1use crate::models::{DatasourceId, Dependency, FileReference, PackageData, PackageType, Party};
25use crate::parser_warn as warn;
26use crate::parsers::utils::{
27 MAX_ITERATION_COUNT, RecursionGuard, read_file_to_string, split_name_email, truncate_field,
28};
29use packageurl::PackageUrl;
30use std::path::Path;
31use toml::Value;
32
33use super::PackageParser;
34use super::license_normalization::{
35 DeclaredLicenseMatchMetadata, build_declared_license_data, empty_declared_license_data,
36 normalize_spdx_expression,
37};
38
39const FIELD_PACKAGE: &str = "package";
40const FIELD_NAME: &str = "name";
41const FIELD_VERSION: &str = "version";
42const FIELD_LICENSE: &str = "license";
43const FIELD_LICENSE_FILE: &str = "license-file";
44const FIELD_AUTHORS: &str = "authors";
45const FIELD_REPOSITORY: &str = "repository";
46const FIELD_HOMEPAGE: &str = "homepage";
47const FIELD_DEPENDENCIES: &str = "dependencies";
48const FIELD_DEV_DEPENDENCIES: &str = "dev-dependencies";
49const FIELD_DEV_DEPENDENCIES_LEGACY: &str = "dev_dependencies";
50const FIELD_BUILD_DEPENDENCIES: &str = "build-dependencies";
51const FIELD_BUILD_DEPENDENCIES_LEGACY: &str = "build_dependencies";
52const FIELD_DESCRIPTION: &str = "description";
53const FIELD_KEYWORDS: &str = "keywords";
54const FIELD_CATEGORIES: &str = "categories";
55const FIELD_RUST_VERSION: &str = "rust-version";
56const FIELD_EDITION: &str = "edition";
57const FIELD_README: &str = "readme";
58const FIELD_PUBLISH: &str = "publish";
59
60pub struct CargoParser;
65
66impl PackageParser for CargoParser {
67 const PACKAGE_TYPE: PackageType = PackageType::Cargo;
68
69 fn extract_packages(path: &Path) -> Vec<PackageData> {
70 let toml_content = match read_cargo_toml(path) {
71 Ok(content) => content,
72 Err(_) => return Vec::new(),
73 };
74
75 let package = toml_content.get(FIELD_PACKAGE).and_then(|v| v.as_table());
76
77 let name = package
78 .and_then(|p| p.get(FIELD_NAME))
79 .and_then(|v| v.as_str())
80 .map(|s| truncate_field(s.to_string()));
81
82 let version = package
83 .and_then(|p| p.get(FIELD_VERSION))
84 .and_then(|v| v.as_str())
85 .map(|s| truncate_field(s.to_string()));
86
87 let raw_license = package
88 .and_then(|p| p.get(FIELD_LICENSE))
89 .and_then(|v| v.as_str())
90 .map(|s| truncate_field(s.to_string()));
91 let file_references = extract_file_references(&toml_content);
92 let (declared_license_expression, declared_license_expression_spdx, license_detections) =
93 raw_license
94 .as_deref()
95 .and_then(normalize_spdx_expression)
96 .map(|normalized| {
97 build_declared_license_data(
98 normalized,
99 DeclaredLicenseMatchMetadata::single_line(
100 raw_license.as_deref().unwrap_or_default(),
101 ),
102 )
103 })
104 .unwrap_or_else(empty_declared_license_data);
105
106 let extracted_license_statement = raw_license.clone();
107
108 let dependencies = extract_dependencies_for_scopes(&toml_content, &[FIELD_DEPENDENCIES]);
109 let dev_dependencies = extract_dependencies_for_scopes(
110 &toml_content,
111 &[FIELD_DEV_DEPENDENCIES, FIELD_DEV_DEPENDENCIES_LEGACY],
112 );
113 let build_dependencies = extract_dependencies_for_scopes(
114 &toml_content,
115 &[FIELD_BUILD_DEPENDENCIES, FIELD_BUILD_DEPENDENCIES_LEGACY],
116 );
117
118 let purl = create_package_url(&name, &version);
119
120 let homepage_url = package
121 .and_then(|p| p.get(FIELD_HOMEPAGE))
122 .and_then(|v| v.as_str())
123 .map(|s| truncate_field(s.to_string()))
124 .or_else(|| {
125 name.as_ref()
126 .map(|n| format!("https://crates.io/crates/{}", n))
127 });
128
129 let repository_url = package
130 .and_then(|p| p.get(FIELD_REPOSITORY))
131 .and_then(|v| v.as_str())
132 .map(|s| truncate_field(s.to_string()));
133 let download_url = None;
