in Ten Lessons – 05: Lists, Dictionaries, and Data


Covered in this lesson:
Lists / Dictionaries / Data Visualisation / External Data

This tutorial introduces Python lists and dictionaries. These datatypes will unlock powerful ways to manage and manipulate collections of elements as opposed to individual values. Following a review of the basics, you’ll look at employing these techniques to visualise data, and reading in values from external files.

Complete list of lessons


Unlike the datatypes you have encountered thus far, lists hold multiple values. To illustrate, consider these two variables:

student = 'Sam'
age = 24

Here, 'Sam' represents a student name – or more technically speaking, some string data. Recording Sam’s age requires an additional age variable. However, one can assign both values to a single variable using a list-type approach:

student = ['Sam', 24]

The square brackets contain both the string and integer values. You may never have used a list before, but can likely make some sense of the syntax? More on syntax shortly, though. You may also be wondering: wouldn’t it be simpler to stick with the two separate variables? Perhaps. It really depends on the application. Lists are ordered, and ordering is significant in many situations – for example, in this sequence of rainbow colours:

rainbow = [

Moreover, lists (and dictionaries) are particularly powerful when combined with loop statements. Performing repetitive operations on diverse datasets is a common programming challenge. As an example: you are tasked with generating a population pyramid for each of the world’s (195, or so) countries. Provided that you can write some loop that plots the chart using a list of Chinese population figures, the list values can be switched out for Transnistria.

Working with Lists

To familiarise yourself with defining, accessing, and modifying lists, create a new sketch. Save this as “rainbow_list” and add the following code:

rainbow = ['blue', 'orange', 'yellow']

The rainbow is currently missing a few colours, and the sequence is incorrect, but this will be amended as we progress. Run the code and observe the Console output:

Printing the rainbow variable displays all three values (along with square brackets and commas). In many instances, though, it’s an individual element that you wish to retrieve. To display a given value, specify its position (index) in square brackets. Begin by printing the first element. Note, however, that Python list indices begin at zero:


Run the sketch to confirm that the Console displays blue.

One can think of each element’s position as its offset from the start of the list. The second element, orange, has an index of 1. The last element in this list, yellow, has an index of 2.

print(rainbow[1])   # displays orange
print(rainbow[2])   # displays yellow

Maybe this syntax remind you of slice notation? Well, it works the same way! For example, the last element can be accessed using -1, while a subset of elements can be extracted using a colon. Add the following code:

print(rainbow[-1])  # displays yellow
print(rainbow[-2])  # displays orange
print(rainbow[0:2]) # displays ['blue', 'orange']

Should you specify an index beyond the bounds of the list – say, rainbow[3] or greater – the Console will display an IndexError message.

Modifying lists

Many lists are dynamic in nature. Consider a game like Breakout (image below). A list stores the bricks comprising the wall; hitting a brick removes it from this list. In some levels, additional bricks may appear during play, thereby inserting new list elements.

LBreakout2 – an open source Breakout clone.
Wilinckx [GPL], from Wikimedia Commons

I am assuming you have played some variant of this game and are probably aware that, upon destruction, select bricks drop power-ups. Moreover, bricks come in different colours, some may be invisible, and others take multiple hits to destroy. These additional properties can all be programmed using lists of lists. Yes – lists can contain other lists, which can, in turn, can contain further nested lists … but more on that later.

To modify an existing element, reassign a new value like you would any other variable but include the index in square brackets. Take a practical example: red must replace blue as the first value in the rainbow list. To accomplish this, add the following lines to the bottom of your “rainbow_list” sketch:

rainbow[0] = 'red'
print(rainbow)      # ['red', 'orange', 'yellow']

The Processing reference includes a number of List Methods. Correctly speaking, these are standard (as opposed to Python features, functional in any Python environment. What follows below are descriptions for several such methods along with code to add to your working sketch. Each example builds on the code before it, so you’ll need to work through all of them, entering each line as you progress.


Adds an element to the end of a list.

print(rainbow)      # red, orange, yellow, blue


Adds one list to the end of another list.

colors = ['indigo', 'violet']
# red, orange, yellow, blue, indigo, violet


Returns the index (position as an integer) for the argument provided; if there are multiple matches, this represents the lowest/first instance. If there is no matching value, the Console reports a ValueError message.

yellowindex = rainbow.index('yellow')
print(yellowindex)  # 2


The insert method accepts two arguments: the first is the index/position to insert the element; the second is the value.

rainbow.insert(3, 'green')
# red, orange, yellow, green, blue, indigo, violet


The pop method accepts a single argument: the index/position of an element to remove. The pop-ed value is returned, should you need to utilise it for some other operation.

i = rainbow.pop(5)  # removes indigo and assigns it to i
or, to just remove indigo:
print(i)            # indigo
# red, orange, yellow, green, blue, violet

However, the argument is optional. If you provide none, Python removes the last element.

rainbow.pop()       # removes violet
# red, orange, yellow, green, blue


Removes the first element with a value that matches the argument.

# red, orange, yellow, green, blue, indigo, violet
# red, orange, yellow, green, blue, violet

Python does offer other list methods, but the above should suffice, for now at least. Any decent Python reference should cover the others – failing that, there’s always your favourite search engine. If you are looking to reorder list elements, there are the reverse() and alphanumerical sort() methods.

Rainbow Sequence Task

Time to consolidate what has been covered thus far. In this challenge, you will apply various list techniques to shuffle the colour bands of an incorrectly sequenced rainbow (orange, followed by violet, then blue, red, yellow, green, and indigo). Wait – what the heck is indigo, anyway? According to the dictionary, indigo is a “purplish-blue colour”, and violet is a “blueish-purple colour” 😕. On that point, why is there no purple band in the rainbow?

