iOS SDK - Programmatically generate a PDF file

Using the CoreGraphics framework is tedious work, in my honest opinion, when it comes to programmatically drawing a PDF file.

I would like to programmatically create a PDF, using various objects from views throughout my app.

I am interested to know if there are any good PDF tutorials around for the iOS SDK, maybe a drop in library.

I've seen this tutorial, PDF Creation Tutorial, but it was mostly written in C. Looking for more Objective-C style. This also seems like a ridiculous way to write to a PDF file, having to calculate where lines and other objects will be placed.

void CreatePDFFile (CGRect pageRect, const char *filename) 
{	
    // This code block sets up our PDF Context so that we can draw to it
    CGContextRef pdfContext;
    CFStringRef path;
    CFURLRef url;
    CFMutableDictionaryRef myDictionary = NULL;

    // Create a CFString from the filename we provide to this method when we call it
    path = CFStringCreateWithCString (NULL, filename,
                                      kCFStringEncodingUTF8);

    // Create a CFURL using the CFString we just defined
    url = CFURLCreateWithFileSystemPath (NULL, path,
                                         kCFURLPOSIXPathStyle, 0);
    CFRelease (path);
    // This dictionary contains extra options mostly for 'signing' the PDF
    myDictionary = CFDictionaryCreateMutable(NULL, 0,
                                             &kCFTypeDictionaryKeyCallBacks,
                                             &kCFTypeDictionaryValueCallBacks);

    CFDictionarySetValue(myDictionary, kCGPDFContextTitle, CFSTR("My PDF File"));
    CFDictionarySetValue(myDictionary, kCGPDFContextCreator, CFSTR("My Name"));
    // Create our PDF Context with the CFURL, the CGRect we provide, and the above defined dictionary
    pdfContext = CGPDFContextCreateWithURL (url, &pageRect, myDictionary);
    // Cleanup our mess
    CFRelease(myDictionary);
    CFRelease(url);
    // Done creating our PDF Context, now it's time to draw to it
    
    // Starts our first page
    CGContextBeginPage (pdfContext, &pageRect);
    
    // Draws a black rectangle around the page inset by 50 on all sides
    CGContextStrokeRect(pdfContext, CGRectMake(50, 50, pageRect.size.width - 100, pageRect.size.height - 100));
    
    // This code block will create an image that we then draw to the page
    const char *picture = "Picture";
    CGImageRef image;
    CGDataProviderRef provider;
    CFStringRef picturePath;
    CFURLRef pictureURL;
    
    picturePath = CFStringCreateWithCString (NULL, picture,
                                             kCFStringEncodingUTF8);
    pictureURL = CFBundleCopyResourceURL(CFBundleGetMainBundle(), picturePath, CFSTR("png"), NULL);
    CFRelease(picturePath);
    provider = CGDataProviderCreateWithURL (pictureURL);
    CFRelease (pictureURL);
    image = CGImageCreateWithPNGDataProvider (provider, NULL, true, kCGRenderingIntentDefault);
    CGDataProviderRelease (provider);
    CGContextDrawImage (pdfContext, CGRectMake(200, 200, 207, 385),image);
    CGImageRelease (image);
    // End image code
    
    // Adding some text on top of the image we just added
    CGContextSelectFont (pdfContext, "Helvetica", 16, kCGEncodingMacRoman);
    CGContextSetTextDrawingMode (pdfContext, kCGTextFill);
    CGContextSetRGBFillColor (pdfContext, 0, 0, 0, 1);
    const char *text = "Hello World!";
    CGContextShowTextAtPoint (pdfContext, 260, 390, text, strlen(text));
    // End text
    
    // We are done drawing to this page, let's end it
    // We could add as many pages as we wanted using CGContextBeginPage/CGContextEndPage
    CGContextEndPage (pdfContext);
    
    // We are done with our context now, so we release it
    CGContextRelease (pdfContext);
}

EDIT: Here's an example on GitHub using libHaru in an iPhone project.

A couple things...

First, there is a bug with CoreGraphics PDF generation in iOS that results in corrupted PDFs. I know this issue exists up to and including iOS 4.1 (I haven't tested iOS 4.2). The issue is related to fonts and only shows up if you include text in your PDF. The symptom is that, when generating the PDF, you'll see errors in the debug console that look like this:

<Error>: can't get CIDs for glyphs for 'TimesNewRomanPSMT'

The tricky aspect is that the resulting PDF will render fine in some PDF readers, but fail to render in other places. So, if you have control over the software that will be used to open your PDF, you may be able to ignore this issue (e.g., if you only intend to display the PDF on the iPhone or Mac desktops, then you should be fine using CoreGraphics). However, if you need to create a PDF that works anywhere, then you should take a closer look at this issue. Here's some additional info:

http://www.iphonedevsdk.com/forum/iphone-sdk-development/15505-pdf-font-problem-cant-get-cids-glyphs.html#post97854

As a workaround, I've used libHaru successfully on iPhone as a replacement for CoreGraphics PDF generation. It was a little tricky getting libHaru to build with my project initially, but once I got my project setup properly, it worked fine for my needs.

