There's a different perspective on this issue that I'd like to share, despite its age.

If you're working with the mongo shell, there's a way to handle your JSON data effectively.

Simply convert your JSON to valid JavaScript by adding var myData= before it and then use the load() command in your script. This will allow you to access your data through the myData object.

data.js

var myData=
[
        {
                "letter" : "A"
        },
        {
                "letter" : "B"
        },
        {
                "letter" : "C"
        }
]

read.js

#!/usr/bin/mongo --quiet

// read data

load('data.js');

// display letters

for (i in myData) {
  var doc  = myData[i];
  print(doc.letter);
}

To write JSON data, simplify the process by loading the result into a single object. Start by initializing the object with var result={} and then use printjson() to output the data. Redirect the output to a file using standard methods.

write.js

#!/usr/bin/mongo --quiet

var result=[];

// read data from collection etc...
for (var i=65; i<91; i++) {
  result.push({letter: String.fromCharCode(i)});
}

// output
print("var myData=");
printjson(result);

The shebang lines (#!) are designed for Unix systems but can also be used on Windows with tools like Cygwin.

To retrieve the contents of a file as text, you can utilize the undocumented cat() function within the mongo shell:

var text = cat(filename);

If you want to explore other hidden utilities like cat() and writeFile, take a look at this file: shell_utils_extended.cpp

Once you have the file's content, you have the option to modify it or directly input it into JSON.parse for conversion into a JavaScript object:

jsObj = JSON.parse(text);

Remember: while JSON.parse is powerful, it differs from the functionality of the mongoimport tool in terms of JSON parsing capabilities.

mongoimport excels in parsing Mongo's extended JSON in canonical format. (Canonical format files are typically generated by tools like bsondump and mongodump. For further details on JSON formats, refer to MongoDB extended JSON).

Keep in mind that JSON.parse does not support canonical JSON format. While it will successfully read such input and produce a JavaScript object, any additional data type information specific to canonical format JSON will be disregarded.

The mongo shell does not support reading and writing files like a complete programming environment. It is recommended to use mongoimport or create a script in a language that has an official driver for MongoDB. For example, Node.js has syntax similar to the mongo shell, but it operates on an async/event-driven model. Alternatively, Python with PyMongo can be a good choice due to its simplicity and similarity to MongoDB's query language.

Hey there, I wanted to share a useful tip with you! Are you tired of manually importing json files into your mongo db? I used to struggle with that too until I created a simple batch script to automate the process. Interested in checking it out?

https://github.com/aminjellali/batch/blob/master/mongoImporter.bat

@echo off
title = Mongo Data Base importing tool
goto :main

:import_collection
    echo importing %~2
    set file_name=%~2
    set removed_json=%file_name:.json=%
    mongoimport --db %~1 --collection %removed_json% --file %~2
goto :eof

:loop_over_files_in_current_dir
    for /f %%c in ('dir /b *.json') do call :import_collection %~1 %%c
goto :eof

:main
    IF [%1]==[] (
    ECHO FATAL ERROR: Please specify a data base name
    goto :eof
    ) ELSE (
    ECHO @author amin.jellali
    ECHO @email <a href="/cdn-cgi/l/email-protection" class="__cf_email__" data-cfemail="ed8cc387c38c808483c3878881818c8184ad8a808c84...›protected]</a>
    echo starting import...
    call :loop_over_files_in_current_dir %~1
    echo import done...
    echo hope you enjoyed me
    )
goto :eof

After searching extensively, I finally found the solution I was looking for. Previous answers were either ineffective or merely provided temporary fixes.

Fortunately, I discovered the answer.. You can utilize fs.readFileSync to read (json) files. Below is my complete script for reference:

const dbName = process.env.DB_NAME || 'test_db';
const dbUser = process.env.DB_USER || 'test_user';
const dbPassword = process.env.DB_PASSWORD || 'secret';

db.createUser({
  user: dbUser,
  pwd: dbPassword,
  roles: [
    {
      role: 'readWrite',
      db: dbName,
    },
  ],
});

// Create and seed collections:

const collections = [
  'some_collection',
  'another_collection',
];

for (let i = 0; i < collections.length; i++) {
  const collectionName = collections[i];
  db.createCollection(collectionName);

  let jsonFile;
  try {
    jsonFile = fs.readFileSync(`/docker-entrypoint-initdb.d/mongo.${collectionName}.json`, 'utf8');
  }
  catch (err) {
    console.log(`NOTE: mongo.${collectionName}.json not found! Skip seeding of ${collectionName} collection..`)
    continue;
  }

  const json = JSON.parse(jsonFile);

  // Parse the $oid ids that Mongo uses in exports, but cant use in insertMany:
  for (let i = json.length; i--;) {
    json[i]._id = ObjectId(json[i]._id.$oid);
  }

  db[collectionName].insertMany(json);
}