前言
时光飞逝，从事Android系统开发已经两年了，总想写点什么来安慰自己。思考了非常久总是无法下笔，认为没什么好写的。如今最终决定写一些符合大多数人需求的东西，想必使用过Android手机的人们一定对“图库”（下面简称Gallery）这个应用非常熟悉。在Android市场里面有各种关于图库的应用，他们的最初原型事实上就是Android系统原生“图库”，仅仅是做了不同的差异化而已（UI差异化）。在研究Gallery源代码之前，我们须要对设计模式有一定的了解，依据自己对Gallery的了解，Gallery的设计就好比是一座设计精良的而且高效运转的机器（32个攒）。毫不夸张地说，在Android市场里，没有一款“图库”应用的设计设计可以和Gallery媲美。接下来的一段时间，就让我们共同来揭开Gallery的神奇面纱。
数据载入
在研究Gallery之前，我们还是来赞赏一下Gallery的总体效果，详细见图1-1所看到的：
图1-1
首先我们先来看一下Gallery的发展历史，在Android2.3之前Android系统的“图库”名为Gallery3D，在Android2.3之后系统将之前的Gallery3D更改为Gallery2，一直沿用到眼下最新版本号（4.4），Gallery2在UI和功能上面做了质的飞跃，是眼下Android源代码中很优秀的模块，对于Android应用开发人员来说是很好的开源项目，当中的设计新思想和设计模式都值得我们借鉴。
如今回到我们研究的主题-数据载入，我们先来看一下Gallery2在源代码中的路径(package/app/Gallery2/),在该路径下包括了“图库”使用的资源和源代码。我们在设计一款软件的时候首先考虑的是数据的存储和訪问，因此我们也依照这种设计思路来探究Gallery2的数据载入过程。讲到这儿略微提一下我分析源代码的方式，可能大家对Android源代码略微有一点了解的同学应该知道，Android源代码是很庞大的，因此选择分析程序的切入点大致能够分为两类：第一种是依照操程序操作步骤分析源代码——适用于界面跳转清晰的程序；另外一种是依据打印的Log信息分析程序的执行逻辑——适用于复杂的操作逻辑。
首先我们先来看一下BucketHelper.java类（/src/com/android/gallery3d/data/BucketHelper.java），该类主要是负责读取MediaProvider数据库中Image和Video数据，详细代码例如以下所看到的：
package com.android.gallery3d.data; import android.annotation.TargetApi;
import android.content.ContentResolver;
import android.database.Cursor;
import android.net.Uri;
import android.provider.MediaStore.Files;
import android.provider.MediaStore.Files.FileColumns;
import android.provider.MediaStore.Images;
import android.provider.MediaStore.Images.ImageColumns;
import android.provider.MediaStore.Video;
import android.util.Log; import com.android.gallery3d.common.ApiHelper;
import com.android.gallery3d.common.Utils;
import com.android.gallery3d.util.ThreadPool.JobContext; import java.util.ArrayList;
import java.util.Arrays;
import java.util.Comparator;
import java.util.HashMap; class BucketHelper { private static final String TAG = "BucketHelper";
private static final String EXTERNAL_MEDIA = "external"; // BUCKET_DISPLAY_NAME is a string like "Camera" which is the directory
// name of where an image or video is in. BUCKET_ID is a hash of the path
// name of that directory (see computeBucketValues() in MediaProvider for
// details). MEDIA_TYPE is video, image, audio, etc.
// BUCKET_DISPLAY_NAME字段为文件文件夹名称 BUCKET_ID字段为文件夹路径（path）的HASH值
// The "albums" are not explicitly recorded in the database, but each image
// or video has the two columns (BUCKET_ID, MEDIA_TYPE). We define an
// "album" to be the collection of images/videos which have the same value
// for the two columns.
// "专辑"的划分方式为：当文件具有同样的文件夹（BUCKET_ID）和多媒体类型（MEDIA_TYPE）即属于同一专辑
// The goal of the query (used in loadSubMediaSetsFromFilesTable()) is to
// find all albums, that is, all unique values for (BUCKET_ID, MEDIA_TYPE).
// In the meantime sort them by the timestamp of the latest image/video in
// each of the album.