From left to right: indigo, dark-violet, and purple. According to your web browser.

Purple is a combination of two spectral colours. There is no wavelength of purple light – it only exists as a combination of red and blue waves. Violet, however, is an actual spectral colour with its own wavelength of approximately 380–420 nanometres. Indigo is positioned somewhere between blue and violet, but exactly where – or if at all – is a matter for debate. In his famous prism experiments, Isaac Newtown defined seven rainbow colours, squeezing indigo in just before violet. You may wonder, why seven colours from a blended array spanning the visible spectrum? This because seven had occult significance. It’s no coincidence that there are seven colours in the rainbow, seven days of the week, and seven musical notes that make up the Western major scale. Today, though, colour scientists are inclined to divide the spectrum at violet and blue, leaving no room for indigo.

In these lessons, we will drop indigo in favour of a six colour rainbow (just like Pink Floyd did).

The Dark Side of the Moon album cover depicts a prism splitting a beam of white light into its constituent colours. Pink Floyd's designer, Storm Thorgerson, opted for a six-colour, indigo-less rainbow.
Capitol Records [Copyright], from, via Wikipedia

Create a new sketch and save it as “rainbow_sequence”. Copy-paste in the following code:


bands = [
  '#FF9900', # orange
  '#6633FF', # violet
  '#0099FF', # blue
  '#FF0000', # red
  '#FFFF00', # yellow
  '#00FF00', # green

fill(bands[0]); rect(0,100, width,50)
fill(bands[1]); rect(0,150, width,50)
fill(bands[2]); rect(0,200, width,50)
fill(bands[3]); rect(0,250, width,50)
fill(bands[4]); rect(0,300, width,50)
fill(bands[5]); rect(0,350, width,50)
Incorrectly sequenced rainbow.

Now, insert the following code just above fill/rect lines:

'''# move violet
violet = bands[_]

'''# move blue
blueindex = bands.index('_______')
bands.insert(4, bands.pop(blueindex))'''

'''# switch orange and red
bands.insert(bands.index('_______'), bands.___(_))'''

fill(bands[0]); rect(0,100, width,50)

Remove the multi-line commented blocks (''') a section time. The goal is to figure out what values and methods replace the underscores. The correct result looks like this:

Once complete, be sure to save this file. You will be editing it further in the next challenge.

Loops + Lists

Loops can be programmed to work with lists, potentially saving on countless lines of manual instruction. As a case a point, let us return to the Breakout game example. Rendering each brick requires at least as many lines of code as there are elements. For example:

fill(bricks[0]);  rect(0,  0, 30,10) # brick A1
fill(bricks[1]);  rect(30, 0, 30,10) # brick A2
fill(bricks[2]);  rect(60, 0, 30,10) # brick A3
fill(bricks[59]); rect(270,50,30,10) # brick F10

This is hardly efficient, nor can it handle a list that is continually adapting in length. In a previous lesson covering for loops, you looked at iterating integer values using a range() function. To perform something similar on a list, you will first need to determine its current length.

Create a new sketch and save it as “iterating_lists”. Add the following code:

rainbow = ['red', 'orange', 'yellow', 'green', 'blue', 'violet']
print( len(rainbow) )

The len() function accepts a list as an argument and returns its length. Run the sketch to confirm this.

The len() function returns a list length of 6.

As the len() function returns an integer – in this case, six – it can serve as an argument for a range() function. Add this for loop to the bottom of your code:


for i in range( len(rainbow) ):

Run the sketch. With each iteration of the loop, i is incremented by 1. On the first iteration, the print line displays the value of rainbow[0], followed by rainbow[1] on the second iteration, and so on until reaching rainbow[5].

Thus far, your loops have relied on the range() function. It’s employed again here to make the code more relatable. However, Python regards loops as inherently iterable, meaning that one can omit the range function altogether. Comment out the previous loop and replace it with a range-less alternative:

for i in range( len(rainbow) ):
for band in rainbow:

Run the code. The Console prints the same list of values. But, what if you wished to display this output instead?

0: red
1: orange
2: yellow
3: green
4: blue
5: violet

Without the i variable from the first version of the loop, you have just the element value but no count. One approach is to adapt the earlier loop. For example:

for i in range( len(rainbow) ):
    print( ('%s: %s') % (i, rainbow[i]) )

This way, the i value is used to print the index and retrieve values from the list. However, Python offers another approach using an enumerate() function. This is best explained using a practical example. Add the following to the bottom of your working code:

for i,v in enumerate(rainbow):
    print( ('%s: %s') % (i, v) )

Recall, though, that i and v are variable names. Use whatever you feel is most descriptive – for example:

for number,colour in enumerate(rainbow):
    print( ('%s: %s') % (number, colour) )

In this instance, I have opted for the British spelling of colour to avoid confusing the variable with Processing’s color() function.

No single approach (range, enumerate, etc.) is necessarily better than another. What you elect to use will depend on your coding style and what you need to accomplish.

Rainbow Sequence Loop Task

Reopen your “rainbow_sequence” task from earlier. Your challenge is to convert the following code into something that uses a loop instead:


fill(bands[0]); rect(0,100, width,50)
fill(bands[1]); rect(0,150, width,50)
fill(bands[2]); rect(0,200, width,50)
fill(bands[3]); rect(0,250, width,50)
fill(bands[4]); rect(0,300, width,50)
fill(bands[5]); rect(0,350, width,50)

Ensure that the visual result remains the same:

An enumerate() is, perhaps, the more elegant solution. If you ever find yourself weighing-up different approaches, take a quick read through The Zen of Python – a list of 19 one-line principles to help guide your coding.

