Skip to main content

provenant/parsers/
poetry_lock.rs

1//! Parser for Poetry poetry.lock lockfiles.
2//!
3//! Extracts resolved dependency information from Poetry lockfiles which use TOML format
4//! to store resolved versions and metadata for Python dependencies.
5//!
6//! # Supported Formats
7//! - poetry.lock (TOML-based lockfile with package metadata)
8//!
9//! # Key Features
10//! - Direct vs transitive dependency tracking via `is_direct` flag
11//! - Dependency groups support (main, dev, etc.) via scope field
12//! - Dependency resolution with exact versions
13//! - Package URL (purl) generation for PyPI packages
14//! - Extra dependencies and optional package handling
15//!
16//! # Implementation Notes
17//! - Uses TOML parsing via `toml` crate
18//! - All lockfile versions are pinned (`is_pinned: Some(true)`)
19//! - Graceful error handling with `warn!()` logs
20//! - Integrates with Python parser utilities for PyPI URL building
21
22use std::collections::HashMap;
23use std::path::Path;
24
25use crate::parser_warn as warn;
26use packageurl::PackageUrl;
27use toml::Value as TomlValue;
28use toml::map::Map as TomlMap;
29
30use crate::models::{
31    DatasourceId, Dependency, PackageData, PackageType, ResolvedPackage, Sha256Digest,
32};
33use crate::parsers::python::{build_pypi_urls, read_toml_file};
34use crate::parsers::utils::{MAX_ITERATION_COUNT, truncate_field};
35
36use super::PackageParser;
37
38const FIELD_PACKAGE: &str = "package";
39const FIELD_METADATA: &str = "metadata";
40const FIELD_NAME: &str = "name";
41const FIELD_VERSION: &str = "version";
42const FIELD_PYTHON_VERSIONS: &str = "python-versions";
43const FIELD_DEPENDENCIES: &str = "dependencies";
44const FIELD_EXTRAS: &str = "extras";
45const FIELD_LOCK_VERSION: &str = "lock-version";
46
47/// Poetry lockfile parser for poetry.lock files.
48///
49/// Extracts pinned Python package dependencies from Poetry-managed projects.
50pub struct PoetryLockParser;
51
52impl PackageParser for PoetryLockParser {
53    const PACKAGE_TYPE: PackageType = PackageType::Pypi;
54
55    fn is_match(path: &Path) -> bool {
56        path.file_name()
57            .and_then(|name| name.to_str())
58            .map(|name| name == "poetry.lock")
59            .unwrap_or(false)
60    }
61
62    fn extract_packages(path: &Path) -> Vec<PackageData> {
63        let toml_content = match read_toml_file(path) {
64            Ok(content) => content,
65            Err(e) => {
66                warn!("Failed to read poetry.lock at {:?}: {}", path, e);
67                return vec![default_package_data()];
68            }
69        };
70
71        vec![parse_poetry_lock(&toml_content)]
72    }
73}
74
75fn parse_poetry_lock(toml_content: &TomlValue) -> PackageData {
76    let packages = toml_content
77        .get(FIELD_PACKAGE)
78        .and_then(|value| value.as_array())
79        .cloned()
80        .unwrap_or_default();
81
82    let metadata = toml_content
83        .get(FIELD_METADATA)
84        .and_then(|value| value.as_table());
85
86    let mut dependencies = Vec::new();
87    for package in packages.iter().take(MAX_ITERATION_COUNT) {
88        if let Some(package_table) = package.as_table()
89            && let Some(dependency) = build_dependency_from_package(package_table)
90        {
91            dependencies.push(dependency);
92        }
93    }
94
95    PackageData {
96        package_type: Some(PoetryLockParser::PACKAGE_TYPE),
97        namespace: None,
98        name: None,
99        version: None,
100        qualifiers: None,
101        subpath: None,
102        primary_language: Some("Python".to_string()),
103        description: None,
104        release_date: None,
105        parties: Vec::new(),
106        keywords: Vec::new(),
107        homepage_url: None,
108        download_url: None,
109        size: None,
110        sha1: None,
111        md5: None,
112        sha256: None,
113        sha512: None,
114        bug_tracking_url: None,
115        code_view_url: None,
116        vcs_url: None,
117        copyright: None,
118        holder: None,
119        declared_license_expression: None,
120        declared_license_expression_spdx: None,
121        license_detections: Vec::new(),
122        other_license_expression: None,
123        other_license_expression_spdx: None,
124        other_license_detections: Vec::new(),
125        extracted_license_statement: None,
126        notice_text: None,
127        source_packages: Vec::new(),
128        file_references: Vec::new(),
129        is_private: false,
130        is_virtual: false,
131        extra_data: build_metadata_extra_data(metadata),
132        dependencies,
133        repository_homepage_url: None,
134        repository_download_url: None,
135        api_data_url: None,
136        datasource_id: Some(DatasourceId::PypiPoetryLock),
137        purl: None,
138    }
139}
140
141fn build_metadata_extra_data(
142    metadata: Option<&TomlMap<String, TomlValue>>,
143) -> Option<HashMap<String, serde_json::Value>> {
144    let mut extra_data = HashMap::new();
145
