推薦先看幾篇文章：
1.Android 熱修復(fù)使用Gradle Plugin1.5改造Nuwa插件
2.GradleTransformAPI的基本使用
3.Transform官方文檔
4.一起玩轉(zhuǎn)Android項(xiàng)目中的字節(jié)碼(強(qiáng)烈推薦)
5.Android編譯流程和Gradle使用
引用說(shuō)明
本文章主要作個(gè)人總結(jié)記錄，感謝CSDNAndroid高級(jí)架構(gòu)的資料和文章，如有侵權(quán)，請(qǐng)聯(lián)系刪除。
知識(shí)點(diǎn)
1. 打包過(guò)程
要了解Transform，首先我們需要知道Gradle構(gòu)建一個(gè)安卓應(yīng)用，會(huì)經(jīng)過(guò)哪些步驟。
1.png
上圖是谷歌官網(wǎng)給出的一個(gè)典型的apk構(gòu)建的過(guò)程，比較概括。主要包括兩個(gè)過(guò)程，首先是編譯過(guò)程，編譯的內(nèi)容包括本工程的文件以及依賴(lài)的各種庫(kù)文件，編譯的輸出包括dex文件和編譯后的資源文件。然后是打包過(guò)程。配合Keystore對(duì)第一步的輸出進(jìn)行簽名對(duì)齊，生成最終的apk文件。
2.png
上面這張圖對(duì)上面的步驟以及每步用到的工具進(jìn)行了細(xì)分，概括如下：
- Java編譯器對(duì)工程本身的java代碼進(jìn)行編譯，這些java代碼有三個(gè)來(lái)源：app的源代碼，由資源文件生成的R文件(aapt工具)，以及有aidl文件生成的java接口文件(aidl工具)。產(chǎn)出為.class文件。
- .class文件和依賴(lài)的三方庫(kù)文件通過(guò)dex工具生成Delvik虛擬機(jī)可執(zhí)行的.dex文件，可能有一個(gè)或多個(gè)，包含了所有的class信息，包括項(xiàng)目自身的class和依賴(lài)的class。產(chǎn)出為.dex文件。
- apkbuilder工具將.dex文件和編譯后的資源文件生成未經(jīng)簽名對(duì)齊的apk文件。這里編譯后的資源文件包括兩部分，一是由aapt編譯產(chǎn)生的編譯后的資源文件，二是依賴(lài)的三方庫(kù)里的資源文件。產(chǎn)出為未經(jīng)簽名的.apk文件。
- 分別由Jarsigner和zipalign對(duì)apk文件進(jìn)行簽名和對(duì)齊，生成最終的apk文件。
2. 什么是Transform
簡(jiǎn)單介紹：
從android-build-tool:gradle:1.5開(kāi)始，gradle插件包含了一個(gè)叫Transform的API，這個(gè)API允許第三方插件在class文件轉(zhuǎn)為為dex文件前操作編譯好的class文件，這個(gè)API的目標(biāo)就是簡(jiǎn)化class文件的自定義的操作而不用對(duì)Task進(jìn)行處理。
作用域：
Transform是作用在.class編譯后，打包成.dex前，可以對(duì).class和resource進(jìn)行再處理的部分。
為了印證，我們隨便建立一個(gè)項(xiàng)目Build的一次。
image
可以很清楚的看到，原生就帶了一系列Transform供使用。那么這些Transform是怎么組織在一起的呢，我們用一張圖表示：
image
每個(gè)Transform其實(shí)都是一個(gè)gradle task，Android編譯器中的TaskManager將每個(gè)Transform串連起來(lái)，第一個(gè)Transform接收來(lái)自javac編譯的結(jié)果，以及已經(jīng)拉取到在本地的第三方依賴(lài)（jar. aar），還有resource資源，注意，這里的resource并非android項(xiàng)目中的res資源，而是asset目錄下的資源。 這些編譯的中間產(chǎn)物，在Transform組成的鏈條上流動(dòng)，每個(gè)Transform節(jié)點(diǎn)可以對(duì)class進(jìn)行處理再傳遞給下一個(gè)Transform。我們常見(jiàn)的混淆，Desugar等邏輯，它們的實(shí)現(xiàn)如今都是封裝在一個(gè)個(gè)Transform中，而我們自定義的Transform，會(huì)插入到這個(gè)Transform鏈條的最前面。
但其實(shí)，上面這幅圖，只是展示Transform的其中一種情況。而Transform其實(shí)可以有兩種輸入，一種是消費(fèi)型的，當(dāng)前Transform需要將消費(fèi)型型輸出給下一個(gè)Transform，另一種是引用型的，當(dāng)前Transform可以讀取這些輸入，而不需要輸出給下一個(gè)Transform，比如Instant Run就是通過(guò)這種方式，檢查兩次編譯之間的diff的。
最終，我們定義的Transform會(huì)被轉(zhuǎn)化成一個(gè)個(gè)TransformTask，在Gradle編譯時(shí)調(diào)用。
TransformManager.class
/**
* Adds a Transform.
