Skip to main content

provenant/output_schema/
file_info.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4use serde::{Deserialize, Serialize, Serializer};
5use serde_json::Map;
6
7use super::author::OutputAuthor;
8use super::copyright::OutputCopyright;
9use super::email::OutputEmail;
10use super::holder::OutputHolder;
11use super::license_detection::OutputLicenseDetection;
12use super::license_match::OutputMatch;
13use super::license_policy_entry::OutputLicensePolicyEntry;
14use super::package_data::OutputPackageData;
15use super::serde_helpers::insert_json;
16use super::tallies::OutputTallies;
17use super::url::OutputURL;
18
19#[derive(Debug, Clone, Deserialize)]
20pub struct OutputFileInfo {
21    #[serde(default)]
22    pub name: String,
23    #[serde(default)]
24    pub base_name: String,
25    #[serde(default)]
26    pub extension: String,
27    pub path: String,
28    #[serde(rename = "type")]
29    pub file_type: crate::models::FileType,
30    pub mime_type: Option<String>,
31    pub file_type_label: Option<String>,
32    #[serde(default)]
33    pub size: u64,
34    pub date: Option<String>,
35    pub sha1: Option<String>,
36    pub md5: Option<String>,
37    pub sha256: Option<String>,
38    pub sha1_git: Option<String>,
39    pub programming_language: Option<String>,
40    #[serde(default)]
41    pub package_data: Vec<OutputPackageData>,
42    #[serde(rename = "detected_license_expression_spdx")]
43    pub license_expression: Option<String>,
44    #[serde(default)]
45    pub license_detections: Vec<OutputLicenseDetection>,
46    #[serde(default, skip_serializing_if = "Vec::is_empty")]
47    pub license_clues: Vec<OutputMatch>,
48    pub percentage_of_license_text: Option<f64>,
49    #[serde(default)]
50    pub copyrights: Vec<OutputCopyright>,
51    #[serde(default)]
52    pub holders: Vec<OutputHolder>,
53    #[serde(default)]
54    pub authors: Vec<OutputAuthor>,
55    #[serde(default, skip_serializing_if = "Vec::is_empty")]
56    pub emails: Vec<OutputEmail>,
57    #[serde(default)]
58    pub urls: Vec<OutputURL>,
59    #[serde(default)]
60    pub for_packages: Vec<String>,
61    #[serde(default)]
62    pub scan_errors: Vec<String>,
63    pub license_policy: Option<Vec<OutputLicensePolicyEntry>>,
64    pub is_generated: Option<bool>,
65    pub is_binary: Option<bool>,
66    pub is_text: Option<bool>,
67    pub is_archive: Option<bool>,
68    pub is_media: Option<bool>,
69    pub is_source: Option<bool>,
70    pub is_script: Option<bool>,
71    pub files_count: Option<usize>,
72    pub dirs_count: Option<usize>,
73    pub size_count: Option<u64>,
74    pub source_count: Option<usize>,
75    #[serde(default, skip_serializing_if = "is_false")]
76    pub is_legal: bool,
77    #[serde(default, skip_serializing_if = "is_false")]
78    pub is_manifest: bool,
79    #[serde(default, skip_serializing_if = "is_false")]
80    pub is_readme: bool,
81    #[serde(default, skip_serializing_if = "is_false")]
82    pub is_top_level: bool,
83    #[serde(default, skip_serializing_if = "is_false")]
84    pub is_key_file: bool,
85    #[serde(default, skip_serializing_if = "is_false")]
86    pub is_community: bool,
87    #[serde(default, skip_serializing_if = "Vec::is_empty")]
88    pub facets: Vec<String>,
89    pub tallies: Option<OutputTallies>,
90}
91
92impl OutputFileInfo {
93    pub(crate) fn should_serialize_info_surface(&self) -> bool {
94        self.date.is_some()
95            || self.sha1.is_some()
96            || self.md5.is_some()
97            || self.sha256.is_some()
98            || self.sha1_git.is_some()
99            || self.mime_type.is_some()
100            || self.file_type_label.is_some()
101            || self.programming_language.is_some()
102            || self.is_binary.is_some()
103            || self.is_text.is_some()
104            || self.is_archive.is_some()
105            || self.is_media.is_some()
106            || self.is_source.is_some()
107            || self.is_script.is_some()
108            || self.files_count.is_some()
109            || self.dirs_count.is_some()
110            || self.size_count.is_some()
111    }
112
113    pub(crate) fn should_serialize_license_surface(&self) -> bool {
114        self.license_expression.is_some()
115            || !self.license_detections.is_empty()
116            || !self.license_clues.is_empty()
117            || self.percentage_of_license_text.is_some()
118    }
119
120    pub(crate) fn detected_license_expression_spdx(&self) -> Option<String> {
121        crate::utils::spdx::combine_license_expressions_preserving_structure(
122            self.license_detections