//
// The order of columns below is important: it must match to the index in
// MediaStore.
private static final String[] PROJECTION_BUCKET = {
ImageColumns.BUCKET_ID,
FileColumns.MEDIA_TYPE,
ImageColumns.BUCKET_DISPLAY_NAME}; // The indices should match the above projections.
private static final int INDEX_BUCKET_ID = 0;
private static final int INDEX_MEDIA_TYPE = 1;
private static final int INDEX_BUCKET_NAME = 2; // We want to order the albums by reverse chronological order. We abuse the
// "WHERE" parameter to insert a "GROUP BY" clause into the SQL statement.
// The template for "WHERE" parameter is like:
// SELECT ... FROM ... WHERE (%s)
// and we make it look like:
// SELECT ... FROM ... WHERE (1) GROUP BY 1,(2)
// The "(1)" means true. The "1,(2)" means the first two columns specified
// after SELECT. Note that because there is a ")" in the template, we use
// "(2" to match it.
private static final String BUCKET_GROUP_BY = "1) GROUP BY 1,(2"; private static final String BUCKET_ORDER_BY = "MAX(datetaken) DESC"; // Before HoneyComb there is no Files table. Thus, we need to query the
// bucket info from the Images and Video tables and then merge them
// together.
//
// A bucket can exist in both tables. In this case, we need to find the
// latest timestamp from the two tables and sort ourselves. So we add the
// MAX(date_taken) to the projection and remove the media_type since we
// already know the media type from the table we query from.
private static final String[] PROJECTION_BUCKET_IN_ONE_TABLE = {
ImageColumns.BUCKET_ID,
"MAX(datetaken)",
ImageColumns.BUCKET_DISPLAY_NAME}; // We keep the INDEX_BUCKET_ID and INDEX_BUCKET_NAME the same as
// PROJECTION_BUCKET so we can reuse the values defined before.
private static final int INDEX_DATE_TAKEN = 1; // When query from the Images or Video tables, we only need to group by BUCKET_ID.
private static final String BUCKET_GROUP_BY_IN_ONE_TABLE = "1) GROUP BY (1"; public static BucketEntry[] loadBucketEntries(
JobContext jc, ContentResolver resolver, int type) {
if (ApiHelper.HAS_MEDIA_PROVIDER_FILES_TABLE) {//当API1>= 11(即Android3.0版本号之后)
return loadBucketEntriesFromFilesTable(jc, resolver, type);//获取MediaScanner数据库中多媒体文件（图片和视频）的文件夹路径和文件夹名称
} else {//Android3.0之前版本号
return loadBucketEntriesFromImagesAndVideoTable(jc, resolver, type);
}
} private static void updateBucketEntriesFromTable(JobContext jc,
ContentResolver resolver, Uri tableUri, HashMap<Integer, BucketEntry> buckets) {
Cursor cursor = resolver.query(tableUri, PROJECTION_BUCKET_IN_ONE_TABLE,
BUCKET_GROUP_BY_IN_ONE_TABLE, null, null);
if (cursor == null) {
Log.w(TAG, "cannot open media database: " + tableUri);
return;
}
try {
while (cursor.moveToNext()) {
int bucketId = cursor.getInt(INDEX_BUCKET_ID);
int dateTaken = cursor.getInt(INDEX_DATE_TAKEN);
BucketEntry entry = buckets.get(bucketId);
if (entry == null) {
entry = new BucketEntry(bucketId, cursor.getString(INDEX_BUCKET_NAME));
buckets.put(bucketId, entry);
entry.dateTaken = dateTaken;
} else {
entry.dateTaken = Math.max(entry.dateTaken, dateTaken);
}
}
} finally {
Utils.closeSilently(cursor);
}
} private static BucketEntry[] loadBucketEntriesFromImagesAndVideoTable(
JobContext jc, ContentResolver resolver, int type) {
HashMap<Integer, BucketEntry> buckets = new HashMap<Integer, BucketEntry>(64);
