Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

How to parse items from a large JSON stream in JavaScript?

So I have downloaded the Wikidata JSON dump and it's about 90GB, too large to load into memory. It consists of a simple JSON structure like this:

[
  item,
  item,
  item,
  ...
]

Each "item" looks something like this:

{
  "type": "item",
  "id": "Q23",
  "labels": {
    "<lang>": obj
  },
  "descriptions": {
    "<lang>": {
      "language": "<lang>",
      "value": "<string>"
    },
  },
  "aliases": {
    "<key>": [
      obj,
      obj,
    ],
  },
  "claims": {
    "<keyID>": [
       {
        "mainsnak": {
          "snaktype": "value",
          "property": "<keyID>",
          "datavalue": {
            "value": {
              "entity-type": "<type>",
              "numeric-id": <num>,
              "id": "<id>"
            },
            "type": "wikibase-entityid"
          },
          "datatype": "wikibase-item"
        },
        "type": "statement",
        "id": "<anotherId>",
        "rank": "preferred",
        "references": [
          {
            "hash": "<hash>",
            "snaks": {
              "<keyIDX>": [
                {
                  "snaktype": "value",
                  "property": "P854",
                  "datavalue": obj,
                  "datatype": "url"
                }
              ]
            },
            "snaks-order": [
              "<propID>"
            ]
          }
        ]
      }
    ]
  },
  "sitelinks": {
    "<lang>wiki": {
      "site": "<lang>wiki",
      "title": "<string>",
      "badges": []
    }
  }
}

The JSON stream is configured like this:

const fs   = require('fs')
const zlib = require('zlib')
const { parser } = require('stream-json')

let stream = fs.createReadStream('./wikidata/latest-all.json.gz')
stream
  .pipe(zlib.createGunzip())
  .pipe(parser())
  .on('data', buildItem)

function buildItem(data) {
  switch (data.name) {
    case `startArray`:
      break
    case `startObject`:
      break
    case `startKey`:
      break
    case `stringChunk`:
      break
    case `endKey`:
      break
    case `keyValue`:
      break
    case `startString`:
      break
    case `endString`:
      break
    case `stringValue`:
      break
    case `endObject`:
      break
    case `endArray`:
      break
  }
}

Notice the buildItem has the key information, it shows that the JSON stream emits objects like this (these are the logs):

{ name: 'startArray' }
{ name: 'startObject' }
{ name: 'startKey' }
{ name: 'startString' }
{ name: 'stringValue', value: 'type' }
{ name: 'endString' }
...

How do you parse this into item objects like the above? Parsing this linear stream into a tree is very difficult to comprehend.

A sample of output from the JSON stream is here, which you could use to test a parser if it helps.

like image 437
Lance Avatar asked Oct 28 '20 20:10

Lance


People also ask

How do you parse JSON in JavaScript?

Use the JavaScript function JSON. parse() to convert text into a JavaScript object: const obj = JSON. parse('{"name":"John", "age":30, "city":"New York"}');

How do I process a large JSON file?

There are some excellent libraries for parsing large JSON files with minimal resources. One is the popular GSON library. It gets at the same effect of parsing the file as both stream and object. It handles each record as it passes, then discards the stream, keeping memory usage low.

What are the methods to parse JSON?

parse() The JSON. parse() method parses a JSON string, constructing the JavaScript value or object described by the string. An optional reviver function can be provided to perform a transformation on the resulting object before it is returned.

How to parse large JSON files into Java objects?

Guide on How to Parse Large Json files into Java Objects. Use Gson Streaming to Read very large Json file and covert it into objects. Gson is a very popular API for parsing JSON strings into Objects. The parse method, provided by Gson is good for reading the entire JSON string and parse it into Java Objects in one go.

How to stream a JSON file in Gson?

In order to Stream a Json file Gson provides JsonReader class. Using this class, we can navigate through JSON objects and nested objects and iteratively convert them into java Objects. Next is the example, where we Read the large JSON file, iterate through its contents, and parse them into objects.

How to parse a JSON array instead of a JavaScript Object?

When using the JSON.parse() on a JSON derived from an array, the method will return a JavaScript array, instead of a JavaScript object. Example const text = '["Ford", "BMW", "Audi", "Fiat"]';

What is jsonstream in Node JS?

JSONStream presents .parse () and .stringify () methods that provide Node.js Transform streams (aka "through" streams) through which large operations can be broken down into smaller, resource-friendly operations.


2 Answers

Using Built-In Functions (StreamArray)

stream-json already has built-in functions that converts streams into objects (in this case, you're looking for StreamArray). You may want to use the built-in functions, as they've been coded with performance in mind.

