Checkpoint lwcc development Changed tactics with the preprocessor. Instead of getting clever and trying to do things the "fast" way, instead, just tokenize the whole input and process it that way. Also, set up so the preprocessor and compiler can be integrated instead of having to have a specifically correct output for the preprocessed file. Also removed the subdirectories in the lwcc directory. It made things more complicated than they needed to be.

all:
	hg rm manual/* || true
	rm -f manual/*.pdf manual/*.html
	docbook2html -o manual manual.docbook.sgml
	docbook2html -u manual.docbook.sgml && mv manual.docbook.html manual/manual.html
	docbook2pdf -u manual.docbook.sgml && mv manual.docbook.pdf manual/manual.pdf
	rm -f manual.docbook.html
	hg add manual/*