to top

Content Provider Basics

一个内容提供者访问数据的中央资源库。提供者是应用程序的一部分,提供自己的操作数据的UI。然而,内容提供者主要是被其他应用程序引用,通过提供者客户对象访问提供者。提供者和提供者客户端为数据提供一个一致的,标准的接口,也处理进程间的联系和数据安全访问。

本文讨论下面几个方面的基础:

  • 内容提供类如何工作。
  • 从内容提供者获取数据的API。
  • 向内容提供者你插入、更新、删除数据的API。
  • 便利的使用内容提供者的API。

Overview

提供者向应用程序呈现数据就像一个或多张表,就像是在关系数据库里一样。一行显示一些数据类型的实例,列的每行显示实例数据集合的独立数据。

例如:一个在android平台里内建的提供者是用户词典,用来存储用户想保存的非标准的拼写。表一说明了在提供者的表里数据是如何的:

Table 1: Sample user dictionary table.

word app id frequency locale _ID
mapreduce user1 100 en_US 1
precompiler user14 200 fr_FR 2
applet user2 225 fr_CA 3
const user1 255 pt_BR 4
int user5 100 en_UK 5

在表1里,每行的单词都是在标准词典里找不到的。没列有一些和单词相关的数据,比如:它所属的区域。列头是列的名字。引用列的区域,你可以引用它的locale列。对于提供者来讲,自动认为列的_ID是“主键”。

注意:提供者并不是一定需要主键,它不需要使用_ID 作为主键的列名如果已经存在。然而,如果你想绑定一个提供者到一个ListView,一个列名是需要_ID的。对于这个需要的解释,在这里:Displaying query results。

访问一个提供者

一个应用程序通过客户类ContentResolver访问内容提供者的数据。这个函数有一个在提供者对象里同名的函数,是一个ContentProvider子类的实例。函数ContentResolver提供”CRUD”的基础(创建,获取,更新,删除)函数。

拥有提供者的应用程序进程的ContentResolver对象和ContentProvider对象可以自动处理进程间的数据交换。

注意:访问一个提供者,应用程序通常在manifest文件里请求指定的权限。在Content Provider Permissions一节里有更多细节。

例如,从User Dictionary Provider获取单词的一列和语言环境,你调用ContentResolver.query().query()函数调用定义的函数[ ContentProvider.query()]。下面的代码显示的是一个[ ContentResolver.query()]调用:

// Queries the user dictionary and returns results
mCursor = getContentResolver().query(
    UserDictionary.Words.CONTENT_URI,   // The content URI of the words table
    mProjection,                        // The columns to return for each row
    mSelectionClause                    // Selection criteria
    mSelectionArgs,                     // Selection criteria
    mSortOrder);                        // The sort order for the returned rows

Table 2 shows how the arguments to query(Uri,projection,selection,selectionArgs,sortOrder) match an SQL SELECT statement:

Table 2: Query() compared to SQL query.

query() argument SELECT keyword/parameter Notes
Uri FROM table_name Uri maps to the table in the provider named table_name.
projection col,col,col,... projection is an array of columns that should be included for each row retrieved.
selection WHERE col = value selection specifies the criteria for selecting rows.
selectionArgs (No exact equivalent. Selection arguments replace ? placeholders in the selection clause.)
sortOrder ORDER BY col,col,... sortOrder specifies the order in which rows appear in the returned Cursor.

Content URIs

一个内容URI是一个定于数据的URI。内容URI包含一个提供者的符号名(它的权限)和一个指向一个表(一个路径)的名字。当你调用客户端函数来访问表的时候,这个URI是参数的一个。

上面的代码,常量CONTENT_URI包含user dictionary的单词表的内容URI。对象ContentResolver 解析URI的权限,使用它。ContentResolver能分派查询参数来更正提供者。

使用内容URI的部分路径来选择想访问的表。提供者的每个外在的表都有一个路径。

之前的代码,“words”表完整的URI是:

content://user_dictionary/words

字符串user_dictionary是提供者的权限,字符串words是表的路径。字符串content://(配置)通常指示标识这项是一个内容URI。

提供者允许你通过附加在URI后的一个ID值来访问表的一行。例如,从获取_ID是4的一行,你可以使用内容URI:

