Skip to main content

provenant/parsers/
uv_lock.rs

1use std::collections::{HashMap, HashSet, VecDeque};
2use std::path::Path;
3
4use crate::parser_warn as warn;
5use packageurl::PackageUrl;
6use serde_json::Value as JsonValue;
7use toml::Value as TomlValue;
8use toml::map::Map as TomlMap;
9
10use crate::models::{DatasourceId, Dependency, PackageData, PackageType, ResolvedPackage};
11use crate::parsers::python::read_toml_file;
12
13use super::PackageParser;
14
15const FIELD_PACKAGE: &str = "package";
16const FIELD_NAME: &str = "name";
17const FIELD_VERSION: &str = "version";
18const FIELD_SOURCE: &str = "source";
19const FIELD_DEPENDENCIES: &str = "dependencies";
20const FIELD_OPTIONAL_DEPENDENCIES: &str = "optional-dependencies";
21const FIELD_DEV_DEPENDENCIES: &str = "dev-dependencies";
22const FIELD_METADATA: &str = "metadata";
23const FIELD_REQUIRES_DIST: &str = "requires-dist";
24const FIELD_REQUIRES_DEV: &str = "requires-dev";
25const FIELD_METADATA_OPTIONAL_DEPENDENCIES: &str = "optional-dependencies";
26const FIELD_MARKER: &str = "marker";
27const FIELD_EXTRA: &str = "extra";
28const FIELD_SPECIFIER: &str = "specifier";
29const FIELD_REVISION: &str = "revision";
30const FIELD_REQUIRES_PYTHON: &str = "requires-python";
31const FIELD_RESOLUTION_MARKERS: &str = "resolution-markers";
32const FIELD_MANIFEST: &str = "manifest";
33
34pub struct UvLockParser;
35
36#[derive(Clone, Debug, Default)]
37struct DirectDependencyInfo {
38    extracted_requirement: Option<String>,
39    scope: Option<String>,
40    is_runtime: bool,
41    is_optional: bool,
42    extra_data: Option<HashMap<String, JsonValue>>,
43    source_key: Option<String>,
44}
45
46#[derive(Clone, Debug)]
47struct DependencyEdge {
48    name: String,
49    extracted_requirement: Option<String>,
50    scope: Option<String>,
51    is_runtime: bool,
52    is_optional: bool,
53    source_key: Option<String>,
54    extra_data: Option<HashMap<String, JsonValue>>,
55}
56
57impl PackageParser for UvLockParser {
58    const PACKAGE_TYPE: PackageType = PackageType::Pypi;
59
60    fn is_match(path: &Path) -> bool {
61        path.file_name()
62            .and_then(|name| name.to_str())
63            .is_some_and(|name| name == "uv.lock")
64    }
65
66    fn extract_packages(path: &Path) -> Vec<PackageData> {
67        let toml_content = match read_toml_file(path) {
68            Ok(content) => content,
69            Err(e) => {
70                warn!("Failed to read uv.lock at {:?}: {}", path, e);
71                return vec![default_package_data()];
72            }
73        };
74
75        vec![parse_uv_lock(&toml_content)]
76    }
77}
78
79fn parse_uv_lock(toml_content: &TomlValue) -> PackageData {
80    let packages = toml_content
81        .get(FIELD_PACKAGE)
82        .and_then(TomlValue::as_array)
83        .cloned()
84        .unwrap_or_default();
85
86    if packages.is_empty() {
87        return default_package_data();
88    }
89
90    let package_tables: Vec<&TomlMap<String, TomlValue>> =
91        packages.iter().filter_map(TomlValue::as_table).collect();
92
93    if package_tables.is_empty() {
94        return default_package_data();
95    }
96
97    let root_index = find_root_package_index(&package_tables);
98    let package_lookup = build_package_lookup(&package_tables);
99
100    let direct_infos = root_index
101        .and_then(|index| package_tables.get(index).copied())
102        .map(collect_root_direct_dependencies)
103        .unwrap_or_default();
104
105    let runtime_roots: Vec<(String, Option<String>)> = direct_infos
106        .iter()
107        .filter(|(_, info)| info.is_runtime)
108        .map(|(name, info)| (name.clone(), info.source_key.clone()))
109        .collect();
110    let dev_roots: Vec<(String, Option<String>)> = direct_infos
111        .iter()
112        .filter(|(_, info)| !info.is_runtime && !info.is_optional)
