Skip to main content

provenant/parsers/nuget/
nuspec.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4use std::fs::File;
5use std::io::BufReader;
6use std::path::Path;
7
8use crate::models::{DatasourceId, Dependency, PackageData, PackageType};
9use crate::parser_warn as warn;
10use packageurl::PackageUrl;
11use quick_xml::Reader;
12use quick_xml::events::Event;
13
14use super::super::PackageParser;
15use super::super::license_normalization::{
16    empty_declared_license_data, normalize_spdx_declared_license,
17};
18use super::super::utils::{MAX_ITERATION_COUNT, truncate_field};
19use super::{
20    build_nuget_description, build_nuget_party, build_nuget_purl, build_nuget_urls,
21    check_file_size, default_package_data, insert_extra_string, parse_repository_metadata,
22};
23
24pub struct NuspecParser;
25
26impl PackageParser for NuspecParser {
27    const PACKAGE_TYPE: PackageType = PackageType::Nuget;
28
29    fn is_match(path: &Path) -> bool {
30        path.extension()
31            .and_then(|ext| ext.to_str())
32            .is_some_and(|ext| ext == "nuspec")
33    }
34
35    fn extract_packages(path: &Path) -> Vec<PackageData> {
36        if let Err(e) = check_file_size(path) {
37            warn!("{}", e);
38            return vec![default_package_data(Some(DatasourceId::NugetNuspec))];
39        }
40
41        let file = match File::open(path) {
42            Ok(f) => f,
43            Err(e) => {
44                warn!("Failed to open .nuspec at {:?}: {}", path, e);
45                return vec![default_package_data(Some(DatasourceId::NugetNuspec))];
46            }
47        };
48
49        let reader = BufReader::new(file);
50        let mut xml_reader = Reader::from_reader(reader);
51        xml_reader.config_mut().trim_text(true);
52
53        let mut name = None;
54        let mut version = None;
55        let mut summary = None;
56        let mut description = None;
57        let mut title = None;
58        let mut homepage_url = None;
59        let mut parties = Vec::new();
60        let mut dependencies = Vec::new();
61        let mut extracted_license_statement = None;
62        let mut license_type = None;
63        let mut copyright = None;
64        let mut vcs_url = None;
65        let mut repository_branch = None;
66        let mut repository_commit = None;
67
68        let mut buf = Vec::new();
69        let mut current_element = String::new();
70        let mut in_metadata = false;
71        let mut in_dependencies = false;
72        let mut current_group_framework = None;
73        let mut iteration_count: usize = 0;
74
75        loop {
76            iteration_count += 1;
77            if iteration_count > MAX_ITERATION_COUNT {
78                warn!(
79                    "Iteration limit exceeded in .nuspec at {:?}; stopping at {} items",
80                    path, MAX_ITERATION_COUNT
81                );
82                break;
83            }
84            match xml_reader.read_event_into(&mut buf) {
85                Ok(Event::Start(e)) => {
86                    let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
87                    current_element = tag_name.clone();
88
89                    if tag_name == "metadata" {
90                        in_metadata = true;
91                    } else if tag_name == "dependencies" && in_metadata {
92                        in_dependencies = true;
93                    } else if tag_name == "group" && in_dependencies {
94                        current_group_framework = e
95                            .attributes()
96                            .filter_map(|a| a.ok())
97                            .find(|attr| attr.key.as_ref() == b"targetFramework")
98                            .and_then(|attr| String::from_utf8(attr.value.to_vec()).ok());
99                    } else if tag_name == "repository" && in_metadata {
100                        let repository = parse_repository_metadata(&e);
101                        vcs_url = repository.vcs_url;
102                        repository_branch = repository.branch;
103                        repository_commit = repository.commit;
104                    } else if tag_name == "license" && in_metadata {
105                        license_type = e
106                            .attributes()
107                            .filter_map(|a| a.ok())
108                            .find(|attr| attr.key.as_ref() == b"type")
109                            .and_then(|attr| String::from_utf8(attr.value.to_vec()).ok());
110                    }
111                }
112                Ok(Event::Empty(e)) => {
113                    let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
114
115                    if tag_name == "dependency" && in_dependencies {
116                        if let Some(dep) =
117                            parse_nuspec_dependency(&e, current_group_framework.as_deref())
118                        {
119                            dependencies.push(dep);
120                        }
121                    } else if tag_name == "repository" && in_metadata {