Uri singleUri = ContentUri.withAppendedId(UserDictionary.Words.CONTENT_URI,4);

通常获取一个行集合的时候使用id并且想要更新或者删除它们中的一个。

注意:类Uri和Uri.Builder包含便利的函数来从一个字符串格式化Uri对象。ContetnUris包含一个便利的函数withAppendedId()来向URI最近一个id。之前的片段是用来追加id到Userdictionary。

从提供者里获取数据

从提供者里获取数据,例子使用User Dictionary Provider。 为了清晰,这一节的代码段在“UI线程”里调用ContentResolver.query()。实际的代码,然而,你需要在分开的线程里做异步查询。可以使用CursorLoader实现,更多的信息在Loaders 指南。代码很短;他们没有显示一个完整的应用。 获取数据,有以下的两个步骤: 1、需要提供者允许读访问。 2、发送一个query到提供者的代码。

请求读访问权限

从提供者获取数据,应用程序需要“读许可”。不可以在运行的时候申请;在你的manifest文件里声明,使用元素扩展权限名,它是提供者定义的。 当你在manifest里声明元素,实际上你就是申请权限。当用户安装引用程序,也就隐式的授予了请求。 为你使用的提供者查找确切的读访问权限的名字,就像其它提供者的访问权限的名字一样,参考提供者文档。 在Content Provider Permissions一节,有关于访问提供者的权限的更多信息。 User Dictionary Provider在manifest文件里定义android.permission.READ_USER_DICTIONARY权限,应用程序想从提供者读取需要这个权限。

构造查询

下一步获取数据时提供者构造一个查询。第一个片段为访问User Dictionary Provider定义了一些变量:


// A "projection" defines the columns that will be returned for each row
String[] mProjection =
{
    UserDictionary.Words._ID,    // Contract class constant for the _ID column name
    UserDictionary.Words.WORD,   // Contract class constant for the word column name
    UserDictionary.Words.LOCALE  // Contract class constant for the locale column name
};

// Defines a string to contain the selection clause
String mSelectionClause = null;

// Initializes an array to contain selection arguments
String[] mSelectionArgs = {""};

下面的代码显示如何使用ContentResolver.query(),例子使用。提供者客户端查询是类似SQL查询,它返回一个列的集合,选择标准集,排序命令。 查询需要返回的列的集合被疑个Projection(变量mProjection)调用。知道获取行的表达式是分到一个选择语句和一个选择参数里的。选择语句是一个逻辑、布尔值、列名、值(变量mSelection)复合表达式。如果你指定替换参数?来代替一个值,查询函数从选择参数数列里获取值(变量mSelectionArgs)。 下面的片段,如果用户不输入一个单词,设置为null,查询返回提供者里所有的单词。如果用户输入一个单词,设置 UserDictionary.Words.Word + " = ?" 并且选择参数数组的第一个元素设置为用户输入的。

/*
 * This defines a one-element String array to contain the selection argument.
 */
String[] mSelectionArgs = {""};

// Gets a word from the UI
mSearchString = mSearchWord.getText().toString();

// Remember to insert code here to check for invalid or malicious input.

// If the word is the empty string, gets everything
if (TextUtils.isEmpty(mSearchString)) {
    // Setting the selection clause to null will return all words
    mSelectionClause = null;
    mSelectionArgs[0] = "";

} else {
    // Constructs a selection clause that matches the word that the user entered.
    mSelectionClause = UserDictionary.Words.WORD + " = ?";

    // Moves the user's input string to the selection arguments.
    mSelectionArgs[0] = mSearchString;

}

// Does a query against the table and returns a Cursor object
mCursor = getContentResolver().query(
    UserDictionary.Words.CONTENT_URI,  // The content URI of the words table
    mProjection,                       // The columns to return for each row
    mSelectionClause                   // Either null, or the word the user entered
    mSelectionArgs,                    // Either empty, or the string the user entered
    mSortOrder);                       // The sort order for the returned rows

// Some providers return null if an error occurs, others throw an exception
if (null == mCursor) {
    /*
     * Insert code here to handle the error. Be sure not to use the cursor! You may want to
     * call android.util.Log.e() to log this error.
     *
     */
// If the Cursor is empty, the provider found no matches
} else if (mCursor.getCount() < 1) {