After grasping the fundamentals of Python lists, the next step is to tackle lists of lists.

Data Visualisation

Data visualisation is a recurring theme in these lessons. It relates neatly to a lot of the coding content and makes for some intriguing – and often, enlightening – visual output. Writing Processing code provides you with full control over visual output, so you’re longer limited to whatever Excel can conjure. Instead, you get to explore novel ways to visualise data – ranging from highly abstract (like something you’d see in an art gallery) to highly informative, or anything in between.

At various points, you’ll be provided brief introductions to useful ‘data viz’ concepts. Lists-of-lists are a means of managing multidimensional data, so now seems an opportune time to review the role of dimension in data visualisation. Before moving onto writing any code, though, we’ll look at a hypothetical scenario demonstrating how list data translates into visual output.


Computers are remarkably efficient data processing tools. It’s not surprising to discover that VisiCalc, a spreadsheet application released in 1979, is considered the world’s first killer application. Killer applications lead consumers to adopt new hard- or software platforms. Video game examples include the Atari 2600 port of Space Invaders, which the quadrupled console’s sales. Tetris is credited as the Nintendo Gameboy’s killer app and to date remains the product line’s top-selling game of all time.

Email is another killer app, albeit more of a messaging protocol. Before email, many people felt they didn’t need an Internet connection, or even a computer for that matter. Shortly after email went mainstream, web browsers converted many remaining online holdouts. Websites have been tracking visitor traffic since the early days of the Web. Nowadays, smart devices gather information from machines, animals, and people. We gather ever vaster quantities of data, yet much of it remains unused. Data visualisation presents opportunities to better utilise it – conveying concepts in a more accessible manner and even providing inputs to experiment with different scenarios.

Raw data is typically structured in a tabular type arrangement. It can be handwritten or captured electronically (CSV files, databases, etc.), but ultimately, can be dumped into a spreadsheet without too much hassle. However, scrolling through endless rows and columns is hardly an effective or engaging approach to analysis. Graphs help by presenting data more insightfully, also making it easier to present findings to others. Lotus 1-2-3, VisiCalc’s usurper, introduced several graphing features, which have been common spreadsheet fixtures ever since. If your spreadsheet software lacks the chart you need, you’ll likely find an app/web-app or programming library suited to the task. Plotly and matplotlib are well worth exploring.

Visualising Tetris Scores

In this section, we’ll be analysing some gameplay data. The focus is the visualisation of this data. There is no need to write any code, but you will need to read through some. In the next section, you will apply this theory in Processing.

In 2010, the inaugural Classic Tetris World Championship (CTWC) took place in Los Angeles, California. The organisers opted for the NES (8-bit Nintendo Entertainment System) version of Tetris, played on original hardware connected to a CRT television screen. The event has run annually ever since, although the venue has moved to Portland, Oregon. In 2017, one could qualify with a score of around 500,000. Now, suppose you wished to enter the upcoming CTWC. You’ve bought a console, cartridge, and CRT television, and have trained for months. However, your high scores seem to have plateaued. To help analyse and (hopefully) improve your scores, you decide to visualise your performance using Python.

To begin, you create a list containing all the dates you have played.

dates = [ 'Jan 28', 'Feb 04', ... ]

You then write some code that plots these values along a single axis labelled “Date”.

From this one-dimensional plot, you can determine how frequently you play. For a solid blue line of dots, you’d have to have been playing daily. There are significant gaps on either end (you purchased the equipment in late January; now it’s almost mid-May) as well as irregular intervals of non-play, but more prominently so towards the start. On average, you are playing around six days a month, and this is increasing. Statisticians would refer to this as an example of univariate analysis because we are concerned with a single variable (the dates you have played). Such analysis is useful for describing features like central tendency (mean, median, mode) and dispersion (range, variance, standard deviation).

Bivariate analysis is the simultaneous analysis of two variables; this helps identify relationships (if any) between them. In this instance, we will explore the correlation between dates and scores.

You decide to add another dimension to your list (a list in a list) so that each element contains two values: a date, and the highest score you accomplished that day. Think of it like this: you added a (“score”) dimension to your data that is now mirrored in your code. Programmers will often refer to 1-dimensional, 2-dimensional – or other higher dimensional – lists to describe the formations of multidimensional arrays.

scores = [
  ['Jan 28', 120000],
  ['Feb 04', 80000],

To visualise this, you must add a dimension to your plot. You elect to use a scatterplot, placing each dot against a horizontal (“Date”) and vertical (“Score”) axis.

It would appear that playing more frequently leads to more erratic high scores. Perhaps, what is most noteworthy, is that you perform best after stretches of no play (about week’s break). The good news is that your all-time high scores do seem to be improving.

You have recorded a lot of data – including the time of day, environmental factors, and more … most of it superfluous – but while poring over these details you realise that there may be a third variable in play: coffee. You add another value to your sub-lists; this is analogous to adding a column to an existing spreadsheet or table. The additional value indicates how many millilitres of coffee you drank (up to 2 cups, or 500ml) before posting a high score for that day.

scores = [
  ['Jan 28', 120000, 220],
  ['Feb 04', 80000,  260],

To visually accommodate the new coffee values, one can simulate depth. You add a third (“Coffee”) axis to the scatterplot.

