Skip to main content

provenant/parsers/
bazel.rs

1// SPDX-FileCopyrightText: nexB Inc. and others
2// SPDX-FileCopyrightText: Provenant contributors
3// SPDX-License-Identifier: Apache-2.0
4// Derived from ScanCode Toolkit (Apache-2.0); modified. See NOTICE.
5
6//! Bazel BUILD file parser
7//!
8//! Extracts package metadata from Bazel BUILD files using Starlark (Python-like) syntax.
9//!
10//! ## Features
11//! - Parses Starlark syntax using starlark_syntax
12//! - Extracts build rules ending with "binary" or "library" (e.g., cc_binary, cc_library)
13//! - Extracts name and licenses fields from rule arguments
14//! - Falls back to parent directory name if no rules found
15//! - **Supports multiple packages**: `extract_packages()` returns all rules (100% parity)
16//!
17//! ## Usage
18//! - `extract_first_package()` - Returns first package (convenience method)
19//! - `extract_packages()` - Returns ALL packages (recommended for BUILD files)
20//!
21//! ## Reference
22//! Python implementation: `reference/scancode-toolkit/src/packagedcode/build.py` (BazelBuildHandler)
23
24use crate::models::{DatasourceId, Dependency, PackageData, PackageType};
25use crate::parsers::utils::{RecursionGuard, capped_iteration_limit, truncate_field};
26use packageurl::PackageUrl;
27use serde_json::{Map as JsonMap, Value as JsonValue};
28use std::path::Path;
29
30use crate::parser_warn as warn;
31use starlark_syntax::syntax::ast;
32use starlark_syntax::syntax::{AstModule, Dialect};
33
34use super::PackageParser;
35use super::metadata::ParserMetadata;
36
37type StarlarkCallArgs = ast::CallArgsP<ast::AstNoPayload>;
38const SCANCODE_SIMPLE_TOP_LEVEL_KEY: &str = "scancode_simple_top_level";
39
40struct StarlarkCall<'a> {
41    func: &'a ast::AstExpr,
42    args: &'a StarlarkCallArgs,
43}
44
45pub struct BazelBuildParser;
46
47impl PackageParser for BazelBuildParser {
48    const PACKAGE_TYPE: PackageType = PackageType::Bazel;
49
50    fn metadata() -> Vec<ParserMetadata> {
51        vec![ParserMetadata {
52            description: "Bazel BUILD file",
53            file_patterns: &["**/BUILD"],
54            package_type: "bazel",
55            primary_language: "",
56            documentation_url: Some("https://bazel.build/"),
57        }]
58    }
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 == "BUILD")
64    }
65
66    fn extract_packages(path: &Path) -> Vec<PackageData> {
67        match parse_bazel_build(path) {
68            Ok(packages) if !packages.is_empty() => packages,
69            Ok(_) => vec![fallback_package_data(path)],
70            Err(e) => {
71                warn!("Failed to parse Bazel BUILD file {:?}: {}", path, e);
72                vec![fallback_package_data(path)]
73            }
74        }
75    }
76}
77
78/// Parse a Bazel BUILD file and extract all package data
79fn parse_bazel_build(path: &Path) -> Result<Vec<PackageData>, String> {
80    let content =
81        crate::parsers::utils::read_file_to_string(path, None).map_err(|e| e.to_string())?;
82    let module = parse_starlark_module("<BUILD>", content)?;
83    let scancode_simple_top_level = is_scancode_simple_top_level_module(&module);
84
85    let mut packages = Vec::new();
86
87    let statements = top_level_statements(&module);
88    let limit = capped_iteration_limit(statements.len(), "bazel BUILD top-level statements");
89    for statement in statements.iter().take(limit) {
90        if let Some(mut package_data) = extract_package_from_statement(statement) {
91            set_scancode_simple_top_level(&mut package_data, scancode_simple_top_level);
92            packages.push(package_data);
93        }
94    }
95
96    Ok(packages)
97}
98
99/// Extract package data from a single AST statement
100fn extract_package_from_statement(statement: &ast::AstStmt) -> Option<PackageData> {
101    let call = extract_call(statement)?;
102    let rule_name = extract_call_name(&call)?;
103
104    if !check_rule_name_ending(rule_name) {
105        return None;
106    }
107
108    let name = extract_string_kwarg(&call, "name")?;
109    let licenses = extract_string_list_kwarg(&call, "licenses");
110    let purl = build_bazel_purl(&name, None).map(truncate_field);
111
112    Some(PackageData {
113        package_type: Some(BazelBuildParser::PACKAGE_TYPE),
