How to export all rows from Datatables using Ajax?

Question

I am using new feature in Datatables: "HTML5 export buttons". I am loading data with Ajax.

https://datatables.net/extensions/buttons/examples/html5/simple.html

The problem is that it only export the page that is currently displayed.

I am exporting like this:

buttons: [
    {
        extend: 'pdfHtml5',
        text: 'PDF',
        exportOptions: {
            "columns": ':visible',
        }
    },
]

How can I export all rows?

Answer 1

You need to tell the AJAX function to get all data, then do the export but cancel the actual draw so that all of that data isn't loading into the DOM. The full data will still exist in memory for the DataTables API though, so you need to refresh it to the way it was before the export.

var oldExportAction = function (self, e, dt, button, config) {
    if (button[0].className.indexOf('buttons-excel') >= 0) {
        if ($.fn.dataTable.ext.buttons.excelHtml5.available(dt, config)) {
            $.fn.dataTable.ext.buttons.excelHtml5.action.call(self, e, dt, button, config);
        }
        else {
            $.fn.dataTable.ext.buttons.excelFlash.action.call(self, e, dt, button, config);
        }
    } else if (button[0].className.indexOf('buttons-print') >= 0) {
        $.fn.dataTable.ext.buttons.print.action(e, dt, button, config);
    }
};

var newExportAction = function (e, dt, button, config) {
    var self = this;
    var oldStart = dt.settings()[0]._iDisplayStart;

    dt.one('preXhr', function (e, s, data) {
        // Just this once, load all data from the server...
        data.start = 0;
        data.length = 2147483647;

        dt.one('preDraw', function (e, settings) {
            // Call the original action function 
            oldExportAction(self, e, dt, button, config);

            dt.one('preXhr', function (e, s, data) {
                // DataTables thinks the first item displayed is index 0, but we're not drawing that.
                // Set the property to what it was before exporting.
                settings._iDisplayStart = oldStart;
                data.start = oldStart;
            });

            // Reload the grid with the original page. Otherwise, API functions like table.cell(this) don't work properly.
            setTimeout(dt.ajax.reload, 0);

            // Prevent rendering of the full data to the DOM
            return false;
        });
    });

    // Requery the server with the new one-time export settings
    dt.ajax.reload();
};

and:

    buttons: [
        {
            extend: 'excel',
            action: newExportAction
        },

Answer 2

This button definition worked for me in a scrolled table (instead of paging):

{
  text: 'PDF',
  action: function(e, dt, button, config) {
    dt.one('preXhr', function(e, s, data) {
      data.length = -1;
    }).one('draw', function(e, settings, json, xhr) {
      var pdfButtonConfig = $.fn.DataTable.ext.buttons.pdfHtml5;
      var addOptions = { exportOptions: { "columns" : ":visible" }};

      $.extend(true,pdfButtonConfig,addOptions);
      pdfButtonConfig.action(e, dt, button, pdfButtonConfig);
    }).draw();
  }
}

It will force the DataTable to request all rows for the current filtering for one request. Then it calls the desired action of the Export button directly. The variable addOptions can be used to alter the standard configuration of the export button.

You might run into problems though if you have a lot of rows as they are all loaded into the DOM.

Answer 3

I know this is an old question, however for anyone struggling with this, here's my solution.

Variables:

var downloading = false,
    downloadTimestamp = null;

---

Download button definition:

buttons: [{
    text: '<span class="glyphicon glyphicon-save-file" aria-hidden="true"></span>',
    titleAttr: 'CSV',
    className: 'downloadCSV',
    action: function(e, dt, node, config) {
        if (downloading === false) { //if download is in progress, do nothing, else
            node.attr('disabled', 'disabled'); //disable download button to prevent multi-click, probably some sort of *busy* indicator is a good idea

            downloading = true; //set downloading status to *true*

            dt.ajax.reload(); //re-run *DataTables* AJAX query with current filter and sort applied
        }
    }
}]

---

Ajax definition:

ajax: {
    url: ajaxURL,
    type: 'POST',
    data: function(data) {
        data.timestamp = new Date().getTime(); //add timestamp to data to be sent, it's going to be useful when retrieving produced file server-side

        downloadTimestamp = data.timestamp; //save timestamp in local variable for use with GET request when retrieving produced file client-side

        if (downloading === true) { //if download button was clicked
            data.download = true; //tell server to prepare data for download
            downloading = data.draw; //set which *DataTable* draw is actually a request to produce file for download
        }

        return { data: JSON.stringify(data) }; //pass data to server for processing
    }
}

---

'preDrawCallback' function:

preDrawCallback: function(settings) {
    if (settings.iDraw === downloading) { //if returned *DataTable* draw matches file request draw value
        downloading = false; //set downloading flag to false

        $('.downloadCSV').removeAttr('disabled'); //enable download button

        window.location.href = ajaxURL + '?' + $.param({ ts: downloadTimestamp }); //navigate to AJAX URL with timestamp as parameter to trigger file download. Or You can have hidden IFrame and set its *src* attribute to the address above.

        return false; //as it is file request, table should not be re-drawn
    }
}

---

Server-side:

if(download == false), then server executes SELECT columns FROM tables WHERE rowNumber BETWEEN firstRow AND lastRow and outputs result for normal display within DataTable.

if(download == true), then server executes SELECT columns FROM tables and stores all rows formatted as CSV file (or any other file format depending on what Your server environment is capable to produce) server-side for later retrieval by GET request.

