Skip to main content

provenant/parsers/
yarn_pnp.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4use std::collections::{HashMap, HashSet};
5use std::path::Path;
6
7use crate::models::{DatasourceId, Dependency, PackageData, PackageType};
8use crate::parser_warn as warn;
9use crate::parsers::utils::{capped_iteration_limit, npm_purl, truncate_field};
10
11use super::PackageParser;
12
13pub struct YarnPnpParser;
14
15impl PackageParser for YarnPnpParser {
16    const PACKAGE_TYPE: PackageType = PackageType::Npm;
17
18    fn is_match(path: &Path) -> bool {
19        path.file_name().and_then(|name| name.to_str()) == Some(".pnp.cjs")
20    }
21
22    fn extract_packages(path: &Path) -> Vec<PackageData> {
23        let content = match crate::parsers::utils::read_file_to_string(path, None) {
24            Ok(content) => content,
25            Err(error) => {
26                warn!("Failed to read .pnp.cjs at {:?}: {}", path, error);
27                return vec![default_package_data()];
28            }
29        };
30
31        match parse_yarn_pnp(&content) {
32            Ok(package_data) => vec![package_data],
33            Err(error) => {
34                warn!("Failed to parse .pnp.cjs at {:?}: {}", path, error);
35                vec![default_package_data()]
36            }
37        }
38    }
39
40    fn metadata() -> Vec<super::metadata::ParserMetadata> {
41        vec![super::metadata::ParserMetadata {
42            description: "yarn plug and play runtime state",
43            file_patterns: &["**/.pnp.cjs"],
44            package_type: "npm",
45            primary_language: "JavaScript",
46            documentation_url: Some("https://yarnpkg.com/features/pnp"),
47        }]
48    }
49}
50
51fn default_package_data() -> PackageData {
52    PackageData {
53        package_type: Some(YarnPnpParser::PACKAGE_TYPE),
54        primary_language: Some("JavaScript".to_string()),
55        datasource_id: Some(DatasourceId::YarnPnpCjs),
56        ..Default::default()
57    }
58}
59
60fn parse_yarn_pnp(content: &str) -> Result<PackageData, String> {
61    let json_text = extract_raw_runtime_state_json(content)
62        .ok_or_else(|| "RAW_RUNTIME_STATE object not found in .pnp.cjs".to_string())?;
63    let runtime_state: serde_json::Value = serde_json::from_str(json_text)
64        .map_err(|error| format!("invalid RAW_RUNTIME_STATE JSON: {error}"))?;
65
66    let registry_entries = runtime_state
67        .get("packageRegistryData")
68        .and_then(serde_json::Value::as_array)
69        .ok_or_else(|| "packageRegistryData missing from RAW_RUNTIME_STATE".to_string())?;
70
71    let root_refs = registry_entries
72        .iter()
73        .find_map(parse_root_dependency_map)
74        .unwrap_or_default();
75    let mut seen_locators = HashSet::new();
76    let mut dependencies = Vec::new();
77
78    let entries_limit = capped_iteration_limit(registry_entries.len(), "yarn PnP registry entries");
79    for entry in registry_entries.iter().take(entries_limit) {
80        let Some(locator) = entry.get(0).and_then(serde_json::Value::as_str) else {
81            continue;
82        };
83        if !seen_locators.insert(locator.to_string()) {
84            continue;
85        }
86        let Some((name, reference)) = split_locator(locator) else {
87            continue;
88        };
89
90        let version = reference
91            .strip_prefix("npm:")
92            .map(|v| truncate_field(v.to_string()));
93        dependencies.push(Dependency {
94            purl: npm_purl(
95                truncate_field(name.to_string()).as_str(),
96                version.as_deref(),
97            ),
98            extracted_requirement: Some(truncate_field(reference.to_string())),
99            scope: Some("dependencies".to_string()),
100            is_runtime: Some(true),
101            is_optional: Some(false),
102            is_pinned: Some(version.is_some()),
103            is_direct: Some(
104                root_refs
105                    .get(name)
106                    .is_some_and(|root_ref| root_ref == reference),
107            ),
108            resolved_package: None,
109            extra_data: Some(HashMap::from([(
110                "locator".to_string(),
111                serde_json::Value::String(truncate_field(locator.to_string())),
112            )])),
113        });
114    }
115
116    let mut package = default_package_data();
117    package.dependencies = dependencies;
118    package.extra_data = Some(HashMap::from([(
119        "package_registry_entries".to_string(),
120        serde_json::Value::from(registry_entries.len()),
121    )]));
122    Ok(package)
123}
124
125fn parse_root_dependency_map(entry: &serde_json::Value) -> Option<HashMap<String, String>> {
126    if !entry.get(0).is_some_and(serde_json::Value::is_null) {
127        return None;
128    }
129
130    let dependencies = entry.get(1)?.get("packageDependencies")?;
131    Some(parse_dependency_pairs(dependencies))
132}
133
134fn parse_dependency_pairs(value: &serde_json::Value) -> HashMap<String, String> {
135    if let Some(array) = value.as_array() {
136        let limit = capped_iteration_limit(array.len(), "yarn PnP dependency pairs");
137        return array
138            .iter()
139            .take(limit)
140            .filter_map(|pair| {
141                let pair = pair.as_array()?;
142                let name = pair.first()?.as_str()?;
143                let reference = pair.get(1)?.as_str()?;
144                Some((
145                    truncate_field(name.to_string()),
146                    truncate_field(reference.to_string()),
147                ))
148            })
149            .collect();
150    }
151
152    let object_limit = capped_iteration_limit(
153        value.as_object().map_or(0, serde_json::Map::len),
154        "yarn PnP dependency object",
155    );
156    value
157        .as_object()
158        .into_iter()
159        .flatten()
160        .take(object_limit)
161        .filter_map(|(name, reference)| {
162            reference.as_str().map(|reference| {
163                (
164                    truncate_field(name.clone()),
165                    truncate_field(reference.to_string()),
166                )
167            })
168        })
169        .collect()
170}
171
172fn split_locator(locator: &str) -> Option<(&str, &str)> {
173    let split_at = locator.rfind('@')?;
174    if split_at == 0 {
175        return None;
176    }
177    Some((&locator[..split_at], &locator[split_at + 1..]))
178}
179
180fn extract_raw_runtime_state_json(content: &str) -> Option<&str> {
181    let marker = "const RAW_RUNTIME_STATE =";
182    let marker_index = content.find(marker)?;
183    let after_marker = &content[marker_index + marker.len()..];
184    let open_index = after_marker.find('{')?;
185    let json_start = marker_index + marker.len() + open_index;
186
187    let mut depth = 0usize;
188    let mut in_string = false;
189    let mut escaped = false;
190
191    for (offset, ch) in content[json_start..].char_indices() {
192        if in_string {
193            if escaped {
194                escaped = false;
195            } else if ch == '\\' {
196                escaped = true;
197            } else if ch == '"' {
198                in_string = false;
199            }
200            continue;
201        }
202
203        match ch {
204            '"' => in_string = true,
205            '{' => depth += 1,
206            '}' => {
207                depth = depth.saturating_sub(1);
208                if depth == 0 {
209                    let end = json_start + offset + ch.len_utf8();
210                    return Some(&content[json_start..end]);
211                }
212            }
213            _ => {}
214        }
215    }
216
217    None
218}