Skip to main content

provenant/output_schema/
file_info.rs

1use serde::{Deserialize, Serialize, Serializer};
2use serde_json::Map;
3
4use super::author::OutputAuthor;
5use super::copyright::OutputCopyright;
6use super::email::OutputEmail;
7use super::holder::OutputHolder;
8use super::license_detection::OutputLicenseDetection;
9use super::license_match::OutputMatch;
10use super::license_policy_entry::OutputLicensePolicyEntry;
11use super::package_data::OutputPackageData;
12use super::serde_helpers::insert_json;
13use super::tallies::OutputTallies;
14use super::url::OutputURL;
15
16#[derive(Debug, Clone, Deserialize)]
17pub struct OutputFileInfo {
18    pub name: String,
19    pub base_name: String,
20    pub extension: String,
21    pub path: String,
22    #[serde(rename = "type")]
23    pub file_type: crate::models::FileType,
24    pub mime_type: Option<String>,
25    pub file_type_label: Option<String>,
26    pub size: u64,
27    pub date: Option<String>,
28    pub sha1: Option<String>,
29    pub md5: Option<String>,
30    pub sha256: Option<String>,
31    pub sha1_git: Option<String>,
32    pub programming_language: Option<String>,
33    #[serde(default)]
34    pub package_data: Vec<OutputPackageData>,
35    #[serde(rename = "detected_license_expression_spdx")]
36    pub license_expression: Option<String>,
37    #[serde(default)]
38    pub license_detections: Vec<OutputLicenseDetection>,
39    #[serde(default, skip_serializing_if = "Vec::is_empty")]
40    pub license_clues: Vec<OutputMatch>,
41    pub percentage_of_license_text: Option<f64>,
42    #[serde(default)]
43    pub copyrights: Vec<OutputCopyright>,
44    #[serde(default)]
45    pub holders: Vec<OutputHolder>,
46    #[serde(default)]
47    pub authors: Vec<OutputAuthor>,
48    #[serde(default, skip_serializing_if = "Vec::is_empty")]
49    pub emails: Vec<OutputEmail>,
50    #[serde(default)]
51    pub urls: Vec<OutputURL>,
52    #[serde(default)]
53    pub for_packages: Vec<String>,
54    #[serde(default)]
55    pub scan_errors: Vec<String>,
56    pub license_policy: Option<Vec<OutputLicensePolicyEntry>>,
57    pub is_generated: Option<bool>,
58    pub is_binary: Option<bool>,
59    pub is_text: Option<bool>,
60    pub is_archive: Option<bool>,
61    pub is_media: Option<bool>,
62    pub is_source: Option<bool>,
63    pub is_script: Option<bool>,
64    pub files_count: Option<usize>,
65    pub dirs_count: Option<usize>,
66    pub size_count: Option<u64>,
67    pub source_count: Option<usize>,
68    #[serde(default, skip_serializing_if = "is_false")]
69    pub is_legal: bool,
70    #[serde(default, skip_serializing_if = "is_false")]
71    pub is_manifest: bool,
72    #[serde(default, skip_serializing_if = "is_false")]
73    pub is_readme: bool,
74    #[serde(default, skip_serializing_if = "is_false")]
75    pub is_top_level: bool,
76    #[serde(default, skip_serializing_if = "is_false")]
77    pub is_key_file: bool,
78    #[serde(default, skip_serializing_if = "is_false")]
79    pub is_community: bool,
80    #[serde(default, skip_serializing_if = "Vec::is_empty")]
81    pub facets: Vec<String>,
82    pub tallies: Option<OutputTallies>,
83}
84
85impl OutputFileInfo {
86    pub(crate) fn should_serialize_info_surface(&self) -> bool {
87        self.date.is_some()
88            || self.sha1.is_some()
89            || self.md5.is_some()
90            || self.sha256.is_some()
91            || self.sha1_git.is_some()
92            || self.mime_type.is_some()
93            || self.file_type_label.is_some()
94            || self.programming_language.is_some()
95            || self.is_binary.is_some()
96            || self.is_text.is_some()
97            || self.is_archive.is_some()
98            || self.is_media.is_some()
99            || self.is_source.is_some()
100            || self.is_script.is_some()
101            || self.files_count.is_some()
102            || self.dirs_count.is_some()
103            || self.size_count.is_some()
104    }
105
106    pub(crate) fn should_serialize_license_surface(&self) -> bool {
107        self.license_expression.is_some()
108            || !self.license_detections.is_empty()
109            || !self.license_clues.is_empty()
110            || self.percentage_of_license_text.is_some()
111    }
112
113    pub(crate) fn detected_license_expression_spdx(&self) -> Option<String> {
114        crate::utils::spdx::combine_license_expressions(
115            self.license_detections
116                .iter()
117                .filter(|detection| !detection.license_expression_spdx.is_empty())
