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    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
107impl Serialize for OutputFileInfo {
108    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
109    where
110        S: Serializer,
111    {
112        let mut map = Map::new();
113        insert_json(&mut map, "path", &self.path)?;
114        insert_json(&mut map, "type", &self.file_type)?;
115        insert_json(&mut map, "name", &self.name)?;
116        insert_json(&mut map, "base_name", &self.base_name)?;
117        insert_json(&mut map, "extension", &self.extension)?;
118        insert_json(&mut map, "size", self.size)?;
119
120        if self.should_serialize_info_surface() {
121            insert_json(&mut map, "date", &self.date)?;
122            insert_json(&mut map, "sha1", self.sha1.as_ref())?;
123            insert_json(&mut map, "md5", self.md5.as_ref())?;
124            insert_json(&mut map, "sha256", self.sha256.as_ref())?;
125            insert_json(&mut map, "sha1_git", self.sha1_git.as_ref())?;
126            insert_json(&mut map, "mime_type", &self.mime_type)?;
127            insert_json(&mut map, "file_type", &self.file_type_label)?;
128            insert_json(&mut map, "programming_language", &self.programming_language)?;
129            insert_json(&mut map, "is_binary", self.is_binary)?;
130            insert_json(&mut map, "is_text", self.is_text)?;
131            insert_json(&mut map, "is_archive", self.is_archive)?;
132            insert_json(&mut map, "is_media", self.is_media)?;
133            insert_json(&mut map, "is_source", self.is_source)?;
134            insert_json(&mut map, "is_script", self.is_script)?;
135            insert_json(&mut map, "files_count", self.files_count)?;
136            insert_json(&mut map, "dirs_count", self.dirs_count)?;
137            insert_json(&mut map, "size_count", self.size_count)?;
138        }
139
140        insert_json(&mut map, "package_data", &self.package_data)?;
141        insert_json(
142            &mut map,
143            "detected_license_expression_spdx",
144            &self.license_expression,
145        )?;
146        insert_json(&mut map, "license_detections", &self.license_detections)?;
147        if !self.license_clues.is_empty() {
148            insert_json(&mut map, "license_clues", &self.license_clues)?;
149        }
150        if self.percentage_of_license_text.is_some() {
151            insert_json(
152                &mut map,
153                "percentage_of_license_text",
154                self.percentage_of_license_text,
155            )?;
156        }
157        insert_json(&mut map, "copyrights", &self.copyrights)?;
158        insert_json(&mut map, "holders", &self.holders)?;
159        insert_json(&mut map, "authors", &self.authors)?;
160        if !self.emails.is_empty() {
161            insert_json(&mut map, "emails", &self.emails)?;
162        }
163        insert_json(&mut map, "urls", &self.urls)?;
164        insert_json(&mut map, "for_packages", &self.for_packages)?;
165        insert_json(&mut map, "scan_errors", &self.scan_errors)?;
166        if self.license_policy.is_some() {
167            insert_json(&mut map, "license_policy", &self.license_policy)?;
168        }
169        if self.is_generated.is_some() {
170            insert_json(&mut map, "is_generated", self.is_generated)?;
171        }
172        if self.source_count.is_some() {
173            insert_json(&mut map, "source_count", self.source_count)?;
174        }
175        if self.is_legal {
176            insert_json(&mut map, "is_legal", self.is_legal)?;
177        }
178        if self.is_manifest {
179            insert_json(&mut map, "is_manifest", self.is_manifest)?;
180        }
181        if self.is_readme {
182            insert_json(&mut map, "is_readme", self.is_readme)?;
183        }
184        if self.is_top_level {
185            insert_json(&mut map, "is_top_level", self.is_top_level)?;
186        }
187        if self.is_key_file {
188            insert_json(&mut map, "is_key_file", self.is_key_file)?;
189        }
190        if self.is_community {
191            insert_json(&mut map, "is_community", self.is_community)?;
192        }
193        if !self.facets.is_empty() {
194            insert_json(&mut map, "facets", &self.facets)?;
195        }
196        if self.tallies.is_some() {
197            insert_json(&mut map, "tallies", &self.tallies)?;
198        }
199
200        map.serialize(serializer)
201    }
202}
203
204impl From<&crate::models::FileInfo> for OutputFileInfo {
205    fn from(value: &crate::models::FileInfo) -> Self {
206        Self {
207            name: value.name.clone(),
208            base_name: value.base_name.clone(),
209            extension: value.extension.clone(),
210            path: value.path.clone(),
211            file_type: value.file_type.clone(),
212            mime_type: value.mime_type.clone(),
213            file_type_label: value.file_type_label.clone(),
214            size: value.size,
215            date: value.date.clone(),
216            sha1: value.sha1.as_ref().map(|d| d.as_hex()),
217            md5: value.md5.as_ref().map(|d| d.as_hex()),
218            sha256: value.sha256.as_ref().map(|d| d.as_hex()),
219            sha1_git: value.sha1_git.as_ref().map(|d| d.as_hex()),
220            programming_language: value.programming_language.clone(),
221            package_data: value
222                .package_data
223                .iter()
224                .map(OutputPackageData::from)
225                .collect(),
226            license_expression: value.license_expression.clone(),
227            license_detections: value
228                .license_detections
229                .iter()
230                .map(OutputLicenseDetection::from)
231                .collect(),
232            license_clues: value.license_clues.iter().map(OutputMatch::from).collect(),
233            percentage_of_license_text: value.percentage_of_license_text,
234            copyrights: value.copyrights.iter().map(OutputCopyright::from).collect(),
235            holders: value.holders.iter().map(OutputHolder::from).collect(),
236            authors: value.authors.iter().map(OutputAuthor::from).collect(),
237            emails: value.emails.iter().map(OutputEmail::from).collect(),
238            urls: value.urls.iter().map(OutputURL::from).collect(),
239            for_packages: value
240                .for_packages
241                .iter()
