provenant/parsers/debian/
dsc.rs1use std::collections::HashMap;
5use std::path::Path;
6
7use crate::models::{DatasourceId, PackageData, PackageType};
8use crate::parser_warn as warn;
9use crate::parsers::rfc822;
10use crate::parsers::utils::{MAX_ITERATION_COUNT, split_name_email, truncate_field};
11
12use super::utils::{build_debian_purl, make_party, parse_dependency_field};
13use super::{PACKAGE_TYPE, default_package_data, read_or_default};
14use crate::parsers::PackageParser;
15
16pub struct DebianDscParser;
18
19impl PackageParser for DebianDscParser {
20 const PACKAGE_TYPE: PackageType = PACKAGE_TYPE;
21
22 fn is_match(path: &Path) -> bool {
23 path.extension().and_then(|e| e.to_str()) == Some("dsc")
24 }
25
26 fn extract_packages(path: &Path) -> Vec<PackageData> {
27 let content = read_or_default!(path, ".dsc file", DatasourceId::DebianSourceControlDsc);
28
29 vec![parse_dsc_content(&content)]
30 }
31}
32
33crate::register_parser!(
34 "Debian source control file (.dsc)",
35 &["**/*.dsc"],
36 "deb",
37 "",
38 Some("https://www.debian.org/doc/debian-policy/ch-controlfields.html"),
39);
40
41enum PgpParseState {
42 Normal,
43 PgpHeader,
44 PgpBody,
45 PgpSignature,
46}
47
48fn strip_pgp_signature(content: &str) -> String {
49 let mut result = String::new();
50 let mut state = PgpParseState::Normal;
51 let mut count = 0usize;
52
53 for line in content.lines() {
54 count += 1;
55 if count > MAX_ITERATION_COUNT {
56 warn!("strip_pgp_signature: exceeded MAX_ITERATION_COUNT lines, stopping");
57 break;
58 }
59 if line.starts_with("-----BEGIN PGP SIGNED MESSAGE-----") {
60 state = PgpParseState::PgpHeader;
61 continue;
62 }
63 if line.starts_with("-----BEGIN PGP SIGNATURE-----") {
64 state = PgpParseState::PgpSignature;
65 continue;
66 }
67 if line.starts_with("-----END PGP SIGNATURE-----") {
68 state = PgpParseState::Normal;
69 continue;
70 }
71 match state {
72 PgpParseState::PgpHeader => {
73 if line.starts_with("Hash:") {
74 continue;
75 }
76 if line.is_empty() && result.is_empty() {
77 state = PgpParseState::PgpBody;
78 continue;
79 }
80 }
81 PgpParseState::PgpSignature => continue,
82 PgpParseState::Normal | PgpParseState::PgpBody => {}
83 }
84 result.push_str(line);
85 result.push('\n');
86 }
87
88 result
89}
90
91fn parse_dsc_content(content: &str) -> PackageData {
92 let clean_content = strip_pgp_signature(content);
93 let metadata = rfc822::parse_rfc822_content(&clean_content);
94 let headers = &metadata.headers;
95
96 let name = rfc822::get_header_first(headers, "source").map(truncate_field);
97 let version = rfc822::get_header_first(headers, "version").map(truncate_field);
98 let architecture = rfc822::get_header_first(headers, "architecture").map(truncate_field);
99 let namespace = Some("debian".to_string());
100
101 let mut package = PackageData {
102 datasource_id: Some(DatasourceId::DebianSourceControlDsc),
103 package_type: Some(PACKAGE_TYPE),
104 namespace: namespace.clone(),
105 name: name.clone(),
106 version: version.clone(),
107 description: rfc822::get_header_first(headers, "description").map(truncate_field),
108 homepage_url: rfc822::get_header_first(headers, "homepage").map(truncate_field),
109 vcs_url: rfc822::get_header_first(headers, "vcs-git").map(truncate_field),
110 code_view_url: rfc822::get_header_first(headers, "vcs-browser").map(truncate_field),
111 ..Default::default()
112 };
113
114 if let (Some(n), Some(v)) = (&name, &version) {
116 package.purl = build_debian_purl(n, Some(v), namespace.as_deref(), architecture.as_deref());
117 }
118
119 if let Some(n) = &name
121 && let Some(source_purl) = build_debian_purl(n, None, namespace.as_deref(), None)
122 {
123 package.source_packages.push(source_purl);
124 }
125
126 if let Some(maintainer) = rfc822::get_header_first(headers, "maintainer") {
127 let (name_opt, email_opt) = split_name_email(&maintainer);
128 package
129 .parties
130 .push(make_party(None, "maintainer", name_opt, email_opt));
131 }
132
133 if let Some(uploaders_str) = rfc822::get_header_first(headers, "uploaders") {
134 for uploader in uploaders_str.split(',') {
135 let uploader = uploader.trim();
136 if uploader.is_empty() {
137 continue;
138 }
139 let (name_opt, email_opt) = split_name_email(uploader);
140 package
141 .parties
142 .push(make_party(None, "uploader", name_opt, email_opt));
143 }
144 }
145
146 if let Some(build_deps) = rfc822::get_header_first(headers, "build-depends") {
148 package.dependencies.extend(parse_dependency_field(
149 &build_deps,
150 "build",
151 false,
152 false,
153 namespace.as_deref(),
154 ));
155 }
156
157 if let Some(standards) = rfc822::get_header_first(headers, "standards-version") {