122                        let repository = parse_repository_metadata(&e);
123                        vcs_url = repository.vcs_url;
124                        repository_branch = repository.branch;
125                        repository_commit = repository.commit;
126                    }
127                }
128                Ok(Event::Text(e)) => {
129                    if !in_metadata {
130                        continue;
131                    }
132
133                    let text = e.decode().ok().map(|s| s.trim().to_string());
134                    if let Some(text) = text.filter(|s| !s.is_empty()) {
135                        match current_element.as_str() {
136                            "id" => name = Some(text),
137                            "version" => version = Some(text),
138                            "summary" => summary = Some(text),
139                            "description" => description = Some(text),
140                            "title" => title = Some(text),
141                            "projectUrl" => homepage_url = Some(text),
142                            "authors" => {
143                                parties.push(build_nuget_party("author", text));
144                            }
145                            "owners" => {
146                                parties.push(build_nuget_party("owner", text));
147                            }
148                            "license" => {
149                                extracted_license_statement = Some(text);
150                            }
151                            "licenseUrl" if extracted_license_statement.is_none() => {
152                                extracted_license_statement = Some(text);
153                            }
154                            "licenseUrl" => {}
155                            "copyright" => copyright = Some(text),
156                            _ => {}
157                        }
158                    }
159                }
160                Ok(Event::End(e)) => {
161                    let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
162
163                    if tag_name == "metadata" {
164                        in_metadata = false;
165                    } else if tag_name == "dependencies" {
166                        in_dependencies = false;
167                    } else if tag_name == "group" {
168                        current_group_framework = None;
169                    }
170
171                    current_element.clear();
172                }
173                Ok(Event::Eof) => break,
174                Err(e) => {
175                    warn!("Error parsing .nuspec at {:?}: {}", path, e);
176                    return vec![default_package_data(Some(DatasourceId::NugetNuspec))];
177                }
178                _ => {}
179            }
180            buf.clear();
181        }
182
183        let final_description = build_nuget_description(
184            summary.as_deref(),
185            description.as_deref(),
186            title.as_deref(),
187            name.as_deref(),
188        );
189
190        let (repository_homepage_url, repository_download_url, api_data_url) =
191            build_nuget_urls(name.as_deref(), version.as_deref());
192
193        let purl = build_nuget_purl(name.as_deref(), version.as_deref());
194
195        let (declared_license_expression, declared_license_expression_spdx, license_detections) =
196            if license_type.as_deref() == Some("expression") {
197                normalize_spdx_declared_license(extracted_license_statement.as_deref())
198            } else {
199                empty_declared_license_data()
200            };
201
202        let holder = None;
203
204        let mut extra_data = serde_json::Map::new();
205        insert_extra_string(&mut extra_data, "license_type", license_type.clone());
206        if license_type.as_deref() == Some("file") {
207            insert_extra_string(
208                &mut extra_data,
209                "license_file",
210                extracted_license_statement.clone(),
211            );
212        }
213        insert_extra_string(&mut extra_data, "repository_branch", repository_branch);
214        insert_extra_string(&mut extra_data, "repository_commit", repository_commit);
215
216        vec![PackageData {
217            datasource_id: Some(DatasourceId::NugetNuspec),
218            package_type: Some(Self::PACKAGE_TYPE),
219            name: name.map(truncate_field),
220            version: version.map(truncate_field),
221            purl,
222            description: final_description.map(truncate_field),
223            homepage_url: homepage_url.map(truncate_field),
224            parties,
225            dependencies,
226            declared_license_expression,
227            declared_license_expression_spdx,
228            license_detections,
229            extracted_license_statement: extracted_license_statement.map(truncate_field),
230            copyright: copyright.map(truncate_field),
231            holder,
232            vcs_url: vcs_url.map(truncate_field),
233            extra_data: if extra_data.is_empty() {
