This is a C++ implementation of John Cowan's tagsoup HTML parser.
It is more than a translation, because I removed as much of the generic framework as I could, so this code is only useful for converting HTML into XHTML. I also tried to be ruthlessly efficient. Part of that includes (manually) alias'ing pointers, handing off ownership across class boundaries with impunity, and so on. Therefore, if you try to use this code in other packages, make sure you understand the lifetimes of the various pointers; valgrind is your friend here.
Thanks to John Cowan for the original code. Any bugs are mine. The brilliance of TagSoup is in its tables and scanner callbacks. It should therefore not be surprising that that code is basically unchanged, while the scaffolding had the major surgery.
The code in schema.hpp needs a hashtable package. I used "lhash" from OpenSSL, with a few patches so that it can be used standalone (i.e., without the rest of OpenSSL). The "getlhash" script fetches the files from the OpenSSL github repository and patches them.
The rest of the code is distributed under the Apache 2.0 license. It is derived from a version of tagsoup which was obtained under the same license.
Rich Salz rich.salz@gmail.com October, 2012