Some interesting diagonal patterns have appeared, but gauging the positions of the dots is tricky. Moreover, should the data grow more multivariate – i.e. another variable is introduced – you face the problem of having to visualise a fourth spatial dimension. But, there are other visual notions to leverage. Consider, facets. Imagine the three-dimensional scatterplot as a cube (six-sided) with transparent walls. If you held it in your hand, it could be rotated to reveal a distinct graph on each side. Now, suppose that you ‘unfolded’ this cube and it laid flat, then separated each facet and arranged them on a grid. The result is a scatterplot matrix – or, acronymically, a SPLOM.

The orange facet of the 3D scatterplot (left) corresponds to the orange subplot in the SPLOM (right).

A three-dimensional SPLOM is three rows wide and three columns high. Should you wish to add further variables to the dataset, the matrix can expand to accommodate them.

A scatterplot matrix of scagnostics measures for the Boston Housing data set. Note that each sub-plot need not be represented as a scatterplot. Within the matrix, one can select a mix of bar, line, scatterplot, and other chart types.
Sigbert [CC BY-SA 3.0], from Wikimedia Commons

Facets are but one notion, though. Size is another option. You decide to revert to the two-dimensional scatterplot, then represent the “Coffee” dimension by adjusting the size of each dot accordingly.

However, each does not have to be circular. By including a variety of different shapes, one can convey even more information. Then, there’s colour. Your friend Sam – who is also an avid coffee drinker – plans to enter the tournament, too. To compare your progress, you plot Sam’s scores in orange.

Your high scores are plotted in blue; Sam's in orange.

Alas, in focussing on visualisation we have drifted away from code snippets. So, how exactly would one structure a Python list to accommodate the scatterplot above? One approach is a 3-dimensional list:

scores = [
  # your scores
    ['Jan 28', 120000, 220],
    ['Feb 04', 80000,  260],
  # sam's scores
    ['Feb 07', 145000, 150],
    ['Feb 09', 80100,  170],

In this way, each player is a list element containing other nested lists. To return to the spreadsheet analogy: instead of adding another column, you have added a new sheet – that is, one sheet for yourself and one for Sam.

Now look at a 2-dimensional alternative:

scores = [
  ['me', 'Jan 28', 120000, 220],
  ['me', 'Feb 04', 80000,  260],
  ['Sam', 'Feb 07', 145000, 150],
  ['Sam', 'Feb 09', 80100,  170],

In the three-dimensional version, your lists had been contained within one element, and Sam’s within another. Now, each entry must include a name. The result is a lot of repetition. Less repetition is usually better.

If you really wanted to give yourself a headache, you could opt for a 1-dimensional list:

scores = [
  'me', 'Jan 28', 120000, 220, 'me', 'Feb 04', ...

Values within the same category are now placed four positions apart from one another. Sure, you could write some loop that works with this … but, eeew! Right?

Interestingly, the number of dimensions you express visually does not always reflect the number of dimensions that comprise your list. In fact, it’s advisable to avoid using anything beyond a three-dimensional list. To return to the spreadsheet analogy one last time: consider adding another column to your existing sheets instead of creating more spreadsheet files.

Effective data visualisation requires the application of art and science to represent multidimensional information structures within two-dimensional visual displays. These displays could be sheets of paper or computer screens. Additionally, though, screens cater to time. As an example, charts and figures can animate while you view them. Using mouse, keyboard, touch, speech, gesture and other input, viewers can explore data in an interactive fashion. For some inspiration, take a look at Fathom’s project showcase. Ben Fry, the principal of Fathom, is also one of Processing’s co-developers.

Lists of Lists

While this may seem complicated at first, appropriately nested lists make complex data sets easier to manage. In this practical exercise, you will create a bar chart; roughly speaking, something resembling the illustration below.

Bar chart outline. The final result will include additional features such as colour.

Create a new sketch and save it as “lists_of_lists”. Add the following setup code:


h = 50

This is the start of a rainbow-coloured bar chart. The h variable and translate() function define the first bar’s height and starting position, respectively. You will begin with some ‘0-dimensional’ data, working your way up to a 3-dimensional list as you progress. To begin, here is a single integer value:

# 0-dimensional
bands = 6
rect(0,0, 40,h*bands)

Add the above code and run the sketch. The result is a vertical bar representing the number of bands in the rainbow sequence. If bands were equal to seven, the bar would extend beyond the bottom of the display window.

An additional dimension is required to describe the colour of each band. A 1-dimensional variable may be expressed as a list. Add the following lines to the bottom of your working code:

# 1-dimensional
bands = [

To render the bands, include a loop statement:

for i in range(len(bands)):
    rect(0,i*h, 40,h)

Run the sketch. The new strip is drawn precisely over the original bar but is divided into six blocks of colour.

The next step is to extend each block of colour, so as to form the horizontal bars. The width of each bar is to be determined by the brightness of its respective colour. To calculate brightness, one can add the red, green, and blue values together. For example, consider white – it is the brightest ‘colour’ on your screen; is represented in hexadecimal as #FFFFFF; and if converted to percentile values, is expressed as 100% red, 100% green, 100% blue. That is an overall brightness of 300%, or if you prefer to average it out, 300 divide by 3 = 100%.