234                None
235            } else {
236                Some(extra_data.into_iter().collect())
237            },
238            repository_homepage_url,
239            repository_download_url,
240            api_data_url,
241            ..default_package_data(Some(DatasourceId::NugetNuspec))
242        }]
243    }
244}
245
246pub(super) fn parse_nuspec_dependency(
247    element: &quick_xml::events::BytesStart,
248    framework: Option<&str>,
249) -> Option<Dependency> {
250    let mut id = None;
251    let mut version = None;
252    let mut include = None;
253    let mut exclude = None;
254
255    for attr in element.attributes().filter_map(|a| a.ok()) {
256        match attr.key.as_ref() {
257            b"id" => id = String::from_utf8(attr.value.to_vec()).ok(),
258            b"version" => version = String::from_utf8(attr.value.to_vec()).ok(),
259            b"include" => include = String::from_utf8(attr.value.to_vec()).ok(),
260            b"exclude" => exclude = String::from_utf8(attr.value.to_vec()).ok(),
261            _ => {}
262        }
263    }
264
265    let name = id?;
266    let purl = PackageUrl::new("nuget", &name).ok().map(|p| p.to_string());
267
268    let mut extra_data = serde_json::Map::new();
269    if let Some(fw) = framework {
270        extra_data.insert(
271            "framework".to_string(),
272            serde_json::Value::String(fw.to_string()),
273        );
274    }
275    if let Some(inc) = include {
276        extra_data.insert("include".to_string(), serde_json::Value::String(inc));
277    }
278    if let Some(exc) = exclude {
279        extra_data.insert("exclude".to_string(), serde_json::Value::String(exc));
280    }
281
282    Some(Dependency {
283        purl,
284        extracted_requirement: version,
285        scope: Some("dependency".to_string()),
286        is_runtime: Some(true),
287        is_optional: Some(false),
288        is_pinned: Some(false),
289        is_direct: Some(true),
290        resolved_package: None,
291        extra_data: if extra_data.is_empty() {
292            None
293        } else {
294            Some(extra_data.into_iter().collect())
295        },
296    })
297}
298
299pub(super) fn parse_nuspec_content(content: &str) -> Result<PackageData, String> {
300    use quick_xml::Reader;
301
302    let mut xml_reader = Reader::from_str(content);
303    xml_reader.config_mut().trim_text(true);
304
305    let mut name = None;
306    let mut version = None;
307    let mut description = None;
308    let mut homepage_url = None;
309    let mut parties = Vec::new();
310    let mut dependencies = Vec::new();
311    let mut extracted_license_statement = None;
312    let mut license_type = None;
313    let mut copyright = None;
314    let mut vcs_url = None;
315    let mut repository_branch = None;
316    let mut repository_commit = None;
317
318    let mut buf = Vec::new();
319    let mut current_element = String::new();
320    let mut in_metadata = false;
321    let mut in_dependencies = false;
322    let mut current_group_framework = None;
323    let mut iteration_count: usize = 0;
324
325    loop {
326        iteration_count += 1;
327        if iteration_count > MAX_ITERATION_COUNT {
328            return Err(format!(
329                "Iteration limit exceeded parsing .nuspec content; stopping at {} items",
330                MAX_ITERATION_COUNT
331            ));
332        }
333        match xml_reader.read_event_into(&mut buf) {
334            Ok(Event::Start(e)) => {
335                let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
336                current_element = tag_name.clone();
337
338                if tag_name == "metadata" {
339                    in_metadata = true;
340                } else if tag_name == "dependencies" && in_metadata {
341                    in_dependencies = true;
342                } else if tag_name == "group" && in_dependencies {
343                    current_group_framework = e
344                        .attributes()
345                        .filter_map(|a| a.ok())
346                        .find(|attr| attr.key.as_ref() == b"targetFramework")
347                        .and_then(|attr| String::from_utf8(attr.value.to_vec()).ok());
348                } else if tag_name == "repository" && in_metadata {
349                    let repository = parse_repository_metadata(&e);
350                    vcs_url = repository.vcs_url;
351                    repository_branch = repository.branch;
352                    repository_commit = repository.commit;
353                } else if tag_name == "license" && in_metadata {
354                    license_type = e
355                        .attributes()
356                        .filter_map(|a| a.ok())
357                        .find(|attr| attr.key.as_ref() == b"type")
358                        .and_then(|attr| String::from_utf8(attr.value.to_vec()).ok());
359                }
360            }
361            Ok(Event::Empty(e)) => {
362                let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
363