    /*
     * Insert code here to notify the user that the search was unsuccessful. This isn't necessarily
     * an error. You may want to offer the user the option to insert a new row, or re-type the
     * search term.
     */

} else {
    // Insert code here to do something with the results

}

这个查询和SQL语句类似。

SELECT _ID, word, frequency, locale FROM words WHERE word = <userinput> ORDER BY word ASC;

这个SQL语句,实际的列名用于替代合约类的常量。

防止恶意插入

如果通过提供者管理的数据在SQL数据库里,包括外部不可信的数据进入原始的SQL语句会导致SQL注入。 考虑这种情况:

Consider this selection clause:

// Constructs a selection clause by concatenating the user's input to the column name
String mSelectionClause =  "var = " + mUserInput;

如果你这样做,你就允许用户串联恶意的SQL到你的SQL语句里。例如:用户可以为mUserInput输入“nothing; DROP TABLE *;”结果选择语句var = nothing; DROP TABLE *; 当选择语句被认为是一个SQL语句,就会引起提供者擦除SQLite数据库里所有的表(除非提供者设置捕获SQL injection 的语句)。 为了解决这个问题,使用一个有?作为可替换参数的选择语句和一个分开的选择参数数组。这样做,用户输入一个**到查询而不是作为SQL语句的一部分被中断。因为,它没有被视为是SQL,用户输入不可以注入恶意的SQL。使用如下的选择语句,而不是使用级联用户输入的语句。

// Constructs a selection clause with a replaceable parameter
String mSelectionClause =  "var = ?";

设置选择参数数组如下:

// Defines an array to contain the selection arguments
String[] selectionArgs = {""};

设置一个选择参数数组就如下:

// Sets the selection argument to the user's input
selectionArgs[0] = mUserInput;

?是可替换参数,选择查询数组首选的方法是指定一节,甚至提供者不基于一个SQL数据库。

显示查询结果

ContentResolver.query() 函数总是返回一个Cursor ,它包含查询的Projection指定的列,行是匹配查询选择标准的。一个Cursor 对象提供随机的读行权限还有它包含的列。使用Cursor 函数,你可以在结果里遍历行,决定每列的数据类型,获取列外的数据,使用结果的其它属性。一些Cursor 实现当提供者数据变化时自动的更新。或当Cursor 改变是触发观察对象的方法,或者两者都有。 注意:一个提供者可能限制访问列基于对象的属性来生成查询。例如:合约提供者访问一些列来同步适配器,因此它不返回到Activity或一个服务。 如果没有匹配选择标准的行,提供者返回一个Cursor对象,它的Cursor.getCount() 是0(一个空的cursor)。 如果内部发生一个错误,查询结果依靠指定的提供者。可能选择返回null,或者抛出异常。 如果Cursor是行的列表,显示Cursor内容的方法是把它通过SimpleCursorAdapter和一个ListView连接。 下面的代码是前面代码的继续。它创建一个对象SimpleCursorAdapter包含通过查询获取的Cursor,设置这个对象作为ListView的适配器。

// Defines a list of columns to retrieve from the Cursor and load into an output row
String[] mWordListColumns =
{
    UserDictionary.Words.WORD,   // Contract class constant containing the word column name
    UserDictionary.Words.LOCALE  // Contract class constant containing the locale column name
};

// Defines a list of View IDs that will receive the Cursor columns for each row
int[] mWordListItems = { R.id.dictWord, R.id.locale};

// Creates a new SimpleCursorAdapter
mCursorAdapter = new SimpleCursorAdapter(
    getApplicationContext(),               // The application's Context object
    R.layout.wordlistrow,                  // A layout in XML for one row in the ListView
    mCursor,                               // The result from the query
    mWordListColumns,                      // A string array of column names in the cursor
    mWordListItems,                        // An integer array of view IDs in the row layout
    0);                                    // Flags (usually none are needed)

// Sets the adapter for the ListView
mWordList.setAdapter(mCursorAdapter);

注意:备份一个ListView和一个Cursor,curso需要包含一个列名叫_ID。因为,查询显示之前返回的单词表的_ID列,甚至ListView不显示。这也解释了为什么每个表的列有一个_ID。

