autosnippet
Version:
Extract code patterns into a knowledge base for AI coding assistants
230 lines (229 loc) • 7.56 kB
JavaScript
/**
* @module StarlarkParser
* @description Starlark (Python 子集) 轻量解析器 — 从 BUILD / BUILD.bazel / BUCK 文件提取构建目标信息
*
* 支持解析：
* - load() 语句 → 推断语言
* - 目标声明 (swift_library, cc_binary, java_library 等)
* - name / deps / srcs / visibility 字段
*
* 设计策略: 正则 + 逐行状态机（不做宏展开）
*/
// ── 规则名 → 语言映射 ────────────────────────────────
export const RULE_TO_LANGUAGE = {
// Bazel
swift_library: 'swift',
swift_binary: 'swift',
swift_test: 'swift',
cc_library: 'cpp',
cc_binary: 'cpp',
cc_test: 'cpp',
java_library: 'java',
java_binary: 'java',
java_test: 'java',
kt_jvm_library: 'kotlin',
kt_jvm_binary: 'kotlin',
py_library: 'python',
py_binary: 'python',
py_test: 'python',
go_library: 'go',
go_binary: 'go',
go_test: 'go',
rust_library: 'rust',
rust_binary: 'rust',
rust_test: 'rust',
ts_project: 'typescript',
proto_library: 'protobuf',
// Buck2
cxx_library: 'cpp',
cxx_binary: 'cpp',
cxx_test: 'cpp',
android_library: 'kotlin',
android_binary: 'kotlin',
apple_library: 'swift',
apple_binary: 'swift',
python_library: 'python',
python_binary: 'python',
// Pants
python_source: 'python',
python_sources: 'python',
docker_image: 'docker',
};
// ── 正则模式 ────────────────────────────────────────
const LOAD_RE = /^load\(\s*"([^"]+)"\s*,\s*((?:"[^"]+"(?:\s*,\s*)?)+)\s*\)/;
const RULE_CALL_RE = /^(\w+)\(\s*$/;
const NAME_RE = /name\s*=\s*"([^"]+)"/;
const TESTONLY_RE = /testonly\s*=\s*(True|1)/;
const STRING_LIST_RE = (field) => new RegExp(`${field}\\s*=\\s*(?:glob\\(\\s*)?\\[([^\\]]*)\\]`, 's');
const DEP_LABEL_RE = /"((?:\/\/[^"]*|:[^"]*))"/g;
const GLOB_RE = /glob\(\s*\[([^\]]*)\]\s*\)/;
// ── 公开 API ────────────────────────────────────────
/**
* 解析单个 BUILD/BUILD.bazel/BUCK 文件内容
*/
export function parseStarlarkBuildFile(content) {
const result = { targets: [], loads: [] };
const lines = content.split('\n');
// Pass 1: 提取 load 语句
for (const line of lines) {
const trimmed = line.trim();
if (trimmed.startsWith('#')) {
continue;
}
const loadMatch = trimmed.match(LOAD_RE);
if (loadMatch) {
const fullLabel = loadMatch[1];
const symbolsPart = loadMatch[2];
// 解析 repository 和 path
let repository = '';
let path = fullLabel;
if (fullLabel.startsWith('@')) {
const slashIdx = fullLabel.indexOf('//');
if (slashIdx !== -1) {
repository = fullLabel.substring(0, slashIdx);
path = fullLabel.substring(slashIdx);
}
}
const symbols = [];
const symRe = /"([^"]+)"/g;
let symMatch;
while ((symMatch = symRe.exec(symbolsPart)) !== null) {
symbols.push(symMatch[1]);
}
result.loads.push({ repository, path, symbols });
}
}
// Pass 2: 提取目标声明（逐行状态机）
let i = 0;
while (i < lines.length) {
const trimmed = lines[i].trim();
// 跳过注释和空行
if (trimmed.startsWith('#') || trimmed === '') {
i++;
continue;
}
// 检测规则调用开始: rule_name(
const ruleMatch = trimmed.match(RULE_CALL_RE);
if (ruleMatch) {
const rule = ruleMatch[1];
// 收集整个调用块直到匹配的 )
const blockLines = [trimmed];
let depth = 1;
let j = i + 1;
while (j < lines.length && depth > 0) {
const bLine = lines[j];
for (const ch of bLine) {
if (ch === '(') {
depth++;
}
if (ch === ')') {
depth--;
}
}
blockLines.push(bLine);
j++;
}
const block = blockLines.join('\n');
const target = parseTargetBlock(rule, block);
if (target) {
result.targets.push(target);
}
i = j;
continue;
}
// 也检测单行调用: rule_name(name = "...")
const inlineMatch = trimmed.match(/^(\w+)\(/);
if (inlineMatch && RULE_TO_LANGUAGE[inlineMatch[1]]) {
// 同样收集到闭合 )
const rule = inlineMatch[1];
const blockLines = [trimmed];
let depth = 0;
for (const ch of trimmed) {
if (ch === '(') {
depth++;
}
if (ch === ')') {
depth--;
}
}
let j = i + 1;
while (j < lines.length && depth > 0) {
const bLine = lines[j];
for (const ch of bLine) {
if (ch === '(') {
depth++;
}
if (ch === ')') {
depth--;
}
}
blockLines.push(bLine);
j++;
}
const block = blockLines.join('\n');
const target = parseTargetBlock(rule, block);
if (target) {
result.targets.push(target);
}
i = j;
continue;
}
i++;
}
return result;
}
// ── 内部解析函数 ────────────────────────────────────
function parseTargetBlock(rule, block) {
const nameMatch = block.match(NAME_RE);
if (!nameMatch) {
return null;
}
return {
rule,
name: nameMatch[1],
srcs: extractStringList(block, 'srcs'),
deps: extractDepLabels(block),
visibility: extractStringList(block, 'visibility'),
testonly: TESTONLY_RE.test(block) ? true : undefined,
};
}
function extractStringList(block, field) {
const listMatch = block.match(STRING_LIST_RE(field));
if (!listMatch) {
return [];
}
let inner = listMatch[1];
// 处理 glob() 模式
const globMatch = inner.match(GLOB_RE);
if (globMatch) {
inner = globMatch[1];
}
const items = [];
const strRe = /"([^"]+)"/g;
let m;
while ((m = strRe.exec(inner)) !== null) {
items.push(m[1]);
}
return items;
}
function extractDepLabels(block) {
const depsMatch = block.match(STRING_LIST_RE('deps'));
if (!depsMatch) {
return [];
}
const inner = depsMatch[1];
const deps = [];
const re = new RegExp(DEP_LABEL_RE.source, 'g');
let m;
while ((m = re.exec(inner)) !== null) {
deps.push(m[1]);
}
// 也捕获非标签格式的字符串依赖
const strRe = /"([^"]+)"/g;
while ((m = strRe.exec(inner)) !== null) {
if (!m[1].startsWith('//') && !m[1].startsWith(':') && !deps.includes(m[1])) {
deps.push(m[1]);
}
}
return deps;
}