118                .map(|detection| detection.license_expression_spdx.clone()),
119        )
120        .or_else(|| {
121            crate::utils::spdx::combine_license_expressions(
122                self.package_data
123                    .iter()
124                    .flat_map(|package_data| package_data.license_detections.iter())
125                    .filter(|detection| !detection.license_expression_spdx.is_empty())
126                    .map(|detection| detection.license_expression_spdx.clone()),
127            )
128        })
129        .or_else(|| {
130            self.license_expression
131                .clone()
132                .filter(|expression| !expression.is_empty())
133        })
134    }
135}
136
137impl Serialize for OutputFileInfo {
138    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
139    where
140        S: Serializer,
141    {
142        let mut map = Map::new();
143        insert_json(&mut map, "path", &self.path)?;
144        insert_json(&mut map, "type", &self.file_type)?;
145        insert_json(&mut map, "name", &self.name)?;
146        insert_json(&mut map, "base_name", &self.base_name)?;
147        insert_json(&mut map, "extension", &self.extension)?;
148        insert_json(&mut map, "size", self.size)?;
149
150        if self.should_serialize_info_surface() {
151            insert_json(&mut map, "date", &self.date)?;
152            insert_json(&mut map, "sha1", self.sha1.as_ref())?;
153            insert_json(&mut map, "md5", self.md5.as_ref())?;
154            insert_json(&mut map, "sha256", self.sha256.as_ref())?;
155            insert_json(&mut map, "sha1_git", self.sha1_git.as_ref())?;
156            insert_json(&mut map, "mime_type", &self.mime_type)?;
157            insert_json(&mut map, "file_type", &self.file_type_label)?;
158            insert_json(&mut map, "programming_language", &self.programming_language)?;
159            insert_json(&mut map, "is_binary", self.is_binary)?;
160            insert_json(&mut map, "is_text", self.is_text)?;
161            insert_json(&mut map, "is_archive", self.is_archive)?;
162            insert_json(&mut map, "is_media", self.is_media)?;
163            insert_json(&mut map, "is_source", self.is_source)?;
164            insert_json(&mut map, "is_script", self.is_script)?;
165            insert_json(&mut map, "files_count", self.files_count)?;
166            insert_json(&mut map, "dirs_count", self.dirs_count)?;
167            insert_json(&mut map, "size_count", self.size_count)?;
168        }
169
170        insert_json(&mut map, "package_data", &self.package_data)?;
171        insert_json(
172            &mut map,
173            "detected_license_expression_spdx",
174            self.detected_license_expression_spdx(),
175        )?;
176        insert_json(&mut map, "license_detections", &self.license_detections)?;
177        if self.should_serialize_license_surface() {
178            insert_json(&mut map, "license_clues", &self.license_clues)?;
179        }
180        if self.percentage_of_license_text.is_some() {
181            insert_json(
182                &mut map,
183                "percentage_of_license_text",
184                self.percentage_of_license_text,
185            )?;
186        }
187        insert_json(&mut map, "copyrights", &self.copyrights)?;
188        insert_json(&mut map, "holders", &self.holders)?;
189        insert_json(&mut map, "authors", &self.authors)?;
190        if !self.emails.is_empty() {
191            insert_json(&mut map, "emails", &self.emails)?;
192        }
193        insert_json(&mut map, "urls", &self.urls)?;
194        insert_json(&mut map, "for_packages", &self.for_packages)?;
195        insert_json(&mut map, "scan_errors", &self.scan_errors)?;
196        if self.license_policy.is_some() {
197            insert_json(&mut map, "license_policy", &self.license_policy)?;
198        }
199        if self.is_generated.is_some() {
200            insert_json(&mut map, "is_generated", self.is_generated)?;
201        }
202        if self.source_count.is_some() {
203            insert_json(&mut map, "source_count", self.source_count)?;
204        }
205        if self.is_legal {
206            insert_json(&mut map, "is_legal", self.is_legal)?;
207        }
208        if self.is_manifest {
209            insert_json(&mut map, "is_manifest", self.is_manifest)?;
210        }
211        if self.is_readme {
212            insert_json(&mut map, "is_readme", self.is_readme)?;
213        }
214        if self.is_top_level {