从查询结果里获取数据

不是简单显示你查询的结果,你可使用它们做别的任务。例如:你可以从user dictionary获取拼写并且在其它提供者里查询。可以在Cursor里遍历行:


// Determine the column index of the column named "word"
int index = mCursor.getColumnIndex(UserDictionary.Words.WORD);

/*
 * Only executes if the cursor is valid. The User Dictionary Provider returns null if
 * an internal error occurs. Other providers may throw an Exception instead of returning null.
 */

if (mCursor != null) {
    /*
     * Moves to the next row in the cursor. Before the first movement in the cursor, the
     * "row pointer" is -1, and if you try to retrieve data at that position you will get an
     * exception.
     */
    while (mCursor.moveToNext()) {

        // Gets the value from the column.
        newWord = mCursor.getString(index);

        // Insert code here to process the retrieved word.

        ...

        // end of while loop
    }
} else {

    // Insert code here to report an error if the cursor is null or the provider threw an exception.
}

Cursor实现一系列的“get”函数,为从对象获取不同类型的数据。例如:前面的代码使用函数getString()。getType()函数返回一个表示数据类型的值。

内容提供者权限

一个内容提供者可以指明其它需要访问数据的程序的权限。权限确保用户知道程序想要访问的数据。基于提供者的请求,程序请求权限以此访问提供者。当安装应用程序的时候,终端用户看到请求权限。 如果提供者的程序不指明权限,其它应用程序不可以访问数据。然而,提供者程序的组件拥有所有的读写权,不管有没有指定。 之前提到的,User Dictionary Provider请求android.permission.READ_USER_DICTIONARY 权限来获取数据。提供者为插入、更新、删除数据分开android.permission.WRITE_USER_DICTIONARY 权限。 获取访问提供者访问的权限,一个应用程序需要在manifest文件里有元素 。当Android包管理器安装应用程序,用户必须同意程序请求的所有权限。如果用户允许,包管理器继续安装;如果用户不允许,包管理器终止安装。 下面的 元素请求User Dictionary Provider读访问。

    <uses-permission android:name="android.permission.READ_USER_DICTIONARY">

关于提供者权限的影响,更多的信息在Security and Permissions 。

插入,更新,删除数据

一些你从提供者获取数据的方法,你使用提供者客户端和提供者的ContentProvider 来修改数据。你调用函数**。提供者和提供者客户端自动处理安全和进程间通信。 插入数据 向一个提供者插入数据,你可以调用。这个函数插入一个新的行,返回该行的内容URI。以下代码显示如何向一个User Dictionary Provider插入新行: 这里有代码 代码段不需要添加-ID列,因为列的维护是自动的。如果提供者给每个添加的行分配一个独一无二的-ID值。提供者使用这个值作为表的主键。 newUri 返回的内容URI指明新增加的行,使用如下格式:

content://user_dictionary/words/<id_value>

The <id_value> is the contents of _ID for the new row. Most providers can detect this form of content URI automatically and then perform the requested operation on that particular row.

To get the value of _ID from the returned Uri, call ContentUris.parseId().

Updating data

To update a row, you use a ContentValues object with the updated values just as you do with an insertion, and selection criteria just as you do with a query. The client method you use is ContentResolver.update(). You only need to add values to the ContentValues object for columns you're updating. If you want to clear the contents of a column, set the value to null.

The following snippet changes all the rows whose locale has the language "en" to a have a locale of null. The return value is the number of rows that were updated:

// Defines an object to contain the updated values
ContentValues mUpdateValues = new ContentValues();

// Defines selection criteria for the rows you want to update
String mSelectionClause = UserDictionary.Words.LOCALE +  "LIKE ?";
String[] mSelectionArgs = {"en_%"};

// Defines a variable to contain the number of updated rows
int mRowsUpdated = 0;

...

/*
 * Sets the updated value and updates the selected words.
 */
mUpdateValues.putNull(UserDictionary.Words.LOCALE);

mRowsUpdated = getContentResolver().update(
    UserDictionary.Words.CONTENT_URI,   // the user dictionary content URI
    mUpdateValues                       // the columns to update
    mSelectionClause                    // the column to select on
    mSelectionArgs                      // the value to compare to
);

You should also sanitize user input when you call ContentResolver.update(). To learn more about this, read the section Protecting against malicious input.