To use it, it'd look something like:

const fs   = require('fs')
const zlib = require('zlib')
const { parser } = require('stream-json')
const { streamArray } = require('stream-json/streamers/StreamArray')

let stream = fs.createReadStream('./wikidata/latest-all.json.gz')
stream
  .pipe(zlib.createGunzip())
  .pipe(parser())
  .pipe(streamArray())
  .on('data', d => processData(d.value))

function processData(data) {
  console.log(data)
}

I recommend taking a look at the wiki at https://github.com/uhop/stream-json/wiki for more information, as it has additional functions, particularly for filtering or transformation, which will probably be useful for you, especially if speed is a concern.

like image 188
Steve Avatar answered Oct 13 '22 00:10

Steve


If I understood you correctly you want something like this. I used an ObjectBuilder class that combines all methods to build one JSON object. It uses parentStack to keep track of all objects and arrays. When the object/array is started with startObject/startArray a new JSON object/array is pushed onto the stack. Once this object/array is completed it is popped off of the stack. The last object that is popped off of the stack is the whole item object and can be processed further (in the example below I just print it out).

The current object or array that is currently being constructed is always on top of the stack.

I had to use a subset of the sample you provided because it did not contain a matching number of startObject and endObject items, which resulted into an invalid JSON. I included this subset below the code.

Hopefully, this is what you were looking for :)

(Note, I only wrapped buildItem() function in runSample() function so that I can include the sample JSON at the bottom to make it look neater in this online editor. You can move buildItem() function outside.)

class ObjectBuilder {
    constructor() {
        this.finalObject = undefined;
        this.parentStack = [];

        this.currentKey = undefined;
    }

    hasFinished() {
        return this.finalObject !== undefined;
    }

    getFinalObject() {
        return this.finalObject;
    }

    currentObject() {
        return this.parentStack[this.parentStack.length - 1];
    }

    addValue(val) {
        if (Array.isArray(this.currentObject())) {
            this.currentObject().push(val);
        }
        else {
            this.currentObject()[this.currentKey] = val;
            this.currentKey = undefined;
        }
    }

    processData(data) {
        switch (data.name) {
            case `startKey`:
            case `endKey`:
            case `startString`:
            case `endString`:
            case `stringChunk`:
                // ignore, always followed by [something]Value
                break;

            case `keyValue`:
                this.currentObject()[data.value] = undefined;
                this.currentKey = data.value;
                break;
            case `numberValue`:
                this.addValue(Number(data.value))
                break
            case `stringValue`:
                this.addValue(data.value);
                break;

            case `startObject`:
                let newObject = {};
                if (this.parentStack.length === 0) {
                    // do nothing else, initialises first parent
                }
                else if (Array.isArray(this.currentObject())) {
                    this.currentObject().push(newObject);
                }
                else {
                    this.currentObject()[this.currentKey] = newObject;
                }
                this.parentStack.push(newObject);
                this.currentKey = undefined;
                break;

            case `endObject`:
                let parent = this.parentStack.pop();
                if (this.parentStack.length === 0) {
                    this.finalObject = parent;
                }
                break;

            case `startArray`:
                let newArray = [];
                if (Array.isArray(this.currentObject())) {
                    this.currentObject().push(newArray);
                }
                else {
                    this.currentObject()[this.currentKey] = newArray;
                }
                this.parentStack.push(newArray);
                this.currentKey = undefined;

                break;
            case `endArray`:
                this.parentStack.pop();
                this.currentKey = undefined;
                break;
        }
    }
}

function runSample(streamData) {

    let currentlyProcessing = undefined;
    function buildItem(data) {
        if (currentlyProcessing === undefined && data.name === "endArray") {
            return; // stream ended
        }

        if (currentlyProcessing === undefined) {
            currentlyProcessing = new ObjectBuilder();
        }
        currentlyProcessing.processData(data);

        if (currentlyProcessing.hasFinished()) {
            // Finished building project; do something with it
            let niceOutput = JSON.stringify(currentlyProcessing.getFinalObject(), null, 4);
            console.log(niceOutput);
            currentlyProcessing = undefined;
        }
    }


    // simulate reading stream
    for (let i = 0; i < streamData.length; ++i) {
        if (i === 0) {
            // Skip first chunk as it starts the array of items
            continue;
        }
        buildItem(streamData[i]);
    }
}

