Skip to main content

provenant/parsers/
bazel.rs

1//! Bazel BUILD file parser
2//!
3//! Extracts package metadata from Bazel BUILD files using Starlark (Python-like) syntax.
4//!
5//! ## Features
6//! - Parses Starlark syntax using starlark_syntax
7//! - Extracts build rules ending with "binary" or "library" (e.g., cc_binary, cc_library)
8//! - Extracts name and licenses fields from rule arguments
9//! - Falls back to parent directory name if no rules found
10//! - **Supports multiple packages**: `extract_packages()` returns all rules (100% parity)
11//!
12//! ## Usage
13//! - `extract_first_package()` - Returns first package (convenience method)
14//! - `extract_packages()` - Returns ALL packages (recommended for BUILD files)
15//!
16//! ## Reference
17//! Python implementation: `reference/scancode-toolkit/src/packagedcode/build.py` (BazelBuildHandler)
18
19use crate::models::{DatasourceId, Dependency, PackageData, PackageType};
20use crate::parsers::utils::{MAX_ITERATION_COUNT, truncate_field};
21use packageurl::PackageUrl;
22use serde_json::{Map as JsonMap, Value as JsonValue};
23use std::path::Path;
24
25use crate::parser_warn as warn;
26use starlark_syntax::syntax::ast;
27use starlark_syntax::syntax::module::AstModuleFields;
28use starlark_syntax::syntax::{AstModule, Dialect};
29
30use super::PackageParser;
31
32type StarlarkCallArgs = ast::CallArgsP<ast::AstNoPayload>;
33const SCANCODE_SIMPLE_TOP_LEVEL_KEY: &str = "scancode_simple_top_level";
34const MAX_RECURSION_DEPTH: usize = 50;
35
36struct StarlarkCall<'a> {
37    func: &'a ast::AstExpr,
38    args: &'a StarlarkCallArgs,
39}
40
41pub struct BazelBuildParser;
42
43impl PackageParser for BazelBuildParser {
44    const PACKAGE_TYPE: PackageType = PackageType::Bazel;
45
46    fn is_match(path: &Path) -> bool {
47        path.file_name()
48            .and_then(|name| name.to_str())
49            .is_some_and(|name| name == "BUILD")
50    }
51
52    fn extract_packages(path: &Path) -> Vec<PackageData> {
53        match parse_bazel_build(path) {
54            Ok(packages) if !packages.is_empty() => packages,
55            Ok(_) => vec![fallback_package_data(path)],
56            Err(e) => {
57                warn!("Failed to parse Bazel BUILD file {:?}: {}", path, e);
58                vec![fallback_package_data(path)]
59            }
60        }
61    }
62}
63
64/// Parse a Bazel BUILD file and extract all package data
65fn parse_bazel_build(path: &Path) -> Result<Vec<PackageData>, String> {
66    let content =
67        crate::parsers::utils::read_file_to_string(path, None).map_err(|e| e.to_string())?;
68    let module = parse_starlark_module("<BUILD>", content)?;
69    let scancode_simple_top_level = is_scancode_simple_top_level_module(&module);
70
71    let mut packages = Vec::new();
72
73    for statement in top_level_statements(&module)
74        .iter()
75        .take(MAX_ITERATION_COUNT)
76    {
77        if let Some(mut package_data) = extract_package_from_statement(statement) {
78            set_scancode_simple_top_level(&mut package_data, scancode_simple_top_level);
79            packages.push(package_data);
80        }
81    }
82
83    Ok(packages)
84}
85
86/// Extract package data from a single AST statement
87fn extract_package_from_statement(statement: &ast::AstStmt) -> Option<PackageData> {
88    let call = extract_call(statement)?;
89    let rule_name = extract_call_name(&call)?;
90
91    if !check_rule_name_ending(rule_name) {
92        return None;
93    }
94
95    let name = extract_string_kwarg(&call, "name")?;
96    let licenses = extract_string_list_kwarg(&call, "licenses");
97    let purl = build_bazel_purl(&name, None).map(truncate_field);
98
99    Some(PackageData {
100        package_type: Some(BazelBuildParser::PACKAGE_TYPE),
101        name: Some(truncate_field(name)),
102        extracted_license_statement: licenses.map(|licenses| truncate_field(licenses.join(", "))),
103        datasource_id: Some(DatasourceId::BazelBuild),
104        purl,
