_index = KeyIndex::new(other_keys);
234        assert_eq!(key_index.check_order_of_indexes(&other_key_index), expected);
235    }
236
237    #[rstest]
238    #[case(
239        KeyIndex::new(vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]),
240        (Key::new("a", DataType::U32), 0)
241    )]
242    #[case(
243        KeyIndex::new(vec![Key::new("a", DataType::U32), Key::new("b", DataType::U32)]),
244        (Key::new("b", DataType::U32), 1)
245    )]
246    fn test_key_index_remove_key(#[case] mut key_index: KeyIndex, #[case] expected: (Key, usize)) {
247        let key = expected.0.clone();
248        assert_eq!(key_index.remove_key(&key), Some(expected));
249    }
250}