114        name: Some(truncate_field(name)),
115        extracted_license_statement: licenses.map(|licenses| truncate_field(licenses.join(", "))),
116        datasource_id: Some(DatasourceId::BazelBuild),
117        purl,
118        ..Default::default()
119    })
120}
121
122/// Check if rule name ends with "binary" or "library"
123fn check_rule_name_ending(rule_name: &str) -> bool {
124    rule_name.ends_with("binary") || rule_name.ends_with("library")
125}
126
127/// Create fallback package data using parent directory name
128fn fallback_package_data(path: &Path) -> PackageData {
129    let name = path
130        .parent()
131        .and_then(|p| p.file_name())
132        .and_then(|n| n.to_str())
133        .map(|s| truncate_field(s.to_string()));
134
135    PackageData {
136        package_type: Some(BazelBuildParser::PACKAGE_TYPE),
137        purl: name
138            .as_deref()
139            .and_then(|name| build_bazel_purl(name, None))
140            .map(truncate_field),
141        name,
142        datasource_id: Some(DatasourceId::BazelBuild),
143        ..Default::default()
144    }
145}
146
147fn set_scancode_simple_top_level(package_data: &mut PackageData, enabled: bool) {
148    let extra_data = package_data.extra_data.get_or_insert_with(Default::default);
149    extra_data.insert(
150        SCANCODE_SIMPLE_TOP_LEVEL_KEY.to_string(),
151        JsonValue::Bool(enabled),
152    );
153}
154
155fn is_scancode_simple_top_level_module(module: &AstModule) -> bool {
156    top_level_statements(module)
157        .iter()
158        .all(is_scancode_simple_top_level_statement)
159}
160
161fn is_scancode_simple_top_level_statement(statement: &ast::AstStmt) -> bool {
162    match &statement.node {
163        ast::StmtP::Expression(expr) => {
164            matches!(&expr.node, ast::ExprP::Call(func, _) if matches!(&func.node, ast::ExprP::Identifier(_)))
165        }
166        _ => true,
167    }
168}
169
170pub struct BazelModuleParser;
171
172impl PackageParser for BazelModuleParser {
173    const PACKAGE_TYPE: PackageType = PackageType::Bazel;
174
175    fn metadata() -> Vec<ParserMetadata> {
176        vec![ParserMetadata {
177            description: "Bazel MODULE.bazel file",
178            file_patterns: &["**/MODULE.bazel"],
179            package_type: "bazel",
180            primary_language: "",
181            documentation_url: Some("https://bazel.build/external/module"),
182        }]
183    }
184
185    fn is_match(path: &Path) -> bool {
186        path.file_name()
187            .and_then(|name| name.to_str())
188            .is_some_and(|name| name == "MODULE.bazel")
189    }
190
191    fn extract_packages(path: &Path) -> Vec<PackageData> {
192        match parse_bazel_module(path) {
193            Ok(package) => vec![package],
194            Err(e) => {
195                warn!("Failed to parse Bazel MODULE.bazel {:?}: {}", path, e);
196                vec![default_bazel_module_package_data()]
197            }
198        }
199    }
200}
201
202fn parse_bazel_module(path: &Path) -> Result<PackageData, String> {
203    let content =
204        crate::parsers::utils::read_file_to_string(path, None).map_err(|e| e.to_string())?;
205    let module = parse_starlark_module("<MODULE.bazel>", content)?;
206
207    let mut package = default_bazel_module_package_data();
208    let mut extra_data = JsonMap::new();
209    let mut dependencies = Vec::new();
210    let mut overrides = Vec::new();
211
212    let statements = top_level_statements(&module);
213    let limit = capped_iteration_limit(statements.len(), "MODULE.bazel top-level statements");
214    for statement in statements.iter().take(limit) {
215        let Some(call) = extract_call(statement) else {
216            continue;
217        };
218
219        let Some(function_name) = extract_call_name(&call) else {
220            continue;
221        };
222
223        match function_name {
224            "module" => {
225                package.name = extract_string_kwarg(&call, "name").map(truncate_field);
226                package.version = extract_string_kwarg(&call, "version").map(truncate_field);
227                package.purl = package
228                    .name