*
* <p>This makes the current transform consumes whatever Streams are currently available and
* creates new ones for the transform output.
*
* <p>his also creates a {@link TransformTask} to run the transform and wire it up with the
* dependencies of the consumed streams.
*
* @param taskFactory the task factory
* @param scope the current scope
* @param transform the transform to add
* @param callback a callback that is run when the task is actually configured
* @param <T> the type of the transform
* @return {@code Optional<AndroidTask<Transform>>} containing the AndroidTask for the given
* transform task if it was able to create it
*/
@NonNull
public <T extends Transform> Optional<TransformTask> addTransform(
@NonNull TaskFactory taskFactory,
@NonNull TransformVariantScope scope,
@NonNull T transform,
@Nullable TransformTask.ConfigActionCallback<T> callback) {
if (!validateTransform(transform)) {
// validate either throws an exception, or records the problem during sync
// so it's safe to just return null here.
return Optional.empty();
}
List<TransformStream> inputStreams = Lists.newArrayList();
String taskName = scope.getTaskName(getTaskNamePrefix(transform));
// get referenced-only streams
List<TransformStream> referencedStreams = grabReferencedStreams(transform);
// find input streams, and compute output streams for the transform.
IntermediateStream outputStream = findTransformStreams(
transform,
scope,
inputStreams,
taskName,
scope.getGlobalScope().getBuildDir());
if (inputStreams.isEmpty() && referencedStreams.isEmpty()) {
// didn't find any match. Means there is a broken order somewhere in the streams.
issueReporter.reportError(
Type.GENERIC,
new EvalIssueException(
String.format(
"Unable to add Transform '%s' on variant '%s': requested streams not available: %s+%s / %s",
transform.getName(),
scope.getFullVariantName(),
transform.getScopes(),
transform.getReferencedScopes(),
transform.getInputTypes())));
return Optional.empty();
}
//noinspection PointlessBooleanExpression
if (DEBUG && logger.isEnabled(LogLevel.DEBUG)) {
logger.debug("ADDED TRANSFORM(" + scope.getFullVariantName() + "):");
logger.debug("\tName: " + transform.getName());
logger.debug("\tTask: " + taskName);
for (TransformStream sd : inputStreams) {
logger.debug("\tInputStream: " + sd);
}
for (TransformStream sd : referencedStreams) {
logger.debug("\tRef'edStream: " + sd);
}
if (outputStream != null) {
logger.debug("\tOutputStream: " + outputStream);
}
}
transforms.add(transform);
// create the task...