Following is ASP JScript code that I've used server-side:

    var timestamp = Number(Request.QueryString('ts')), //if it's a GET request, get timestamp
        tableData = {
            draw: data.draw,
            recordsTotal: 100, //some number static or dynamic
            recordsFiltered: 10, //some number static or dynamic
            data: []
        };
        jsonData = String(Request.Form('data')), //if it's POST request, get data sent by *DataTable* AJAX
        data = jsonData === 'undefined' || jsonData.length === 0 ? null : JSON.parse(jsonData); //do some error checking (optional)

    if(!isNaN(timestamp)) { //check timestamp is valid
        var csvTextKey = 'download-' + timestamp, //this is where timestamp value is used (can be any other unique value)
            csvText = Session(csvTextKey); //obtain saved CSV text from local server-side storage

        if(typeof csvText === 'undefined') { //if CSV text does not exist in local storage, return nothing (or throw error is You wish)
            Response.End();
        }

        //if CSV exists:
        Response.ContentType = 'text/csv'; //set response mime type
        Response.AddHeader('Content-Disposition', 'attachment; filename=test.csv'); //add header to tell browser that content should be downloaded as file and not displayed

        Response.Write(csvText); //send all content to browser

        Response.End(); //stop further server-side code execution
    }

    //if timestamp is not valid then we assume this is POST request, hence data should be either prepared for display or stored for file creation

    if(typeof data !== 'object' || data === null) { //do some more clever error checking
        throw 'data is not an object or is null';
    }

        var recordset = data.download === true ? sqlConnection.Execute('SELECT * FROM #FinalTable') : Utilities.prepAndRunSQLQuery('SELECT * FROM #FinalTable WHERE rowId BETWEEN ? AND ?', [data.start, data.start + data.length], //execute SELECT either for display or for file creation
            headerRow = [],
            sqlHeaderRow = [],
            exportData = [];; 

        if(data.download === true) { //create CSV file (or any other file)
            if(!Array.isArray(data.columns)) {
                throw 'data.columns is not an array';
            }

            for(var i = 0, dataColumnsCount = data.columns.length; i < dataColumnsCount; ++i) {
                var dataColumn = data.columns[i], //get columns data object sent by client
                    title = dataColumn.title, //this is custom property set on client-side (not shown in code above)
                    sqlColumnName = typeof dataColumn.data === 'string' ? dataColumn.data : (typeof dataColumn.data.display === 'string' ? dataColumn.data.display : dataColumn.data['_']); //set SQL table column name variable

                if(typeof title === 'string' && typeof sqlColumnName === 'string' && columnNames.indexOf(sqlColumnName) > -1) { //some more error checking
                    headerRow.push(title);
                    sqlHeaderRow.push(sqlColumnName);
                }
            }

            exportData.push('"' + headerRow.join('","') + '"'); //add table header row to in CSV file format
        }

        while(recordset.EOF === false) { //iterate through recordset
            if(data.download === true) { //if download flag is set build string containing CSV content
                var row = [];

                for(var i = 0, count = sqlHeaderRow.length; i < count; ++i) {
                    row.push(String(recordset.Fields(sqlHeaderRow[i]).Value).replace('"', '""'));
                }

                exportData.push('"' + row.join('","') + '"');
            }

            else { //else format data for display
                var row = {};

                for(var i = 1, fieldsCount = recordset.Fields.Count; i < fieldsCount; ++i) {
                    var field = recordset.Fields(i),
                        name = field.Name,
                        value = field.Value;

                    row[name] = value;
                }

                tableData.data.push(row);
            }

            recordset.MoveNext();
        }

if(data.download === true) { //save CSV content in server-side storage
    Session('download-' + data.timestamp) = exportData.join('\r\n'); //this is where timestamp value is used (can be any other unique value)
}

Response.Write(JSON.stringify(tableData)); //return data for display, if download flag is set, tableData.data = []

Answer 4

The answer of Selcuk will work absolutely fine if we can fix the value of "All" beforehand. Suppose the number of rows is stored in variable row_count. Then

var row_count = $("#row_count").val();
var table = $('#example').DataTable({
    serverSide: true,
    ajax: "/your_ajax_url/",
    lengthMenu: [[25, 100, row_count], [25, 100, "All"]],
    pageLength: 25,
    buttons: [
        {
            extend: 'excel',
            text: '<span class="fa fa-file-excel-o"></span> Excel Export',
            exportOptions: {
                modifier: {
                    search: 'applied',
                    order: 'applied'
                }
            }
        }
    ],
    // other options
}); 

Answer 5

I'm using Datatables Version: 1.10.15, and got @kevenpo's answer to work. I had to modify it a bit to handle our server-side parameters, but that was the only stumbling block. I changed his line: data.length = 2147483647; to data.params[2]= -1; because we stored our server-side parameters in a params sub-array. I have not tested it yet with a very large dataset to see what the performance is, but this is a very clever solution.

Answer 6

According to DataTables documentation there is no way to export all rows when you are using server side:

Special note on server-side processing: When using DataTables in server-side processing mode (serverSide) the selector-modifier has very little effect on the rows selected since all processing (ordering, search etc) is performed at the server. Therefore, the only rows that exist on the client-side are those shown in the table at any one time, and the selector can only select those rows which are on the current page.

I worked this around by adding an 'ALL' parameter to the length menu and training end users to display all records before doing a PDF (or XLS) export:

var table = $('#example').DataTable({
    serverSide: true,
    ajax: "/your_ajax_url/",
    lengthMenu: [[25, 100, -1], [25, 100, "All"]],
    pageLength: 25,
    buttons: [
        {
            extend: 'excel',
            text: '<span class="fa fa-file-excel-o"></span> Excel Export',
            exportOptions: {
                modifier: {
                    search: 'applied',
                    order: 'applied'
                }
            }
        }
    ],
    // other options
});