Processing spreadsheet data in Go

Your managers, all through the hierarchy, love circulating spreadsheets via email. (They simply don’t know better.) How to extract and analyze the relevant data from the daily mess? Go can help.

This article is also available as a video on the Applied Go YouTube channel:

It is the shortened version of a lecture in my upcoming minicourse “Workplace Automation With Go”.

Spreadsheet data is everywhere. You can find it in Excel sheets as well as when downloading business data from a website.

Package encoding/csv from the Go standard library can help you processing that data and produce statistics, reports or other kinds of output from it. Here is how.

Let’s assume we work at a stationery distributor. Every evening, we receive a spreadsheet containing the orders of the day for review. The data looks like this:

Date Order ID Order Item Unit Price Quantity
2017-11-17 1 Ball Pen 1.99 50
2017-11-17 2 Notebook 12.99 10
2017-11-17 3 Binder 4.99 25
2017-11-20 4 Pencil 0.99 100
2017-11-20 5 Sketch Block 2.99 40
2017-11-22 6 Ball Pen 1.99 30
2017-11-23 7 Sketch Block 2.99 20
2017-11-24 8 Ball Pen 1.99 60

We’re interested in some information that is not directly contained in the data; especially, we want to know

  • the total price for each order,
  • the total sales volume,
  • and the number of ball pens sold.

As we get a new copy every day, creating formulas within the spreadsheet is not an option. Instead, we decide to write a small tool to do the calculations for us. Also, the tool shall add the result to the table and write a new spreadsheet file.

But before starting to code, our first step is to export the spreadsheet data to CSV. To make things a bit more complicated, we export the data with a semicolon as the column separator.

(The exact steps vary, depending on the spreadsheet software used.)

The raw CSV data looks like this:

Date;Order ID;Order Item;Unit Price;Quantity
2017-11-17;1;Ball Pen;1.99;50
2017-11-18;5;Sketch Block;2.99;40
2017-11-19;6;Ball Pen;1.99;30
2017-11-19;7;Sketch Block;2.99;20
2017-11-19;8;Ball Pen;1.99;60

We can see a header row and data rows, with data separated by semicolons.

Now let’s dive into the Go code to process the data from this spreadsheet and from all spreadsheets that are still to come.

Reading and processing CSV data with Go

Imports and main

package main

We only use packages from the standard library here.

import (

In main(), we sketch out our program flow:

  • Read the CSV file,
  • calculate the desired numbers, and
  • write the results to a new CSV file.
func main() {
	rows := readOrders("orders.csv")
	rows = calculate(rows)
	writeOrders("ordersReport.csv", rows)

Reading CSV files

As the next step, we need to read in the header row, and then the data rows. The result shall be a two-dimensional slice of strings, or a slice of slices of strings.

readOrders takes a filename and returns a two-dimensional list of spreadsheet cells.

func readOrders(name string) [][]string {

	f, err := os.Open(name)

Usually we would return the error to the caller and handle all errors in function main(). However, this is just a small command-line tool, and so we use log.Fatal() instead, in order to write the error message to the terminal and exit immediately.

	if err != nil {
		log.Fatalf("Cannot open '%s': %s\n", name, err.Error())

After this point, the file has been successfully opened, and we want to ensure that it gets closed when no longer needed, so we add a deferred call to f.Close().

	defer f.Close()

To read in the CSV data, we create a new CSV reader that reads from the input file.

The CSV reader is aware of the CSV data format. It separates the input stream into rows and columns, and returns a slice of slices of strings.

	r := csv.NewReader(f)

We can even adjust the reader to recognize a semicolon, rather than a comma, as the column separator.

	r.Comma = ';'

Read the whole file at once. (We don’t expect large files.)

	rows, err := r.ReadAll()

Again, we check for any error,

	if err != nil {
		log.Fatalln("Cannot read CSV data:", err.Error())

and finally we can return the rows.

	return rows

Process the data

Now that the data is read in, we can loop over the rows, and read from or write to each row slice as needed.

This is where we can extract the desired information: The total price for each order, the total sales volume, and the number of ball pens sold.

calculate takes a spreadsheet, extracts and calculates the desired information, and returns the result as a new spreadsheet.

func calculate(rows [][]string) [][]string {

	sum := 0
	nb := 0

To process the data, we loop over the rows, and read from or write to each row slice as needed.

	for i := range rows {

The first row is the header row. Here, we only want to add a new header for the column that holds the total prices.

		if i == 0 {
			rows[0] = append(rows[0], "Total")

From the next row onwards, we calculate the total price, sum up all prices, and count the number of ball pens being ordered.

This is fairly straightforward, as we know the indexes of the item name, the unit price, and the quantity. The only difficulty is that all columns are string values but we need the price and quantity values as numeric values.