229                    .as_deref()
230                    .and_then(|name| build_bazel_purl(name, package.version.as_deref()))
231                    .map(truncate_field);
232
233                if let Some(repo_name) =
234                    extract_string_kwarg(&call, "repo_name").map(truncate_field)
235                {
236                    extra_data.insert("repo_name".to_string(), JsonValue::String(repo_name));
237                }
238                if let Some(compatibility_level) = extract_int_kwarg(&call, "compatibility_level") {
239                    extra_data.insert(
240                        "compatibility_level".to_string(),
241                        JsonValue::Number(compatibility_level.into()),
242                    );
243                }
244                if let Some(bazel_compatibility) = extract_kwarg_json(&call, "bazel_compatibility")
245                {
246                    extra_data.insert("bazel_compatibility".to_string(), bazel_compatibility);
247                }
248            }
249            "bazel_dep" => {
250                if let Some(dep) = extract_bazel_dependency(&call) {
251                    dependencies.push(dep);
252                }
253            }
254            "archive_override"
255            | "git_override"
256            | "local_path_override"
257            | "single_version_override"
258            | "multiple_version_override" => {
259                overrides.push(extract_override(function_name, &call));
260            }
261            _ => {}
262        }
263    }
264
265    if package.name.is_none() {
266        return Ok(default_bazel_module_package_data());
267    }
268
269    if !overrides.is_empty() {
270        extra_data.insert("overrides".to_string(), JsonValue::Array(overrides));
271    }
272
273    package.dependencies = dependencies;
274    package.extra_data = (!extra_data.is_empty()).then(|| extra_data.into_iter().collect());
275    Ok(package)
276}
277
278fn parse_starlark_module(filename: &str, content: String) -> Result<AstModule, String> {
279    let dialect = Dialect {
280        enable_top_level_stmt: true,
281        ..Dialect::Standard
282    };
283    AstModule::parse(filename, content, &dialect).map_err(|error| error.to_string())
284}
285
286fn top_level_statements(module: &AstModule) -> &[ast::AstStmt] {
287    match &module.statement().node {
288        ast::StmtP::Statements(statements) => statements,
289        _ => std::slice::from_ref(module.statement()),
290    }
291}
292
293fn extract_call(statement: &ast::AstStmt) -> Option<StarlarkCall<'_>> {
294    match &statement.node {
295        ast::StmtP::Expression(expr) => extract_call_expr(expr),
296        ast::StmtP::Assign(assign) => extract_call_expr(&assign.rhs),
297        _ => None,
298    }
299}
300
301fn extract_call_expr(expr: &ast::AstExpr) -> Option<StarlarkCall<'_>> {
302    match &expr.node {
303        ast::ExprP::Call(func, args) => Some(StarlarkCall { func, args }),
304        _ => None,
305    }
306}
307
308fn extract_call_name<'a>(call: &'a StarlarkCall<'_>) -> Option<&'a str> {
309    match &call.func.node {
310        ast::ExprP::Identifier(identifier) => Some(identifier.node.ident.as_str()),
311        _ => None,
312    }
313}
314
315fn extract_named_kwarg<'a>(call: &'a StarlarkCall<'_>, key: &str) -> Option<&'a ast::AstExpr> {
316    call.args
317        .args
318        .iter()
319        .find_map(|argument| match &argument.node {
320            ast::ArgumentP::Named(name, value) if name.node == key => Some(value),
321            _ => None,
322        })
323}
324
325fn extract_string_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<String> {
326    extract_named_kwarg(call, key).and_then(expr_as_string)
327}
328
329fn extract_string_list_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<Vec<String>> {
330    let expr = extract_named_kwarg(call, key)?;
331    let items = match &expr.node {
332        ast::ExprP::List(items) | ast::ExprP::Tuple(items) => items,
333        _ => return None,
334    };
335    let limit = capped_iteration_limit(items.len(), "bazel string-list kwarg items");
336    let values: Vec<_> = items
337        .iter()
338        .take(limit)
339        .filter_map(expr_as_string)
340        .collect();