TransformTask task =
taskFactory.create(
new TransformTask.ConfigAction<>(
scope.getFullVariantName(),
taskName,
transform,
inputStreams,
referencedStreams,
outputStream,
recorder,
callback));
return Optional.ofNullable(task);
}
3.Transform解讀
我們首先先定義一個(gè)自定義的Transform，需要實(shí)現(xiàn)如下方法。
class AspectJTransform extends Transform {
final String NAME = "AjcTransform"
@Override
String getName() {
return NAME
}
@Override
Set<QualifiedContent.ContentType> getInputTypes() {
return TransformManager.CONTENT_CLASS
}
@Override
Set<? super QualifiedContent.Scope> getScopes() {
return TransformManager.SCOPE_FULL_PROJECT
}
@Override
boolean isIncremental() {
return true
}
@Override
void transform(TransformInvocation transformInvocation) throws TransformException, InterruptedException, IOException {
super.transform(transformInvocation)
}
首先我們一項(xiàng)項(xiàng)分析：
(1)Name
@Override
String getName() {
return NAME
}
Name顧名思義，就是我們的Transform名稱(chēng)，再回到我們剛剛Build的流程里：
image
這個(gè)最終的名字是如何構(gòu)成的呢？好像跟我們這邊的定義的名字有區(qū)別。
在gradle plugin的源碼中有一個(gè)叫TransformManager的類(lèi)，這個(gè)類(lèi)管理著所有的Transform的子類(lèi)，里面有一個(gè)方法叫g(shù)etTaskNamePrefix，在這個(gè)方法中就是獲得Task的前綴，以transform開(kāi)頭，之后拼接ContentType，這個(gè)ContentType代表著這個(gè)Transform的輸入文件的類(lèi)型，類(lèi)型主要有兩種，一種是Classes，另一種是Resources，ContentType之間使用And連接，拼接完成后加上With，之后緊跟的就是這個(gè)Transform的Name，name在getName()方法中重寫(xiě)返回即可。代碼如下：
@NonNull
static String getTaskNamePrefix(@NonNull Transform transform) {
StringBuilder sb = new StringBuilder(100);
sb.append("transform");
sb.append(
transform
.getInputTypes()
.stream()
.map(
inputType ->
CaseFormat.UPPER_UNDERSCORE.to(
CaseFormat.UPPER_CAMEL, inputType.name()))
.sorted() // Keep the order stable.
.collect(Collectors.joining("And")));
sb.append("With");
StringHelper.appendCapitalized(sb, transform.getName());
sb.append("For");
return sb.toString();
}
(2)getInputTypes()
先來(lái)看代碼注釋?zhuān)⑨寣?xiě)的很清晰了，必須是CLASSES(0x01),RESOURCES(0x02)之一，相當(dāng)于Transform需要處理的類(lèi)型。
/**
* Returns the type(s) of data that is consumed by the Transform. This may be more than
* one type.
*
* <strong>This must be of type {@link QualifiedContent.DefaultContentType}</strong>
*/
@NonNull
public abstract Set<ContentType> getInputTypes();
----------------------------------
/**
* The type of of the content.
*/
enum DefaultContentType implements ContentType {
/**
* The content is compiled Java code. This can be in a Jar file or in a folder. If
* in a folder, it is expected to in sub-folders matching package names.
*/
CLASSES(0x01),
/** The content is standard Java resources. */
RESOURCES(0x02);
private final int value;
DefaultContentType(int value) {
this.value = value;
}
@Override
public int getValue() {
return value;
}
}
其實(shí)還有一些另外的，還沒(méi)研究
CotentType
(3)getScopes()
先來(lái)看源碼注釋?zhuān)@個(gè)的作用相當(dāng)于用來(lái)Transform表明作用域
/**
* Returns the scope(s) of the Transform. This indicates which scopes the transform consumes.
*/
@NonNull
public abstract Set<Scope> getScopes();
開(kāi)發(fā)一共可以選如下幾種：
/**
* The scope of the content.
*
* <p>
* This indicates what the content represents, so that Transforms can apply to only part(s)
* of the classes or resources that the build manipulates.
