The problem is how to split a texi file into multiple html
files, one per mode, in a single pass.
We have two conflicting goals:
* Avoding funny characters in file names, if possible.
This implies that we cannot use a non-clashing mapping,
since that would create unnecessarily ugly filenames.
* Dealing with nodes with similar but different names
that map to the same filename. These must be made distinct.
* We want to process the texinfo file in a single pass.
Solution: Maintain a table mapping node names to file names.
We call this the file name table. When we see a mode name (in
a menu, xref, or @node header), look for it in the file name
table. If it is found, use the file name in the table for urls.
If not, generate a file name. Try first the simple aesthetic
mapping, where all "funny" characters are mapped to '-'. If there
is no entry yet in the file name table for a node that maps to that
filename, enter the filename in the table, and use it. If there
is a clash, then generate a unique name in some other way.
One way is to use the standard url mangling (i.e. spaces are
mapped to '%20'). Another is to append an integer (i.e. look for
"foo1', then 'Foo2', etc). A third is to use a more complex
mapping - e.g. 'Foo*' gets mapped to 'Foo_star'.
--
--Per Bothner
[EMAIL PROTECTED] http://www.bothner.com/per/
_______________________________________________
Bug-texinfo mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-texinfo