Deleting data

Deleting rows is similar to retrieving row data: you specify selection criteria for the rows you want to delete and the client method returns the number of deleted rows. The following snippet deletes rows whose appid matches "user". The method returns the number of deleted rows.


// Defines selection criteria for the rows you want to delete
String mSelectionClause = UserDictionary.Words.APP_ID + " LIKE ?";
String[] mSelectionArgs = {"user"};

// Defines a variable to contain the number of rows deleted
int mRowsDeleted = 0;

...

// Deletes the words that match the selection criteria
mRowsDeleted = getContentResolver().delete(
    UserDictionary.Words.CONTENT_URI,   // the user dictionary content URI
    mSelectionClause                    // the column to select on
    mSelectionArgs                      // the value to compare to
);

You should also sanitize user input when you call ContentResolver.delete(). To learn more about this, read the section Protecting against malicious input.

Provider Data Types

Content providers can offer many different data types. The User Dictionary Provider offers only text, but providers can also offer the following formats:

  • integer
  • long integer (long)
  • floating point
  • long floating point (double)

Another data type that providers often use is Binary Large OBject (BLOB) implemented as a 64KB byte array. You can see the available data types by looking at the Cursor class "get" methods.

The data type for each column in a provider is usually listed in its documentation. The data types for the User Dictionary Provider are listed in the reference documentation for its contract class UserDictionary.Words (contract classes are described in the section Contract Classes). You can also determine the data type by calling Cursor.getType().

Providers also maintain MIME data type information for each content URI they define. You can use the MIME type information to find out if your application can handle data that the provider offers, or to choose a type of handling based on the MIME type. You usually need the MIME type when you are working with a provider that contains complex data structures or files. For example, the ContactsContract.Data table in the Contacts Provider uses MIME types to label the type of contact data stored in each row. To get the MIME type corresponding to a content URI, call ContentResolver.getType().

The section MIME Type Reference describes the syntax of both standard and custom MIME types.

Alternative Forms of Provider Access

Three alternative forms of provider access are important in application development:

  • Batch access: You can create a batch of access calls with methods in the ContentProviderOperation class, and then apply them with ContentResolver.applyBatch().
  • Asynchronous queries: You should do queries in a separate thread. One way to do this is to use a CursorLoader object. The examples in the Loaders guide demonstrate how to do this.
  • Data access via intents: Although you can't send an intent directly to a provider, you can send an intent to the provider's application, which is usually the best-equipped to modify the provider's data.

Batch access and modification via intents are described in the following sections.

Batch access

Batch access to a provider is useful for inserting a large number of rows, or for inserting rows in multiple tables in the same method call, or in general for performing a set of operations across process boundaries as a transaction (an atomic operation).

To access a provider in "batch mode", you create an array of ContentProviderOperation objects and then dispatch them to a content provider with ContentResolver.applyBatch(). You pass the content provider's authority to this method, rather than a particular content URI, which allows each ContentProviderOperation object in the array to work against a different table. A call to ContentResolver.applyBatch() returns an array of results.

The description of the ContactsContract.RawContacts contract class includes a code snippet that demonstrates batch insertion. The Contact Manager sample application contains an example of batch access in its ContactAdder.java source file.

Data access via intents

Intents can provide indirect access to a content provider. You allow the user to access data in a provider even if your application doesn't have access permissions, either by getting a result intent back from an application that has permissions, or by activating an application that has permissions and letting the user do work in it.

Getting access with temporary permissions

You can access data in a content provider, even if you don't have the proper access permissions, by sending an intent to an application that does have the permissions and receiving back a result intent containing "URI" permissions. These are permissions for a specific content URI that last until the activity that receives them is finished. The application that has permanent permissions grants temporary permissions by setting a flag in the result intent:

Note: These flags don't give general read or write access to the provider whose authority is contained in the content URI. The access is only for the URI itself.

A provider defines URI permissions for content URIs in its manifest, using the android:grantUriPermission attribute of the <provider> element, as well as the <grant-uri-permission> child element of the <provider> element. The URI permissions mechanism is explained in more detail in the Security and Permissions guide, in the section "URI Permissions".

