Wrapping Text With ImageMagick

I’ve already explained that I wanted to create an image slide for a slideshow from some information in a spreadsheet. I decided to use ImageMagick and, in particular, the Wand library for python to help me.

As with all theatre projects this one is collaborative. The original concept for this was James’ and he’d created some examples by hand*. Each slide has some fixed elements and then information specific to the event. Using those he’s created some template images that contain the fixed elements, there are a number of templates depending on the type of event.

Basic SlideEach one has basically the same layout and the same (wavy) line which delineates where I can draw the event elements onto the slide. Then I add the date and time and the title of the event and who is presenting it. All of that is straightforward with either a Caption, or a Text element.

Image AreaThe more challenging part is the lower half of the slide. The spreadsheet contains a link to an event specific image in Google Drive, so I download the image data directly. I want the image to fit into, at most, roughly half of the remaining area, so I resize it keeping the aspect ratio but capping the height and width.

with Image(blob=image_data) as img:
    size = "{}x{}".format(max_width, max_height)
    img.transform(resize=size)
    img_height = img.height # save height
    img_width = img.width # save width
    main_img.composite(img, left=left, top=top)

Slide With Image

Now I want to display the description wrapped around the slide. ImageMagick will easily draw all sorts of amazing text effects (on one line) but text wrapping is more of a challenge. You can use a Caption to wrap text inside a rectangle, but my rectangle now has a corner taken out of it. Also I was asked to truncate the text at a sensible point (e.g. the end of a sentence) if it was too big to fit.

This is the point for some funky arithmetic. Firstly create a Drawing and set the Font you want to use on it. Now text is either going to be next to the picture (a narrow line) or underneath it (a wide one), so using the font size and the height available, and of course the actual image height and width, calculate how many lines you can fit of each width. Now we just need to split the text into that many lines each of the right length.

The simplest way to split the text is to use the python textwrap module but that requires you to specify a fixed maximum number of characters and my font isn’t monospaced. Instead I use the Draw module’s get_font_metrics to do the same job. It requires a dummy image for some reason.

words = para.split(" ")
lines = []
with Image(height=10, width=10) as img:
  line = words.pop(0)
  for word in words:
    line_width = draw.get_font_metrics(img, 
      line + " " + word).text_width
    if line_width < max_width:
      line = line + " " + word
    else:
      lines.append(line)
      line = word
  if line != "":
    lines.append(line)
return lines

Using this I can build up a list of lines of text. Firstly the narrow ones and then the wide ones, discarding any text that won’t fit. Then I can draw them onto the image line by line. Giving me a finished slide.Complete Slide

 

*James is also responsible for the spreadsheet that holds all this data. You should ask him about it, it’s awesomely clever.

Advertisements

From Spreadsheet to Slideshow with a Raspberry Pi

As well as actually spending time playing in the theatre I also occasionally get involved in various computer related projects for one. This particular project was to help set up a rolling slideshow of “coming soon” events that we can display in the foyer of the theatre.

Of course this project lends itself to a raspberry pi so I got mine out and started work. Everybody (it seems) creates a slideshow on their raspberry pi, so there are many solutions to choose from. In this case (you’ll see why in a minute) I wanted a slideshow that looped through pictures in a directory and automatically coped with pictures being added and deleted. For now I’m using Eye of Gnome which does this admirably but this can easily be replaced in future.

So where are the pictures coming from? Well that’s where this gets interesting. The list of upcoming events is kept and managed in a Google Spreadsheet so I needed to take this information and turn it into a set of images for the slideshow. As the information in the spreadsheet is regularly updated I wanted to write a python script that could be run regularly by a cron job to pick up changes and create the new slideshow images.

Google provide an API to Google Drive that allows access to files and file information and they provide a python library to access this.
I followed the instructions here to configure a developers account and then grant access to the APIs you need (in this case Drive and Sheets). I decided to use the OAuth for Server to Server authentication to avoid the need for human interaction, so the other thing I needed to do was to make sure that the service account has access to the files it needs. I did this by sharing the relevant files and directories in Drive with the service account’s email address. I only needed to grant view access to the files as we’re only ever reading the data.

Then on the pi I needed to install a few libraries, the Google API of course, a date parsing module and Image Magick to help create the images.

sudo apt-get install python-pip
sudo pip install --upgrade google-api-python-client
sudo apt-get install python-dev
sudo pip install ciso8601
sudo apt-get install libmagickwand-dev
sudo pip install Wand

Once that’s done I could write the script to download the data. The Google API documentation is full of example code so it’s easy to get this working.
Firstly I retrieve the modified time of the spreadsheet, using the Drive API, to see if it’s changed since we last ran. If it has then I download the data from the spreadsheet and check each row to see if it’s been updated (each row has a unique identifier and a last updated time-stamp). For each updated row I delete any existing image and generate a new one.
Finally I record locally the most recently updated time of the rows so I know when to start from next time.

Note (because it surprised me) that the latest updated time of a row might actually be later than the modified time of the file as there’s sometimes a couple of minutes delay before the file modified time is updated.

That really was all there was to it. Well, okay, there’s actually a few other things going on too, but that’s for another post perhaps.