159 let map = package.extra_data.get_or_insert_with(HashMap::new);
160 map.insert("standards_version".to_string(), standards.into());
161 }
162
163 package
164}
165
166#[cfg(test)]
167mod tests {
168 use super::*;
169 use crate::models::DatasourceId;
170 use std::path::PathBuf;
171
172 #[test]
173 fn test_dsc_parser_is_match() {
174 assert!(DebianDscParser::is_match(&PathBuf::from("package.dsc")));
175 assert!(DebianDscParser::is_match(&PathBuf::from(
176 "adduser_3.118+deb11u1.dsc"
177 )));
178 assert!(!DebianDscParser::is_match(&PathBuf::from("control")));
179 assert!(!DebianDscParser::is_match(&PathBuf::from("package.txt")));
180 }
181
182 #[test]
183 fn test_dsc_parser_adduser() {
184 let path = PathBuf::from("testdata/debian/dsc_files/adduser_3.118+deb11u1.dsc");
185 let package = DebianDscParser::extract_first_package(&path);
186
187 assert_eq!(package.package_type, Some(PACKAGE_TYPE));
188 assert_eq!(package.namespace, Some("debian".to_string()));
189 assert_eq!(package.name, Some("adduser".to_string()));
190 assert_eq!(package.version, Some("3.118+deb11u1".to_string()));
191 assert_eq!(
192 package.purl,
193 Some("pkg:deb/debian/adduser@3.118%2Bdeb11u1?arch=all".to_string())
194 );
195 assert_eq!(
196 package.vcs_url,
197 Some("https://salsa.debian.org/debian/adduser.git".to_string())
198 );
199 assert_eq!(
200 package.code_view_url,
201 Some("https://salsa.debian.org/debian/adduser".to_string())
202 );
203 assert_eq!(
204 package.datasource_id,
205 Some(DatasourceId::DebianSourceControlDsc)
206 );
207
208 assert_eq!(package.parties.len(), 2);
209 assert_eq!(package.parties[0].role, Some("maintainer".to_string()));
210 assert_eq!(
211 package.parties[0].name,
212 Some("Debian Adduser Developers".to_string())
213 );
214 assert_eq!(
215 package.parties[0].email,
216 Some("adduser@packages.debian.org".to_string())
217 );
218 assert_eq!(package.parties[0].r#type, None);
219
220 assert_eq!(package.parties[1].role, Some("uploader".to_string()));
221 assert_eq!(package.parties[1].name, Some("Marc Haber".to_string()));
222 assert_eq!(
223 package.parties[1].email,
224 Some("mh+debian-packages@zugschlus.de".to_string())
225 );
226 assert_eq!(package.parties[1].r#type, None);
227
228 assert_eq!(package.source_packages.len(), 1);
229 assert_eq!(
230 package.source_packages[0],
231 "pkg:deb/debian/adduser".to_string()
232 );
233
234 assert!(!package.dependencies.is_empty());
235 let build_dep_names: Vec<String> = package
236 .dependencies
237 .iter()
238 .filter_map(|d| d.purl.as_ref())
239 .filter(|p| p.contains("po-debconf") || p.contains("debhelper"))
240 .map(|p| p.to_string())
241 .collect();
242 assert!(build_dep_names.len() >= 2);
243 }
244
245 #[test]
246 fn test_dsc_parser_zsh() {
247 let path = PathBuf::from("testdata/debian/dsc_files/zsh_5.7.1-1+deb10u1.dsc");
248 let package = DebianDscParser::extract_first_package(&path);
249
250 assert_eq!(package.name, Some("zsh".to_string()));
251 assert_eq!(package.version, Some("5.7.1-1+deb10u1".to_string()));
252 assert_eq!(package.namespace, Some("debian".to_string()));
253 assert!(package.purl.is_some());
254 assert!(package.purl.as_ref().unwrap().contains("zsh"));
255 assert!(package.purl.as_ref().unwrap().contains("5.7.1"));
256 }
257
258 #[test]
259 fn test_parse_dsc_content_basic() {
260 let content = "Format: 3.0 (native)
261Source: testpkg
262Binary: testpkg
263Architecture: amd64
264Version: 1.0.0
265Maintainer: Test User <test@example.com>
266Standards-Version: 4.5.0
267Build-Depends: debhelper (>= 12)
268Files:
269 abc123 1024 testpkg_1.0.0.tar.xz
270";
271
272 let package = parse_dsc_content(content);
273 assert_eq!(package.name, Some("testpkg".to_string()));
274 assert_eq!(package.version, Some("1.0.0".to_string()));
275 assert_eq!(package.namespace, Some("debian".to_string()));
276 assert_eq!(package.parties.len(), 1);
277 assert_eq!(package.parties[0].name, Some("Test User".to_string()));
278 assert_eq!(
279 package.parties[0].email,
280 Some("test@example.com".to_string())
281 );
282 assert_eq!(package.dependencies.len(), 1);
283 assert!(package.purl.as_ref().unwrap().contains("arch=amd64"));
284 }
285
286 #[test]
287 fn test_parse_dsc_content_with_uploaders() {
288 let content = "Source: mypkg
289Version: 2.0
290Architecture: all
291Maintainer: Main Dev <main@example.com>
292Uploaders: Dev One <dev1@example.com>, Dev Two <dev2@example.com>
293";
294
295 let package = parse_dsc_content(content);
296 assert_eq!(package.parties.len(), 3);
297 assert_eq!(package.parties[0].role, Some("maintainer".to_string()));
298 assert_eq!(package.parties[1].role, Some("uploader".to_string()));
299 assert_eq!(package.parties[2].role, Some("uploader".to_string()));
300 }
301}