How does component-scan work in Spring Framework

[JerryWang]

In Spring configuration xml file, we can define a package for tag component-scan, which tells Spring framework to search all classes within this specified package, to look for those classes which are annotated with @Named or @Component.

I am very curious about how Spring framework achieves this scan, so I have made some debugging to figure it out.

In this blog How to find the exact location where bean configuration file is parsed in Spring framework I have already found the location where the xml configuration file is parsed by Spring framework, so I can directly set breakpoint in found source code.

Here the package to be scanned is parsed from xml file:

And the actual scan is performed in line 87:

Here all classes within the specified package and its children packages are extracted as resource, now I have 7 resources as candidates for scan which makes sense since I have totally 7 classes in the package:

The evaluation to check whether the class has qualified annotation is done in this method:

If the scanned class has at least one annotation ( the annotation written on class is stored in metsadataReader ) which resides in this.includeFilters, then it is considered as a candidate.

By inspecting content of this.includeFilters, we can know that Spring framework considers @Component and @Named as qualified annotation for automatic component scan logic.

Back to my example, since my bean class has @named annotated,

In the runtime, this annotation written in class source code is extracted via reflection, and checked against Spring framework pre-defined annotation set. Here below is how my bean class evaluated as candidate, since it has @Named annotation.