*/
enum Scope implements ScopeType {
/** Only the project (module) content */
PROJECT(0x01),
/** Only the sub-projects (other modules) */
SUB_PROJECTS(0x04),
/** Only the external libraries */
EXTERNAL_LIBRARIES(0x10),
/** Code that is being tested by the current variant, including dependencies */
TESTED_CODE(0x20),
/** Local or remote dependencies that are provided-only */
PROVIDED_ONLY(0x40),
/**
* Only the project's local dependencies (local jars)
*
* @deprecated local dependencies are now processed as {@link #EXTERNAL_LIBRARIES}
*/
@Deprecated
PROJECT_LOCAL_DEPS(0x02),
/**
* Only the sub-projects's local dependencies (local jars).
*
* @deprecated local dependencies are now processed as {@link #EXTERNAL_LIBRARIES}
*/
@Deprecated
SUB_PROJECTS_LOCAL_DEPS(0x08);
一般來(lái)說(shuō)如果是要處理所有class字節(jié)碼，Scope我們一般使用TransformManager.SCOPE_FULL_PROJECT。即
public static final Set<Scope> SCOPE_FULL_PROJECT =
Sets.immutableEnumSet(
Scope.PROJECT,
Scope.SUB_PROJECTS,
Scope.EXTERNAL_LIBRARIES);
還有一些其他的可以參考下。
Scope
(4)isIncremental()
增量編譯開(kāi)關(guān)。
/**
* Returns whether the Transform can perform incremental work.
*
* <p>If it does, then the TransformInput may contain a list of changed/removed/added files, unless
* something else triggers a non incremental run.
*/
public abstract boolean isIncremental();
當(dāng)我們開(kāi)啟增量編譯的時(shí)候，相當(dāng)input包含了changed/removed/added三種狀態(tài)，實(shí)際上還有notchanged。需要做的操作如下：
- NOTCHANGED: 當(dāng)前文件不需處理，甚至復(fù)制操作都不用；
- ADDED、CHANGED: 正常處理，輸出給下一個(gè)任務(wù)；
- REMOVED: 移除outputProvider獲取路徑對(duì)應(yīng)的文件。
(5)transform()
先來(lái)看一下源碼注釋?zhuān)荰ransform處理文件的核心代碼：
/**
* Executes the Transform.
*
* <p>The inputs are packaged as an instance of {@link TransformInvocation}
* <ul>
* <li>The <var>inputs</var> collection of {@link TransformInput}. These are the inputs
* that are consumed by this Transform. A transformed version of these inputs must
* be written into the output. What is received is controlled through
* {@link #getInputTypes()}, and {@link #getScopes()}.</li>
* <li>The <var>referencedInputs</var> collection of {@link TransformInput}. This is
* for reference only and should be not be transformed. What is received is controlled
* through {@link #getReferencedScopes()}.</li>
* </ul>
*
* A transform that does not want to consume anything but instead just wants to see the content
* of some inputs should return an empty set in {@link #getScopes()}, and what it wants to
* see in {@link #getReferencedScopes()}.
*
* <p>Even though a transform's {@link Transform#isIncremental()} returns true, this method may
* be receive <code>false</code> in <var>isIncremental</var>. This can be due to
* <ul>
* <li>a change in secondary files ({@link #getSecondaryFiles()},
* {@link #getSecondaryFileOutputs()}, {@link #getSecondaryDirectoryOutputs()})</li>
* <li>a change to a non file input ({@link #getParameterInputs()})</li>
* <li>an unexpected change to the output files/directories. This should not happen unless
* tasks are improperly configured and clobber each other's output.</li>
* <li>a file deletion that the transform mechanism could not match to a previous input.