To manage the colours in as RGB percentages, one must substitute each hexadecimal string with a list of integers. The result is list of lists – a 2-dimensional array:

# 2-dimensional
bands = [
  [ 100, 0, 0   ],
  [ 100, 60, 0  ],
  [ 100, 100, 0 ],
  [ 0, 100, 0   ],
  [ 0, 60, 100  ],
  [ 40, 20, 100 ]

Add the code above to the end of your working sketch. To access a list element within another list, include a second pair of square brackets. For example, to retrieve the percentage of green in the second (orange) band, it’s:

print( bands[1][1] )    # displays 60

Now, set the colorMode to use values between zero and one hundred, and add a loop to draw the bars:

colorMode(RGB, 100)

for i in range(len(bands)):
    r = bands[i][0]
    g = bands[i][1]
    b = bands[i][2]
    sum = r + g + b
    avg = sum / 3
    fill(avg, avg, avg)
    rect(0,i*h, sum,h)

Run the sketch. The width of each bar is governed by brightness – that is, the sum of the r, g, and b values. The bars are filled with greys to indicate the relative brightness of each colour.

Oddly, the green bar (fourth from the top) is equivalent in brightness/darkness to the red (top) bar. To recall, here is a colour swatch of each:

This has to do with how the human eye perceives colour. We have a greater number of green receptors, so green light appears more prominent. There are ways to compromise for this, but for now, our averaging formula will suffice.

Adapt the existing loop, so that each bar indicates the quantities of primary colour that comprise it:

    b = bands[i][2]
    #sum = r + g + b
    #avg = sum / 3
    #fill(avg, avg, avg)
    #rect(0,i*h, sum,h)
    rect(0,i*h, r,h)
    rect(0+r,i*h, g,h)
    rect(0+r+g,i*h, b,h)

Labels will help elucidate things. To add some to the dataset, one could go another list deeper, for example:

bands = [
  [ [ 100, 0, 0  ], 'red'    ],
  [ [ 100, 60, 0 ], 'orange' ],

print( bands[1][0][1] ) # displays 60

However, the above code just overcomplicates matters. Adding another dimension is overkill; a fourth element is all that is required. Adapt your code as below:

bands = [
  [100, 0, 0, 'red'],
  [100, 60, 0, 'orange'],

for i in range(len(bands)):
    text(bands[i][3], -20,i*h+30)
Completed graph, with labels.

Many lists work just fine with a single dimension – take, for instance, a shopping list. You can think of a two-dimensional list as a grid or table. This makes them useful for plotting 2D graphics. Three- and other higher-dimensional arrays have their place, but before employing such a structure, consider whether adding another position to your 2D array may be more sensible.

Breakout Task

In this challenge, you will recreate a Breakout level. Some code will be provided to work with, and this will include a three-dimensional array. Working with such a list requires a nested loop – that is, a loop inside another loop.

Create a new sketch and save it as “breakout”; then, copy-paste in the code below:


# ball and paddle
ellipse(350,440, 18,18)
rect(300,520, 190,40)

r = '#FF0000' # red
o = '#FF9900' # orange
y = '#FFFF00' # yellow
g = '#00FF00' # green
b = '#0099FF' # blue
p = '#6633FF' # violet

bricks = [
  [ [0,r,1], [1,o,1], [2,y,1], [3,g,1] ], # row 0
  [ [0,o,1], [1,y,1], [2,g,1], [3,b,1] ], # row 1
  [ [0,y,1], [1,g,1], [2,b,1], [3,p,1] ], # row 2
  [ [0,g,1], [1,b,2], [2,p,2], [3,b,1] ], # row 3
  [ [0,b,1], [1,p,2],          [3,g,1] ], # row 4
  [ [0,p,1],                   [3,y,1] ], # row 5
  [                            [3,o,1] ], # row 6
  [ [0,g,1]                            ]  # row 7

In an attempt to make the code more readable, the bricks list has been typed-out in a fashion that reflects the visual positioning of each brick. In the following order, each brick contains a: column-number; fill-colour; and hit-count (indicating how many hits are required to destroy it). Take the first brick as an example:


The brick is positioned in column 0, has a fill of red, and requires one (remaining) hit to destroy. Of course, one can infer the row position from the list in which the brick resides. Add two print() statements to confirm this information:

print( bricks[0] )       # displays the first row of bricks
print( bricks[0][0] )    # displays the first brick

Should you wish to retrieve the colour of the first brick, it’s:

print( bricks[0][0][1] ) # displays #FF0000

Now, you must complete the task as per the result below. Bricks with a hit-count of 2 have a shine effect.

The correctly completed task. Note that the three centre bricks have a shine.

As mentioned already, you will need to employ a nested loop. If you are stumped, perhaps these few lines will get you going?

for row,bricks in enumerate(bricks):

    for brick in bricks:
        x = brick[0]*150

If you are more comfortable with a range() style approach, that should work fine too.


Dictionaries can be thought of as an extension to the Python list concept. Now that you understand lists, grasping dictionaries should be easy. In a list, each value corresponds to a number indicating its position. One can, therefore, describe lists as numerically indexed. With dictionaries, each value is instead associated with a keyword. Dictionaries, therefore, are collections of key-value pairs.

If you want to experiment with the dictionary code that follows, create a new “dictionary” sketch now. However, you may find that reading over the content is enough to grasp it. Then you can begin coding again when you encounter the coffee chart task.

Creating New Dictionaries

Whereas lists are defined within square brackets, dictionaries rely on { braces }. The code below contrasts a list and a dictionary used to store an individual student’s details.

# list
studentlist = ['Sam', 24]

# dictionary
studentdict = {'name':'Sam', 'age':24}

Preceding each dictionary value is a key (in quotes) and a colon. Separating each key-value pair is a comma. You may store as many key-value pairs as you like in a dictionary – although, technically speaking, this is limited by how much free RAM/memory is available on your computer.