113        .map(|(name, info)| (name.clone(), info.source_key.clone()))
114        .collect();
115    let optional_roots: Vec<(String, Option<String>)> = direct_infos
116        .iter()
117        .filter(|(_, info)| info.is_optional)
118        .map(|(name, info)| (name.clone(), info.source_key.clone()))
119        .collect();
120
121    let runtime_reachable =
122        collect_reachable_packages(&package_tables, &package_lookup, &runtime_roots, false);
123    let dev_reachable =
124        collect_reachable_packages(&package_tables, &package_lookup, &dev_roots, true);
125    let optional_reachable =
126        collect_reachable_packages(&package_tables, &package_lookup, &optional_roots, true);
127
128    let mut package_data = default_package_data();
129    package_data.extra_data = build_lock_extra_data(toml_content);
130
131    if let Some(index) = root_index
132        && let Some(root_table) = package_tables.get(index)
133    {
134        package_data.name = root_table
135            .get(FIELD_NAME)
136            .and_then(TomlValue::as_str)
137            .map(normalize_pypi_name);
138        package_data.version = root_table
139            .get(FIELD_VERSION)
140            .and_then(TomlValue::as_str)
141            .map(|value| value.to_string());
142        package_data.is_virtual =
143            package_source_table(root_table).is_some_and(|source| source.contains_key("virtual"));
144        package_data.purl = package_data
145            .name
146            .as_deref()
147            .and_then(|name| create_pypi_purl(name, package_data.version.as_deref()));
148    }
149
150    package_data.dependencies = package_tables
151        .iter()
152        .enumerate()
153        .filter(|(index, _)| Some(*index) != root_index)
154        .filter_map(|(_, package_table)| {
155            build_top_level_dependency(
156                package_table,
157                root_index.is_none(),
158                &direct_infos,
159                &runtime_reachable,
160                &dev_reachable,
161                &optional_reachable,
162                &package_lookup,
163            )
164        })
165        .collect();
166
167    package_data
168}
169
170fn build_top_level_dependency(
171    package_table: &TomlMap<String, TomlValue>,
172    no_root_package: bool,
173    direct_infos: &HashMap<String, DirectDependencyInfo>,
174    runtime_reachable: &HashSet<String>,
175    dev_reachable: &HashSet<String>,
176    optional_reachable: &HashSet<String>,
177    package_lookup: &HashMap<String, Vec<usize>>,
178) -> Option<Dependency> {
179    let name = package_table
180        .get(FIELD_NAME)
181        .and_then(TomlValue::as_str)
182        .map(normalize_pypi_name)?;
183    let version = package_table
184        .get(FIELD_VERSION)
185        .and_then(TomlValue::as_str)
186        .map(|value| value.to_string())?;
187
188    let direct_info = direct_infos.get(&name);
189    let is_direct = direct_info.is_some();
190    let is_runtime = if no_root_package {
191        true
192    } else if let Some(info) = direct_info {
193        info.is_runtime
194    } else if runtime_reachable.contains(&name) {
195        true
196    } else {
197        !dev_reachable.contains(&name) && !optional_reachable.contains(&name)
198    };
199    let is_optional = direct_info.is_some_and(|info| info.is_optional)
200        || (!is_direct && optional_reachable.contains(&name) && !runtime_reachable.contains(&name));
201
202    Some(Dependency {
203        purl: create_pypi_purl(&name, Some(&version)),
204        extracted_requirement: direct_info.and_then(|info| info.extracted_requirement.clone()),
205        scope: direct_info.and_then(|info| info.scope.clone()),
206        is_runtime: Some(is_runtime),
207        is_optional: Some(is_optional),
208        is_pinned: Some(true),
209        is_direct: Some(is_direct),