123                .iter()
124                .filter(|detection| !detection.license_expression_spdx.is_empty())
125                .map(|detection| detection.license_expression_spdx.clone()),
126        )
127        .or_else(|| {
128            crate::utils::spdx::combine_license_expressions_preserving_structure(
129                self.package_data
130                    .iter()
131                    .flat_map(|package_data| package_data.license_detections.iter())
132                    .filter(|detection| !detection.license_expression_spdx.is_empty())
133                    .map(|detection| detection.license_expression_spdx.clone()),
134            )
135        })
136        .or_else(|| {
137            self.license_expression
138                .clone()
139                .filter(|expression| !expression.is_empty())
140        })
141    }
142}
143
144impl Serialize for OutputFileInfo {
145    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
146    where
147        S: Serializer,
148    {
149        let mut map = Map::new();
150        insert_json(&mut map, "path", &self.path)?;
151        insert_json(&mut map, "type", &self.file_type)?;
152        insert_json(&mut map, "name", &self.name)?;
153        insert_json(&mut map, "base_name", &self.base_name)?;
154        insert_json(&mut map, "extension", &self.extension)?;
155        insert_json(&mut map, "size", self.size)?;
156
157        if self.should_serialize_info_surface() {
158            insert_json(&mut map, "date", &self.date)?;
159            insert_json(&mut map, "sha1", self.sha1.as_ref())?;
160            insert_json(&mut map, "md5", self.md5.as_ref())?;
161            insert_json(&mut map, "sha256", self.sha256.as_ref())?;
162            insert_json(&mut map, "sha1_git", self.sha1_git.as_ref())?;
163            insert_json(&mut map, "mime_type", &self.mime_type)?;
164            insert_json(&mut map, "file_type", &self.file_type_label)?;
165            insert_json(&mut map, "programming_language", &self.programming_language)?;
166            insert_json(&mut map, "is_binary", self.is_binary)?;
167            insert_json(&mut map, "is_text", self.is_text)?;
168            insert_json(&mut map, "is_archive", self.is_archive)?;
169            insert_json(&mut map, "is_media", self.is_media)?;
170            insert_json(&mut map, "is_source", self.is_source)?;
171            insert_json(&mut map, "is_script", self.is_script)?;
172            insert_json(&mut map, "files_count", self.files_count)?;
173            insert_json(&mut map, "dirs_count", self.dirs_count)?;
174            insert_json(&mut map, "size_count", self.size_count)?;
175        }
176
177        insert_json(&mut map, "package_data", &self.package_data)?;
178        insert_json(
179            &mut map,
180            "detected_license_expression_spdx",
181            self.detected_license_expression_spdx(),
182        )?;
183        insert_json(&mut map, "license_detections", &self.license_detections)?;
184        if self.should_serialize_license_surface() {
185            insert_json(&mut map, "license_clues", &self.license_clues)?;
186        }
187        if self.percentage_of_license_text.is_some() {
188            insert_json(
189                &mut map,
190                "percentage_of_license_text",
191                self.percentage_of_license_text,
192            )?;
193        }
194        insert_json(&mut map, "copyrights", &self.copyrights)?;
195        insert_json(&mut map, "holders", &self.holders)?;
196        insert_json(&mut map, "authors", &self.authors)?;
197        if !self.emails.is_empty() {
198            insert_json(&mut map, "emails", &self.emails)?;
199        }
200        insert_json(&mut map, "urls", &self.urls)?;
201        insert_json(&mut map, "for_packages", &self.for_packages)?;
202        insert_json(&mut map, "scan_errors", &self.scan_errors)?;
203        if self.license_policy.is_some() {
204            insert_json(&mut map, "license_policy", &self.license_policy)?;
205        }
206        if self.is_generated.is_some() {
207            insert_json(&mut map, "is_generated", self.is_generated)?;
208        }
209        if self.source_count.is_some() {
210            insert_json(&mut map, "source_count", self.source_count)?;
211        }
212        if self.is_legal {
213            insert_json(&mut map, "is_legal", self.is_legal)?;
214        }
215        if self.is_manifest {
216            insert_json(&mut map, "is_manifest", self.is_manifest)?;
217        }
218        if self.is_readme {
219            insert_json(&mut map, "is_readme", self.is_readme)?;
220        }