146    if let Some(metadata) = metadata {
147        if let Some(python_versions) = metadata
148            .get(FIELD_PYTHON_VERSIONS)
149            .and_then(|value| value.as_str())
150            && !python_versions.is_empty()
151        {
152            extra_data.insert(
153                "python_version".to_string(),
154                serde_json::Value::String(truncate_field(python_versions.to_string())),
155            );
156        }
157
158        if let Some(lock_version) = metadata.get(FIELD_LOCK_VERSION) {
159            let lock_version = lock_version
160                .as_str()
161                .map(|value| value.to_string())
162                .or_else(|| lock_version.as_integer().map(|value| value.to_string()));
163
164            if let Some(lock_version) = lock_version
165                && !lock_version.is_empty()
166            {
167                extra_data.insert(
168                    "lock_version".to_string(),
169                    serde_json::Value::String(truncate_field(lock_version)),
170                );
171            }
172        }
173    }
174
175    if extra_data.is_empty() {
176        None
177    } else {
178        Some(extra_data)
179    }
180}
181
182fn build_dependency_from_package(package_table: &TomlMap<String, TomlValue>) -> Option<Dependency> {
183    let name = package_table
184        .get(FIELD_NAME)
185        .and_then(|value| value.as_str())
186        .map(normalize_pypi_name)
187        .map(truncate_field)?;
188
189    let version = package_table
190        .get(FIELD_VERSION)
191        .and_then(|value| value.as_str())
192        .map(|value| truncate_field(value.to_string()))?;
193
194    let purl = create_pypi_purl(&name, Some(&version));
195
196    let resolved_package = build_resolved_package(package_table, &name, &version);
197
198    let poetry_optional = package_table
199        .get("optional")
200        .and_then(|value| value.as_bool())
201        .unwrap_or(false);
202
203    let extra_data = Some(HashMap::from([(
204        "poetry_optional".to_string(),
205        serde_json::Value::Bool(poetry_optional),
206    )]));
207
208    Some(Dependency {
209        purl,
210        extracted_requirement: None,
211        scope: None,
212        is_runtime: None,
213        is_optional: None,
214        is_pinned: Some(true),
215        is_direct: None,
216        resolved_package: Some(Box::new(resolved_package)),
217        extra_data,
218    })
219}
220
221fn build_resolved_package(
222    package_table: &TomlMap<String, TomlValue>,
223    name: &str,
224    version: &str,
225) -> ResolvedPackage {
226    let dependencies = extract_package_dependencies(package_table);
227
228    let urls = build_pypi_urls(Some(name), Some(version));
229
230    let repository_homepage_url = urls.repository_homepage_url.map(truncate_field);
231    let repository_download_url = urls.repository_download_url.map(truncate_field);
232    let api_data_url = urls.api_data_url.map(truncate_field);
233    let purl = urls.purl.map(truncate_field);
234
235    // Extract sha256 hash from files array (first file's hash)
236    let sha256 = extract_sha256_from_files(package_table);
237
238    ResolvedPackage {
239        primary_language: Some("Python".to_string()),
240        download_url: None,
241        sha1: None,
242        sha256: sha256.and_then(|h| Sha256Digest::from_hex(&h).ok()),
243        sha512: None,
244        md5: None,
245        is_virtual: true,
246        extra_data: None,
247        dependencies,
248        repository_homepage_url,
249        repository_download_url,
250        api_data_url,
251        datasource_id: Some(DatasourceId::PypiPoetryLock),
252        purl,
253        ..ResolvedPackage::new(
254            PoetryLockParser::PACKAGE_TYPE,
255            String::new(),
256            truncate_field(name.to_string()),
257            truncate_field(version.to_string()),
258        )
259    }
260}
261
262fn extract_package_dependencies(package_table: &TomlMap<String, TomlValue>) -> Vec<Dependency> {
263    let mut dependencies = Vec::new();
264
265    if let Some(dep_table) = package_table
266        .get(FIELD_DEPENDENCIES)
267        .and_then(|value| value.as_table())
268    {
269        for (dep_name, dep_value) in dep_table.iter().take(MAX_ITERATION_COUNT) {
270            if let Some(dependency) = build_dependency_from_table(dep_name, dep_value) {
271                dependencies.push(dependency);
272            }
273        }
274    }
275
276    if let Some(extras_table) = package_table
277        .get(FIELD_EXTRAS)
278        .and_then(|value| value.as_table())
279    {
280        for (extra_name, extra_values) in extras_table.iter().take(MAX_ITERATION_COUNT) {
281            if let Some(extra_list) = extra_values.as_array() {
282                for extra in extra_list.iter().take(MAX_ITERATION_COUNT) {
283                    if let Some(spec) = extra.as_str()
284                        && let Some(dependency) = build_dependency_from_extra(extra_name, spec)
285                    {
286                        dependencies.push(dependency);