Accessing Dictionaries

To access a dictionary item, specify its key within square brackets.

print( studentdict['name'] )  # displays Sam
print( studentdict['age'] )   # displays 24

To print the entire dictionary, omit the square brackets and key, leaving just the variable name.

print( studentdict )          # {'name': 'Sam', 'age': 24}

You now understand the syntactical differences between lists and dictionaries. Dictionaries also possess their own set of methods. Many list methods – such as append(), extend(), index(), insert(), and remove() – will not work on dictionaries. There are few important dictionary methods you will need in this lesson, which are covered below. For more methods, refer to any decent Python reference.


The keys method will return a list of all the dictionary’s keys.

print( studentdict.keys() )   # ['name', 'age']


The values method will return a list of all the dictionary’s values.

print( studentdict.values() ) # ['Sam', 24]


The .items() method returns a list of all the dictionary’s key-value pairs.

print( studentdict.items() )  # [('name', 'Sam'), ('age', 24)]

This method is especially useful for iterating dictionaries (as you will soon see). Be warned, though: it may return values in a seemingly arbitrary order, i.e. not always the order in which they appeared when defining the dictionary. This has to do with how Python stores dictionaries (a topic beyond the scope of these tutorials).

Round brackets denote a tuple. Tuple can be pronounced as “too-ple” or “tuh-ple” depending on who you want to annoy. Tuples are not covered in this lesson, but for now, consider them as interchangeable with lists. For example:

items = studentdict.items()
print( items[0] )             # ('name', 'Sam')
print( items[0][0] )          # name

Note how tuple elements are also numerically indexed, and how list syntax is used to retrieve values. In a nutshell, the key difference is that tuples, once defined, cannot be modified. For more information, refer to the Processing reference.

Modifying Dictionaries

Dictionaries are dynamic structures. You can add and modify key-value pairs whenever you please. To change an existing value, simply reassign it as you would a list element. Of course, you will use a key name as opposed to a numeric index.

studentdict['age'] = 25
# {'name': 'Sam', 'age': 25}

To add a new key-value pair, follow the same process.

studentdict['id'] = 19011501
print( studentdict )
# {'name': 'Sam', 'id': 19011501, 'age': 25}

To remove a key-value pair, use the del statement.

del studentdict['age']
# {'name': 'Sam', 'id': 19011501}

If you need to add/combine one dictionary with another, refer to the update() method.

Nested Dictionaries

As with lists, dictionary values may comprise a mix of data types and can even include other dictionaries or lists.

# dictionary of lists
students = {
  'names':['Sam', 'Lee'],
  'ages':[24, 18]
print( students['names'][1] ) # displays Lee

Lists can also contain dictionaries.

# list of dictionaries
students = [
  {'name':'Sam', 'age':24},
  {'name':'Lee', 'age':18}
print( students[1]['name'] )  # displays Lee

What you name your keys – i.e. name and age – and how you nest collections of elements should help relate your data to real-world models.

Loops + Dictionaries

As with lists, there are many scenarios where you will want to loop through dictionaries. Considering that a dictionary can hold thousands or even millions of key-value pairs, this is a powerful feature. Because of the key-value system, though, iterating dictionaries is a little different than lists. You can iterate a dictionary’s keys, iterate its values, or iterate its key-value pairs. This is where the keys(), values(), and items() methods prove particularly handy. We will explore an example of each approach. First, let’s print studentdict to see what we are dealing with:

# {'name': 'Sam', 'id': 19011501}

Because the keys() method returns a list, you can use it to iterate a dictionary’s keys.

for k in studentdict.keys():
The Console displays:

On the first iteration, the print line displays the value of studentdict.keys()[0], and on the second iteration, studentdict.keys()[1]. This is confirmed in the Console, which displays “name” then “id”. Key iteration, however, automatically occurs when you combine a loop and a dictionary. If you omit the keys() method in the previous example, the result is the same.

#for k in studentdict.keys():
for k in studentdict:
The Console displays:

If you prefer a more explicit coding style, stick with the keys() method.

The values() method can be used similarly to the keys(). Of course, this returns just the element values.

for v in studentdict.values():
The Console displays:

If you need to retrieve keys and values, use the items() method. In the example below, the loop prints a tuple holding the current iteration’s key-value pair. Because the items() method returns both a key and a value, you must include two variable names (between the for and in of the statement). You may name these whatever you like, but the order of assignment is always: key first, value second.

for k,v in studentdict.items():
The Console displays:
('name', 'Sam')
('id', '19011501')

If you want the loop to retrieve the keys in alphanumerical order, use the sorted() function.

for k,v in sorted( studentdict.items() ):
# prints:
# ('id', 19011501)
# ('name', 'Sam')

The “id” tuple now appears first in the Console (before “name”). The sorted() function accepts additional arguments (such as reverse ordering). The Processing reference includes more information.

Coffee Chart Task

In this task, you will combine dictionaries, a list, and a loop. The result is a chart illustrating different types of coffee – that is, the amount of espresso, hot-water, steamed-milk, and foamed-milk that comprise each type.

Create a new sketch and save it as “coffee_chart”. Add the following setup code:


mug = 110
col = 1
row = 1
coffees = [
  'cafe con leche', 'espresso', 'demi-creme',
  'americano', 'capucchino', 'latte',
  'ristretto', 'macchiato', 'flat white'

for coffee in coffees:
    x = width/4*col
    y = height/4*row

    # mug
    arc(x+55,y, 40, 40, -HALF_PI, HALF_PI)
    arc(x+55,y, 65, 65, -HALF_PI, HALF_PI)
    rect(x-mug/2,y-mug/2, mug,mug)

    if col%3 == 0:
        row += 1
        col = 1
        col += 1

Run the sketch; it renders nine empty mugs.

