provenant/parsers/
rpm_license_files.rs

1//! Parser for RPM license files in /usr/share/licenses/ directories.
2//!
3//! Identifies packages from their license files installed in the standard
4//! /usr/share/licenses/ location, primarily used in Mariner distroless containers.
5
6use crate::models::{DatasourceId, PackageType};
7use std::path::Path;
8
9use crate::models::PackageData;
10use crate::parsers::PackageParser;
11use crate::parsers::utils::truncate_field;
12
13const PACKAGE_TYPE: PackageType = PackageType::Rpm;
14
15/// Parser for RPM license files in /usr/share/licenses/ directories.
16///
17/// Identifies packages from their license files installed in the standard
18/// /usr/share/licenses/ location, primarily used in Mariner distroless containers.
19///
20/// # Supported Formats
21/// - `/usr/share/licenses/*/COPYING*` - COPYING license files
22/// - `/usr/share/licenses/*/LICENSE*` - LICENSE files
23///
24/// # Key Features
25/// - Extracts package name from directory path
26/// - Supports Mariner distroless container convention
27/// - Package URL generation with mariner namespace
28///
29/// # Implementation Notes
30/// - Package name is extracted from the directory between `licenses/` and the filename
31/// - For example: `/usr/share/licenses/openssl/LICENSE` → package name is "openssl"
32/// - Does NOT perform license detection (that's handled by the license scanner)
33/// - datasource_id: "rpm_package_licenses"
34/// - namespace: "mariner"
35pub struct RpmLicenseFilesParser;
36
37impl PackageParser for RpmLicenseFilesParser {
38    const PACKAGE_TYPE: PackageType = PACKAGE_TYPE;
39
40    fn is_match(path: &Path) -> bool {
41        let path_str = path.to_string_lossy();
42
43        // Check if path contains usr/share/licenses/
44        if !path_str.contains("usr/share/licenses/") {
45            return false;
46        }
47
48        // Get the filename
49        if let Some(filename) = path.file_name().and_then(|f| f.to_str()) {
50            // Match files starting with COPYING or LICENSE (case-sensitive)
51            filename.starts_with("COPYING") || filename.starts_with("LICENSE")
52        } else {
53            false
54        }
55    }
56
57    fn extract_packages(path: &Path) -> Vec<PackageData> {
58        // Extract package name from path
59        let path_str = path.to_string_lossy();
60
61        // Split by usr/share/licenses/ and get the next path component
62        let name = if let Some(after_licenses) = path_str.split("usr/share/licenses/").nth(1) {
63            // Get the first path component after licenses/ (the package name)
64            after_licenses
65                .split('/')
66                .next()
67                .map(|s| truncate_field(s.to_string()))
68        } else {
69            None
70        };
71
72        // Build package data
73        let mut pkg = PackageData {
74            package_type: Some(PACKAGE_TYPE),
75            datasource_id: Some(DatasourceId::RpmPackageLicenses),
76            namespace: Some("mariner".to_string()),
77            name: name.clone(),
78            ..Default::default()
79        };
80
81        // Build PURL if we have a name
82        if let Some(ref package_name) = name {
83            use packageurl::PackageUrl;
84            if let Ok(mut purl) = PackageUrl::new(PACKAGE_TYPE.as_str(), package_name)
85                && purl.with_namespace("mariner").is_ok()
86            {
87                pkg.purl = Some(truncate_field(purl.to_string()));
88            }
89        }
90
91        vec![pkg]
92    }
93}
94
95crate::register_parser!(
96    "RPM mariner distroless package license files",
97    &[
98        "*usr/share/licenses/*/COPYING*",
99        "*usr/share/licenses/*/LICENSE*"
100    ],
101    "rpm",
102    "",
103    Some("https://github.com/microsoft/marinara/"),
104);
provenant/parsers/rpm_license_files.rs

provenant/parsers/
rpm_license_files.rs