105        ..Default::default()
106    })
107}
108
109/// Check if rule name ends with "binary" or "library"
110fn check_rule_name_ending(rule_name: &str) -> bool {
111    rule_name.ends_with("binary") || rule_name.ends_with("library")
112}
113
114/// Create fallback package data using parent directory name
115fn fallback_package_data(path: &Path) -> PackageData {
116    let name = path
117        .parent()
118        .and_then(|p| p.file_name())
119        .and_then(|n| n.to_str())
120        .map(|s| truncate_field(s.to_string()));
121
122    PackageData {
123        package_type: Some(BazelBuildParser::PACKAGE_TYPE),
124        purl: name
125            .as_deref()
126            .and_then(|name| build_bazel_purl(name, None))
127            .map(truncate_field),
128        name,
129        datasource_id: Some(DatasourceId::BazelBuild),
130        ..Default::default()
131    }
132}
133
134fn set_scancode_simple_top_level(package_data: &mut PackageData, enabled: bool) {
135    let extra_data = package_data.extra_data.get_or_insert_with(Default::default);
136    extra_data.insert(
137        SCANCODE_SIMPLE_TOP_LEVEL_KEY.to_string(),
138        JsonValue::Bool(enabled),
139    );
140}
141
142fn is_scancode_simple_top_level_module(module: &AstModule) -> bool {
143    top_level_statements(module)
144        .iter()
145        .all(is_scancode_simple_top_level_statement)
146}
147
148fn is_scancode_simple_top_level_statement(statement: &ast::AstStmt) -> bool {
149    match &statement.node {
150        ast::StmtP::Expression(expr) => {
151            matches!(&expr.node, ast::ExprP::Call(func, _) if matches!(&func.node, ast::ExprP::Identifier(_)))
152        }
153        _ => true,
154    }
155}
156
157#[cfg(test)]
158mod tests {
159    use super::*;
160    use crate::models::PackageType;
161    use std::path::PathBuf;
162
163    #[test]
164    fn test_is_match() {
165        assert!(BazelBuildParser::is_match(&PathBuf::from("BUILD")));
166        assert!(BazelBuildParser::is_match(&PathBuf::from("path/to/BUILD")));
167        assert!(!BazelBuildParser::is_match(&PathBuf::from("BUILD.bazel")));
168        assert!(!BazelBuildParser::is_match(&PathBuf::from("build")));
169        assert!(!BazelBuildParser::is_match(&PathBuf::from("BUCK")));
170    }
171
172    #[test]
173    fn test_check_rule_name_ending() {
174        assert!(check_rule_name_ending("cc_binary"));
175        assert!(check_rule_name_ending("cc_library"));
176        assert!(check_rule_name_ending("java_binary"));
177        assert!(check_rule_name_ending("py_library"));
178        assert!(!check_rule_name_ending("filegroup"));
179        assert!(!check_rule_name_ending("load"));
180        assert!(!check_rule_name_ending("cc_test"));
181    }
182
183    #[test]
184    fn test_fallback_package_data() {
185        let path = PathBuf::from("/path/to/myproject/BUILD");
186        let pkg = fallback_package_data(&path);
187        assert_eq!(pkg.package_type, Some(PackageType::Bazel));
188        assert_eq!(pkg.name, Some("myproject".to_string()));
189        assert_eq!(pkg.purl.as_deref(), Some("pkg:bazel/myproject"));
190    }
191
192    #[test]
193    fn test_scancode_simple_top_level_allows_direct_calls() {
194        let module = parse_starlark_module(
195            "<BUILD>",
196            "cc_library(name = \"demo\")\npy_binary(name = \"tool\")\n".to_string(),
197        )
198        .expect("parse BUILD");
199
200        assert!(is_scancode_simple_top_level_module(&module));
201    }
202
203    #[test]
204    fn test_scancode_simple_top_level_rejects_attribute_calls() {
205        let module = parse_starlark_module(
206            "<BUILD>",
207            "selects.config_setting_group(name = \"demo\")\ncc_library(name = \"demo\")\n"
208                .to_string(),
209        )
210        .expect("parse BUILD");
211
212        assert!(!is_scancode_simple_top_level_module(&module));
213    }
214
215    #[test]
216    fn test_scancode_simple_top_level_rejects_non_call_expressions() {
217        let module =
