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