HPDF-1.4.5: Generation of PDF documents

Safe HaskellSafe-Inferred




Quick documentation for the PDF library.

For detailed examples, download the tar.gz package from Hackage and look at the test.hs in folder Test.


Creating a document

When you create a document, you must give some information for the PDF file like the author, the default size (the pages can use different sizes if specified) and if the document is compressed.

So, a standard way to start a PDF document is with:

main :: IO()
main = do
    let rect = PDFRect 0 0 600 400
    runPdf "demo.pdf" (standardDocInfo { author=toPDFString "alpheccar", compressed = False}) rect $ do

where myDocument is generating the pages and is a value of the PDF monad.

Adding pages

You can add pages and specify a hierarchical structure for the pages. This hierarchy is optional. Here is an example of how you could add some pages and specify the table of contents:

myDocument :: PDF () 
myDocument = do
    page1 <- addPage Nothing
    newSection (toPDFString "Section") Nothing Nothing $ do
     newSection (toPDFString "Subsection") Nothing Nothing $ do
        createPageContent page1

when you use addPage you can specify a different size for the page or use the document's default one. In newSection, the two Maybe options are used to style the entry in the PDF table of contents.

There are other functions to add pages with transitions.

Creating the page content

To create content for a page, you have to use a page reference with drawWithPage.

drawWithPage is using a Draw monad value.

Element of the Draw monad are built with geometry, text and color primitives.

createPageContent :: PDFReference PDFPage -> Draw () 
createPageContent page = drawWithPage page $ do 
    strokeColor red
    setWidth 0.5
    stroke $ Rectangle 10 0 200 300


Text is complex. You can use the low level PDFText to create a text in the Draw monad. For instance:

textText :: PDFFont -> PDFString -> Draw ()
textText f t = do
     drawText $ do
         setFont f
         textStart 10 200.0
         leading $ getHeight f
         renderMode FillText
         displayText t
         displayText $ toPDFString "Another little test"

It gives a detailed control on the position of characters and lines but it is too much work.

The library is thus supporting a higher level typesetting system with paragraph styles.

Displaying a formatted text is done with displayFormattedText and using a typesetting monad value:

displayFormattedText (Rectangle (10 :+ 0) (110 :+ 300)) NormalPara Normal $ do 
   paragraph $ do
        txt $ "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor "
        txt $ "incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud "
        txt $ "exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute "
        txt $ "irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla "
        txt $ "pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia "
        txt $ "deserunt mollit anim id est laborum."

The text will be formatted using the NormalPara paragraph style and the Normal style for sentences.

NormalPara is part of an algebraic data type defining some vertical styles (from file test.hs):

data MyVertStyles = NormalPara
                  | CirclePara
                  | BluePara !PDFFloat

and Normal is part of another algebraic data typec (from file test.hs):

data MyParaStyles = Normal
                  | Bold
                  | Crazy
                  | SuperCrazy [Int] [PDFFloat]
                  | DebugStyle
                  | RedRectStyle
                  | BlueStyle

The library is coming with standard styles StandardParagraphStyle and StandardStyle.

Custom styles must be instances of some classes. A ComparableStyle to allow the typesetting algorithm to decide when to group different characters in a span of the same style.

A Style class used for sentence style. And a ParagraphStyle to group together the paragraph style and the sentence style that can be used in this paragraph.

Why the ComparableStyle is used instead of the class Eq ? A style is containing information used for the font (size etc ...) but it can also contain additional information used by styling function (a styling function may draw a decoration). In that latter case, the additional information is changing the look of the sentence but not its layout : the font size is not changed. So, from a text point of view, the PDF text is drawn using the same attributes. But the additional decoration on top of it is changing.

So, ComparableStyle is used to compare the font settings of a style.

The ParagraphStyle is used to change the geometry of the paragraph (the paragraph can be typeset using a circle as shape for instance). This style is also used to style the bounding box.

