Java: `java/xxe` doesn't cover public XML helper methods that parse caller-provided input
- 主要語言
- CodeQL
- 星號
- 10.1k
- 分支
- 2.1k
- 平均合併
- 2 天 15 小時
- 30 天內合併 PR
- 141
描述
This looks like a gap in `java/xxe` around library-style XML helpers. If a public helper takes caller-provided XML as an `InputStream`, `String`, `Reader`, or `URL`, builds a `DocumentBuilderFactory` without XXE hardening, and then calls `newDocumentBuilder().parse(...)`, I wouldn't expect that to fall out of coverage.
CodeQL already seems to model the parser sink and the usual safe configs here, and it still finds other `java/xxe` results in the same codebase, so this doesn't look like a missing sink or a failed analysis. It looks more like these public parsing entry points are not treated as strong enough sources for this query. That's a pretty normal pattern in SDK helpers, XML utility classes, metadata loaders, and config parsers, where the trust boundary is the method parameter rather than a controller or request object.
The aws-sdk-android fix for CVE-2022-4725 is a good example: both `RegionMetadataParser.parse(InputStream)` and `XpathUtils.documentFrom(...)` were fixed by adding the standard `DocumentBuilderFactory` hardening.
Semgrep's `documentbuilderfactory-disallow-doctype-decl-missing` rule catches the same shape by reusing the existing parser-construction sink and checking for missing hardening, which makes this seem like a fairly local modeling issue.
My guess is that `java/xxe` could cover this by recognizing public parse helpers that feed caller-controlled XML directly, or through trivial wrappers, into the existing XXE sinks. I would expect the same hole to show up for similar helper methods built around `SAXParserFactory` and `XMLInputFactory` too.
```java
import java.io.InputStream;
import javax.xml.parsers.DocumentBuilderFactory;
class XmlHelper {
public static void parse(InputStream in) throws Exception {
DocumentBuilderFactory f = DocumentBuilderFactory.newInstance();
var b = f.newDocumentBuilder();
b.parse(in);
}
}
```
貢獻指南
研究方向
從 java/xxe 查詢及其現有的解析器建構 sink 和安全設定建模開始。檢查對將 InputStream、String、Reader 或 URL 值傳遞給 DocumentBuilderFactory.parse 的公開 helper 的涵蓋情況,然後與類似的 SAXParserFactory 和 XMLInputFactory 入口進行比較。完成標準是:測試能夠展示對所述 helper 形態和簡單 wrapper 的 findings,同時不遺失現有結果。
由索引模型根據 Issue 內容生成。
評估
- 技術堆疊
- java
- 領域
- security
- Issue 類型
- 功能
- 難度
- 4/5
- 預估耗時
- 3-5 天
- 活躍度
- 冷清
- 描述清晰度
- 基本清楚
- 新手友好度
- 48/100