218            parse_starlark_module("<BUILD>", "[(cc_binary(name = \"demo\"),)]\n".to_string())
219                .expect("parse BUILD");
220
221        assert!(!is_scancode_simple_top_level_module(&module));
222    }
223}
224
225crate::register_parser!(
226    "Bazel BUILD file",
227    &["**/BUILD"],
228    "bazel",
229    "",
230    Some("https://bazel.build/"),
231);
232
233pub struct BazelModuleParser;
234
235impl PackageParser for BazelModuleParser {
236    const PACKAGE_TYPE: PackageType = PackageType::Bazel;
237
238    fn is_match(path: &Path) -> bool {
239        path.file_name()
240            .and_then(|name| name.to_str())
241            .is_some_and(|name| name == "MODULE.bazel")
242    }
243
244    fn extract_packages(path: &Path) -> Vec<PackageData> {
245        match parse_bazel_module(path) {
246            Ok(package) => vec![package],
247            Err(e) => {
248                warn!("Failed to parse Bazel MODULE.bazel {:?}: {}", path, e);
249                vec![default_bazel_module_package_data()]
250            }
251        }
252    }
253}
254
255fn parse_bazel_module(path: &Path) -> Result<PackageData, String> {
256    let content =
257        crate::parsers::utils::read_file_to_string(path, None).map_err(|e| e.to_string())?;
258    let module = parse_starlark_module("<MODULE.bazel>", content)?;
259
260    let mut package = default_bazel_module_package_data();
261    let mut extra_data = JsonMap::new();
262    let mut dependencies = Vec::new();
263    let mut overrides = Vec::new();
264
265    for statement in top_level_statements(&module)
266        .iter()
267        .take(MAX_ITERATION_COUNT)
268    {
269        let Some(call) = extract_call(statement) else {
270            continue;
271        };
272
273        let Some(function_name) = extract_call_name(&call) else {
274            continue;
275        };
276
277        match function_name {
278            "module" => {
279                package.name = extract_string_kwarg(&call, "name").map(truncate_field);
280                package.version = extract_string_kwarg(&call, "version").map(truncate_field);
281                package.purl = package
282                    .name
283                    .as_deref()
284                    .and_then(|name| build_bazel_purl(name, package.version.as_deref()))
285                    .map(truncate_field);
286
287                if let Some(repo_name) =
288                    extract_string_kwarg(&call, "repo_name").map(truncate_field)
289                {
290                    extra_data.insert("repo_name".to_string(), JsonValue::String(repo_name));
291                }
292                if let Some(compatibility_level) = extract_int_kwarg(&call, "compatibility_level") {
293                    extra_data.insert(
294                        "compatibility_level".to_string(),
295                        JsonValue::Number(compatibility_level.into()),
296                    );
297                }
298                if let Some(bazel_compatibility) = extract_kwarg_json(&call, "bazel_compatibility")
299                {
300                    extra_data.insert("bazel_compatibility".to_string(), bazel_compatibility);
301                }
302            }
303            "bazel_dep" => {
304                if let Some(dep) = extract_bazel_dependency(&call) {
305                    dependencies.push(dep);
306                }
307            }
308            "archive_override"
309            | "git_override"
310            | "local_path_override"
311            | "single_version_override"
312            | "multiple_version_override" => {
313                overrides.push(extract_override(function_name, &call));
314            }
315            _ => {}
316        }
317    }
318
319    if package.name.is_none() {
320        return Ok(default_bazel_module_package_data());
321    }
322
323    if !overrides.is_empty() {
324        extra_data.insert("overrides".to_string(), JsonValue::Array(overrides));
325    }
326