* This should not happen in most case, except in some cases where dependencies have
* changed.</li>
* </ul>
* In such an event, when <var>isIncremental</var> is false, the inputs will not have any
* incremental change information:
* <ul>
* <li>{@link JarInput#getStatus()} will return {@link Status#NOTCHANGED} even though
* the file may be added/changed.</li>
* <li>{@link DirectoryInput#getChangedFiles()} will return an empty map even though
* some files may be added/changed.</li>
* </ul>
*
* @param transformInvocation the invocation object containing the transform inputs.
* @throws IOException if an IO error occurs.
* @throws InterruptedException
* @throws TransformException Generic exception encapsulating the cause.
*/
public void transform(@NonNull TransformInvocation transformInvocation)
throws TransformException, InterruptedException, IOException {
// Just delegate to old method, for code that uses the old API.
//noinspection deprecation
transform(transformInvocation.getContext(), transformInvocation.getInputs(),
transformInvocation.getReferencedInputs(),
transformInvocation.getOutputProvider(),
transformInvocation.isIncremental());
}
大致意思如下，具體大家一定要仔細(xì)看注釋?zhuān)?br>
- 如果拿取了getInputs()的輸入進(jìn)行消費(fèi)，則transform后必須再輸出給下一級(jí)
- 如果拿取了getReferencedInputs()的輸入，則不應(yīng)該被transform。
- 是否增量編譯要以transformInvocation.isIncremental()為準(zhǔn)。
(6)getSecondaryFiles()
上面transform函數(shù)這里還提到了一個(gè)東西叫
secondary files ({@link #getSecondaryFiles()}
網(wǎng)上Transform的講解對(duì)它的提及比較少，先看看注釋?zhuān)臀覀冎敖榻B的一樣，有一系列API：
/**
* Returns a list of additional file(s) that this Transform needs to run. Preferably, use
* {@link #getSecondaryFiles()} API which allow eah secondary file to indicate if changes
* can be handled incrementally or not. This API will treat all additional file change as
* a non incremental event.
*
* <p>Changes to files returned in this list will trigger a new execution of the Transform
* even if the qualified-content inputs haven't been touched.
*
* <p>Any changes to these files will trigger a non incremental execution.
*
* <p>The default implementation returns an empty collection.
*
* @deprecated replaced by {@link #getSecondaryFiles()}
*/
@Deprecated
@NonNull
public Collection<File> getSecondaryFileInputs() {
return ImmutableList.of();
}
/**
* Returns a list of additional file(s) that this Transform needs to run.
*
* <p>Changes to files returned in this list will trigger a new execution of the Transform
* even if the qualified-content inputs haven't been touched.
*
* <p>Each secondary input has the ability to be declared as necessitating a non incremental
* execution in case of change. This Transform can therefore declare which secondary file
* changes it supports in incremental mode.
*
* <p>The default implementation returns an empty collection.
*/
@NonNull
public Collection<SecondaryFile> getSecondaryFiles() {
return ImmutableList.of();
}
/**
* Returns a list of additional (out of streams) file(s) that this Transform creates.
*
* <p>These File instances can only represent files, not directories. For directories, use
* {@link #getSecondaryDirectoryOutputs()}
*
*
* <p>Changes to files returned in this list will trigger a new execution of the Transform
* even if the qualified-content inputs haven't been touched.
*
* <p>Changes to these output files force a non incremental execution.
*
* <p>The default implementation returns an empty collection.
*/
@NonNull
public Collection<File> getSecondaryFileOutputs() {
return ImmutableList.of();
}
/**
* Returns a list of additional (out of streams) directory(ies) that this Transform creates.
*
* <p>These File instances can only represent directories. For files, use
* {@link #getSecondaryFileOutputs()}
*
* <p>Changes to directories returned in this list will trigger a new execution of the Transform
* even if the qualified-content inputs haven't been touched.
*
* <p>Changes to these output directories force a non incremental execution.
*
* <p>The default implementation returns an empty collection.
