Skip to content

Latest commit

 

History

History
207 lines (164 loc) · 6.75 KB

java.md

File metadata and controls

207 lines (164 loc) · 6.75 KB

Build Java Targets

java_library

Build a jar from java source file。

java_library(
    name = 'rpc',
    srcs = glob([
        'src/main/java/**/*.java',
    ], exclude = [
        '*Test.java'
    ]),
    resources = [
        'src/main/resources/log4j.properties',
        'resources/rpc.conf',
        'resources/services.xml'
    ],
    deps = [
        '//common/rpc:rpc_error_code_proto'         # Depends on proto_libraries
        '//common/rpc:rpc_options_proto',
        '//common/net/http/java:http_utilities',    # Depends on other java libraries
        '//thirdparty/java/deps:slf4j',             # Depends on third party java libraries
        '//thirdparty/java/deps:guava',
        '//thirdparty/java/deps:netty-4.0.36.final',
    ]
)
  • resources attribute

resource file to be packed into this jar，you can also use glob function. Blade respect the Maven Standard Directory Layout, treats the resources dir as the root of resource files to be packed into the generated jar. So there are some mapping rules, for examples:

resources file path (str or tuple) path in jar description
resources/log4j.properties log4j.properties starts after the resources dir
src/main/resources/runtime.conf src/main/resources/runtime.conf ditto
('src/main/resources/runtime.conf', 'conf/runtime.conf') conf/runtime.conf manual rename the packed file
//app/global/countries.xml app/countries.xml can't find the root
  • prebuilt attribute

Used to describe a prebuilt jar file, but think twice before use it, usually maven_jar is more preferable.

java_library(
    name = 'parquet-column-gdt',
    prebuilt = True,
    binary_jar = 'parquet-column-1.9.1-SNAPSHOT.jar',
)
  • coverage attribute bool, Whether generate test coverage data for this library. It is useful to be False in some cases such as srcs are generated.

Mutiple kinds of dependancy

For java targets，except the normal deps, they also support exported_deps，provided_deps. There is a notable difference from C++ targets: for java_library, the dependancies are not transtive at the compile phrase. For example, in this BUILD file:

java_library(name = 'A', srcs = 'A.java', deps = ':B')
java_library(name = 'B', srcs = 'B.java', deps = ':C')
java_library(name = 'C', srcs = 'C.java')

The symbols defined in C.java is invisible to A.java, so when you write the deps for a java_library, you must add all of the direct dependencies into the deps. you can see you import list, ensure each library you imported is in the deps list.

  • exported_deps attribute

Each dep in the list will be transitive for the user of this library.

As you already know, in the java (and also other JVM languages such as scala) build rules, deps are not transitve. If a type from dependency appears in the public interface of a library, the users may don't know they should depends one your dependency, use this attribute will be a help.

When a dependency appears in the exported_deps, it will be passed to the source file at the compile phrase.

For the above example, if in B.java, some method used the type defined in C.java，the user of B, A also have to depends on C, or there will be a compile error. if you put C int the exported_deps of B, A will get C as as it compile dependency.

java_library(name = 'B', srcs = 'B.java', exported_deps = ':C')
  • provided_deps attribute

provided_deps is used to describe some libraries which will be provied by the runtime environment, same as the provided in maven scope，there dependencies will be used to compile，but they will not be packed into the final fatjar. The scenaio is hadoop or spark. this attribute can reduce the size of fatjar, and also reduces the conflict with the environment provided libraries.

maven_jar

Use this rule to describe a jar in the maven repository.

maven_jar (
  name = 'hadoop-common-2.7.2-tdw',
  id = 'org.apache.hadoop:hadoop-common:2.7.2-tdw-1.0.1',  # full maven artifact id
)

Besides name and id, maven_jar also provides the following attributes to fit the use of maven repository.

  • classifier attribute Specify classifier, for the same maven id(group:artifact:version), you can use classifier to locate different jar file. Such as hadoop-common-2.2.0.jar and hadoop-common-2.2.0-tests.jar。

  • transitive attribute Whether use transitive maven dependency, the default value is True, blade will download jar and its transitive dependencies; otherwise only the jar file of this target will be downloaded.

In order to avoid duplicated descriptions of artificts with the same id and avoid potential version conflicts, it is recommended to centralized management for maven_jars.

java_fat_library

Merge all java_library/maven_jar, generate a fatjar, can be used for deploy, same as jar-with-dependencies in maven.

java_library(
    name = 'log_process',
    srcs = glob(['**/*.java'], exclude = ['*Test.java']),
    resources = glob(['resources/*']),
    deps = [
        ':log_proto',
        '//log/com/tencent/monitor:monitor',
        '//log/com/tencent/monitor:schema',
        '//thirdparty/java/deps:hbase-client',
        '//thirdparty/java/deps:snappy',
    ],
    provided_deps = [
        '//thirdparty/java/deps:hadoop-common',
        '//thirdparty/java/deps:hadoop-mapreduce',
        '//thirdparty/java/deps:log4j',
        '//thirdparty/java/deps:slf4j',
    ],
)

java_fat_library(
    name = 'log_process_main',
    deps = [
        ':log_process',
        '//thirdparty/java/deps:guava',
    ],
    resources = [
        ('//log/configuration/log4j.properties', 'log4j.properties'),
        ('//log/configuration/sites.xml', 'conf/sites.xml'),
    ],
    exclusions = [
        'com.google.protobuf:*:*',
        'org.apache.spark:*:*',
    ]
)
  • exclusions attribute

Specify maven dependencies to be excludes. The syntax is a list of maven ids (group:artifact:version), also support wildcard, such as: com.google.protobuf:protobuf:\* and com.google.protobuf:\*:\*, but only the tail parts can be wildcard.

java_binary

Build executable from java source files.

java_binary(
    name = 'poppy_java_example',
    srcs = [
        glob('src/com/soso/poppy/tool/*.Java)'
    ],
    deps = [
        '//poppy:poppy_java_client',
        '//poppy:rpc_example_proto',
    ]
)

The results include a fat jar with a wrapper shell script.

java_test

Build and run tests.

java_test(
    name = 'poppy_java_test',
    srcs = [
        glob('test/com/soso/poppy/**/*Test.Java)'
    ],
    deps = [
        '//poppy:poppy_java',
        '//thirdparty/junit:junit',
    ],
)