At this point, the coffees variable is nothing more than a list of names. Replace this with the list of dictionaries below. It is easiest to copy and paste over the existing coffees list.

row = 1
coffees = [
  { 'name':'cafe con leche','espresso':50, 'hotwater':0, 'steamedmilk':30,'foamedmilk':0  },
  { 'name':'espresso',      'espresso':60, 'hotwater':0, 'steamedmilk':0, 'foamedmilk':0  },
  { 'name':'demi-creme',    'espresso':40, 'hotwater':0, 'steamedmilk':40,'foamedmilk':0  },
  { 'name':'americano',     'espresso':60, 'hotwater':30,'steamedmilk':0, 'foamedmilk':0  },
  { 'name':'capucchino',    'espresso':40, 'hotwater':0, 'steamedmilk':30,'foamedmilk':30 },
  { 'name':'latte',         'espresso':35, 'hotwater':0, 'steamedmilk':10,'foamedmilk':30 },
  { 'name':'ristretto',     'espresso':30, 'hotwater':0, 'steamedmilk':0, 'foamedmilk':0  },
  { 'name':'macchiato',     'espresso':40, 'hotwater':0, 'steamedmilk':0, 'foamedmilk':60 },
  { 'name':'flat white',    'espresso':40, 'hotwater':0, 'steamedmilk':60,'foamedmilk':0  }

The challenge is to fill each mug with the relevant proportions of ingredients. Begin with the labels (beneath each mug). The finished result looks like this:

espresso   hot-water   steamed-milk   foamed-milk

This code can easily be adapted to include as many coffee types as you desire, each with its own mug. However, the process of having to write and format the coffees data using Python syntax leaves room for improvement. In the next section, you will look at how Python can read in data from external files.

External Data

You have already utilised external data in the form of images in your Processing sketches. Python – and by extension, Processing – can handle many additional file types. For instance, you could create a game using Processing that incorporates various audio and video files, storing these multimedia assets in your sketch’s “data” directory. For now, though, we will look at text-based files.

Processing is not that great for developing games. There are plenty of better options out there. LÖVE is a neat little 2D-engine programmed in a language called Lua. For something Python-based, you could try Pygame.

A file format is a standardised means of encoding information for storage on a digital medium. There are many such formats, each interpreted differently. For example, applications are encoded in executable formats, such as APK for Android or EXE for Windows. Then, there are multimedia formats, like MP3 for music or JPG for images. You can identify a file’s format by its file extension. Frustratingly, many operating systems hide file extensions – but, if you dig around in your Windows or Mac Finder settings, you can get them to show in your file manager. File extensions are (most commonly) three letters in length, always preceded by a full-stop/period, and tacked onto the end of a file name. In the screenshot below, the file manager reveals the extensions of three files: an audio (.mp3), plain text (.txt) and image (.jpg) file.

Your system relies on the file extensions to open files with an appropriate app, and also to generate icons. If you remove or rename a file extension, this association is lost. Perhaps you have tried to open an MP3 (or JPG) file in a text editor? If so, you’ll know that all you get is a bunch of garbled characters:

An MP3 file opened with Atom (a code editor).

Atom is designed for editing text-encoded files, and therefore attempts to interpret the audio data as characters. While you might spot some intelligible metadata in there somewhere, it’s 99% gobbledygook. Yet, if you open this same file in iTunes, Windows Media Player, or VLC, you hear music.

Some file formats are text-based; this means that you can open them in any text or code editor (like Atom) and make some sense of the content. To clarify: by text-based, I mean plain text – as in, not a Word document. “Plain” means no bold, no italic, no varied fonts, etc. If you have written HTML before, you know what I’m talking about. So – you may be wondering – what is plain text good for? Well, shopping lists … and HTML, XML, CSV, JSON, and programming code, among other things. For instance, Processing files are plain text, albeit with a .pyde file extension.

Suppose that you wish to share a music playlist on your blog. For sharing playlists, many media players provide an ‘export to XML’ feature of some sort. XML (eXtensible Markup Language) files have a .xml extension. To give an idea of what VLC generates, here is a snippet of XML code:

<?xml version="1.0" encoding="UTF-8"?>
<playlist version="1" xmlns="">
  <title>Pink Floyd Playlist</title>
      <title>Speak to Me</title>
      <creator>Pink Floyd</creator>
      <album>The Dark Side of the Moon</album>
      <creator>Pink Floyd</creator>
      <album>The Dark Side of the Moon</album>

It is not important to understand XML for this lesson – but, although you may never have written or viewed any XML before, you can likely make some sense of the playlist’s contents? Contained within an opening and closing pair of track tags (<track>...</track>), you can discern the details of each song. Without getting into the technical details, it is relatively easy to parse this content using JavaScript so that it displays in a webpage (your blog).

Alternatively, you could elect to use JavaScript Object Notation. JSON files are named with a .json extension. The syntax closely resembles that of Python dictionaries:

  "playlist" : {
    "title"    : "Pink Floyd Playlist",
    "track"    : [
        "location"    : ["file:///music/SpeakToMe.mp3"],
        "title"       : "Speak to Me",
        "creator"     : "Pink Floyd",
        "album"       : "The Dark Side of the Moon"
        "trackNum"    : "1"
        "location"    : ["file:///music/Breathe.mp3"],
        "title"       : "Breathe",
        "creator"     : "Pink Floyd",
        "album"       : "The Dark Side of the Moon"
        "trackNum"    : "2"

XML and JSON are excellent formats for exchanging information between applications. Consequently, many programming languages (including Python) provide libraries for generating and parsing XML and JSON files.