221        if self.is_top_level {
222            insert_json(&mut map, "is_top_level", self.is_top_level)?;
223        }
224        if self.is_key_file {
225            insert_json(&mut map, "is_key_file", self.is_key_file)?;
226        }
227        if self.is_community {
228            insert_json(&mut map, "is_community", self.is_community)?;
229        }
230        if !self.facets.is_empty() {
231            insert_json(&mut map, "facets", &self.facets)?;
232        }
233        if self.tallies.is_some() {
234            insert_json(&mut map, "tallies", &self.tallies)?;
235        }
236
237        map.serialize(serializer)
238    }
239}
240
241impl From<&crate::models::FileInfo> for OutputFileInfo {
242    fn from(value: &crate::models::FileInfo) -> Self {
243        Self {
244            name: value.name.clone(),
245            base_name: value.base_name.clone(),
246            extension: value.extension.clone(),
247            path: value.path.clone(),
248            file_type: value.file_type.clone(),
249            mime_type: value.mime_type.clone(),
250            file_type_label: value.file_type_label.clone(),
251            size: value.size,
252            date: value.date.clone(),
253            sha1: value.sha1.as_ref().map(|d| d.as_hex()),
254            md5: value.md5.as_ref().map(|d| d.as_hex()),
255            sha256: value.sha256.as_ref().map(|d| d.as_hex()),
256            sha1_git: value.sha1_git.as_ref().map(|d| d.as_hex()),
257            programming_language: value.programming_language.clone(),
258            package_data: value
259                .package_data
260                .iter()
261                .map(OutputPackageData::from)
262                .collect(),
263            license_expression: value.license_expression.clone(),
264            license_detections: value
265                .license_detections
266                .iter()
267                .map(OutputLicenseDetection::from)
268                .collect(),
269            license_clues: value.license_clues.iter().map(OutputMatch::from).collect(),
270            percentage_of_license_text: value.percentage_of_license_text,
271            copyrights: value.copyrights.iter().map(OutputCopyright::from).collect(),
272            holders: value.holders.iter().map(OutputHolder::from).collect(),
273            authors: value.authors.iter().map(OutputAuthor::from).collect(),
274            emails: value.emails.iter().map(OutputEmail::from).collect(),
275            urls: value.urls.iter().map(OutputURL::from).collect(),
276            for_packages: value
277                .for_packages
278                .iter()
279                .map(|uid| uid.to_string())
280                .collect(),
281            scan_errors: value.scan_errors.clone(),
282            license_policy: value
283                .license_policy
284                .as_ref()
285                .map(|v| v.iter().map(OutputLicensePolicyEntry::from).collect()),
286            is_generated: value.is_generated,
287            is_binary: value.is_binary,
288            is_text: value.is_text,
289            is_archive: value.is_archive,
290            is_media: value.is_media,
291            is_source: value.is_source,
292            is_script: value.is_script,
293            files_count: value.files_count,
294            dirs_count: value.dirs_count,
295            size_count: value.size_count,
296            source_count: value.source_count,
297            is_legal: value.is_legal,
298            is_manifest: value.is_manifest,
299            is_readme: value.is_readme,
300            is_top_level: value.is_top_level,
301            is_key_file: value.is_key_file,
302            is_community: value.is_community,
303            facets: value.facets.clone(),
304            tallies: value.tallies.as_ref().map(OutputTallies::from),
305        }
306    }
307}
308
309impl TryFrom<&OutputFileInfo> for crate::models::FileInfo {
310    type Error = String;
311    fn try_from(value: &OutputFileInfo) -> Result<Self, Self::Error> {
312        let mut package_data = Vec::with_capacity(value.package_data.len());
313        for p in &value.package_data {
314            package_data.push(crate::models::PackageData::try_from(p)?);
315        }
316        let mut license_detections = Vec::with_capacity(value.license_detections.len());
317        for d in &value.license_detections {
318            license_detections.push(crate::models::LicenseDetection::try_from(d)?);
319        }
320        let mut license_clues = Vec::with_capacity(value.license_clues.len());
321        for m in &value.license_clues {
322            license_clues.push(crate::models::Match::try_from(m)?);
