- What is PoDoFo?
- Requirements
- Licensing
- Development quickstart
- String encoding and buffer conventions
- API Stability
- PoDoFo tools
- TODO
- FAQ
- No warranty
- Contributions
- Authors
PoDoFo is a s a free portable C++ library to work with the PDF file format.
PoDoFo provides classes to parse a PDF file and modify its content into memory. The changes can be written back to disk easily. Besides PDF parsing PoDoFo also provides facilities to create your own PDF files from scratch. It currently does not support rendering PDF content.
To build PoDoFo lib you'll need a c++17 compiler, CMake 3.16 and the following libraries:
- freetype2
- fontconfig (required for Unix platforms, optional for Windows)
- OpenSSL
- LibXml2
- zlib
- libjpeg (optional)
- libtiff (optional)
- libpng (optional)
- libidn (optional)
For the most polular toolchains, PoDoFo requires the following minimum versions:
- msvc++ 14.16 (VS 2017 15.9)
- gcc 8.1
- clang/llvm 7.0
It is regularly tested with the following IDE/toolchains versions:
- Visual Studio 2017 15.9
- Visual Studio 2019 16.11
- Visual Studio 2022 17.3
- gcc 9.3.1
- XCode 13.3
- NDK r23b
PoDoFo library is licensed under the LGPL 2.0 or later terms. PoDoFo tools are licensed under the GPL 2.0 or later terms.
Learn by example by looking at the github workflow definitions.
There are examples using vcpkg under
Windows,
brew under macos,
and apt-get under ubuntu, installing all the
required dependencies, bootstrapping the CMake project, building and testing the library. In Windows,
follow the vcpkg quickstart guide to setup the package manager repository first.
It may be also useful to set the enviroment variable VCPKG_DEFAULT_TRIPLET to x64-windows to default installing 64 bit dependencies
and define a VCPKG_INSTALLATION_ROOT variable with the location of the repository as created in the quickstart.
Additionally there's a playground area in the repository where you can have access to pre-build dependencies for some popular architectures/operating systems. Have a look in the Readme there.
All std::strings or std::string_view in the library are intended
to hold UTF-8 encoded string content. PdfString and PdfName constructors
accept UTF-8 encoded strings by default (PdfName accept only characters in the
PdfDocEncoding char set, though). charbuff abd bufferview
instead represent a generic octet buffer.
PoDoFo has an unstable API that is the result of an extensive API review of PoDoFo 0.9.x. It is expected to converge to a stable API as soon as the review process is completed. See API Stability for more details.
Warning: Tools are currently untested and unmaintained
PoDoFo tools are still available in the source tree
but their compilation is disabled by default because they are unsted/unmaintained,
and will not receive support until their status is cleared. It's not recommended to include them in software distributions.
If you want to build them make sure to bootstrap the CMake project with -DPODOFO_ENABLE_TOOLS=TRUE.
Tools are conveniently enabled in the playground
at least to ensure library changes won't break their compilation.
There's a TODO list, or look at the issue tracker.
Q: How do I sign a document?
A: The signing procedure is still as low level as it was before the
ongoing source changes. This is going to change, soon!, with a new high-level
API for signing being worked on, which will be fully unit tested. For now you
should check the podofosign tool (WARNING: untested) which should give
you the idea how to sign documents creating a CMS structure directly with
OpenSSL.
To describe the procedure briefly, one has to fully Implement a PdfSigner,
retrieve or create a PdfSignature field, create an output device (see next question)
and use PoDoFo::SignDocument(doc, device, signer, signature). When signing,
the sequence of calls of PdfSignature works in this way: method PdfSigner::Reset()
is called first, then the PdfSigner::ComputeSignature(buffer, dryrun) is called with
an empty buffer and the dryrun argument set to true. In this call one can just
resize the buffer overestimating the required size for the signature, or just
compute a fake signature that must be saved on the buffer. Then a sequence of
PdfSigner::AppendData(buffer) are called, receiving all the document data to
be signed. A final PdfSigner::ComputeSignature(buffer, dryrun) is called, with
the dryrun parameter set to false. The buffer on this call is cleared (capacity
is not altered) or not accordingly to the value of PdfSigner::SkipBufferClear().
