provenant/parsers/
os_release.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4//! Parser for Linux OS release metadata files.
5//!
6//! Extracts distribution information from `/etc/os-release` and `/usr/lib/os-release`
7//! files which identify the Linux distribution and version.
8//!
9//! # Supported Formats
10//! - `/etc/os-release` (primary location)
11//! - `/usr/lib/os-release` (fallback location)
12//!
13//! # Key Features
14//! - Distribution identification (name, version, ID)
15//! - Namespace mapping (debian, fedora, etc.)
16//! - Pretty name extraction
17//! - Version ID parsing
18//!
19//! # Implementation Notes
20//! - Format: shell-compatible key=value pairs
21//! - Values may be quoted with single or double quotes
22//! - Comments start with #
23//! - Spec: https://www.freedesktop.org/software/systemd/man/os-release.html
24
25use crate::models::{DatasourceId, PackageType};
26use std::collections::HashMap;
27use std::path::Path;
28
29use crate::parser_warn as warn;
30
31use crate::models::PackageData;
32
33use super::PackageParser;
34use super::metadata::ParserMetadata;
35use super::utils::{MAX_ITERATION_COUNT, read_file_to_string, truncate_field};
36
37const PACKAGE_TYPE: PackageType = PackageType::LinuxDistro;
38
39/// Parser for Linux OS release metadata files
40pub struct OsReleaseParser;
41
42impl PackageParser for OsReleaseParser {
43    const PACKAGE_TYPE: PackageType = PACKAGE_TYPE;
44
45    fn metadata() -> Vec<ParserMetadata> {
46        vec![ParserMetadata {
47            description: "Linux OS release metadata file",
48            file_patterns: &["*etc/os-release", "*usr/lib/os-release"],
49            package_type: "linux-distro",
50            primary_language: "",
51            documentation_url: Some(
52                "https://www.freedesktop.org/software/systemd/man/os-release.html",
53            ),
54        }]
55    }
56
57    fn is_match(path: &Path) -> bool {
58        path.to_str()
59            .is_some_and(|p| p.ends_with("/etc/os-release") || p.ends_with("/usr/lib/os-release"))
60    }
61
62    fn extract_packages(path: &Path) -> Vec<PackageData> {
63        let content = match read_file_to_string(path, None) {
64            Ok(c) => c,
65            Err(e) => {
66                warn!("Failed to read os-release file {:?}: {}", path, e);
67                return vec![PackageData {
68                    package_type: Some(PACKAGE_TYPE),
69                    datasource_id: Some(DatasourceId::EtcOsRelease),
70                    ..Default::default()
71                }];
72            }
73        };
74
75        vec![parse_os_release(&content)]
76    }
77}
78
79pub(crate) fn parse_os_release(content: &str) -> PackageData {
80    let fields = parse_key_value_pairs(content);
81
82    let id = fields.get("ID").map(|s| s.as_str()).unwrap_or("");
83    let id_like = fields.get("ID_LIKE").map(|s| s.as_str());
84    let pretty_name = fields
85        .get("PRETTY_NAME")
86        .map(|s| s.to_lowercase())
87        .unwrap_or_default();
88    let version_id = fields.get("VERSION_ID").cloned();
89
90    // Namespace and name mapping logic from Python reference
91    let (namespace, name) = determine_namespace_and_name(id, id_like, &pretty_name);
92
93    let homepage_url = fields.get("HOME_URL").cloned().map(truncate_field);
94    let bug_tracking_url = fields.get("BUG_REPORT_URL").cloned().map(truncate_field);
95    let code_view_url = fields.get("SUPPORT_URL").cloned().map(truncate_field);
96
97    PackageData {
98        package_type: Some(PACKAGE_TYPE),
99        namespace: Some(truncate_field(namespace.to_string())),
100        name: Some(truncate_field(name.to_string())),
101        version: version_id.map(truncate_field),
102        homepage_url,
103        bug_tracking_url,
104        code_view_url,
105        datasource_id: Some(DatasourceId::EtcOsRelease),
106        ..Default::default()
107    }
108}
109
110fn determine_namespace_and_name<'a>(
111    id: &'a str,
112    id_like: Option<&'a str>,
113    pretty_name: &'a str,
114) -> (&'a str, &'a str) {
115    match id {
116        "debian" => {
117            let name = if pretty_name.contains("distroless") {
118                "distroless"
119            } else {
120                "debian"
121            };
122            ("debian", name)
123        }
124        "ubuntu" if id_like == Some("debian") => ("debian", "ubuntu"),
125        id if id.starts_with("fedora") || id_like == Some("fedora") => {
126            let name = id_like.unwrap_or(id);
127            (id, name)
128        }
129        _ => {
130            let name = id_like.unwrap_or(id);
131            (id, name)
132        }
133    }
134}
135
136fn parse_key_value_pairs(content: &str) -> HashMap<String, String> {
137    let mut fields = HashMap::new();
138
139    for line in content.lines().take(MAX_ITERATION_COUNT) {
140        let line = line.trim();
141
142        // Skip empty lines and comments
143        if line.is_empty() || line.starts_with('#') {
144            continue;
145        }
146
147        // Parse KEY=VALUE format
148        if let Some((key, value)) = line.split_once('=') {
149            let key = key.trim().to_string();
150            let value = unquote(value.trim());
151            fields.insert(key, value);
152        }
153    }
154
155    fields
156}
157
158fn unquote(s: &str) -> String {
159    let s = s.trim();
160    if (s.starts_with('"') && s.ends_with('"')) || (s.starts_with('\'') && s.ends_with('\'')) {
161        s[1..s.len() - 1].to_string()
162    } else {
163        s.to_string()
164    }
165}
provenant/parsers/os_release.rs

provenant/parsers/
os_release.rs
