An Extremely Basic Guide to Writing PDFs by Hand

The goal of this post is to teach you how to write a very basic PDF containing some shapes and some text. The goal is not to teach you how PDFs work in any level of detail - I’m certainly not the person to be teaching you that & it’s been done well by people before me.

If you want to actually learn how to write PDFs I recommend the following resources:

I’ll be teaching you how to build a PDF containing text, rectangles and circles. Like I said, very rudimentary. NOTE: all code samples are written in Typescript.

Here are some important details that I should find a better place for:


Below is the template that we’ll be using. For now, you’re safe to ignore everything bar the {{WHATEVER}} bits - these are placeholders for the stuff that you’ll be writing


      1 0 obj
          << /Type /Catalog
              /Pages 2 0 R

      2 0 obj
          << /Type /Pages
             /Kids [3 0 R]
             /Count 1
             /MediaBox [0 0 {{PAGE_WIDTH}} {{PAGE_HEIGHT}}]

      3 0 obj
          <<  /Type /Page
              /Parent 2 0 R
              << /Font
                  << /F1
                      << /Type /Font
                         /Subtype /Type1
                         /BaseFont /ArialMT
              /Contents [ {{OBJECT_REFERENCES}} ]


      0 {{NUM_OBJECTS}}
      0000000000 65535 f
          << /Root 1 0 R
             /Size {{NUM_OBJECTS}}

{{PAGE_WIDTH}} and {{PAGE_HEIGHT}} are the width and height of your document in points (pt). You can convert from px to pt with this helper

function pxToPt(px: number): number {
          return (px * 72) / 96

{{OBJECT_DEFINITIONS}} are all of the objects (text and shapes) that you want to define. These look like this

4 0 obj

This says "create an object with ID 4 and generation 0 and set it’s value to whatever {{OBJECT_VALUE}} is". The generation part doesn’t seem to be very important. More on this later.

{{OBJECT_REFERENCES}} are “references” to the above objects. You need to explicitly tell the PDF interpreter which objects are in which page via the /Contents array. Say we have two objects with IDs 4 and 5 in our page, the references would look like this

/Contents [ 4 0 R
                  5 0 R ]

The line break isn’t needed but it makes things a little clearer.

Finally, {{NUM_OBJECTS}} is… the number of objects in the document. This is the number of objects you defined, including the three object definitions included in the template, plus one. Why the one? Who knows. I’m sure it’s in the spec.

Defining your own object

What your object is depends on what you put into the body of an object declaration. The stuff that comprises an object is defined inside something called a stream

4 0 obj
      << /Length {{STREAM_LENGTH_IN_BYTES}} >>

First we let the PDF interpreter know how long the stream is. This is the length of every byte inside stream and endstream, bar trailing spaces.

Trap for young players

There’s a good chance that this isn’t equal to the length of the string that you dump in there. It needs to be the number of bytes. Unicode characters, which count as one character in Javascript strings, are actually made up of multiple bytes. You can count the bytes using a node buffer

$ node
      > let unicodeString = 'ü'
      > unicodeString.length
      > Buffer.from(unicodeString).length

The stream content itself is a sequence of postfix commands. It can also be binary content, but we don’t need to worry about that.

Adding text

The following snippet creates black text with the content Hello, world! at coordinates 0, 0 with a font size of 16pt

4 0 obj
      << /Length 57 >>
          0 0 0 rg
          /F1 16 Tf
          0 0 Td
          (Hello, world!) Tj

Let’s break this down piece by piece:

0 0 0 rg

The rg function in the first line of the body sets the color of the text. PDFs use what’s known as the rg color system. It’s essentially the same as rgb, but scaled 0 - 1. To convert, just divide each rgb value by 360

function rgbColorspaceToRgColorspace(c: Color.Rgb): Color.Rg {
          return { r: c.r / 360, g: c.g / 360, b: c.b / 360 }

So in the above example, we are setting the color to black.

/F1 16 Tf

If you go back to the template, you’ll see /F1 inside object 3 just before we specify the font. The Tf command sets the font & the font size — so we’re specifying 16 point Arial.

0 0 Td

This specifies the location to render the text. Again, the origin of the coordinate system in PDFs is the lower left hand corner, in contrast with most systems that you’ll be used to. This means that we’re going to draw the text in the lower left hand corner of the page.

(Hello, world!) Tj

Finally, this is what actually renders the text. Instead of using quotes to represent strings, PDFs use parentheses. Don’t ask me why.

<< /Length 57 >>

const stream = `BT 0 0 0 rg /F1 16 Tf 0 0 Td (Hello, world!) Tj ET`;
      console.log(Buffer.from(stream).length); // 57

So our above example is 57 bytes — hence the length declaration.