210        resolved_package: Some(Box::new(build_resolved_package(
211            package_table,
212            package_lookup,
213        ))),
214        extra_data: direct_info.and_then(|info| info.extra_data.clone()),
215    })
216}
217
218fn build_resolved_package(
219    package_table: &TomlMap<String, TomlValue>,
220    package_lookup: &HashMap<String, Vec<usize>>,
221) -> ResolvedPackage {
222    let name = package_table
223        .get(FIELD_NAME)
224        .and_then(TomlValue::as_str)
225        .map(normalize_pypi_name)
226        .unwrap_or_default();
227    let version = package_table
228        .get(FIELD_VERSION)
229        .and_then(TomlValue::as_str)
230        .map(|value| value.to_string())
231        .unwrap_or_default();
232
233    let (_, repository_download_url, api_data_url, purl) =
234        build_pypi_urls(Some(&name), Some(&version));
235    let repository_homepage_url = Some(format!("https://pypi.org/project/{}", name));
236    let (download_url, sha256) = extract_artifact_metadata(package_table);
237
238    ResolvedPackage {
239        primary_language: Some("Python".to_string()),
240        download_url,
241        sha1: None,
242        sha256,
243        sha512: None,
244        md5: None,
245        is_virtual: true,
246        extra_data: build_package_extra_data(package_table),
247        dependencies: collect_package_dependency_edges(package_table)
248            .into_iter()
249            .map(|edge| edge_to_dependency(edge, package_lookup))
250            .collect(),
251        repository_homepage_url,
252        repository_download_url,
253        api_data_url,
254        datasource_id: Some(DatasourceId::PypiUvLock),
255        purl,
256        ..ResolvedPackage::new(UvLockParser::PACKAGE_TYPE, String::new(), name, version)
257    }
258}
259
260fn edge_to_dependency(
261    edge: DependencyEdge,
262    package_lookup: &HashMap<String, Vec<usize>>,
263) -> Dependency {
264    let is_pinned = edge
265        .source_key
266        .as_ref()
267        .map(|_| !package_lookup.contains_key(&edge.name))
268        .unwrap_or(false);
269
270    Dependency {
271        purl: create_pypi_purl(&edge.name, None),
272        extracted_requirement: edge.extracted_requirement,
273        scope: edge.scope,
274        is_runtime: Some(edge.is_runtime),
275        is_optional: Some(edge.is_optional),
276        is_pinned: Some(is_pinned),
277        is_direct: Some(true),
278        resolved_package: None,
279        extra_data: edge.extra_data,
280    }
281}
282
283fn collect_root_direct_dependencies(
284    root_table: &TomlMap<String, TomlValue>,
285) -> HashMap<String, DirectDependencyInfo> {
286    let mut infos = HashMap::new();
287    let metadata = root_table.get(FIELD_METADATA).and_then(TomlValue::as_table);
288    let runtime_requirements = metadata
289        .and_then(|metadata| metadata.get(FIELD_REQUIRES_DIST))
290        .map(parse_requirement_metadata_array)
291        .unwrap_or_default();
292    let dev_requirements = metadata
293        .and_then(|metadata| metadata.get(FIELD_REQUIRES_DEV))
294        .and_then(TomlValue::as_table)
295        .map(parse_requirement_metadata_table)
296        .unwrap_or_default();
297    let optional_requirements = metadata
298        .and_then(|metadata| metadata.get(FIELD_METADATA_OPTIONAL_DEPENDENCIES))
299        .and_then(TomlValue::as_table)
300        .map(parse_requirement_metadata_table)
301        .unwrap_or_default();
302
303    for edge in collect_dependency_edges_from_array(
304        root_table
305            .get(FIELD_DEPENDENCIES)
306            .and_then(TomlValue::as_array),
307        None,
308        true,
309        false,
310        runtime_requirements.get("__runtime__"),
311    ) {
312        merge_direct_dependency_info(&mut infos, edge);
313    }
314
315    if let Some(optional_table) = root_table
316        .get(FIELD_OPTIONAL_DEPENDENCIES)