The other attributes like distance between two lines etc ... are controlled in the typesetting monad.

setParaStyle (BluePara 0)
setFirstPassTolerance 500
unstyledGlue 6 0.33 0
paragraph $ do
      txt $ "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. "

Inside a paragraph, it is possible to change the line style and create new paragraphs:

paragraph $ do
    txt $ "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor"
    setStyle Bold
    txt $ " incididunt ut labore et dolore magna aliqua. "

When charts are created, it is often useful to be able to display captions, labels etc ... The position of the box containing the text is relative to some specific points in the drawing. To ease with this use-case, an additional function is provided : drawTextBox


The typesetting is similar to the TeX one with kern, glues and boxes. So, it means that any drawing can be used as a letter since any drawing can be contained in a box. The operators to draw boxes, glues are part of the MonadStyle monad. The Draw value can be transformed into a box with mkDrawBox.

The paragraph and the typesetting monad are instances of this class. So, boxes, glues, kerns can be used in horizontal mode (paragraph) or vertical mode (typesetting monad).


Building shapes inside the draw monad is easy. For instance:

strokeColor red
stroke $ Rectangle 0 (200 :+ 100)
fillColor blue
fill $ Ellipse 100 100 300 200
fillAndStroke $ RoundRectangle 32 32 200 200 600 400

you can also create paths.

In addition to color, other attributes can be changed:

withNewContext $ do
    setWidth 2
    setDash $ DashPattern [3] 0
    stroke $ Rectangle 0 (200 :+ 100)

withNewContext is saving and restoring the settings.

Shapes can be filled with shading patterns:

paintWithShading (RadialShading 0 0 50 0 0 600 (Rgb 1 0 0) (Rgb 0 0 1)) (addShape $ Rectangle 0 (300 :+ 300))
paintWithShading (AxialShading 300 300 600 400 (Rgb 1 0 0) (Rgb 0 0 1)) (addShape $ Ellipse 300 300 600 400)

Note that in above example, addShape is used. You can't use stroke or fill. You are just adding a shape to a path.

More complex patterns can also be used to fill the shapes. In below example we are filling shapes with a complex drawing defined with a Draw monad value.

patternTest :: PDFReference PDFPage -> PDF ()
patternTest page = do
     p <- createUncoloredTiling 0 0 100 50 100 50 ConstantSpacing pattern
     cp <- createColoredTiling 0 0 100 50 100 50 ConstantSpacing cpattern
     drawWithPage page $ do
         strokeColor green
         setUncoloredFillPattern p (Rgb 1 0 0)
         fillAndStroke $ Ellipse 0 0 300 300
         setColoredFillPattern cp
         fillAndStroke $ Ellipse 300 300 600 400

       pattern = do
           stroke (Ellipse 0 0 100 50)
       cpattern = do
           strokeColor (Rgb 0 0 1)
           stroke (Ellipse 0 0 100 50) 

X Form

You can share an object between different pages of a document. It helps reducing the size of the document is the shared drawing is big. An object can be a Draw monad value. But it can be a JPEG picture too.

r <- createPDFXForm 0 0 200 200 lineStyle
drawWithPage page6 $ do
     drawXObject r

in the above example, lineStyle is a Draw() value.


It is possible to embed JPEG images in the document.

testImage ::  JpegFile -> PDFReference PDFPage -> PDF ()
testImage jpgf page =  do
    jpg <- createPDFJpeg jpgf
    drawWithPage page $ do
      withNewContext $ do
          setFillAlpha 0.4
          drawXObject jpg
      withNewContext $ do
           applyMatrix $ rotate (Degree 20)
           applyMatrix $ translate (200 :+ 200)
           applyMatrix $ scale 2 2
           drawXObject jpg

The JpegFile value must be created in the IO monad with:

Right jpg <- readJpegFile "logo.jpg"  

The haskell code is just extracting the size of the image from the file. The image is not decoded.


A pdf page can contain several kind of annotations like links, notes etc ... For instance, to define and display a link:

newAnnotation (URLLink (toPDFString "Go to my blog") [0,0,200,100] "http://www.alpheccar.org" True)


The PDF format is full of extensions. Depending on the viewer that you use some extensions may not be supported. It is always a good thing to test on a few viewers if you use complex features.

Mobile viewers (tablets and phones) are generally focusing on a more portable and more restricted set of features. So, you may not be able to display you document on a mobile device if you use complex features.

So, I repeat : test.