215            insert_json(&mut map, "is_top_level", self.is_top_level)?;
216        }
217        if self.is_key_file {
218            insert_json(&mut map, "is_key_file", self.is_key_file)?;
219        }
220        if self.is_community {
221            insert_json(&mut map, "is_community", self.is_community)?;
222        }
223        if !self.facets.is_empty() {
224            insert_json(&mut map, "facets", &self.facets)?;
225        }
226        if self.tallies.is_some() {
227            insert_json(&mut map, "tallies", &self.tallies)?;
228        }
229
230        map.serialize(serializer)
231    }
232}
233
234impl From<&crate::models::FileInfo> for OutputFileInfo {
235    fn from(value: &crate::models::FileInfo) -> Self {
236        Self {
237            name: value.name.clone(),
238            base_name: value.base_name.clone(),
239            extension: value.extension.clone(),
240            path: value.path.clone(),
241            file_type: value.file_type.clone(),
242            mime_type: value.mime_type.clone(),
243            file_type_label: value.file_type_label.clone(),
244            size: value.size,
245            date: value.date.clone(),
246            sha1: value.sha1.as_ref().map(|d| d.as_hex()),
247            md5: value.md5.as_ref().map(|d| d.as_hex()),
248            sha256: value.sha256.as_ref().map(|d| d.as_hex()),
249            sha1_git: value.sha1_git.as_ref().map(|d| d.as_hex()),
250            programming_language: value.programming_language.clone(),
251            package_data: value
252                .package_data
253                .iter()
254                .map(OutputPackageData::from)
255                .collect(),
256            license_expression: value.license_expression.clone(),
257            license_detections: value
258                .license_detections
259                .iter()
260                .map(OutputLicenseDetection::from)
261                .collect(),
262            license_clues: value.license_clues.iter().map(OutputMatch::from).collect(),
263            percentage_of_license_text: value.percentage_of_license_text,
264            copyrights: value.copyrights.iter().map(OutputCopyright::from).collect(),
265            holders: value.holders.iter().map(OutputHolder::from).collect(),
266            authors: value.authors.iter().map(OutputAuthor::from).collect(),
267            emails: value.emails.iter().map(OutputEmail::from).collect(),
268            urls: value.urls.iter().map(OutputURL::from).collect(),
269            for_packages: value
270                .for_packages
271                .iter()
272                .map(|uid| uid.to_string())
273                .collect(),
274            scan_errors: value.scan_errors.clone(),
275            license_policy: value
276                .license_policy
277                .as_ref()
278                .map(|v| v.iter().map(OutputLicensePolicyEntry::from).collect()),
279            is_generated: value.is_generated,
280            is_binary: value.is_binary,
281            is_text: value.is_text,
282            is_archive: value.is_archive,
283            is_media: value.is_media,
284            is_source: value.is_source,
285            is_script: value.is_script,
286            files_count: value.files_count,
287            dirs_count: value.dirs_count,
288            size_count: value.size_count,
289            source_count: value.source_count,
290            is_legal: value.is_legal,
291            is_manifest: value.is_manifest,
292            is_readme: value.is_readme,
293            is_top_level: value.is_top_level,
294            is_key_file: value.is_key_file,
295            is_community: value.is_community,
296            facets: value.facets.clone(),
297            tallies: value.tallies.as_ref().map(OutputTallies::from),
298        }
299    }
300}
301
302impl TryFrom<&OutputFileInfo> for crate::models::FileInfo {
303    type Error = String;
304    fn try_from(value: &OutputFileInfo) -> Result<Self, Self::Error> {
305        let mut package_data = Vec::with_capacity(value.package_data.len());
306        for p in &value.package_data {
307            package_data.push(crate::models::PackageData::try_from(p)?);
308        }
309        let mut license_detections = Vec::with_capacity(value.license_detections.len());
310        for d in &value.license_detections {
311            license_detections.push(crate::models::LicenseDetection::try_from(d)?);
312        }
313        let mut license_clues = Vec::with_capacity(value.license_clues.len());
314        for m in &value.license_clues {