317        .and_then(TomlValue::as_table)
318    {
319        for (group, value) in optional_table {
320            let requirement_map = optional_requirements.get(group);
321            for edge in collect_dependency_edges_from_array(
322                value.as_array(),
323                Some(group.to_string()),
324                false,
325                true,
326                requirement_map,
327            ) {
328                merge_direct_dependency_info(&mut infos, edge);
329            }
330        }
331    }
332
333    if let Some(dev_table) = root_table
334        .get(FIELD_DEV_DEPENDENCIES)
335        .and_then(TomlValue::as_table)
336    {
337        for (group, value) in dev_table {
338            let requirement_map = dev_requirements.get(group);
339            for edge in collect_dependency_edges_from_array(
340                value.as_array(),
341                Some(group.to_string()),
342                false,
343                false,
344                requirement_map,
345            ) {
346                merge_direct_dependency_info(&mut infos, edge);
347            }
348        }
349    }
350
351    infos
352}
353
354fn merge_direct_dependency_info(
355    infos: &mut HashMap<String, DirectDependencyInfo>,
356    edge: DependencyEdge,
357) {
358    let name = edge.name.clone();
359    let new_info = direct_info_from_edge(edge);
360
361    if let Some(existing) = infos.get_mut(&name) {
362        existing.is_runtime |= new_info.is_runtime;
363        existing.is_optional &= new_info.is_optional;
364
365        if existing.extracted_requirement.is_none() {
366            existing.extracted_requirement = new_info.extracted_requirement.clone();
367        }
368
369        existing.scope = merge_scope(existing.scope.as_ref(), new_info.scope.as_ref());
370        existing.extra_data =
371            merge_optional_json_maps(existing.extra_data.take(), new_info.extra_data);
372
373        if existing.source_key != new_info.source_key {
374            existing.source_key = None;
375        }
376    } else {
377        infos.insert(name, new_info);
378    }
379}
380
381fn merge_scope(current: Option<&String>, new: Option<&String>) -> Option<String> {
382    match (current, new) {
383        (None, None) => None,
384        (None, Some(_)) | (Some(_), None) => None,
385        (Some(left), Some(right)) if left == right => Some(left.clone()),
386        _ => None,
387    }
388}
389
390fn merge_optional_json_maps(
391    current: Option<HashMap<String, JsonValue>>,
392    new: Option<HashMap<String, JsonValue>>,
393) -> Option<HashMap<String, JsonValue>> {
394    match (current, new) {
395        (None, None) => None,
396        (Some(map), None) | (None, Some(map)) => Some(map),
397        (Some(mut current), Some(new)) => {
398            for (key, value) in new {
399                current.entry(key).or_insert(value);
400            }
401            Some(current)
402        }
403    }
404}
405
406fn direct_info_from_edge(edge: DependencyEdge) -> DirectDependencyInfo {
407    DirectDependencyInfo {
408        extracted_requirement: edge.extracted_requirement,
409        scope: edge.scope,
410        is_runtime: edge.is_runtime,
411        is_optional: edge.is_optional,
412        extra_data: edge.extra_data,
413        source_key: edge.source_key,
414    }
415}
416
417fn collect_package_dependency_edges(
418    package_table: &TomlMap<String, TomlValue>,
419) -> Vec<DependencyEdge> {
420    let mut edges = Vec::new();
421
422    edges.extend(collect_dependency_edges_from_array(
423        package_table
424            .get(FIELD_DEPENDENCIES)
425            .and_then(TomlValue::as_array),
426        None,
427        true,
428        false,
429        None,
430    ));
431
432    if let Some(optional_table) = package_table
433        .get(FIELD_OPTIONAL_DEPENDENCIES)
434        .and_then(TomlValue::as_table)