341    (!values.is_empty()).then_some(values)
342}
343
344fn extract_bool_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<bool> {
345    extract_named_kwarg(call, key).and_then(expr_as_bool)
346}
347
348fn extract_int_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<i64> {
349    extract_named_kwarg(call, key).and_then(expr_as_i64)
350}
351
352fn extract_kwarg_json(call: &StarlarkCall<'_>, key: &str) -> Option<JsonValue> {
353    extract_named_kwarg(call, key)
354        .and_then(|expr| expr_to_json(expr, &mut RecursionGuard::depth_only()))
355}
356
357fn extract_bazel_dependency(call: &StarlarkCall<'_>) -> Option<Dependency> {
358    let name = extract_string_kwarg(call, "name").map(truncate_field)?;
359    let version = extract_string_kwarg(call, "version").map(truncate_field);
360    let is_dev = extract_bool_kwarg(call, "dev_dependency").unwrap_or(false);
361    let mut extra_data = JsonMap::new();
362
363    let fields = ["repo_name", "max_compatibility_level", "registry"];
364    let limit = capped_iteration_limit(fields.len(), "bazel dependency extra fields");
365    for field in fields.iter().take(limit) {
366        if let Some(value) = extract_kwarg_json(call, field) {
367            extra_data.insert(field.to_string(), value);
368        }
369    }
370
371    Some(Dependency {
372        purl: build_bazel_purl(&name, version.as_deref()).map(truncate_field),
373        extracted_requirement: version.clone(),
374        scope: Some(if is_dev { "dev" } else { "dependencies" }.to_string()),
375        is_runtime: Some(!is_dev),
376        is_optional: Some(is_dev),
377        is_pinned: Some(version.is_some()),
378        is_direct: Some(true),
379        resolved_package: None,
380        extra_data: (!extra_data.is_empty()).then(|| extra_data.into_iter().collect()),
381    })
382}
383
384fn extract_override(kind: &str, call: &StarlarkCall<'_>) -> JsonValue {
385    let mut override_map = JsonMap::new();
386    override_map.insert("kind".to_string(), JsonValue::String(kind.to_string()));
387    let limit = capped_iteration_limit(call.args.args.len(), "bazel override arguments");
388    for argument in call.args.args.iter().take(limit) {
389        if let ast::ArgumentP::Named(name, value) = &argument.node
390            && let Some(value) = expr_to_json(value, &mut RecursionGuard::depth_only())
391        {
392            override_map.insert(name.node.clone(), value);
393        }
394    }
395    JsonValue::Object(override_map)
396}
397
398fn expr_as_string(expr: &ast::AstExpr) -> Option<String> {
399    match &expr.node {
400        ast::ExprP::Literal(ast::AstLiteral::String(value)) => Some(value.node.clone()),
401        _ => None,
402    }
403}
404
405fn expr_as_bool(expr: &ast::AstExpr) -> Option<bool> {
406    match &expr.node {
407        ast::ExprP::Identifier(identifier) => match identifier.node.ident.as_str() {
408            "True" => Some(true),
409            "False" => Some(false),
410            _ => None,
411        },
412        _ => None,
413    }
414}
415
416fn expr_as_i64(expr: &ast::AstExpr) -> Option<i64> {
417    match &expr.node {
418        ast::ExprP::Literal(ast::AstLiteral::Int(value)) => value.node.to_string().parse().ok(),
419        _ => None,
420    }
421}
422
423fn expr_to_json(expr: &ast::AstExpr, guard: &mut RecursionGuard<()>) -> Option<JsonValue> {
424    if guard.descend() {
425        return None;
426    }
427    let result = match &expr.node {
428        ast::ExprP::Literal(ast::AstLiteral::String(value)) => {
429            Some(JsonValue::String(value.node.clone()))
430        }
431        ast::ExprP::Literal(ast::AstLiteral::Int(value)) => value
432            .node
433            .to_string()
434            .parse::<i64>()
435            .ok()
436            .map(|value| JsonValue::Number(value.into()))
437            .or_else(|| Some(JsonValue::String(value.node.to_string()))),
438        ast::ExprP::Literal(ast::AstLiteral::Float(value)) => {
439            serde_json::Number::from_f64(value.node).map(JsonValue::Number)
440        }
441        ast::ExprP::Identifier(identifier) => match identifier.node.ident.as_str() {
442            "True" => Some(JsonValue::Bool(true)),
443            "False" => Some(JsonValue::Bool(false)),