We know that column 2 contains the item name.

		item := rows[i][2]

Another obstacle we are facing here is that the prices are floating-point values but for financial calculations, we want to use precise integer calculation only. Luckily, the strings and strconv packages have got us covered.

Column 3 contains the price. Remove the decimal point using strings.Replace(), and turn the value into an integer (representing the value in cents) using strconv.Atoi.

		price, err := strconv.Atoi(strings.Replace(rows[i][3], ".", "", -1))
		if err != nil {
			log.Fatalf("Cannot retrieve price of %s: %s\n", item, err)

Column 4 contains the ordered quantity. Again, we convert the value into an integer.

		qty, err := strconv.Atoi(rows[i][4])
		if err != nil {
			log.Fatalf("Cannot retrieve quantity of %s: %s\n", item, err)

Calculate the total and append it to the current row.

		total := price * qty

We use a helper function to turn the total value (an integer) back into a floating-point value with two decimals, represented as a string (see below).

		rows[i] = append(rows[i], intToFloatString(total))

Update the total sum

		sum += total

and the # of ball pens.

		if item == "Ball Pen" {
			nb += qty

Here we append two new rows. The first one shows the total sum, and the second one shows the number of ball pens ordered.

	rows = append(rows, []string{"", "", "", "Sum", "", intToFloatString(sum)})
	rows = append(rows, []string{"", "", "", "Ball Pens", fmt.Sprint(nb), ""})

Return the new spreadsheet.

	return rows

intToFloatString takes an integer n and calculates the floating point value representing n/100 as a string.

func intToFloatString(n int) string {
	intgr := n / 100
	frac := n - intgr*100
	return fmt.Sprintf("%d.%d", intgr, frac)

Write the new CSV data

Finally, we write the result to a new file, using os.Create() and a CSV writer that knows how to turn the slice of slices of strings back into a proper CSV file.

Note that we do not set the separator to semicolon here, as we want to create a standard CSV format this time.

writeOrders takes a filename and a spreadsheet and writes the spreadsheet as CSV to the file.

func writeOrders(name string, rows [][]string) {

	f, err := os.Create(name)
	if err != nil {
		log.Fatalf("Cannot open '%s': %s\n", name, err.Error())

We are going to write to a file, so any errors are relevant and need to be logged. Hence the anonymous func instead of a one-liner.

	defer func() {
		e := f.Close()
		if e != nil {
			log.Fatalf("Cannot close '%s': %s\n", name, e.Error())

	w := csv.NewWriter(f)
	err = w.WriteAll(rows)

When running this code, the output file should look like this:

Date,Order ID,Order Item,Unit Price,Quantity,Total
2017-11-17,1,Ball Pen,1.99,50,99.50
2017-11-18,5,Sketch Block,2.99,40,119.60
2017-11-19,6,Ball Pen,1.99,30,59.70
2017-11-19,7,Sketch Block,2.99,20,59.80
2017-11-19,8,Ball Pen,1.99,60,119.40
,,,Ball Pens,140,

And we can open it in our spreadsheet app, or in a CSV viewer, to get a nicely formatted table.

Date Order ID Order Item Unit Price Quantity Total
2017-11-17 1 Ball Pen 1.99 50 99.50
2017-11-17 2 Notebook 12.99 10 129.90
2017-11-17 3 Binder 4.99 25 124.75
2017-11-18 4 Pencil 0.99 100 99.0
2017-11-18 5 Sketch Block 2.99 40 119.60
2017-11-19 6 Ball Pen 1.99 30 59.70
2017-11-19 7 Sketch Block 2.99 20 59.80
2017-11-19 8 Ball Pen 1.99 60 119.40
Sum 811.65
Ball Pens 140

Here we can see our new Totals column, as well as the two new rows that show the overall sum and the number of ball pens ordered.

How to get and run the code

Step 1: go get the code. Note the -d flag that prevents auto-installing the binary into $GOPATH/bin.

This time, also note the /... postfix that downloads all files, not only those imported by the main package.

go get -d

Step 2: cd to the source code directory.

cd $GOPATH/src/

Step 3. Run the binary.

go run spreadsheet.go

You should then find a file named ordersReport.csv in the current directory. Verify that it contains the expected result.

Q&A: Why CSV?

I use CSV here, rather than the file formats used by Excel or Open/Libre Office or Numbers, in order to stay as flexible and vendor-independent as possible. If you specifically want to work with Excel sheets, a quick search on GitHub should return a couple of useful third-party libraries. I have not used any of them yet, so I can neither share any experience nor recommend a particular one.

Wikipedia: Comma-separated values - Details about the CSV format.

Happy coding!

comments powered by Disqus