435    {
436        for (group, value) in optional_table {
437            edges.extend(collect_dependency_edges_from_array(
438                value.as_array(),
439                Some(group.to_string()),
440                false,
441                true,
442                None,
443            ));
444        }
445    }
446
447    if let Some(dev_table) = package_table
448        .get(FIELD_DEV_DEPENDENCIES)
449        .and_then(TomlValue::as_table)
450    {
451        for (group, value) in dev_table {
452            edges.extend(collect_dependency_edges_from_array(
453                value.as_array(),
454                Some(group.to_string()),
455                false,
456                false,
457                None,
458            ));
459        }
460    }
461
462    edges
463}
464
465fn collect_dependency_edges_from_array(
466    values: Option<&Vec<TomlValue>>,
467    scope: Option<String>,
468    is_runtime: bool,
469    is_optional: bool,
470    requirement_map: Option<&HashMap<String, String>>,
471) -> Vec<DependencyEdge> {
472    values
473        .into_iter()
474        .flatten()
475        .filter_map(|value| {
476            build_dependency_edge(
477                value,
478                scope.clone(),
479                is_runtime,
480                is_optional,
481                requirement_map,
482            )
483        })
484        .collect()
485}
486
487fn build_dependency_edge(
488    value: &TomlValue,
489    scope: Option<String>,
490    is_runtime: bool,
491    is_optional: bool,
492    requirement_map: Option<&HashMap<String, String>>,
493) -> Option<DependencyEdge> {
494    let table = value.as_table()?;
495    let name = table
496        .get(FIELD_NAME)
497        .and_then(TomlValue::as_str)
498        .map(normalize_pypi_name)?;
499
500    let mut extra_data = HashMap::new();
501    if let Some(marker) = table.get(FIELD_MARKER).and_then(TomlValue::as_str) {
502        extra_data.insert(
503            FIELD_MARKER.to_string(),
504            JsonValue::String(marker.to_string()),
505        );
506    }
507    if let Some(extra_value) = table.get(FIELD_EXTRA) {
508        let json_value = toml_value_to_json(extra_value);
509        extra_data.insert(FIELD_EXTRA.to_string(), json_value);
510    }
511
512    let source_key = table
513        .get(FIELD_SOURCE)
514        .and_then(TomlValue::as_table)
515        .and_then(source_table_key);
516    if let Some(source) = table.get(FIELD_SOURCE) {
517        extra_data.insert(FIELD_SOURCE.to_string(), toml_value_to_json(source));
518    }
519
520    let extracted_requirement = requirement_map
521        .and_then(|map| map.get(&name).cloned())
522        .or_else(|| {
523            table
524                .get(FIELD_SPECIFIER)
525                .and_then(TomlValue::as_str)
526                .map(|value| value.to_string())
527        });
528
529    Some(DependencyEdge {
530        name,
531        extracted_requirement,
532        scope,
533        is_runtime,
534        is_optional,
535        source_key,
536        extra_data: (!extra_data.is_empty()).then_some(extra_data),
537    })
538}
539
540fn parse_requirement_metadata_array(value: &TomlValue) -> HashMap<String, HashMap<String, String>> {
541    let mut grouped = HashMap::new();
542    let runtime = value
543        .as_array()
544        .map(|values| parse_requirement_entries(values))
545        .unwrap_or_default();
546    grouped.insert("__runtime__".to_string(), runtime);
547    grouped
548}
549
550fn parse_requirement_metadata_table(
551    table: &TomlMap<String, TomlValue>,
552) -> HashMap<String, HashMap<String, String>> {
553    table
554        .iter()
555        .map(|(group, value)| {
556            (
557                group.to_string(),
558                value
559                    .as_array()
560                    .map(|values| parse_requirement_entries(values))