287                    }
288                }
289            }
290        }
291    }
292
293    dependencies
294}
295
296fn build_dependency_from_table(dep_name: &str, dep_value: &TomlValue) -> Option<Dependency> {
297    let (requirement, is_optional) = match dep_value {
298        TomlValue::String(value) => (Some(truncate_field(value.to_string())), false),
299        TomlValue::Table(table) => (
300            table
301                .get(FIELD_VERSION)
302                .and_then(|value| value.as_str())
303                .map(|value| truncate_field(value.to_string())),
304            table
305                .get("optional")
306                .and_then(|value| value.as_bool())
307                .unwrap_or(false),
308        ),
309        _ => (None, false),
310    };
311
312    let normalized_name = normalize_pypi_name(dep_name);
313    let purl = create_pypi_purl(&normalized_name, None);
314
315    Some(Dependency {
316        purl,
317        extracted_requirement: requirement,
318        scope: Some(truncate_field(FIELD_DEPENDENCIES.to_string())),
319        is_runtime: Some(true),
320        is_optional: Some(is_optional),
321        is_pinned: Some(false),
322        is_direct: Some(true),
323        resolved_package: None,
324        extra_data: None,
325    })
326}
327
328fn build_dependency_from_extra(extra_name: &str, spec: &str) -> Option<Dependency> {
329    let (name, requirement) = parse_poetry_dependency_spec(spec)?;
330    let purl = create_pypi_purl(&name, None);
331
332    Some(Dependency {
333        purl,
334        extracted_requirement: requirement,
335        scope: Some(truncate_field(extra_name.to_string())),
336        is_runtime: None,
337        is_optional: Some(true),
338        is_pinned: Some(false),
339        is_direct: Some(true),
340        resolved_package: None,
341        extra_data: None,
342    })
343}
344
345fn parse_poetry_dependency_spec(spec: &str) -> Option<(String, Option<String>)> {
346    let trimmed = spec.trim();
347    if trimmed.is_empty() {
348        return None;
349    }
350
351    if let Some(paren_pos) = trimmed.find(" (") {
352        let name_part = trimmed[..paren_pos].trim();
353        let requirement_part = trimmed[paren_pos + 2..].trim();
354        let requirement = requirement_part.trim_end_matches(')').trim();
355        if name_part.is_empty() {
356            return None;
357        }
358        let normalized_name = truncate_field(normalize_pypi_name(name_part));
359        let requirement = if requirement.is_empty() {
360            None
361        } else {
362            Some(truncate_field(requirement.to_string()))
363        };
364        return Some((normalized_name, requirement));
365    }
366
367    Some((truncate_field(normalize_pypi_name(trimmed)), None))
368}
369
370fn normalize_pypi_name(name: &str) -> String {
371    name.trim().to_ascii_lowercase()
372}
373
374fn create_pypi_purl(name: &str, version: Option<&str>) -> Option<String> {
375    if name.contains('[') || name.contains(']') {
376        return Some(truncate_field(build_manual_pypi_purl(name, version)));
377    }
378
379    if let Ok(mut purl) = PackageUrl::new(PoetryLockParser::PACKAGE_TYPE.as_str(), name) {
380        if let Some(version) = version
381            && purl.with_version(version).is_err()
382        {
383            return None;
384        }
385        return Some(truncate_field(purl.to_string()));
386    }
387
388    Some(truncate_field(build_manual_pypi_purl(name, version)))
389}
390
391fn build_manual_pypi_purl(name: &str, version: Option<&str>) -> String {
392    let encoded_name = encode_pypi_name(name);
393    let mut purl = format!("pkg:pypi/{}", encoded_name);
394    if let Some(version) = version
395        && !version.is_empty()
396    {
397        purl.push('@');
398        purl.push_str(version);
399    }
400    purl
401}
402
403fn encode_pypi_name(name: &str) -> String {
404    name.replace('[', "%5b").replace(']', "%5d")
405}
406
407fn extract_sha256_from_files(package_table: &TomlMap<String, TomlValue>) -> Option<String> {
408    package_table
409        .get("files")
410        .and_then(|files| files.as_array())
411        .and_then(|files_array| files_array.first())
412        .and_then(|first_file| first_file.as_table())
413        .and_then(|file_table| file_table.get("hash"))
414        .and_then(|hash_value| hash_value.as_str())
415        .and_then(|hash_str| {
416            hash_str
417                .strip_prefix("sha256:")
418                .map(|s| truncate_field(s.to_string()))
419        })
420}
421
422fn default_package_data() -> PackageData {
423    PackageData {
424        package_type: Some(PoetryLockParser::PACKAGE_TYPE),
425        primary_language: Some("Python".to_string()),
426        datasource_id: Some(DatasourceId::PypiPoetryLock),
427        ..Default::default()
428    }
429}
430
431crate::register_parser!(
432    "Poetry lockfile",
433    &["**/poetry.lock"],
434    "pypi",
435    "Python",
436    Some("https://python-poetry.org/docs/basic-usage/#installing-with-poetrylock"),
437);