444            "None" => Some(JsonValue::Null),
445            _ => None,
446        },
447        ast::ExprP::List(elts) | ast::ExprP::Tuple(elts) => {
448            let limit = capped_iteration_limit(elts.len(), "bazel list/tuple elements");
449            Some(JsonValue::Array(
450                elts.iter()
451                    .take(limit)
452                    .filter_map(|e| expr_to_json(e, guard))
453                    .collect(),
454            ))
455        }
456        ast::ExprP::Dict(items) => {
457            let mut map = JsonMap::new();
458            let limit = capped_iteration_limit(items.len(), "bazel dict entries");
459            for (key, value) in items.iter().take(limit) {
460                let Some(key) = expr_as_string(key) else {
461                    continue;
462                };
463                if let Some(value) = expr_to_json(value, guard) {
464                    map.insert(key, value);
465                }
466            }
467            Some(JsonValue::Object(map))
468        }
469        _ => None,
470    };
471    guard.ascend();
472    result
473}
474
475fn build_bazel_purl(name: &str, version: Option<&str>) -> Option<String> {
476    let mut purl = PackageUrl::new("bazel", name).ok()?;
477    if let Some(version) = version.filter(|value| !value.trim().is_empty()) {
478        purl.with_version(version).ok()?;
479    }
480    Some(purl.to_string())
481}
482
483fn default_bazel_module_package_data() -> PackageData {
484    PackageData {
485        package_type: Some(BazelModuleParser::PACKAGE_TYPE),
486        datasource_id: Some(DatasourceId::BazelModule),
487        ..Default::default()
488    }
489}
490
491#[cfg(test)]
492mod tests {
493    use super::*;
494    use crate::models::PackageType;
495    use std::path::PathBuf;
496
497    #[test]
498    fn test_is_match() {
499        assert!(BazelBuildParser::is_match(&PathBuf::from("BUILD")));
500        assert!(BazelBuildParser::is_match(&PathBuf::from("path/to/BUILD")));
501        assert!(!BazelBuildParser::is_match(&PathBuf::from("BUILD.bazel")));
502        assert!(!BazelBuildParser::is_match(&PathBuf::from("build")));
503        assert!(!BazelBuildParser::is_match(&PathBuf::from("BUCK")));
504    }
505
506    #[test]
507    fn test_check_rule_name_ending() {
508        assert!(check_rule_name_ending("cc_binary"));
509        assert!(check_rule_name_ending("cc_library"));
510        assert!(check_rule_name_ending("java_binary"));
511        assert!(check_rule_name_ending("py_library"));
512        assert!(!check_rule_name_ending("filegroup"));
513        assert!(!check_rule_name_ending("load"));
514        assert!(!check_rule_name_ending("cc_test"));
515    }
516
517    #[test]
518    fn test_fallback_package_data() {
519        let path = PathBuf::from("/path/to/myproject/BUILD");
520        let pkg = fallback_package_data(&path);
521        assert_eq!(pkg.package_type, Some(PackageType::Bazel));
522        assert_eq!(pkg.name, Some("myproject".to_string()));
523        assert_eq!(pkg.purl.as_deref(), Some("pkg:bazel/myproject"));
524    }
525
526    #[test]
527    fn test_scancode_simple_top_level_allows_direct_calls() {
528        let module = parse_starlark_module(
529            "<BUILD>",
530            "cc_library(name = \"demo\")\npy_binary(name = \"tool\")\n".to_string(),
531        )
532        .expect("parse BUILD");
533
534        assert!(is_scancode_simple_top_level_module(&module));
535    }
536
537    #[test]
538    fn test_scancode_simple_top_level_rejects_attribute_calls() {
539        let module = parse_starlark_module(
540            "<BUILD>",
541            "selects.config_setting_group(name = \"demo\")\ncc_library(name = \"demo\")\n"
542                .to_string(),
543        )
544        .expect("parse BUILD");
545
546        assert!(!is_scancode_simple_top_level_module(&module));
547    }
548
549    #[test]
550    fn test_scancode_simple_top_level_rejects_non_call_expressions() {
551        let module =
552            parse_starlark_module("<BUILD>", "[(cc_binary(name = \"demo\"),)]\n".to_string())
553                .expect("parse BUILD");
554
555        assert!(!is_scancode_simple_top_level_module(&module));
556    }
557}