561                    .unwrap_or_default(),
562            )
563        })
564        .collect()
565}
566
567fn parse_requirement_entries(values: &[TomlValue]) -> HashMap<String, String> {
568    values
569        .iter()
570        .filter_map(|value| {
571            let table = value.as_table()?;
572            let name = table
573                .get(FIELD_NAME)
574                .and_then(TomlValue::as_str)
575                .map(normalize_pypi_name)?;
576            let specifier = table
577                .get(FIELD_SPECIFIER)
578                .and_then(TomlValue::as_str)
579                .map(|value| value.to_string())?;
580            Some((name, specifier))
581        })
582        .collect()
583}
584
585fn collect_reachable_packages(
586    package_tables: &[&TomlMap<String, TomlValue>],
587    package_lookup: &HashMap<String, Vec<usize>>,
588    roots: &[(String, Option<String>)],
589    include_non_runtime_edges: bool,
590) -> HashSet<String> {
591    let mut visited = HashSet::new();
592    let mut queue: VecDeque<(String, Option<String>)> = roots.iter().cloned().collect();
593
594    while let Some((name, source_key)) = queue.pop_front() {
595        let Some(index) =
596            match_package_index(package_tables, package_lookup, &name, source_key.as_deref())
597        else {
598            continue;
599        };
600
601        let Some(package_table) = package_tables.get(index) else {
602            continue;
603        };
604
605        let package_name = package_table
606            .get(FIELD_NAME)
607            .and_then(TomlValue::as_str)
608            .map(normalize_pypi_name)
609            .unwrap_or(name);
610
611        if !visited.insert(package_name.clone()) {
612            continue;
613        }
614
615        let edges = if include_non_runtime_edges {
616            collect_package_dependency_edges(package_table)
617        } else {
618            collect_dependency_edges_from_array(
619                package_table
620                    .get(FIELD_DEPENDENCIES)
621                    .and_then(TomlValue::as_array),
622                None,
623                true,
624                false,
625                None,
626            )
627        };
628
629        for edge in edges {
630            queue.push_back((edge.name, edge.source_key));
631        }
632    }
633
634    visited
635}
636
637fn build_package_lookup(
638    package_tables: &[&TomlMap<String, TomlValue>],
639) -> HashMap<String, Vec<usize>> {
640    let mut lookup: HashMap<String, Vec<usize>> = HashMap::new();
641    for (index, package_table) in package_tables.iter().enumerate() {
642        if let Some(name) = package_table
643            .get(FIELD_NAME)
644            .and_then(TomlValue::as_str)
645            .map(normalize_pypi_name)
646        {
647            lookup.entry(name).or_default().push(index);
648        }
649    }
650    lookup
651}
652
653fn match_package_index(
654    package_tables: &[&TomlMap<String, TomlValue>],
655    package_lookup: &HashMap<String, Vec<usize>>,
656    name: &str,
657    source_key: Option<&str>,
658) -> Option<usize> {
659    let candidates = package_lookup.get(name)?;
660    if candidates.len() == 1 {
661        return candidates.first().copied();
662    }
663
664    let source_key = source_key?;
665    candidates.iter().copied().find(|index| {
666        package_tables
667            .get(*index)
668            .and_then(|table| package_source_table(table))
669            .and_then(source_table_key)
670            .as_deref()
671            == Some(source_key)
672    })
673}
674
675fn find_root_package_index(package_tables: &[&TomlMap<String, TomlValue>]) -> Option<usize> {
676    if let Some(index) = package_tables.iter().position(|table| {
677        package_source_table(table)
678            .and_then(local_source_path)
679            .is_some_and(|path| path == ".")