*/
@NonNull
public Collection<File> getSecondaryDirectoryOutputs() {
return ImmutableList.of();
}
跟據(jù)字面意思理解，除了主輸入/輸出流之外，Transform還可以額外定義另外的流供下個(gè)使用，不過(guò)我們平時(shí)用到的不多，可以根據(jù)系統(tǒng)自帶的Transform源碼看看它輸出了啥，比如ProguardTransform:
public class ProGuardTransform extends BaseProguardAction {
......
private final ImmutableList<File> secondaryFileOutputs;
......
public ProGuardTransform(@NonNull VariantScope variantScope) {
......
secondaryFileOutputs = ImmutableList.of(printMapping, printSeeds, printUsage);
}
@NonNull
@Override
public Collection<SecondaryFile> getSecondaryFiles() {
final List<SecondaryFile> files = Lists.newArrayList();
if (testedMappingFile != null && testedMappingFile.isFile()) {
files.add(SecondaryFile.nonIncremental(testedMappingFile));
} else if (testMappingConfiguration != null) {
files.add(SecondaryFile.nonIncremental(testMappingConfiguration));
}
// the config files
files.add(SecondaryFile.nonIncremental(getAllConfigurationFiles()));
return files;
}
@NonNull
@Override
public Collection<File> getSecondaryFileOutputs() {
return secondaryFileOutputs;
}
可以看到，它實(shí)際上是對(duì)mapping文件額外的配置，相當(dāng)于如注釋一樣，是相對(duì)于主流額外新一個(gè)流，實(shí)際開(kāi)發(fā)中我們用的較少。
(7)isCacheable()
按照慣例，先看注釋?zhuān)?/p>
/**
* Returns if this transform's outputs should be cached. Please read {@link
* org.gradle.api.tasks.CacheableTask} Javadoc if you would like to make your transform
* cacheable.
*/
public boolean isCacheable() {
return false;
}
如果我們的transform需要被緩存，則為true。
它被TransformTask所用到：
@CacheableTask
public class TransformTask extends StreamBasedTask implements Context {
@Override
public void execute(@NonNull TransformTask task) {
task.transform = transform;
task.consumedInputStreams = consumedInputStreams;
task.referencedInputStreams = referencedInputStreams;
task.outputStream = outputStream;
task.setVariantName(variantName);
task.recorder = recorder;
if (configActionCallback != null) {
configActionCallback.callback(transform, task);
}
//這一句，如果設(shè)置為true，則緩存輸出
task.getOutputs().cacheIf(t -> transform.isCacheable());
task.registerConsumedAndReferencedStreamInputs();
}
4.Transform編寫(xiě)模板
(1)無(wú)增量編譯：
class AspectJTransform extends Transform {
final String NAME = "AjcTransform"
@Override
String getName() {
return NAME
}
@Override
Set<QualifiedContent.ContentType> getInputTypes() {
return TransformManager.CONTENT_CLASS
}
@Override
Set<? super QualifiedContent.Scope> getScopes() {
return TransformManager.SCOPE_FULL_PROJECT
}
@Override
boolean isIncremental() {
return false
}
@Override
void transform(TransformInvocation transformInvocation) throws TransformException, InterruptedException, IOException {
super.transform(transformInvocation)
//OutputProvider管理輸出路徑，如果消費(fèi)型輸入為空，你會(huì)發(fā)現(xiàn)OutputProvider == null
TransformOutputProvider outputProvider = transformInvocation.getOutputProvider();
transformInvocation.inputs.each { TransformInput input ->
input.jarInputs.each { JarInput jarInput ->
//處理Jar
processJarInput(jarInput, outputProvider)
}
input.directoryInputs.each { DirectoryInput directoryInput ->
//處理源碼文件
processDirectoryInputs(directoryInput, outputProvider)
}
}
}
