sbom_tools/matching/
aliases.rs1use serde::{Deserialize, Serialize};
4use std::collections::{HashMap, HashSet};
5
6#[derive(Debug, Clone, Default, Serialize, Deserialize)]
8pub struct AliasTable {
9 alias_to_canonical: HashMap<String, String>,
11 canonical_to_aliases: HashMap<String, HashSet<String>>,
13}
14
15impl AliasTable {
16 #[must_use]
18 pub fn new() -> Self {
19 Self::default()
20 }
21
22 #[must_use]
24 pub fn with_builtins() -> Self {
25 let mut table = Self::new();
26 table.load_builtins();
27 table
28 }
29
30 fn load_builtins(&mut self) {
32 self.add_aliases("pkg:pypi/pillow", &["PIL", "python-pillow", "pillow"]);
34 self.add_aliases("pkg:pypi/scikit-learn", &["sklearn", "scikit_learn"]);
35 self.add_aliases(
36 "pkg:pypi/beautifulsoup4",
37 &["bs4", "BeautifulSoup", "beautifulsoup"],
38 );
39 self.add_aliases("pkg:pypi/pyyaml", &["yaml", "PyYAML"]);
40 self.add_aliases(
41 "pkg:pypi/opencv-python",
42 &["cv2", "opencv-python-headless", "opencv"],
43 );
44 self.add_aliases("pkg:pypi/python-dateutil", &["dateutil"]);
45 self.add_aliases("pkg:pypi/attrs", &["attr"]);
46 self.add_aliases("pkg:pypi/importlib-metadata", &["importlib_metadata"]);
47 self.add_aliases("pkg:pypi/typing-extensions", &["typing_extensions"]);
48 self.add_aliases("pkg:pypi/zipp", &["zipfile"]);
49
50 self.add_aliases(
52 "pkg:npm/lodash",
53 &["lodash-es", "lodash.merge", "lodash.get"],
54 );
55 self.add_aliases("pkg:npm/react", &["react-dom"]);
56 self.add_aliases("pkg:npm/webpack", &["webpack-cli"]);
57
58 self.add_aliases(
60 "yaml-parser",
61 &["pyyaml", "js-yaml", "serde_yaml", "gopkg.in/yaml"],
62 );
63 self.add_aliases("json-parser", &["serde_json", "json", "encoding/json"]);
64 }
65
66 pub fn add_aliases(&mut self, canonical: &str, aliases: &[&str]) {
68 let canonical_lower = canonical.to_lowercase();
69
70 self.alias_to_canonical
72 .insert(canonical_lower.clone(), canonical_lower.clone());
73
74 let alias_set = self
76 .canonical_to_aliases
77 .entry(canonical_lower.clone())
78 .or_default();
79 alias_set.insert(canonical_lower.clone());
80
81 for alias in aliases {
83 let alias_lower = alias.to_lowercase();
84 self.alias_to_canonical
85 .insert(alias_lower.clone(), canonical_lower.clone());
86 if let Some(set) = self.canonical_to_aliases.get_mut(&canonical_lower) {
87 set.insert(alias_lower);
88 }
89 }
90 }
91
92 #[must_use]
94 pub fn get_canonical(&self, alias: &str) -> Option<String> {
95 self.alias_to_canonical.get(&alias.to_lowercase()).cloned()
96 }
97
98 #[must_use]
100 pub fn is_alias(&self, canonical: &str, name: &str) -> bool {
101 let canonical_lower = canonical.to_lowercase();
102 let name_lower = name.to_lowercase();
103
104 self.canonical_to_aliases.get(&canonical_lower).is_some_and(|aliases| aliases.contains(&name_lower))
105 }
106
107 #[must_use]
109 pub fn get_aliases(&self, canonical: &str) -> Option<&HashSet<String>> {
110 self.canonical_to_aliases.get(&canonical.to_lowercase())
111 }
112
113 pub fn load_json(&mut self, json: &str) -> Result<(), serde_json::Error> {
115 let entries: Vec<AliasEntry> = serde_json::from_str(json)?;
116 for entry in entries {
117 let aliases: Vec<&str> = entry.aliases.iter().map(std::string::String::as_str).collect();
118 self.add_aliases(&entry.canonical, &aliases);
119 }
120 Ok(())
121 }
122
123 pub fn to_json(&self) -> Result<String, serde_json::Error> {
125 let entries: Vec<AliasEntry> = self
126 .canonical_to_aliases
127 .iter()
128 .map(|(canonical, aliases)| AliasEntry {
129 canonical: canonical.clone(),
130 aliases: aliases.iter().cloned().collect(),
131 })
132 .collect();
133 serde_json::to_string_pretty(&entries)
134 }
135}
136
137#[derive(Debug, Serialize, Deserialize)]
139struct AliasEntry {
140 canonical: String,
141 aliases: Vec<String>,
142}
143
144#[cfg(test)]
145mod tests {
146 use super::*;
147
148 #[test]
149 fn test_alias_lookup() {
150 let table = AliasTable::with_builtins();
151
152 assert_eq!(
154 table.get_canonical("PIL"),
155 Some("pkg:pypi/pillow".to_lowercase())
156 );
157
158 assert_eq!(
160 table.get_canonical("sklearn"),
161 Some("pkg:pypi/scikit-learn".to_lowercase())
162 );
163 }
164
165 #[test]
166 fn test_is_alias() {
167 let table = AliasTable::with_builtins();
168
169 assert!(table.is_alias("pkg:pypi/pillow", "PIL"));
170 assert!(table.is_alias("pkg:pypi/pillow", "pillow"));
171 assert!(!table.is_alias("pkg:pypi/pillow", "numpy"));
172 }
173
174 #[test]
175 fn test_custom_aliases() {
176 let mut table = AliasTable::new();
177 table.add_aliases("my-package", &["my_package", "mypackage"]);
178
179 assert_eq!(
180 table.get_canonical("my_package"),
181 Some("my-package".to_string())
182 );
183 assert!(table.is_alias("my-package", "mypackage"));
184 }
185}