242                .map(|uid| uid.to_string())
243                .collect(),
244            scan_errors: value.scan_errors.clone(),
245            license_policy: value
246                .license_policy
247                .as_ref()
248                .map(|v| v.iter().map(OutputLicensePolicyEntry::from).collect()),
249            is_generated: value.is_generated,
250            is_binary: value.is_binary,
251            is_text: value.is_text,
252            is_archive: value.is_archive,
253            is_media: value.is_media,
254            is_source: value.is_source,
255            is_script: value.is_script,
256            files_count: value.files_count,
257            dirs_count: value.dirs_count,
258            size_count: value.size_count,
259            source_count: value.source_count,
260            is_legal: value.is_legal,
261            is_manifest: value.is_manifest,
262            is_readme: value.is_readme,
263            is_top_level: value.is_top_level,
264            is_key_file: value.is_key_file,
265            is_community: value.is_community,
266            facets: value.facets.clone(),
267            tallies: value.tallies.as_ref().map(OutputTallies::from),
268        }
269    }
270}
271
272impl TryFrom<&OutputFileInfo> for crate::models::FileInfo {
273    type Error = String;
274    fn try_from(value: &OutputFileInfo) -> Result<Self, Self::Error> {
275        let mut package_data = Vec::with_capacity(value.package_data.len());
276        for p in &value.package_data {
277            package_data.push(crate::models::PackageData::try_from(p)?);
278        }
279        let mut license_detections = Vec::with_capacity(value.license_detections.len());
280        for d in &value.license_detections {
281            license_detections.push(crate::models::LicenseDetection::try_from(d)?);
282        }
283        let mut license_clues = Vec::with_capacity(value.license_clues.len());
284        for m in &value.license_clues {
285            license_clues.push(crate::models::Match::try_from(m)?);
286        }
287        let mut copyrights = Vec::with_capacity(value.copyrights.len());
288        for c in &value.copyrights {
289            copyrights.push(crate::models::Copyright::try_from(c)?);
290        }
291        let mut holders = Vec::with_capacity(value.holders.len());
292        for h in &value.holders {
293            holders.push(crate::models::Holder::try_from(h)?);
294        }
295        let mut authors = Vec::with_capacity(value.authors.len());
296        for a in &value.authors {
297            authors.push(crate::models::Author::try_from(a)?);
298        }
299        let mut emails = Vec::with_capacity(value.emails.len());
300        for e in &value.emails {
301            emails.push(crate::models::OutputEmail::try_from(e)?);
302        }
303        let mut urls = Vec::with_capacity(value.urls.len());
304        for u in &value.urls {
305            urls.push(crate::models::OutputURL::try_from(u)?);
306        }
307        let license_policy = value
308            .license_policy
309            .as_ref()
310            .map(|v| {
311                v.iter()
312                    .map(crate::models::LicensePolicyEntry::try_from)
313                    .collect::<Result<Vec<_>, _>>()
314            })
315            .transpose()?;
316        Ok(Self {
317            name: value.name.clone(),
318            base_name: value.base_name.clone(),
319            extension: value.extension.clone(),
320            path: value.path.clone(),
321            file_type: value.file_type.clone(),
322            mime_type: value.mime_type.clone(),
323            file_type_label: value.file_type_label.clone(),
324            size: value.size,
325            date: value.date.clone(),
326            sha1: value
327                .sha1
328                .as_ref()
329                .map(|s| crate::models::Sha1Digest::from_hex(s))
330                .transpose()
331                .map_err(|e| format!("invalid sha1: {}", e))?,
332            md5: value
333                .md5
334                .as_ref()
335                .map(|s| crate::models::Md5Digest::from_hex(s))
336                .transpose()
337                .map_err(|e| format!("invalid md5: {}", e))?,
338            sha256: value
339                .sha256
340                .as_ref()
341                .map(|s| crate::models::Sha256Digest::from_hex(s))
342                .transpose()
343                .map_err(|e| format!("invalid sha256: {}", e))?,
344            sha1_git: value
345                .sha1_git
346                .as_ref()
347                .map(|s| crate::models::GitSha1::from_hex(s))
348                .transpose()
349                .map_err(|e| format!("invalid sha1_git: {}", e))?,
350            programming_language: value.programming_language.clone(),
351            package_data,
352            license_expression: value.license_expression.clone(),
353            license_detections,
354            license_clues,
355            percentage_of_license_text: value.percentage_of_license_text,
356            copyrights,
357            holders,
358            authors,
359            emails,
360            urls,
361            for_packages: value
362                .for_packages
363                .iter()
364                .map(|s| crate::models::PackageUid::from_raw(s.clone()))
365                .collect(),
366            scan_errors: value.scan_errors.clone(),
367            license_policy,
368            is_generated: value.is_generated,
369            is_binary: value.is_binary,
370            is_text: value.is_text,
371            is_archive: value.is_archive,
372            is_media: value.is_media,
373            is_source: value.is_source,
374            is_script: value.is_script,
375            files_count: value.files_count,
376            dirs_count: value.dirs_count,
377            size_count: value.size_count,
378            source_count: value.source_count,
379            is_legal: value.is_legal,
380            is_manifest: value.is_manifest,
381            is_readme: value.is_readme,
382            is_top_level: value.is_top_level,
383            is_key_file: value.is_key_file,
384            is_community: value.is_community,
385            facets: value.facets.clone(),
386            tallies: value
387                .tallies
388                .as_ref()
389                .map(crate::models::Tallies::try_from)
390                .transpose()?,
391        })
392    }
393}