Generating Grids of Images in LaTeX with datatool and tikz

This is a quick one but I thought it would be worth sharing as it took me quite a bit of digging to find all the tools I needed to get what I wanted done.


I wanted to generate grids of related images on various pages from PDF files. Since I was having trouble getting matplotlib to embed PDFs instead of PNGs I figured I’d give LaTeX a shot.

Note: this is an artificial example, I don’t actually make pdfs of paintings. It was easier to get fun jpgs for this example but it works just as well with PDFs.



Here is an example of the resulting PDF.

For each artist name, I have a folder containing their self-portrait /portraits and /examples with an example of their paintings in the format {artist_name}.jpg.

For each row in the CSV, I want to make a page in the document with a column for each artist containing their portrait and an example of their paintings.

The full code is below and I will briefly outline some of the main components and tools I found useful, of course there is much more available than what I will discuss.


datatool is really nice for reading, looking up, and iterating through tabular data files.

Loading database

  • \DTLsetseparator{,} pretty self-explanatory, tells the package which character to use to separate the columns.

  • \DTLloaddb[noheader]{artists}{artists.csv} loads the file artists.csv into the macro artists. I used the option noheader to tell it my CSV does not have a header row with column labels. Columns are then given the default names Column1, Column2, etc.

Looping over rows

  • \DTLforeach{artists}{}{loop body} iterates over the rows of the CSV. Since I don’t care about the column names here, the second argument is left empty. But you can use it to assign macros to column names. Say we have a header column a,b,c we can assign commands to them as:

  • DTLforeach{artists}{\adam=a, \bob=b, \charlie=c} so now we can access columns by name.`

  • Also useful, the iteration index in the for-loop is stored in \DTLcurrentindex.

Looping over rows

  • \DTLforeachkeyinrow{\artist}{loop body} loops over the columns in the current row. Must be used inside a row \DTLforeach. Assigns the value of the current row to the first argument, in this case, artist.
  • You can also get the current index of the column in \dtlcol.

FYI there are a lot more useful tools in the documentation such as conditionals \DTLiffirstrow which let you treat specific rows specially.


I also needed to do some string splitting to build paths to the appropriate files so I used the xstring package.

  • \StrSplit{string to split}{position}{left}{right} this command splits the first argument at the position in the second argument and stores the left part and right part in the last two arguments.

Checking for existence of file

Some of the files I was going to include would sometimes be absent and I wanted a way to handle this case without crashing the compilation. Thankfully latex has a nice conditional to help.

  • \IfFileExists{filename}{true-branch}{false-branch} so I can include the graphics in the true branch and do nothing or put a placeholder in the flase branch.


This is a very well-known package so I don’t think I need to go into too much detail here but I simply placed each graphic as a node to form a grid with the following syntax:

\node (name) at (0, 0) {\includegraphics{pic.pdf}};

In order to properly position the images in a grid, I used the column index \dtlcol and a constant factor to select a row in my image grid.

I also wanted to get rid of any padding around the images which is accomplished with the following arguments when creating the tikz picture:

\begin{tikzpicture}[every node/.style={inner sep=0,outer sep=0}]

Of course, I could have used subfig or subcaption packages but I wanted to use tikz in case I wanted any additional drawing on the immages such as frames and arrows.

Miscellaneous command line tools

In the process, I also learned about two very nice BASH tools.

  • My file list CSV contained underscores which datatools and after reading underscores in filenames are not good practice, I decided to rename my files containing dashes to underscores. For this, I used a bash find and replace with a for loop:

$ for x in ls; do mv $x ${x//_/-}; done

The syntax is ${variable//find/replace}.

  • Another cool one I didn’t know about is basename which returns the basename of a path:
$ basename /path/to/file.txt
$ file.txt
  • When generating my images there was a long list of files I had to process with a bash script which was taking some time. So I discovered the GNU command parallel docs which on Mac does not come default but is a quick brew install parallel away and is really cool.

The basic usage is:

$ cat filelist.txt | parallel -in {} -out {./}.out

Here filelist.txt contains a list of files we need to process and sends each line to and parallel takes care of launching each call in a different process. The {} are called string replacements which we can use to manipulate the format of the input. For example {./} strips the file suffix so we can make the output suffix be .out.

Full code





			\begin{tikzpicture}[every node/.style={inner sep=0,outer sep=0}]
				%get the ligand name
								% name
					\node () at (6*\dtlcol-5,0) ;
				% portrait, leaving a space after the macro allows parser to end macro nicely.
				\IfFileExists{portraits/\name .jpg}{
					% portrait
					\node () at (6*\dtlcol-5 ,5) {\includegraphics[width=0.2\textwidth]{portraits/\name .jpg}};}
%					% painting
				\IfFileExists{examples/\name .jpg}{
					\node () at (6*\dtlcol-5,15) {\includegraphics[width=0.2\textwidth]{examples/\name .jpg}};}


Of course, I am open to suggestions for better ways to approach this problem!