path_list.rs

  1use std::{
  2    path::{Path, PathBuf},
  3    sync::Arc,
  4};
  5
  6use itertools::Itertools;
  7use util::paths::SanitizedPath;
  8
  9/// A list of absolute paths, in a specific order.
 10///
 11/// The paths are stored in lexicographic order, so that they can be compared to
 12/// other path lists without regard to the order of the paths.
 13///
 14/// The paths can be retrieved in the original order using `ordered_paths()`.
 15#[derive(Default, PartialEq, Eq, Debug, Clone)]
 16pub struct PathList {
 17    /// The paths, in lexicographic order.
 18    paths: Arc<[PathBuf]>,
 19    /// The order in which the paths were provided.
 20    ///
 21    /// See `ordered_paths()` for a way to get the paths in the original order.
 22    order: Arc<[usize]>,
 23}
 24
 25#[derive(Debug)]
 26pub struct SerializedPathList {
 27    pub paths: String,
 28    pub order: String,
 29}
 30
 31impl PathList {
 32    pub fn new<P: AsRef<Path>>(paths: &[P]) -> Self {
 33        let mut indexed_paths: Vec<(usize, PathBuf)> = paths
 34            .iter()
 35            .enumerate()
 36            .map(|(ix, path)| (ix, SanitizedPath::new(path).into()))
 37            .collect();
 38        indexed_paths.sort_by(|(_, a), (_, b)| a.cmp(b));
 39        let order = indexed_paths.iter().map(|e| e.0).collect::<Vec<_>>().into();
 40        let paths = indexed_paths
 41            .into_iter()
 42            .map(|e| e.1)
 43            .collect::<Vec<_>>()
 44            .into();
 45        Self { order, paths }
 46    }
 47
 48    pub fn is_empty(&self) -> bool {
 49        self.paths.is_empty()
 50    }
 51
 52    /// Get the paths in lexicographic order.
 53    pub fn paths(&self) -> &[PathBuf] {
 54        self.paths.as_ref()
 55    }
 56
 57    /// Get the order in which the paths were provided.
 58    pub fn order(&self) -> &[usize] {
 59        self.order.as_ref()
 60    }
 61
 62    /// Get the paths in the original order.
 63    pub fn ordered_paths(&self) -> impl Iterator<Item = &PathBuf> {
 64        self.order
 65            .iter()
 66            .zip(self.paths.iter())
 67            .sorted_by_key(|(i, _)| **i)
 68            .map(|(_, path)| path)
 69    }
 70
 71    pub fn is_lexicographically_ordered(&self) -> bool {
 72        self.order.iter().enumerate().all(|(i, &j)| i == j)
 73    }
 74
 75    pub fn deserialize(serialized: &SerializedPathList) -> Self {
 76        let mut paths: Vec<PathBuf> = if serialized.paths.is_empty() {
 77            Vec::new()
 78        } else {
 79            serialized.paths.split('\n').map(PathBuf::from).collect()
 80        };
 81
 82        let mut order: Vec<usize> = serialized
 83            .order
 84            .split(',')
 85            .filter_map(|s| s.parse().ok())
 86            .collect();
 87
 88        if !paths.is_sorted() || order.len() != paths.len() {
 89            order = (0..paths.len()).collect();
 90            paths.sort();
 91        }
 92
 93        Self {
 94            paths: paths.into(),
 95            order: order.into(),
 96        }
 97    }
 98
 99    pub fn serialize(&self) -> SerializedPathList {
100        use std::fmt::Write as _;
101
102        let mut paths = String::new();
103        for path in self.paths.iter() {
104            if !paths.is_empty() {
105                paths.push('\n');
106            }
107            paths.push_str(&path.to_string_lossy());
108        }
109
110        let mut order = String::new();
111        for ix in self.order.iter() {
112            if !order.is_empty() {
113                order.push(',');
114            }
115            write!(&mut order, "{}", *ix).unwrap();
116        }
117        SerializedPathList { paths, order }
118    }
119}
120
121#[cfg(test)]
122mod tests {
123    use super::*;
124
125    #[test]
126    fn test_path_list() {
127        let list1 = PathList::new(&["a/d", "a/c"]);
128        let list2 = PathList::new(&["a/c", "a/d"]);
129
130        assert_eq!(list1.paths(), list2.paths(), "paths differ");
131        assert_eq!(list1.order(), &[1, 0], "list1 order incorrect");
132        assert_eq!(list2.order(), &[0, 1], "list2 order incorrect");
133
134        let list1_deserialized = PathList::deserialize(&list1.serialize());
135        assert_eq!(list1_deserialized, list1, "list1 deserialization failed");
136
137        let list2_deserialized = PathList::deserialize(&list2.serialize());
138        assert_eq!(list2_deserialized, list2, "list2 deserialization failed");
139
140        assert_eq!(
141            list1.ordered_paths().collect_array().unwrap(),
142            [&PathBuf::from("a/d"), &PathBuf::from("a/c")],
143            "list1 ordered paths incorrect"
144        );
145        assert_eq!(
146            list2.ordered_paths().collect_array().unwrap(),
147            [&PathBuf::from("a/c"), &PathBuf::from("a/d")],
148            "list2 ordered paths incorrect"
149        );
150    }
151
152    #[test]
153    fn test_path_list_ordering() {
154        let list = PathList::new(&["b", "a", "c"]);
155        assert_eq!(
156            list.paths(),
157            &[PathBuf::from("a"), PathBuf::from("b"), PathBuf::from("c")]
158        );
159        assert_eq!(list.order(), &[1, 0, 2]);
160        assert!(!list.is_lexicographically_ordered());
161
162        let serialized = list.serialize();
163        let deserialized = PathList::deserialize(&serialized);
164        assert_eq!(deserialized, list);
165
166        assert_eq!(
167            deserialized.ordered_paths().collect_array().unwrap(),
168            [
169                &PathBuf::from("b"),
170                &PathBuf::from("a"),
171                &PathBuf::from("c")
172            ]
173        );
174
175        let list = PathList::new(&["b", "c", "a"]);
176        assert_eq!(
177            list.paths(),
178            &[PathBuf::from("a"), PathBuf::from("b"), PathBuf::from("c")]
179        );
180        assert_eq!(list.order(), &[2, 0, 1]);
181        assert!(!list.is_lexicographically_ordered());
182
183        let serialized = list.serialize();
184        let deserialized = PathList::deserialize(&serialized);
185        assert_eq!(deserialized, list);
186
187        assert_eq!(
188            deserialized.ordered_paths().collect_array().unwrap(),
189            [
190                &PathBuf::from("b"),
191                &PathBuf::from("c"),
192                &PathBuf::from("a"),
193            ]
194        );
195    }
196}