134
135 let api_data_url = generate_cargo_api_url(&name, &version);
136
137 let repository_homepage_url = name
138 .as_ref()
139 .map(|n| format!("https://crates.io/crates/{}", n));
140
141 let repository_download_url = match (&name, &version) {
142 (Some(n), Some(v)) => Some(format!(
143 "https://crates.io/api/v1/crates/{}/{}/download",
144 n, v
145 )),
146 _ => None,
147 };
148
149 let description = package
150 .and_then(|p| p.get(FIELD_DESCRIPTION))
151 .and_then(|v| v.as_str())
152 .map(|s| truncate_field(s.trim().to_string()));
153
154 let keywords = extract_keywords_and_categories(&toml_content);
155
156 let extra_data = extract_extra_data(&toml_content);
157 let is_private = package
158 .and_then(|p| p.get(FIELD_PUBLISH))
159 .is_some_and(|value| matches!(value, Value::Boolean(false)));
160 vec![PackageData {
161 package_type: Some(Self::PACKAGE_TYPE),
162 namespace: None,
163 name,
164 version,
165 qualifiers: None,
166 subpath: None,
167 primary_language: Some("Rust".to_string()),
168 description,
169 release_date: None,
170 parties: extract_parties(&toml_content),
171 keywords,
172 homepage_url,
173 download_url,
174 size: None,
175 sha1: None,
176 md5: None,
177 sha256: None,
178 sha512: None,
179 bug_tracking_url: None,
180 code_view_url: None,
181 vcs_url: repository_url,
182 copyright: None,
183 holder: None,
184 declared_license_expression,
185 declared_license_expression_spdx,
186 license_detections,
187 other_license_expression: None,
188 other_license_expression_spdx: None,
189 other_license_detections: Vec::new(),
190 extracted_license_statement,
191 notice_text: None,
192 source_packages: Vec::new(),
193 file_references,
194 is_private,
195 is_virtual: false,
196 extra_data,
197 dependencies: [dependencies, dev_dependencies, build_dependencies].concat(),
198 repository_homepage_url,
199 repository_download_url,
200 api_data_url,
201 datasource_id: Some(DatasourceId::CargoToml),
202 purl,
203 }]
204 }
205
206 fn is_match(path: &Path) -> bool {
207 path.file_name()
208 .and_then(|name| name.to_str())
209 .is_some_and(|name| name.eq_ignore_ascii_case("cargo.toml"))
210 }
211}
212
213fn read_cargo_toml(path: &Path) -> Result<Value, String> {
215 let content =
216 read_file_to_string(path, None).map_err(|e| format!("Failed to read file: {}", e))?;
217
218 toml::from_str(&content).map_err(|e| format!("Failed to parse TOML: {}", e))
219}
220
221fn generate_cargo_api_url(name: &Option<String>, _version: &Option<String>) -> Option<String> {
222 const REGISTRY: &str = "https://crates.io/api/v1/crates";
223 name.as_ref().map(|name| format!("{}/{}", REGISTRY, name))
224}
225
226fn create_package_url(name: &Option<String>, version: &Option<String>) -> Option<String> {
227 name.as_ref().and_then(|name| {
228 let mut package_url = match PackageUrl::new(CargoParser::PACKAGE_TYPE.as_str(), name) {
229 Ok(p) => p,
230 Err(e) => {
231 warn!(
232 "Failed to create PackageUrl for cargo package '{}': {}",
233 name, e
234 );
235 return None;
236 }
237 };
238
239 if let Some(v) = version
240 && let Err(e) = package_url.with_version(v)
241 {
242 warn!(
243 "Failed to set version '{}' for cargo package '{}': {}",
244 v, name, e
245 );
246 return None;
247 }
248
249 Some(package_url.to_string())
250 })
251}
252
253fn extract_parties(toml_content: &Value) -> Vec<Party> {
255 let mut parties = Vec::new();
256
257 if let Some(package) = toml_content.get(FIELD_PACKAGE).and_then(|v| v.as_table())
258 && let Some(authors) = package.get(FIELD_AUTHORS).and_then(|v| v.as_array())
259 {
260 for author in authors.iter().take(MAX_ITERATION_COUNT) {
261 if let Some(author_str) = author.as_str() {
262 let (name, email) = split_name_email(author_str);