Second, depending on the format/layout of your PDF, you might consider using Interface Builder to create a view that serves as a "template" for your PDF output. You would then write code to load the view, fill in any data (e.g., set text for UILabels, etc.), then render the individual elements of the view into the PDF. In other words, use IB to specify coordinates, fonts, images, etc. and write code to render various elements (e.g., UILabel, UIImageView, etc.) in a generic way so you don't have to hard-code everything. I used this approach and it worked out great for my needs. Again, this may or may not make sense for your situation depending on the formatting/layout needs of your PDF.

EDIT: (response to 1st comment)

My implementation is part of a commercial product meaning that I can't share the full code, but I can give a general outline:

I created a .xib file with a view and sized the view to 850 x 1100 (my PDF was targeting 8.5 x 11 inches, so this makes it easy to translate to/from design-time coordinates).

In code, I load the view:

- (UIView *)loadTemplate
{
    NSArray *nib = [[NSBundle mainBundle] loadNibNamed:@"ReportTemplate" owner:self options:nil];
    for (id view in nib) {
        if ([view isKindOfClass: [UIView class]]) {
            return view;
        }
    }

    return nil;
}

I then fill in various elements. I used tags to find the appropriate elements, but you could do this other ways. Example:

UILabel *label = (UILabel *)[templateView viewWithTag:TAG_FIRST_NAME];
if (label != nil) {
    label.text = (firstName != nil) ? firstName : @"None";

Then I call a function to render the view to the PDF file. This function recursively walks the view hierarchy and renders each subview. For my project, I need to support only Label, ImageView, and View (for nested views):

- (void)addObject:(UIView *)view
{
    if (view != nil && !view.hidden) {
        if ([view isKindOfClass:[UILabel class]]) {
            [self addLabel:(UILabel *)view];
        } else if ([view isKindOfClass:[UIImageView class]]) {
            [self addImageView:(UIImageView *)view];
        } else if ([view isKindOfClass:[UIView class]]) {
            [self addContainer:view];
        }
    }
}

As an example, here's my implementation of addImageView (HPDF_ functions are from libHaru):

- (void)addImageView:(UIImageView *)imageView
{
    NSData *pngData = UIImagePNGRepresentation(imageView.image);
    if (pngData != nil) {
        HPDF_Image image = HPDF_LoadPngImageFromMem(_pdf, [pngData bytes], [pngData length]);
        if (image != NULL) {
            CGRect destRect = [self rectToPDF:imageView.frame];

            float x = destRect.origin.x;
            float y = destRect.origin.y - destRect.size.height;
            float width = destRect.size.width;
            float height = destRect.size.height;

            HPDF_Page page = HPDF_GetCurrentPage(_pdf);
            HPDF_Page_DrawImage(page, image, x, y, width, height);
        }
    }
}

Hopefully that gives you the idea.

It's a late reply, but as i struggled a lot with pdf generation, i considered it worthwhile to share my views. Instead of Core graphics, to create a context, you can also use UIKit methods to generate a pdf.

Apple has documented it well in the drawing and printing guide.

The PDF functions on iOS are all CoreGraphics based, which makes sense, because the draw primitives in iOS are also CoreGraphics based. If you want to be rendering 2D primitives directly into a PDF, you'll have to use CoreGraphics.

There are some shortcuts for objects that live in UIKit as well, like images. Drawing a PNG to a PDF context still requires the call to CGContextDrawImage, but you can do it with the CGImage that you can get from a UIImage, e.g.:

UIImage * myPNG = [UIImage imageNamed:@"mypng.png"];
CGContextDrawImage (pdfContext, CGRectMake(200, 200, 207, 385), [myPNG CGImage]);

If you want more guidance on overall design, please be more explicit about what you mean by "various objects throughout your app" and what you're trying to accomplish.

Late, but possibly helpful to others. It sounds like a PDF template style of operation might be the approach you want rather than building the PDF in code. Since you want to email it off anyway, you are connected to the net so you could use something like the Docmosis cloud service which is effectively a mail-merge service. Send it the data/images to merge with your template. That type of approach has the benefit of a lot less code and offloading most of the processing from your iPad app. I've seen it used in an iPad app and it was nice.