const streamData = [{"name": "startArray"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "type"},{"name": "endKey"},{"name": "keyValue","value": "type"},{"name": "startString"},{"name": "stringChunk","value": "item"},{"name": "endString"},{"name": "stringValue","value": "item"},{"name": "startKey"},{"name": "stringChunk","value": "id"},{"name": "endKey"},{"name": "keyValue","value": "id"},{"name": "startString"},{"name": "stringChunk","value": "Q31"},{"name": "endString"},{"name": "stringValue","value": "Q31"},{"name": "startKey"},{"name": "stringChunk","value": "labels"},{"name": "endKey"},{"name": "keyValue","value": "labels"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "el"},{"name": "endKey"},{"name": "keyValue","value": "el"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "el"},{"name": "endString"},{"name": "stringValue","value": "el"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "Β"},{"name": "stringChunk","value": "έ"},{"name": "stringChunk","value": "λ"},{"name": "stringChunk","value": "γ"},{"name": "stringChunk","value": "ι"},{"name": "stringChunk","value": "ο"},{"name": "endString"},{"name": "stringValue","value": "Βέλγιο"},{"name": "endObject"},{"name": "startKey"},{"name": "stringChunk","value": "ay"},{"name": "endKey"},{"name": "keyValue","value": "ay"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "ay"},{"name": "endString"},{"name": "stringValue","value": "ay"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "Bilkiya"},{"name": "endString"},{"name": "stringValue","value": "Bilkiya"},{"name": "endObject"},{"name": "startKey"},{"name": "stringChunk","value": "pnb"},{"name": "endKey"},{"name": "keyValue","value": "pnb"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "pnb"},{"name": "endString"},{"name": "stringValue","value": "pnb"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "ب"},{"name": "stringChunk","value": "ی"},{"name": "stringChunk","value": "ل"},{"name": "stringChunk","value": "ج"},{"name": "stringChunk","value": "ی"},{"name": "stringChunk","value": "م"},{"name": "endString"},{"name": "stringValue","value": "بیلجیم"},{"name": "endObject"},{"name": "endObject"},{"name": "endObject"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "type"},{"name": "endKey"},{"name": "keyValue","value": "type"},{"name": "startString"},{"name": "stringChunk","value": "item"},{"name": "endString"},{"name": "stringValue","value": "item"},{"name": "startKey"},{"name": "stringChunk","value": "id"},{"name": "endKey"},{"name": "keyValue","value": "id"},{"name": "startString"},{"name": "stringChunk","value": "Q31"},{"name": "endString"},{"name": "stringValue","value": "Q31"},{"name": "startKey"},{"name": "stringChunk","value": "labels"},{"name": "endKey"},{"name": "keyValue","value": "labels"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "el"},{"name": "endKey"},{"name": "keyValue","value": "el"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "el"},{"name": "endString"},{"name": "stringValue","value": "el"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "Β"},{"name": "stringChunk","value": "έ"},{"name": "stringChunk","value": "λ"},{"name": "stringChunk","value": "γ"},{"name": "stringChunk","value": "ι"},{"name": "stringChunk","value": "ο"},{"name": "endString"},{"name": "stringValue","value": "Βέλγιο"},{"name": "endObject"},{"name": "startKey"},{"name": "stringChunk","value": "ay"},{"name": "endKey"},{"name": "keyValue","value": "ay"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "ay"},{"name": "endString"},{"name": "stringValue","value": "ay"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "Bilkiya"},{"name": "endString"},{"name": "stringValue","value": "Bilkiya"},{"name": "endObject"},{"name": "startKey"},{"name": "stringChunk","value": "pnb"},{"name": "endKey"},{"name": "keyValue","value": "pnb"},{"name": "startObject"},{"name": "startKey"},{"name": "stringChunk","value": "language"},{"name": "endKey"},{"name": "keyValue","value": "language"},{"name": "startString"},{"name": "stringChunk","value": "pnb"},{"name": "endString"},{"name": "stringValue","value": "pnb"},{"name": "startKey"},{"name": "stringChunk","value": "value"},{"name": "endKey"},{"name": "keyValue","value": "value"},{"name": "startString"},{"name": "stringChunk","value": "ب"},{"name": "stringChunk","value": "ی"},{"name": "stringChunk","value": "ل"},{"name": "stringChunk","value": "ج"},{"name": "stringChunk","value": "ی"},{"name": "stringChunk","value": "م"},{"name": "endString"},{"name": "stringValue","value": "بیلجیم"},{"name": "endObject"},{"name": "startKey"},{"name": "stringChunk","value": "nestedArray"},{"name": "endKey"},{"name": "keyValue","value": "nestedArray"},{"name": "startArray"},{"name": "stringValue","value": "a"},{"name": "stringValue","value": "b"},{"name": "startArray"},{"name": "stringValue","value": "c"},{"name": "startObject"},{"name": "keyValue","value": "another object"},{"name": "stringValue","value": "d"},{"name": "endObject"},{"name": "stringValue","value": "e"},{"name": "endArray"},{"name": "stringValue","value": "b"},{"name": "endArray"},{"name": "endObject"},{"name": "endObject"},{"name": "endArray"}];

runSample(streamData);
like image 35
Luka Kralj Avatar answered Oct 13 '22 00:10

Luka Kralj