解析在Android 11MB的大JSON

Question

我在解析与GSON和杰克逊的Android大约11MB的JSON大问题。问题是发生内存不足错误异常并且堆大小不足以完成此过程。 这是我的纸模型类

public class Paper { 

public int primaryKey; 

public String title; 

public int entry; 

public Boolean favourite; 

public String comment; 

public int opt; 

public int score; 
} 

这是我的响应模型类

public class Response { 

public List<Paper> papers; 

} 

这是我的JSON字符串

{"Paper":[[{"abstract":"Not Available","title":"A Fully Intraocular 0.0169mm<sup>2<\/sup>/pixel 512-Channel Self-Calibrating Epiretinal Prosthesis in 65nm CMOS","primaryKey":3,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"A Scalable 2.9mW 1Mb/s eTextiles Body Area Network Transceiver with Remotely Powered Sensors and Bi-Directional Data Communication","primaryKey":14,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"A 0.18µm CMOS SoC for a 100m-Range 10fps 200×96-Pixel Time-of-Flight Depth Sensor","primaryKey":20,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"A 12b 1.6GS/s 40mW DAC in 40nm CMOS with >70dB SFDR over Entire Nyquist Bandwidth","primaryKey":26,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"All-Digital Hybrid Temperature Sensor Network for Dense Thermal Monitoring","primaryKey":49,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"32Gb/s Data-Interpolator Receiver with 2-Tap DFE in 28nm CMOS","primaryKey":51,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"A 93% Efficiency Reconfigurable Switched-Capacitor DC-DC Converter Using On-Chip Ferroelectric Capacitors","primaryKey":60,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1},{"abstract":"Not Available","title":"A 45nm CMOS Near-Field Communication Radio with 0.15A/m RX Sensitivity and 4mA Current Consumption in Card Emulation Mode","primaryKey":61,"entry":9,"score":-1,"comment":null,"favourite":false,"opt":1}]]} 

我不知道我做了一个错误。我对文件的含义不了了之。

Answer 1

在完整输入读取之前，使用流解析器并尝试在读取内容时处理内容。这样可以避免在内存中保存完整的结构。

例如，如果您的输入JSON是一个巨大的数组，您可以按元素处理输入元素。

Answer 2

你也可以试试Genson，检查数据绑定API（new Genson().deserialize(json, ToClass.class)）或直接使用流api。

这取决于你想要做什么。 如果你需要内存中的所有数据，那么没有太多的选择，你需要增加它。 如果你可以在读取它时处理内容，并且不需要将它永久地存储在内存中，那么对于流式API（非常高性能且使用极少的内存）它就可以正常工作。

编辑 解决方案

1. 你想要让你拥有更小的工作要做使用数据绑定。首先看看这个项目https://github.com/joelittlejohn/jsonschema2pojo它允许你基于json例子生成java bean类。如果您正在执行多个ser/deser，那么当您有生成的类时只需执行new Genson().deserialize(json, MyGeneratedClass.class);，则应该重用Genson实例以获得更好的性能。

2. 如果您不想使用1），如果结构在运行时更改（这会阻止生成类），或者仍然存在内存问题，请使用Gensons streaming api。 它是高效，高性能和易于使用的内存。