327    package.dependencies = dependencies;
328    package.extra_data = (!extra_data.is_empty()).then(|| extra_data.into_iter().collect());
329    Ok(package)
330}
331
332fn parse_starlark_module(filename: &str, content: String) -> Result<AstModule, String> {
333    let dialect = Dialect {
334        enable_top_level_stmt: true,
335        ..Dialect::Standard
336    };
337    AstModule::parse(filename, content, &dialect).map_err(|error| error.to_string())
338}
339
340fn top_level_statements(module: &AstModule) -> &[ast::AstStmt] {
341    match &module.statement().node {
342        ast::StmtP::Statements(statements) => statements,
343        _ => std::slice::from_ref(module.statement()),
344    }
345}
346
347fn extract_call(statement: &ast::AstStmt) -> Option<StarlarkCall<'_>> {
348    match &statement.node {
349        ast::StmtP::Expression(expr) => extract_call_expr(expr),
350        ast::StmtP::Assign(assign) => extract_call_expr(&assign.rhs),
351        _ => None,
352    }
353}
354
355fn extract_call_expr(expr: &ast::AstExpr) -> Option<StarlarkCall<'_>> {
356    match &expr.node {
357        ast::ExprP::Call(func, args) => Some(StarlarkCall { func, args }),
358        _ => None,
359    }
360}
361
362fn extract_call_name<'a>(call: &'a StarlarkCall<'_>) -> Option<&'a str> {
363    match &call.func.node {
364        ast::ExprP::Identifier(identifier) => Some(identifier.node.ident.as_str()),
365        _ => None,
366    }
367}
368
369fn extract_named_kwarg<'a>(call: &'a StarlarkCall<'_>, key: &str) -> Option<&'a ast::AstExpr> {
370    call.args
371        .args
372        .iter()
373        .find_map(|argument| match &argument.node {
374            ast::ArgumentP::Named(name, value) if name.node == key => Some(value),
375            _ => None,
376        })
377}
378
379fn extract_string_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<String> {
380    extract_named_kwarg(call, key).and_then(expr_as_string)
381}
382
383fn extract_string_list_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<Vec<String>> {
384    let expr = extract_named_kwarg(call, key)?;
385    let items = match &expr.node {
386        ast::ExprP::List(items) | ast::ExprP::Tuple(items) => items,
387        _ => return None,
388    };
389    let values: Vec<_> = items
390        .iter()
391        .take(MAX_ITERATION_COUNT)
392        .filter_map(expr_as_string)
393        .collect();
394    (!values.is_empty()).then_some(values)
395}
396
397fn extract_bool_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<bool> {
398    extract_named_kwarg(call, key).and_then(expr_as_bool)
399}
400
401fn extract_int_kwarg(call: &StarlarkCall<'_>, key: &str) -> Option<i64> {
402    extract_named_kwarg(call, key).and_then(expr_as_i64)
403}
404
405fn extract_kwarg_json(call: &StarlarkCall<'_>, key: &str) -> Option<JsonValue> {
406    extract_named_kwarg(call, key).and_then(|expr| expr_to_json(expr, 0))
407}
408
409fn extract_bazel_dependency(call: &StarlarkCall<'_>) -> Option<Dependency> {
410    let name = extract_string_kwarg(call, "name").map(truncate_field)?;
411    let version = extract_string_kwarg(call, "version").map(truncate_field);
412    let is_dev = extract_bool_kwarg(call, "dev_dependency").unwrap_or(false);
413    let mut extra_data = JsonMap::new();
414
415    for field in ["repo_name", "max_compatibility_level", "registry"]
416        .iter()
417        .take(MAX_ITERATION_COUNT)
418    {
419        if let Some(value) = extract_kwarg_json(call, field) {
420            extra_data.insert(field.to_string(), value);
421        }
422    }
423
424    Some(Dependency {
425        purl: build_bazel_purl(&name, version.as_deref()).map(truncate_field),
426        extracted_requirement: version.clone(),
427        scope: Some(if is_dev { "dev" } else { "dependencies" }.to_string()),
428        is_runtime: Some(!is_dev),
429        is_optional: Some(is_dev),
430        is_pinned: Some(version.is_some()),
431        is_direct: Some(true),
432        resolved_package: None,