263 parties.push(Party {
264 r#type: None,
265 role: Some("author".to_string()),
266 name,
267 email,
268 url: None,
269 organization: None,
270 organization_url: None,
271 timezone: None,
272 });
273 }
274 }
275 if authors.len() > MAX_ITERATION_COUNT {
276 warn!(
277 "Authors array has {} entries, capping at MAX_ITERATION_COUNT ({})",
278 authors.len(),
279 MAX_ITERATION_COUNT
280 );
281 }
282 }
283
284 parties
285}
286
287fn is_cargo_version_pinned(version_str: &str) -> bool {
294 let trimmed = version_str.trim();
295
296 if trimmed.is_empty() {
298 return false;
299 }
300
301 if trimmed.contains('^')
303 || trimmed.contains('~')
304 || trimmed.contains('>')
305 || trimmed.contains('<')
306 || trimmed.contains('*')
307 || trimmed.contains('=')
308 {
309 return false;
310 }
311
312 trimmed.matches('.').count() >= 2
316}
317
318fn extract_dependencies(toml_content: &Value, scope: &str) -> Vec<Dependency> {
319 use serde_json::json;
320
321 let mut dependencies = Vec::new();
322
323 let is_runtime = !scope.ends_with("dev-dependencies") && !scope.ends_with("build-dependencies");
325
326 if let Some(deps_table) = toml_content.get(scope).and_then(|v| v.as_table()) {
327 if deps_table.len() > MAX_ITERATION_COUNT {
328 warn!(
329 "Dependency table '{}' has {} entries, capping at MAX_ITERATION_COUNT ({})",
330 scope,
331 deps_table.len(),
332 MAX_ITERATION_COUNT
333 );
334 }
335 for (name, value) in deps_table.iter().take(MAX_ITERATION_COUNT) {
336 let (extracted_requirement, is_optional, extra_data_map, is_pinned) = match value {
337 Value::String(version_str) => {
338 let pinned = is_cargo_version_pinned(version_str);
340 (
341 Some(version_str.to_string()),
342 false,
343 std::collections::HashMap::new(),
344 pinned,
345 )
346 }
347 Value::Table(table) => {
348 let version = table
350 .get("version")
351 .and_then(|v| v.as_str())
352 .map(String::from);
353
354 let pinned = version.as_ref().is_some_and(|v| is_cargo_version_pinned(v));
355
356 let is_optional = table
357 .get("optional")
358 .and_then(|v| v.as_bool())
359 .unwrap_or(false);
360
361 let mut extra_data = std::collections::HashMap::new();
362
363 for (key, val) in table {
365 match key.as_str() {
366 "version" => {
367 if let Some(v) = val.as_str() {
369 extra_data.insert("version".to_string(), json!(v));
370 }
371 }
372 "features" => {
373 if let Some(features_array) = val.as_array() {
375 let features: Vec<String> = features_array
376 .iter()
377 .filter_map(|f| f.as_str().map(String::from))
378 .collect();
379 extra_data.insert("features".to_string(), json!(features));
380 }
381 }
382 "optional" => {
383 }
385 _ => {
386 if let Some(s) = val.as_str() {
388 extra_data.insert(key.clone(), json!(s));
389 } else if let Some(b) = val.as_bool() {
390 extra_data.insert(key.clone(), json!(b));
391 } else if let Some(i) = val.as_integer() {
392 extra_data.insert(key.clone(), json!(i));
393 }
394 }
395 }
396 }
397
398 (version, is_optional, extra_data, pinned)
399 }
400 _ => {
401 continue;
403 }
404 };
405
406 if extracted_requirement.is_some() || !extra_data_map.is_empty() {
408 let purl = match PackageUrl::new(CargoParser::PACKAGE_TYPE.as_str(), name) {
409 Ok(p) => p.to_string(),
410 Err(e) => {
411 warn!(
412 "Failed to create PackageUrl for cargo dependency '{}': {}",
413 name, e
414 );
415 continue; }
417 };
418
419 dependencies.push(Dependency {
420 purl: Some(purl),
421 extracted_requirement,
422 scope: Some(scope.to_string()),
423 is_runtime: Some(is_runtime),
424 is_optional: Some(is_optional),
425 is_pinned: Some(is_pinned),
426 is_direct: Some(true),
427 resolved_package: None,
428 extra_data: if extra_data_map.is_empty() {
429 None
430 } else {
431 Some(extra_data_map)