Q: PdfMemDocument::SaveUpdate() or PoDoFo::SignDocument() write only a partial
file: why so and why there's no mechanism to seamlessly handle the incremental
update as it was in PoDoFo? What should be done to correctly update/sign the
document?
A: The previous mechanism in PoDoFo required enablement of document for incremental updates, which is a decision step which I believe should be not be necessary. Also:
- In case of file loaded document it still required to perform the update in the same file, and the check was performed on the path of the files being operated to, which is unsafe;
- In case of buffers worked for one update/signing operation but didn't work for following operations.
The idea is to implement a more explicit mechanism that makes more clear and/or enforces the fact that the incremental update must be performed on the same file in case of file loaded documents or that underlying buffer will grow following subsequent operations in case of buffer loaded documents. Before that, as a workaround, a couple of examples showing the correct operation to either update or sign a document (file or buffer loaded) are presented. Save an update on a file loaded document, by copying the source to another destination:
string inputPath;
string outputPath;
auto input = std::make_shared<FileStreamDevice>(inputPath);
FileStreamDevice output(outputPath, FileMode::Create);
input->CopyTo(output);
PdfMemDocument doc;
doc.LoadFromDevice(input);
doc.SaveUpdate(output);
Sign a buffer loaded document:
bufferview inputBuffer;
charbuff outputBuffer;
auto input = std::make_shared<SpanStreamDevice>(inputBuffer);
BufferStreamDevice output(outputBuffer);
input->CopyTo(output);
PdfMemDocument doc;
doc.LoadFromDevice(input);
// Retrieve/create the signature, create the signer, ...
PoDoFo::SignDocument(doc, output, signer, signature);
Q: Can I sign a document a second time?
A: Yes, this is tested, but to make sure this will work you'll to re-parse the document a second time, as re-using the already loaded document is still untested (this may change later). For example do as it follows:
bufferview inputBuffer;
charbuff outputBuffer;
auto input = std::make_shared<SpanStreamDevice>(inputBuffer);
BufferStreamDevice output(outputBuffer);
input->CopyTo(output);
{
PdfMemDocument doc;
doc.LoadFromDevice(input);
// Retrieve/create the signature, create the signer, ...
PoDoFo::SignDocument(doc, output, signer, signature);
}
input = std::make_shared<SpanStreamDevice>(output);
{
PdfMemDocument doc;
doc.LoadFromDevice(input);
// Retrieve/create the signature, create the signer, ...
PoDoFo::SignDocument(doc, output, signer, signature);
}
PoDoFo may or may not work for your needs and comes with absolutely no warranty. Serious bugs, including security flaws, may be fixed at arbitrary timeframes, or not fixed at all.
Please subscribe to the project mailing list which is still followed by several of the original developers of PoDoFo. A gitter community has also been created to ease some more informal chatter. If you find a bug and know how to fix it, or you want to add a small feature, you're welcome to send a pull request, providing it follows the coding style of the project. As a minimum requisite, any contribution should be:
- valuable for a multitude of people and not only self relevant for the contributor;
- consistent with surrounding code and not result in unpredictable behavior and/or bugs.
Other reasons for the rejection, or hold, of a pull request may be:
- the change doesn't fit the scope of PoDoFo;
- the change shows lack of knowledge/mastery of the PDF specification and/or C++ language;
- the change breaks automatic tests performed by the maintainer;
- general lack of time in reviewing and merging the change.
If you need to implement a bigger feature or refactor, ask first if it was already planned. The feature may be up for grabs, meaning that it's open for external contributions. Please write in the relevant issue that you started to work on that, to receive some feedback/coordination. If it's not, it means that the refactor/feature is planned to be implemented later by the maintainer(s). If the feature is not listed in the issues, add it and/or create a discussion to receive some feedback and discuss some basic design choices.
PoDoFo is currently developed and maintained by Francesco Pretto, together with Dominik Seichter and others. See the file AUTHORS.md for more details. Please don't use personal emails for technical support requests, but create github issues instead.