680    }) {
681        return Some(index);
682    }
683
684    package_tables.iter().position(|table| {
685        package_source_table(table)
686            .is_some_and(|source| source.contains_key("editable") || source.contains_key("virtual"))
687    })
688}
689
690fn local_source_path(source_table: &TomlMap<String, TomlValue>) -> Option<&str> {
691    source_table
692        .get("virtual")
693        .and_then(TomlValue::as_str)
694        .or_else(|| source_table.get("editable").and_then(TomlValue::as_str))
695}
696
697fn build_lock_extra_data(toml_content: &TomlValue) -> Option<HashMap<String, JsonValue>> {
698    let mut extra_data = HashMap::new();
699
700    if let Some(version) = toml_content
701        .get(FIELD_VERSION)
702        .and_then(TomlValue::as_integer)
703    {
704        extra_data.insert(
705            "lockfile_version".to_string(),
706            JsonValue::String(version.to_string()),
707        );
708    }
709
710    if let Some(revision) = toml_content
711        .get(FIELD_REVISION)
712        .and_then(TomlValue::as_integer)
713    {
714        extra_data.insert(
715            FIELD_REVISION.to_string(),
716            JsonValue::String(revision.to_string()),
717        );
718    }
719
720    if let Some(requires_python) = toml_content
721        .get(FIELD_REQUIRES_PYTHON)
722        .and_then(TomlValue::as_str)
723    {
724        extra_data.insert(
725            "requires_python".to_string(),
726            JsonValue::String(requires_python.to_string()),
727        );
728    }
729
730    if let Some(markers) = toml_content.get(FIELD_RESOLUTION_MARKERS) {
731        extra_data.insert(
732            FIELD_RESOLUTION_MARKERS.to_string(),
733            toml_value_to_json(markers),
734        );
735    }
736
737    if let Some(manifest) = toml_content.get(FIELD_MANIFEST) {
738        extra_data.insert(FIELD_MANIFEST.to_string(), toml_value_to_json(manifest));
739    }
740
741    (!extra_data.is_empty()).then_some(extra_data)
742}
743
744fn build_package_extra_data(
745    package_table: &TomlMap<String, TomlValue>,
746) -> Option<HashMap<String, JsonValue>> {
747    let mut extra_data = HashMap::new();
748
749    if let Some(source) = package_table.get(FIELD_SOURCE) {
750        extra_data.insert(FIELD_SOURCE.to_string(), toml_value_to_json(source));
751    }
752
753    if let Some(metadata) = package_table.get(FIELD_METADATA) {
754        extra_data.insert(FIELD_METADATA.to_string(), toml_value_to_json(metadata));
755    }
756
757    (!extra_data.is_empty()).then_some(extra_data)
758}
759
760fn extract_artifact_metadata(
761    package_table: &TomlMap<String, TomlValue>,
762) -> (Option<String>, Option<String>) {
763    if let Some(sdist_table) = package_table.get("sdist").and_then(TomlValue::as_table) {
764        let download_url = sdist_table
765            .get("url")
766            .and_then(TomlValue::as_str)
767            .map(|value| value.to_string());
768        let sha256 = sdist_table
769            .get("hash")
770            .and_then(TomlValue::as_str)
771            .and_then(strip_sha256_prefix);
772        if download_url.is_some() || sha256.is_some() {
773            return (download_url, sha256);
774        }
775    }
776
777    let wheel_table = package_table
778        .get("wheels")
779        .and_then(TomlValue::as_array)
780        .and_then(|wheels| wheels.first())
781        .and_then(TomlValue::as_table);
782
783    let download_url = wheel_table
784        .and_then(|table| table.get("url"))
785        .and_then(TomlValue::as_str)
786        .map(|value| value.to_string());
787    let sha256 = wheel_table
788        .and_then(|table| table.get("hash"))
789        .and_then(TomlValue::as_str)
790        .and_then(strip_sha256_prefix);
791
792    (download_url, sha256)
793}
794
795fn strip_sha256_prefix(value: &str) -> Option<String> {
796    value.strip_prefix("sha256:").map(|hash| hash.to_string())
797}
798
799fn package_source_table(
800    package_table: &TomlMap<String, TomlValue>,