432 },
433 });
434 }
435 }
436 }
437
438 dependencies
439}
440
441fn extract_dependencies_for_scopes(toml_content: &Value, scopes: &[&str]) -> Vec<Dependency> {
442 scopes
443 .iter()
444 .flat_map(|scope| extract_dependencies(toml_content, scope))
445 .collect()
446}
447
448fn extract_keywords_and_categories(toml_content: &Value) -> Vec<String> {
450 let mut keywords = Vec::new();
451
452 if let Some(package) = toml_content.get(FIELD_PACKAGE).and_then(|v| v.as_table()) {
453 if let Some(kw_array) = package.get(FIELD_KEYWORDS).and_then(|v| v.as_array()) {
454 if kw_array.len() > MAX_ITERATION_COUNT {
455 warn!(
456 "Keywords array has {} entries, capping at MAX_ITERATION_COUNT ({})",
457 kw_array.len(),
458 MAX_ITERATION_COUNT
459 );
460 }
461 for kw in kw_array.iter().take(MAX_ITERATION_COUNT) {
462 if let Some(kw_str) = kw.as_str() {
463 keywords.push(truncate_field(kw_str.to_string()));
464 }
465 }
466 }
467
468 if let Some(cat_array) = package.get(FIELD_CATEGORIES).and_then(|v| v.as_array()) {
469 if cat_array.len() > MAX_ITERATION_COUNT {
470 warn!(
471 "Categories array has {} entries, capping at MAX_ITERATION_COUNT ({})",
472 cat_array.len(),
473 MAX_ITERATION_COUNT
474 );
475 }
476 for cat in cat_array.iter().take(MAX_ITERATION_COUNT) {
477 if let Some(cat_str) = cat.as_str() {
478 keywords.push(truncate_field(cat_str.to_string()));
479 }
480 }
481 }
482 }
483
484 keywords
485}
486
487fn extract_file_references(toml_content: &Value) -> Vec<FileReference> {
488 let mut file_references = Vec::new();
489
490 if let Some(package) = toml_content
491 .get(FIELD_PACKAGE)
492 .and_then(|value| value.as_table())
493 {
494 for path in [
495 package
496 .get(FIELD_LICENSE_FILE)
497 .and_then(|value| value.as_str()),
498 package.get(FIELD_README).and_then(|value| value.as_str()),
499 ]
500 .into_iter()
501 .flatten()
502 {
503 if file_references
504 .iter()
505 .any(|reference: &FileReference| reference.path == path)
506 {
507 continue;
508 }
509
510 file_references.push(FileReference {
511 path: path.to_string(),
512 size: None,
513 sha1: None,
514 md5: None,
515 sha256: None,
516 sha512: None,
517 extra_data: None,
518 });
519 }
520 }
521
522 file_references
523}
524
525fn toml_to_json(value: &toml::Value, guard: &mut RecursionGuard<()>) -> serde_json::Value {
526 if guard.descend() {
527 warn!("TOML nesting depth exceeded, returning Null");
528 return serde_json::Value::Null;
529 }
530 let result = match value {
531 toml::Value::String(s) => serde_json::json!(s),
532 toml::Value::Integer(i) => serde_json::json!(i),
533 toml::Value::Float(f) => serde_json::json!(f),
534 toml::Value::Boolean(b) => serde_json::json!(b),
535 toml::Value::Array(a) => {
536 serde_json::Value::Array(a.iter().map(|v| toml_to_json(v, guard)).collect())
537 }
538 toml::Value::Table(t) => {
539 let map: serde_json::Map<String, serde_json::Value> = t
540 .iter()
541 .map(|(k, v)| (k.clone(), toml_to_json(v, guard)))
542 .collect();
543 serde_json::Value::Object(map)
544 }
545 toml::Value::Datetime(d) => serde_json::json!(d.to_string()),
546 };
547 guard.ascend();
548 result
549}
550
551fn extract_extra_data(
553 toml_content: &Value,
554) -> Option<std::collections::HashMap<String, serde_json::Value>> {
555 use serde_json::json;
556 let mut extra_data = std::collections::HashMap::new();
557
558 if let Some(package) = toml_content.get(FIELD_PACKAGE).and_then(|v| v.as_table()) {
559 if package.len() > MAX_ITERATION_COUNT {
560 warn!(
561 "Package table has {} entries, exceeding MAX_ITERATION_COUNT ({})",
562 package.len(),
563 MAX_ITERATION_COUNT
564 );
565 }
566 if let Some(rust_version_value) = package.get(FIELD_RUST_VERSION) {
567 if let Some(rust_version_str) = rust_version_value.as_str() {