To keep things simple, we will look at reading in data from another text-type format: CSV. Comma-Separated Value files are, perhaps, the simplest means of formatting plain-text data. The file extension is – yup, you guessed it – .csv. Because of CSV’s simplicity, you can often forgo using a library. Each line of a CSV file is an entry. Each entry consists of one or more fields, separated by commas. Here is the same abridged playlist in CSV format:

file:///music/SpeakToMe.mp3,Speak to Me,Pink Floyd,The Dark Side of the Moon,1
file:///music/Breathe.mp3,Breathe,Pink Floyd,The Dark Side of the Moon,2

The first line of the file contains the field headings. Lines two and three provide the details of tracks 1 and 2 respectively. You will find that your spreadsheet software (Microsoft Excel, LibreOffice Calc, or similar) will associate itself with any files bearing the extension .csv. Opening any CSV file in a spreadsheet displays the information in the typical row-and-column arrangement. This is useful for preparing CSV data – but be aware that none of the styling (cell-sizes, font colours, etc.) will be retained once saved back to CSV.

Opening the playlist.csv file in LibreOffice Calc.

CSV files do not always rely on a comma to delimit each field. For instance, tab and space-separated values are standard, too.

To read in text-based files, Processing uses a loadStrings() function. If you want to experiment with the CSV code that follows, you can create a new “csv” sketch now. However, you may find that reading over the content is enough to grasp it. Then, you can begin coding again when you encounter the Game Sales Chart task.

As with all external assets, the playlist.csv must be placed in the “data” sub-directory of your sketch. The loadStrings() function accepts a single argument: the path of your text file. It returns this file as a list of strings, each representing an individual line.

csv = loadStrings('playlist.csv')
The entire csv list printed to the Console.

To separate out each line (the list’s elements) use a loop.

csv = loadStrings('playlist.csv')

for entry in csv:
The playlist.csv file printed line-by-line using a loop.

The split() method can now be used to dice up each line into further lists. This works using the delimiter argument of your preference – in this case, a comma.

csv = loadStrings('playlist.csv')

for entry in csv:
    print( entry.split(',') )
Each entry printed as a list of strings (note the square brackets at the start and end of each line).

The u tags indicate unicode character representation. This is not important for now; unicode is effectively string data and everything will behave the same, regardless. To print the title of each track, retrieve the field with an index of [1].

csv = loadStrings('playlist.csv')

for entry in csv:
    #print( entry.split(',') )
    print( entry.split(',')[1] )

CSV, JSON, and XML each have their strengths and weaknesses. The beauty of CSV lies in its simplicity, but it does not support hierarchically-structured data. XML is an established, widely-supported, and flexible data exchange format, but it can turn out overly-complex and bloated at times. JSON is somewhat of a middle-ground, with a syntax that’s more compact than XML; JSON is also growing increasingly popular on the Web. You should weigh up the relative merits of each when considering what is best for your projects.

Game Sales Chart Task

In this final challenge, you will generate a bar chart of the best selling games of all time. Create a new sketch and save it as “game_sales_chart”.

The games list has been sourced from a Wikipedia article (which may well have shuffled since) and converted from an HTML table to a tab-delimited CSV file. Download a copy for your sketch and place it in the “data” sub-directory:


Add some basic setup code:

csv = loadStrings('list_of_best-selling_video_games.csv')

I have opted for tab-separated values. It’s highly unlikely that any game titles or studio/publisher names will contain tab characters, but there may be commas that could interfere with the split() step. Here are the first few lines of the of the CSV file; the tabs do not always form perfect columns, but, more importantly, there is a single (invisible) tab character between each field.

Rank 	Sales 	Title 	Developer 	Publisher
1 	170000000 	Tetris 	Elektronorgtechnica 	Spectrum HoloByte
2 	144000000 	Minecraft  	Mojang 	Mojang

You may want to open the CSV file for yourself to inspect the values. There are fifty games in all. The respective sales figure will determine the width of each bar in your chart. You will need to perform some mathematical calculation to scale the bars relative to the display window – but, while the “Sales” figures appear to be numbers, they are actually stored as ‘text’. In other words, you cannot perform mathematical operations on string data. To demonstrate, add this code to the bottom of your sketch:

tetrisentry = csv[1].split('\t')
tetrissales = tetrisentry[1]
print( tetrissales )
The Console displays: 170000000

Okay, so everything seems fine, for now. You retrieved the Tetris sales figure and printed it to the Console. Next, try some arithmetic:

print( tetrissales + 1 )
TypeError: cannot concatenate 'unicode' and 'int' objects

As reported in the Console, Python is unable to add together the unicode and integer data. Fortunately, there is an easy fix. One must convert the unicode data to something numeric. The int() and float() functions convert various data types to integer and floating point values, respectively.

print( int(tetrissales) + 1 )

Run the sketch. The Console is now error-free and displays 170000001.

Now, complete the chart as per the screenshot below.

Begin with a loop that prints each entry. Then, get the labels displaying (before the bars). Once you have labels, add plain white bars of the correct width, then finish off with the rainbow sequence effect.

Lesson 06

That’s it for lesson 05! Lists and dictionaries are relatively straight-forward, though combining collections of values with loops is a trickier concept to grasp. That said, what you have learned here is vital for what lies ahead – both in these lessons and beyond. In the next tutorial, we’ll zoom-in further, like, to pixel level. You’ll pick up some neat new skills, like how to read values off pixels to create your very own Photoshop-esque filters.

Begin Lesson 06: Pixels and Graphics

Complete list of lessons