if ((type & MediaObject.MEDIA_TYPE_IMAGE) != 0) {
updateBucketEntriesFromTable(
jc, resolver, Images.Media.EXTERNAL_CONTENT_URI, buckets);
}
if ((type & MediaObject.MEDIA_TYPE_VIDEO) != 0) {
updateBucketEntriesFromTable(
jc, resolver, Video.Media.EXTERNAL_CONTENT_URI, buckets);
}
BucketEntry[] entries = buckets.values().toArray(new BucketEntry[buckets.size()]);
Arrays.sort(entries, new Comparator<BucketEntry>() {
@Override
public int compare(BucketEntry a, BucketEntry b) {
// sorted by dateTaken in descending order
return b.dateTaken - a.dateTaken;
}
});
return entries;
} private static BucketEntry[] loadBucketEntriesFromFilesTable(
JobContext jc, ContentResolver resolver, int type) {
Uri uri = getFilesContentUri(); Cursor cursor = resolver.query(uri,
PROJECTION_BUCKET, BUCKET_GROUP_BY,
null, BUCKET_ORDER_BY);
if (cursor == null) {
Log.w(TAG, "cannot open local database: " + uri);
return new BucketEntry[0];
}
ArrayList<BucketEntry> buffer = new ArrayList<BucketEntry>();
int typeBits = 0;
if ((type & MediaObject.MEDIA_TYPE_IMAGE) != 0) {
typeBits |= (1 << FileColumns.MEDIA_TYPE_IMAGE);
}
if ((type & MediaObject.MEDIA_TYPE_VIDEO) != 0) {
typeBits |= (1 << FileColumns.MEDIA_TYPE_VIDEO);
}
try {
while (cursor.moveToNext()) {
if ((typeBits & (1 << cursor.getInt(INDEX_MEDIA_TYPE))) != 0) {
BucketEntry entry = new BucketEntry(
cursor.getInt(INDEX_BUCKET_ID),
cursor.getString(INDEX_BUCKET_NAME));//构造元数据BucketEntry if (!buffer.contains(entry)) {
buffer.add(entry);//加入数据信息
}
}
if (jc.isCancelled()) return null;
}
} finally {
Utils.closeSilently(cursor);
}
return buffer.toArray(new BucketEntry[buffer.size()]);
} private static String getBucketNameInTable(
ContentResolver resolver, Uri tableUri, int bucketId) {
String selectionArgs[] = new String[] {String.valueOf(bucketId)};
Uri uri = tableUri.buildUpon()
.appendQueryParameter("limit", "1")
.build();
Cursor cursor = resolver.query(uri, PROJECTION_BUCKET_IN_ONE_TABLE,
"bucket_id = ?", selectionArgs, null);
try {
if (cursor != null && cursor.moveToNext()) {
return cursor.getString(INDEX_BUCKET_NAME);
}
} finally {
Utils.closeSilently(cursor);
}
return null;
} @TargetApi(ApiHelper.VERSION_CODES.HONEYCOMB)
private static Uri getFilesContentUri() {
return Files.getContentUri(EXTERNAL_MEDIA);
} public static String getBucketName(ContentResolver resolver, int bucketId) {
if (ApiHelper.HAS_MEDIA_PROVIDER_FILES_TABLE) {
String result = getBucketNameInTable(resolver, getFilesContentUri(), bucketId);
return result == null ? "" : result;
} else {
String result = getBucketNameInTable(
resolver, Images.Media.EXTERNAL_CONTENT_URI, bucketId);
if (result != null) return result;
result = getBucketNameInTable(
resolver, Video.Media.EXTERNAL_CONTENT_URI, bucketId);
return result == null ? "" : result;
}
} public static class BucketEntry {
public String bucketName;
public int bucketId;
public int dateTaken; public BucketEntry(int id, String name) {
bucketId = id;
bucketName = Utils.ensureNotNull(name);
} @Override
public int hashCode() {
return bucketId;
} @Override
public boolean equals(Object object) {
if (!(object instanceof BucketEntry)) return false;
BucketEntry entry = (BucketEntry) object;
return bucketId == entry.bucketId;
}
}
}
接下来我们再来看看BucketHelper类的调用关系的时序图，详细如1-2所看到的：
图1-2
到眼下为止我们大致了解了Gallery数据载入的一个大体流程，接下来的文章将分析Album数据的读取以及数据封装。