323        }
324        let mut copyrights = Vec::with_capacity(value.copyrights.len());
325        for c in &value.copyrights {
326            copyrights.push(crate::models::Copyright::try_from(c)?);
327        }
328        let mut holders = Vec::with_capacity(value.holders.len());
329        for h in &value.holders {
330            holders.push(crate::models::Holder::try_from(h)?);
331        }
332        let mut authors = Vec::with_capacity(value.authors.len());
333        for a in &value.authors {
334            authors.push(crate::models::Author::try_from(a)?);
335        }
336        let mut emails = Vec::with_capacity(value.emails.len());
337        for e in &value.emails {
338            emails.push(crate::models::OutputEmail::try_from(e)?);
339        }
340        let mut urls = Vec::with_capacity(value.urls.len());
341        for u in &value.urls {
342            urls.push(crate::models::OutputURL::try_from(u)?);
343        }
344        let license_policy = value
345            .license_policy
346            .as_ref()
347            .map(|v| {
348                v.iter()
349                    .map(crate::models::LicensePolicyEntry::try_from)
350                    .collect::<Result<Vec<_>, _>>()
351            })
352            .transpose()?;
353        Ok(Self {
354            name: value.name.clone(),
355            base_name: value.base_name.clone(),
356            extension: value.extension.clone(),
357            path: value.path.clone(),
358            file_type: value.file_type.clone(),
359            mime_type: value.mime_type.clone(),
360            file_type_label: value.file_type_label.clone(),
361            size: value.size,
362            date: value.date.clone(),
363            sha1: value
364                .sha1
365                .as_ref()
366                .map(|s| crate::models::Sha1Digest::from_hex(s))
367                .transpose()
368                .map_err(|e| format!("invalid sha1: {}", e))?,
369            md5: value
370                .md5
371                .as_ref()
372                .map(|s| crate::models::Md5Digest::from_hex(s))
373                .transpose()
374                .map_err(|e| format!("invalid md5: {}", e))?,
375            sha256: value
376                .sha256
377                .as_ref()
378                .map(|s| crate::models::Sha256Digest::from_hex(s))
379                .transpose()
380                .map_err(|e| format!("invalid sha256: {}", e))?,
381            sha1_git: value
382                .sha1_git
383                .as_ref()
384                .map(|s| crate::models::GitSha1::from_hex(s))
385                .transpose()
386                .map_err(|e| format!("invalid sha1_git: {}", e))?,
387            programming_language: value.programming_language.clone(),
388            package_data,
389            license_expression: value.license_expression.clone(),
390            license_detections,
391            license_clues,
392            percentage_of_license_text: value.percentage_of_license_text,
393            copyrights,
394            holders,
395            authors,
396            emails,
397            urls,
398            for_packages: value
399                .for_packages
400                .iter()
401                .map(|s| crate::models::PackageUid::from_raw(s.clone()))
402                .collect(),
403            scan_errors: value.scan_errors.clone(),
404            scan_diagnostics: crate::models::diagnostics_from_legacy_scan_errors(
405                &value.scan_errors,
406            ),
407            license_policy,
408            is_generated: value.is_generated,
409            is_binary: value.is_binary,
410            is_text: value.is_text,
411            is_archive: value.is_archive,
412            is_media: value.is_media,
413            is_source: value.is_source,
414            is_script: value.is_script,
415            files_count: value.files_count,
416            dirs_count: value.dirs_count,
417            size_count: value.size_count,
418            source_count: value.source_count,
419            is_legal: value.is_legal,
420            is_manifest: value.is_manifest,
421            is_readme: value.is_readme,
422            is_top_level: value.is_top_level,
423            is_key_file: value.is_key_file,
424            is_community: value.is_community,
425            facets: value.facets.clone(),
426            tallies: value
427                .tallies
428                .as_ref()
429                .map(crate::models::Tallies::try_from)
430                .transpose()?,
431        })
432    }
433}