315            license_clues.push(crate::models::Match::try_from(m)?);
316        }
317        let mut copyrights = Vec::with_capacity(value.copyrights.len());
318        for c in &value.copyrights {
319            copyrights.push(crate::models::Copyright::try_from(c)?);
320        }
321        let mut holders = Vec::with_capacity(value.holders.len());
322        for h in &value.holders {
323            holders.push(crate::models::Holder::try_from(h)?);
324        }
325        let mut authors = Vec::with_capacity(value.authors.len());
326        for a in &value.authors {
327            authors.push(crate::models::Author::try_from(a)?);
328        }
329        let mut emails = Vec::with_capacity(value.emails.len());
330        for e in &value.emails {
331            emails.push(crate::models::OutputEmail::try_from(e)?);
332        }
333        let mut urls = Vec::with_capacity(value.urls.len());
334        for u in &value.urls {
335            urls.push(crate::models::OutputURL::try_from(u)?);
336        }
337        let license_policy = value
338            .license_policy
339            .as_ref()
340            .map(|v| {
341                v.iter()
342                    .map(crate::models::LicensePolicyEntry::try_from)
343                    .collect::<Result<Vec<_>, _>>()
344            })
345            .transpose()?;
346        Ok(Self {
347            name: value.name.clone(),
348            base_name: value.base_name.clone(),
349            extension: value.extension.clone(),
350            path: value.path.clone(),
351            file_type: value.file_type.clone(),
352            mime_type: value.mime_type.clone(),
353            file_type_label: value.file_type_label.clone(),
354            size: value.size,
355            date: value.date.clone(),
356            sha1: value
357                .sha1
358                .as_ref()
359                .map(|s| crate::models::Sha1Digest::from_hex(s))
360                .transpose()
361                .map_err(|e| format!("invalid sha1: {}", e))?,
362            md5: value
363                .md5
364                .as_ref()
365                .map(|s| crate::models::Md5Digest::from_hex(s))
366                .transpose()
367                .map_err(|e| format!("invalid md5: {}", e))?,
368            sha256: value
369                .sha256
370                .as_ref()
371                .map(|s| crate::models::Sha256Digest::from_hex(s))
372                .transpose()
373                .map_err(|e| format!("invalid sha256: {}", e))?,
374            sha1_git: value
375                .sha1_git
376                .as_ref()
377                .map(|s| crate::models::GitSha1::from_hex(s))
378                .transpose()
379                .map_err(|e| format!("invalid sha1_git: {}", e))?,
380            programming_language: value.programming_language.clone(),
381            package_data,
382            license_expression: value.license_expression.clone(),
383            license_detections,
384            license_clues,
385            percentage_of_license_text: value.percentage_of_license_text,
386            copyrights,
387            holders,
388            authors,
389            emails,
390            urls,
391            for_packages: value
392                .for_packages
393                .iter()
394                .map(|s| crate::models::PackageUid::from_raw(s.clone()))
395                .collect(),
396            scan_errors: value.scan_errors.clone(),
397            license_policy,
398            is_generated: value.is_generated,
399            is_binary: value.is_binary,
400            is_text: value.is_text,
401            is_archive: value.is_archive,
402            is_media: value.is_media,
403            is_source: value.is_source,
404            is_script: value.is_script,
405            files_count: value.files_count,
406            dirs_count: value.dirs_count,
407            size_count: value.size_count,
408            source_count: value.source_count,
409            is_legal: value.is_legal,
410            is_manifest: value.is_manifest,
411            is_readme: value.is_readme,
412            is_top_level: value.is_top_level,
413            is_key_file: value.is_key_file,
414            is_community: value.is_community,
415            facets: value.facets.clone(),
416            tallies: value
417                .tallies
418                .as_ref()
419                .map(crate::models::Tallies::try_from)
420                .transpose()?,
421        })
422    }
423}