void processJarInput(JarInput jarInput, TransformOutputProvider outputProvider) {
File dest = outputProvider.getContentLocation(
jarInput.getFile().getAbsolutePath(),
jarInput.getContentTypes(),
jarInput.getScopes(),
Format.JAR)
//TODO do some transform
//將修改過(guò)的字節(jié)碼copy到dest，就可以實(shí)現(xiàn)編譯期間干預(yù)字節(jié)碼的目的了
FileUtils.copyFiley(jarInput.getFile(), dest)
}
void processDirectoryInputs(DirectoryInput directoryInput, TransformOutputProvider outputProvider) {
File dest = outputProvider.getContentLocation(directoryInput.getName(),
directoryInput.getContentTypes(), directoryInput.getScopes(),
Format.DIRECTORY)
//建立文件夾
FileUtils.forceMkdir(dest)
//TODO do some transform
//將修改過(guò)的字節(jié)碼copy到dest，就可以實(shí)現(xiàn)編譯期間干預(yù)字節(jié)碼的目的了
FileUtils.copyDirectory(directoryInput.getFile(), dest)
}
}
這里只是實(shí)現(xiàn)了簡(jiǎn)單的拷貝，具體怎么處理可以根據(jù)需求出發(fā)
(2)帶增量編譯(推薦)：
class AspectJTransform extends Transform {
final String NAME = "AjcTransform"
@Override
String getName() {
return NAME
}
@Override
Set<QualifiedContent.ContentType> getInputTypes() {
return TransformManager.CONTENT_CLASS
}
@Override
Set<? super QualifiedContent.Scope> getScopes() {
return TransformManager.SCOPE_FULL_PROJECT
}
@Override
boolean isIncremental() {
return true
}
@Override
void transform(TransformInvocation transformInvocation) throws TransformException, InterruptedException, IOException {
super.transform(transformInvocation)
boolean isIncremental = transformInvocation.isIncremental()
//OutputProvider管理輸出路徑，如果消費(fèi)型輸入為空，你會(huì)發(fā)現(xiàn)OutputProvider == null
TransformOutputProvider outputProvider = transformInvocation.getOutputProvider()
if (!isIncremental) {
//不需要增量編譯，先清除全部
outputProvider.deleteAll()
}
transformInvocation.getInputs().each { TransformInput input ->
input.jarInputs.each { JarInput jarInput ->
//處理Jar
processJarInputWithIncremental(jarInput, outputProvider, isIncremental)
}
input.directoryInputs.each { DirectoryInput directoryInput ->
//處理文件
processDirectoryInputWithIncremental(directoryInput, outputProvider, isIncremental)
}
}
}
void processJarInputWithIncremental(JarInput jarInput, TransformOutputProvider outputProvider, boolean isIncremental) {
File dest = outputProvider.getContentLocation(
jarInput.getFile().getAbsolutePath(),
jarInput.getContentTypes(),
jarInput.getScopes(),
Format.JAR)
if (isIncremental) {
//處理增量編譯
processJarInputWhenIncremental(jarInput, dest)
} else {
//不處理增量編譯
processJarInput(jarInput, dest)
}
}
void processJarInput(JarInput jarInput, File dest) {
transformJarInput(jarInput, dest)
}
void processJarInputWhenIncremental(JarInput jarInput, File dest) {
switch (jarInput.status) {
case Status.NOTCHANGED:
break
case Status.ADDED:
case Status.CHANGED:
//處理有變化的
transformJarInputWhenIncremental(jarInput.getFile(), dest, jarInput.status)
break
case Status.REMOVED:
//移除Removed
if (dest.exists()) {
FileUtils.forceDelete(dest)
}
break
}
}
void transformJarInputWhenIncremental(JarInput jarInput, File dest, Status status) {
if (status == Status.CHANGED) {
//Changed的狀態(tài)需要先刪除之前的
if (dest.exists()) {
FileUtils.forceDelete(dest)
}
}
//真正transform的地方
transformJarInput(jarInput, dest)
}
void transformJarInput(JarInput jarInput, File dest) {
