CodeQL documentation

About CodeQL

CodeQL is the analysis engine used by developers to automate security checks, and by security researchers to perform variant analysis.

In CodeQL, code is treated like data. Security vulnerabilities, bugs, and other errors are modeled as queries that can be executed against databases extracted from code. You can run the standard CodeQL queries, written by GitHub researchers and community contributors, or write your own to use in custom analyses. Queries that find potential bugs highlight the result directly in the source file.

About variant analysis

Variant analysis is the process of using a known security vulnerability as a seed to find similar problems in your code. It’s a technique that security engineers use to identify potential vulnerabilities, and ensure these threats are properly fixed across multiple codebases.

Querying code using CodeQL is the most efficient way to perform variant analysis. You can use the standard CodeQL queries to identify seed vulnerabilities, or find new vulnerabilities by writing your own custom CodeQL queries. Then, develop or iterate over the query to automatically find logical variants of the same bug that could be missed using traditional manual techniques.

CodeQL analysis

CodeQL analysis consists of three steps:

  1. Preparing the code, by creating a CodeQL database
  2. Running CodeQL queries against the database
  3. Interpreting the query results

Database creation

To create a database, CodeQL first extracts a single relational representation of each source file in the codebase.

For compiled languages, extraction works by monitoring the normal build process. Each time a compiler is invoked to process a source file, a copy of that file is made, and all relevant information about the source code is collected. This includes syntactic data about the abstract syntax tree and semantic data about name binding and type information.

For interpreted languages, the extractor runs directly on the source code, resolving dependencies to give an accurate representation of the codebase.

There is one extractor for each language supported by CodeQL to ensure that the extraction process is as accurate as possible. For multi-language codebases, databases are generated one language at a time.

After extraction, all the data required for analysis (relational data, copied source files, and a language-specific database schema, which specifies the mutual relations in the data) is imported into a single directory, known as a CodeQL database.

Query execution

After you’ve created a CodeQL database, one or more queries are executed against it. CodeQL queries are written in a specially-designed object-oriented query language called QL. You can run the queries checked out from the CodeQL repo (or custom queries that you’ve written yourself) using the CodeQL for VS Code extension or the CodeQL CLI. For more information about queries, see “Learning CodeQL.”

Query results

The final step converts results produced during query execution into a form that is more meaningful in the context of the source code. That is, the results are interpreted in a way that highlights the potential issue that the queries are designed to find.

Queries contain metadata properties that indicate how the results should be interpreted. For instance, some queries display a simple message at a single location in the code. Others display a series of locations that represent steps along a data-flow or control-flow path, along with a message explaining the significance of the result. Queries that don’t have metadata are not interpreted—their results are output as a table and not displayed in the source code.

Following interpretation, results are output for code review and triaging. In CodeQL for Visual Studio Code, interpreted query results are automatically displayed in the source code. Results generated by the CodeQL CLI can be output into a number of different formats for use with different tools.

About CodeQL databases

CodeQL databases contain queryable data extracted from a codebase, for a single language at a particular point in time. The database contains a full, hierarchical representation of the code, including a representation of the abstract syntax tree, the data flow graph, and the control flow graph.

Each language has its own unique database schema that defines the relations used to create a database. The schema provides an interface between the initial lexical analysis during the extraction process, and the actual complex analysis using CodeQL. The schema specifies, for instance, that there is a table for every language construct.

For each language, the CodeQL libraries define classes to provide a layer of abstraction over the database tables. This provides an object-oriented view of the data which makes it easier to write queries.

For example, in a CodeQL database for a Java program, two key tables are:

  • The expressions table containing a row for every single expression in the source code that was analyzed during the build process.
  • The statements table containing a row for every single statement in the source code that was analyzed during the build process.

The CodeQL library defines classes to provide a layer of abstraction over each of these tables (and the related auxiliary tables): Expr and Stmt.