568 extra_data.insert("rust_version".to_string(), json!(rust_version_str));
569 } else if rust_version_value
570 .as_table()
571 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
572 {
573 extra_data.insert("rust-version".to_string(), json!("workspace"));
574 }
575 }
576
577 if let Some(edition_value) = package.get(FIELD_EDITION) {
579 if let Some(edition_str) = edition_value.as_str() {
580 extra_data.insert("rust_edition".to_string(), json!(edition_str));
581 } else if edition_value
582 .as_table()
583 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
584 {
585 extra_data.insert("edition".to_string(), json!("workspace"));
586 }
587 }
588
589 if let Some(documentation) = package.get("documentation").and_then(|v| v.as_str()) {
591 extra_data.insert("documentation_url".to_string(), json!(documentation));
592 }
593
594 if let Some(license_file) = package.get(FIELD_LICENSE_FILE).and_then(|v| v.as_str()) {
596 extra_data.insert("license_file".to_string(), json!(license_file));
597 }
598
599 if let Some(readme_value) = package.get(FIELD_README) {
600 if let Some(readme_file) = readme_value.as_str() {
601 extra_data.insert("readme_file".to_string(), json!(readme_file));
602 } else if let Some(readme_enabled) = readme_value.as_bool() {
603 extra_data.insert("readme".to_string(), json!(readme_enabled));
604 } else if readme_value
605 .as_table()
606 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
607 {
608 extra_data.insert("readme".to_string(), json!("workspace"));
609 }
610 }
611
612 if let Some(publish_value) = package.get(FIELD_PUBLISH) {
613 extra_data.insert(
614 "publish".to_string(),
615 toml_to_json(publish_value, &mut RecursionGuard::depth_only()),
616 );
617 }
618
619 if let Some(version_value) = package.get(FIELD_VERSION)
622 && version_value
623 .as_table()
624 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
625 {
626 extra_data.insert("version".to_string(), json!("workspace"));
627 }
628
629 if let Some(license_value) = package.get(FIELD_LICENSE)
631 && license_value
632 .as_table()
633 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
634 {
635 extra_data.insert("license".to_string(), json!("workspace"));
636 }
637
638 if let Some(homepage_value) = package.get(FIELD_HOMEPAGE)
640 && homepage_value
641 .as_table()
642 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
643 {
644 extra_data.insert("homepage".to_string(), json!("workspace"));
645 }
646
647 if let Some(repository_value) = package.get(FIELD_REPOSITORY)
649 && repository_value
650 .as_table()
651 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
652 {
653 extra_data.insert("repository".to_string(), json!("workspace"));
654 }
655
656 if let Some(categories_value) = package.get(FIELD_CATEGORIES)
658 && categories_value
659 .as_table()
660 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
661 {
662 extra_data.insert("categories".to_string(), json!("workspace"));
663 }
664
665 if let Some(authors_value) = package.get(FIELD_AUTHORS)
667 && authors_value
668 .as_table()
669 .is_some_and(|t| t.get("workspace") == Some(&toml::Value::Boolean(true)))
670 {
671 extra_data.insert("authors".to_string(), json!("workspace"));
672 }
673 }
674
675 if let Some(workspace_value) = toml_content.get("workspace") {
677 extra_data.insert(
678 "workspace".to_string(),
679 toml_to_json(workspace_value, &mut RecursionGuard::depth_only()),
680 );
681 }
682
683 if extra_data.is_empty() {
684 None
685 } else {
686 Some(extra_data)
687 }
688}
689
690crate::register_parser!(
691 "Rust Cargo.toml manifest",
692 &["**/Cargo.toml", "**/cargo.toml"],
693 "cargo",
694 "Rust",
695 Some("https://doc.rust-lang.org/cargo/reference/manifest.html"),
696);