For example, you can retrieve data for a contact in the Contacts Provider, even if you don't have the READ_CONTACTS permission. You might want to do this in an application that sends e-greetings to a contact on his or her birthday. Instead of requesting READ_CONTACTS, which gives you access to all of the user's contacts and all of their information, you prefer to let the user control which contacts are used by your application. To do this, you use the following process:

  1. Your application sends an intent containing the action ACTION_PICK and the "contacts" MIME type CONTENT_ITEM_TYPE, using the method startActivityForResult().
  2. Because this intent matches the intent filter for the People app's "selection" activity, the activity will come to the foreground.
  3. In the selection activity, the user selects a contact to update. When this happens, the selection activity calls setResult(resultcode, intent) to set up a intent to give back to your application. The intent contains the content URI of the contact the user selected, and the "extras" flags FLAG_GRANT_READ_URI_PERMISSION. These flags grant URI permission to your app to read data for the contact pointed to by the content URI. The selection activity then calls finish() to return control to your application.
  4. Your activity returns to the foreground, and the system calls your activity's onActivityResult() method. This method receives the result intent created by the selection activity in the People app.
  5. With the content URI from the result intent, you can read the contact's data from the Contacts Provider, even though you didn't request permanent read access permission to the provider in your manifest. You can then get the contact's birthday information or his or her email address and then send the e-greeting.

Using another application

A simple way to allow the user to modify data to which you don't have access permissions is to activate an application that has permissions and let the user do the work there.

For example, the Calendar application accepts an ACTION_INSERT intent, which allows you to activate the application's insert UI. You can pass "extras" data in this intent, which the application uses to pre-populate the UI. Because recurring events have a complex syntax, the preferred way of inserting events into the Calendar Provider is to activate the Calendar app with an ACTION_INSERT and then let the user insert the event there.

Contract Classes

A contract class defines constants that help applications work with the content URIs, column names, intent actions, and other features of a content provider. Contract classes are not included automatically with a provider; the provider's developer has to define them and then make them available to other developers. Many of the providers included with the Android platform have corresponding contract classes in the package android.provider.

For example, the User Dictionary Provider has a contract class UserDictionary containing content URI and column name constants. The content URI for the "words" table is defined in the constant UserDictionary.Words.CONTENT_URI. The UserDictionary.Words class also contains column name constants, which are used in the example snippets in this guide. For example, a query projection can be defined as:

String[] mProjection =
{
    UserDictionary.Words._ID,
    UserDictionary.Words.WORD,
    UserDictionary.Words.LOCALE
};

Another contract class is ContactsContract for the Contacts Provider. The reference documentation for this class includes example code snippets. One of its subclasses, ContactsContract.Intents.Insert, is a contract class that contains constants for intents and intent data.

MIME Type Reference

Content providers can return standard MIME media types, or custom MIME type strings, or both.

MIME types have the format

type/subtype

For example, the well-known MIME type text/html has the text type and the html subtype. If the provider returns this type for a URI, it means that a query using that URI will return text containing HTML tags.

Custom MIME type strings, also called "vendor-specific" MIME types, have more complex type and subtype values. The type value is always

vnd.android.cursor.dir

for multiple rows, or

vnd.android.cursor.item

for a single row.

The subtype is provider-specific. The Android built-in providers usually have a simple subtype. For example, the when the Contacts application creates a row for a telephone number, it sets the following MIME type in the row:

vnd.android.cursor.item/phone_v2

Notice that the subtype value is simply phone_v2.

Other provider developers may create their own pattern of subtypes based on the provider's authority and table names. For example, consider a provider that contains train timetables. The provider's authority is com.example.trains, and it contains the tables Line1, Line2, and Line3. In response to the content URI

content://com.example.trains/Line1

for table Line1, the provider returns the MIME type

vnd.android.cursor.dir/vnd.example.line1

In response to the content URI

content://com.example.trains/Line2/5

for row 5 in table Line2, the provider returns the MIME type

vnd.android.cursor.item/vnd.example.line2

Most content providers define contract class constants for the MIME types they use. The Contacts Provider contract class ContactsContract.RawContacts, for example, defines the constant CONTENT_ITEM_TYPE for the MIME type of a single raw contact row.

Content URIs for single rows are described in the section Content URIs.