433        extra_data: (!extra_data.is_empty()).then(|| extra_data.into_iter().collect()),
434    })
435}
436
437fn extract_override(kind: &str, call: &StarlarkCall<'_>) -> JsonValue {
438    let mut override_map = JsonMap::new();
439    override_map.insert("kind".to_string(), JsonValue::String(kind.to_string()));
440    for argument in call.args.args.iter().take(MAX_ITERATION_COUNT) {
441        if let ast::ArgumentP::Named(name, value) = &argument.node
442            && let Some(value) = expr_to_json(value, 0)
443        {
444            override_map.insert(name.node.clone(), value);
445        }
446    }
447    JsonValue::Object(override_map)
448}
449
450fn expr_as_string(expr: &ast::AstExpr) -> Option<String> {
451    match &expr.node {
452        ast::ExprP::Literal(ast::AstLiteral::String(value)) => Some(value.node.clone()),
453        _ => None,
454    }
455}
456
457fn expr_as_bool(expr: &ast::AstExpr) -> Option<bool> {
458    match &expr.node {
459        ast::ExprP::Identifier(identifier) => match identifier.node.ident.as_str() {
460            "True" => Some(true),
461            "False" => Some(false),
462            _ => None,
463        },
464        _ => None,
465    }
466}
467
468fn expr_as_i64(expr: &ast::AstExpr) -> Option<i64> {
469    match &expr.node {
470        ast::ExprP::Literal(ast::AstLiteral::Int(value)) => value.node.to_string().parse().ok(),
471        _ => None,
472    }
473}
474
475fn expr_to_json(expr: &ast::AstExpr, depth: usize) -> Option<JsonValue> {
476    if depth > MAX_RECURSION_DEPTH {
477        return None;
478    }
479    match &expr.node {
480        ast::ExprP::Literal(ast::AstLiteral::String(value)) => {
481            Some(JsonValue::String(value.node.clone()))
482        }
483        ast::ExprP::Literal(ast::AstLiteral::Int(value)) => value
484            .node
485            .to_string()
486            .parse::<i64>()
487            .ok()
488            .map(|value| JsonValue::Number(value.into()))
489            .or_else(|| Some(JsonValue::String(value.node.to_string()))),
490        ast::ExprP::Literal(ast::AstLiteral::Float(value)) => {
491            serde_json::Number::from_f64(value.node).map(JsonValue::Number)
492        }
493        ast::ExprP::Identifier(identifier) => match identifier.node.ident.as_str() {
494            "True" => Some(JsonValue::Bool(true)),
495            "False" => Some(JsonValue::Bool(false)),
496            "None" => Some(JsonValue::Null),
497            _ => None,
498        },
499        ast::ExprP::List(elts) | ast::ExprP::Tuple(elts) => Some(JsonValue::Array(
500            elts.iter()
501                .take(MAX_ITERATION_COUNT)
502                .filter_map(|e| expr_to_json(e, depth + 1))
503                .collect(),
504        )),
505        ast::ExprP::Dict(items) => {
506            let mut map = JsonMap::new();
507            for (key, value) in items.iter().take(MAX_ITERATION_COUNT) {
508                let Some(key) = expr_as_string(key) else {
509                    continue;
510                };
511                if let Some(value) = expr_to_json(value, depth + 1) {
512                    map.insert(key, value);
513                }
514            }
515            Some(JsonValue::Object(map))
516        }
517        _ => None,
518    }
519}
520
521fn build_bazel_purl(name: &str, version: Option<&str>) -> Option<String> {
522    let mut purl = PackageUrl::new("bazel", name).ok()?;
523    if let Some(version) = version.filter(|value| !value.trim().is_empty()) {
524        purl.with_version(version).ok()?;
525    }
526    Some(purl.to_string())
527}
528
529fn default_bazel_module_package_data() -> PackageData {
530    PackageData {
531        package_type: Some(BazelModuleParser::PACKAGE_TYPE),
532        datasource_id: Some(DatasourceId::BazelModule),
533        ..Default::default()
534    }
535}
536
537crate::register_parser!(
538    "Bazel MODULE.bazel file",
539    &["**/MODULE.bazel"],
540    "bazel",
541    "",
542    Some("https://bazel.build/external/module"),
543);