364                if tag_name == "dependency" && in_dependencies {
365                    if let Some(dep) =
366                        parse_nuspec_dependency(&e, current_group_framework.as_deref())
367                    {
368                        dependencies.push(dep);
369                    }
370                } else if tag_name == "repository" && in_metadata {
371                    let repository = parse_repository_metadata(&e);
372                    vcs_url = repository.vcs_url;
373                    repository_branch = repository.branch;
374                    repository_commit = repository.commit;
375                }
376            }
377            Ok(Event::Text(e)) => {
378                if !in_metadata {
379                    continue;
380                }
381
382                let text = e.decode().ok().map(|s| s.trim().to_string());
383                if let Some(text) = text.filter(|s| !s.is_empty()) {
384                    match current_element.as_str() {
385                        "id" => name = Some(text),
386                        "version" => version = Some(text),
387                        "description" => description = Some(text),
388                        "projectUrl" => homepage_url = Some(text),
389                        "authors" => {
390                            parties.push(build_nuget_party("author", text));
391                        }
392                        "owners" => {
393                            parties.push(build_nuget_party("owner", text));
394                        }
395                        "license" => {
396                            extracted_license_statement = Some(text);
397                        }
398                        "licenseUrl" if extracted_license_statement.is_none() => {
399                            extracted_license_statement = Some(text);
400                        }
401                        "licenseUrl" => {}
402                        "copyright" => copyright = Some(text),
403                        _ => {}
404                    }
405                }
406            }
407            Ok(Event::End(e)) => {
408                let tag_name = String::from_utf8_lossy(e.name().as_ref()).to_string();
409
410                if tag_name == "metadata" {
411                    in_metadata = false;
412                } else if tag_name == "dependencies" {
413                    in_dependencies = false;
414                } else if tag_name == "group" {
415                    current_group_framework = None;
416                }
417
418                current_element.clear();
419            }
420            Ok(Event::Eof) => break,
421            Err(e) => {
422                return Err(format!("XML parsing error: {}", e));
423            }
424            _ => {}
425        }
426        buf.clear();
427    }
428
429    let (repository_homepage_url, repository_download_url, api_data_url) =
430        build_nuget_urls(name.as_deref(), version.as_deref());
431
432    let (declared_license_expression, declared_license_expression_spdx, license_detections) =
433        if license_type.as_deref() == Some("expression") {
434            normalize_spdx_declared_license(extracted_license_statement.as_deref())
435        } else {
436            empty_declared_license_data()
437        };
438
439    let holder = None;
440
441    let mut extra_data = serde_json::Map::new();
442    insert_extra_string(&mut extra_data, "license_type", license_type.clone());
443    if license_type.as_deref() == Some("file") {
444        insert_extra_string(
445            &mut extra_data,
446            "license_file",
447            extracted_license_statement.clone(),
448        );
449    }
450    insert_extra_string(&mut extra_data, "repository_branch", repository_branch);
451    insert_extra_string(&mut extra_data, "repository_commit", repository_commit);
452
453    Ok(PackageData {
454        datasource_id: Some(DatasourceId::NugetNupkg),
455        package_type: Some(super::nupkg::NupkgParser::PACKAGE_TYPE),
456        name: name.map(truncate_field),
457        version: version.map(truncate_field),
458        description: description.map(truncate_field),
459        homepage_url: homepage_url.map(truncate_field),
460        parties,
461        dependencies,
462        declared_license_expression,
463        declared_license_expression_spdx,
464        license_detections,
465        extracted_license_statement: extracted_license_statement.map(truncate_field),
466        copyright: copyright.map(truncate_field),
467        holder,
468        vcs_url: vcs_url.map(truncate_field),
469        extra_data: if extra_data.is_empty() {
470            None
471        } else {
472            Some(extra_data.into_iter().collect())
473        },
474        repository_homepage_url,
475        repository_download_url,
476        api_data_url,
477        ..default_package_data(Some(DatasourceId::NugetNupkg))
478    })
479}
480
481crate::register_parser!(
482    ".NET .nuspec package specification",
483    &["**/*.nuspec"],
484    "nuget",
485    "C#",
486    Some("https://learn.microsoft.com/en-us/nuget/reference/nuspec"),
487);