801) -> Option<&TomlMap<String, TomlValue>> {
802    package_table
803        .get(FIELD_SOURCE)
804        .and_then(TomlValue::as_table)
805}
806
807fn source_table_key(source_table: &TomlMap<String, TomlValue>) -> Option<String> {
808    ["registry", "editable", "virtual", "git"]
809        .into_iter()
810        .find_map(|key| {
811            source_table
812                .get(key)
813                .and_then(TomlValue::as_str)
814                .map(|value| format!("{}:{}", key, value))
815        })
816}
817
818fn build_pypi_urls(
819    name: Option<&str>,
820    version: Option<&str>,
821) -> (
822    Option<String>,
823    Option<String>,
824    Option<String>,
825    Option<String>,
826) {
827    let repository_homepage_url = name.map(|value| format!("https://pypi.org/project/{}", value));
828
829    let repository_download_url = name.and_then(|value| {
830        version.map(|ver| {
831            format!(
832                "https://pypi.org/packages/source/{}/{}/{}-{}.tar.gz",
833                &value[..1.min(value.len())],
834                value,
835                value,
836                ver
837            )
838        })
839    });
840
841    let api_data_url = name.map(|value| {
842        if let Some(ver) = version {
843            format!("https://pypi.org/pypi/{}/{}/json", value, ver)
844        } else {
845            format!("https://pypi.org/pypi/{}/json", value)
846        }
847    });
848
849    let purl = name.and_then(|value| create_pypi_purl(value, version));
850
851    (
852        repository_homepage_url,
853        repository_download_url,
854        api_data_url,
855        purl,
856    )
857}
858
859fn normalize_pypi_name(name: &str) -> String {
860    name.trim().to_ascii_lowercase()
861}
862
863fn create_pypi_purl(name: &str, version: Option<&str>) -> Option<String> {
864    if name.contains('[') || name.contains(']') {
865        return Some(build_manual_pypi_purl(name, version));
866    }
867
868    if let Ok(mut purl) = PackageUrl::new(UvLockParser::PACKAGE_TYPE.as_str(), name) {
869        if let Some(version) = version
870            && purl.with_version(version).is_err()
871        {
872            return None;
873        }
874        return Some(purl.to_string());
875    }
876
877    Some(build_manual_pypi_purl(name, version))
878}
879
880fn build_manual_pypi_purl(name: &str, version: Option<&str>) -> String {
881    let encoded_name = name.replace('[', "%5b").replace(']', "%5d");
882    let mut purl = format!("pkg:pypi/{}", encoded_name);
883    if let Some(version) = version
884        && !version.is_empty()
885    {
886        purl.push('@');
887        purl.push_str(version);
888    }
889    purl
890}
891
892fn toml_value_to_json(value: &TomlValue) -> JsonValue {
893    match value {
894        TomlValue::String(value) => JsonValue::String(value.clone()),
895        TomlValue::Integer(value) => JsonValue::String(value.to_string()),
896        TomlValue::Float(value) => JsonValue::String(value.to_string()),
897        TomlValue::Boolean(value) => JsonValue::Bool(*value),
898        TomlValue::Datetime(value) => JsonValue::String(value.to_string()),
899        TomlValue::Array(values) => {
900            JsonValue::Array(values.iter().map(toml_value_to_json).collect())
901        }
902        TomlValue::Table(values) => JsonValue::Object(
903            values
904                .iter()
905                .map(|(key, value)| (key.clone(), toml_value_to_json(value)))
906                .collect(),
907        ),
908    }
909}
910
911fn default_package_data() -> PackageData {
912    PackageData {
913        package_type: Some(UvLockParser::PACKAGE_TYPE),
914        primary_language: Some("Python".to_string()),
915        datasource_id: Some(DatasourceId::PypiUvLock),
916        ..Default::default()
917    }
918}
919
920crate::register_parser!(
921    "uv lockfile",
922    &["**/uv.lock"],
923    "pypi",
924    "Python",
925    Some("https://docs.astral.sh/uv/concepts/projects/layout/"),
926);