//TODO do some transform
//將修改過(guò)的字節(jié)碼copy到dest，就可以實(shí)現(xiàn)編譯期間干預(yù)字節(jié)碼的目的了
FileUtils.copyFile(jarInput.getFile(), dest)
}
void processDirectoryInputWithIncremental(DirectoryInput directoryInput, TransformOutputProvider outputProvider, boolean isIncremental) {
File dest = outputProvider.getContentLocation(
directoryInput.getFile().getAbsolutePath(),
directoryInput.getContentTypes(),
directoryInput.getScopes(),
Format.DIRECTORY)
if (isIncremental) {
//處理增量編譯
processDirectoryInputWhenIncremental(directoryInput, dest)
} else {
processDirectoryInput(directoryInput, dest)
}
}
void processDirectoryInputWhenIncremental(DirectoryInput directoryInput, File dest) {
FileUtils.forceMkdir(dest)
String srcDirPath = directoryInput.getFile().getAbsolutePath()
String destDirPath = dest.getAbsolutePath()
Map<File, Status> fileStatusMap = directoryInput.getChangedFiles()
fileStatusMap.each { Map.Entry<File, Status> entry ->
File inputFile = entry.getKey()
Status status = entry.getValue()
String destFilePath = inputFile.getAbsolutePath().replace(srcDirPath, destDirPath)
File destFile = new File(destFilePath)
switch (status) {
case Status.NOTCHANGED:
break
case Status.REMOVED:
if (destFile.exists()) {
FileUtils.forceDelete(destFile)
}
break
case Status.ADDED:
case Status.CHANGED:
FileUtils.touch(destFile)
transformSingleFile(inputFile, destFile, srcDirPath)
break
}
}
}
void processDirectoryInput(DirectoryInput directoryInput, File dest) {
transformDirectoryInput(directoryInput, dest)
}
void transformDirectoryInput(DirectoryInput directoryInput, File dest) {
//TODO do some transform
//將修改過(guò)的字節(jié)碼copy到dest，就可以實(shí)現(xiàn)編譯期間干預(yù)字節(jié)碼的目的了
FileUtils.copyDirectory(directoryInput.getFile(), dest)
}
void transformSingleFile(File inputFile, File destFile, String srcDirPath) {
FileUtils.copyFile(inputFile, destFile)
}
}
5.Transform注冊(cè)和使用
定義一個(gè)插件
class AspectJWeaverPlugin implements Plugin<Project> {
@Override
void apply(Project project) {
boolean hasApp = project.getPlugins().hasPlugin(AppPlugin.class)
if (hasApp) {
def appExtension = project.getExtensions().getByType(AppExtension.class)
appExtension.registerTransform(new AspectJTransform(), Collections.EMPTY_LIST)
}
}
}
6.Transform優(yōu)化
一般就三種：
- 增量編譯
- 并發(fā)編譯
- include... exclude...縮小transform范圍
這里講一下并發(fā)編譯，簡(jiǎn)單實(shí)現(xiàn)如下：
WaitableExecutor waitableExecutor = WaitableExecutor.useGlobalSharedThreadPool()
......
transformInvocation.getInputs().each { TransformInput input ->
input.jarInputs.each { JarInput jarInput ->
//多線程處理Jar
waitableExecutor.execute(new Callable<Object>() {
@Override
Object call() throws Exception {
processJarInputWithIncremental(jarInput, outputProvider, isIncremental)
return null
}
})
}
input.directoryInputs.each { DirectoryInput directoryInput ->
//多線程處理文件
waitableExecutor.execute(new Callable<Object>() {
@Override
Object call() throws Exception {
processDirectoryInputWithIncremental(directoryInput, outputProvider, isIncremental)
return null
}
})
}
}
//等待所有任務(wù)結(jié)束
waitableExecutor.waitForTasksWithQuickFail(true)
7.Transform用途
結(jié)合ASM、AspectJ、javassit等字節(jié)碼處理框架進(jìn)行AOP編程，具體后面會(huì)講到
