====================================================================== PolyglotMan (nee RosettaMan) is a filter for UNIX manual pages. It takes as input man pages for a variety of UNIX flavors and produces as output a variety of file formats. Currently PolyglotMan accepts man pages from the following flavors of UNIX: Hewlett-Packard HP-UX, AT&T System V, SunOS, Sun Solaris, OSF/1, DEC Ultrix, SGI IRIX, Linux, SCO, FreeBSD; and produces output for the following formats: printable ASCII only (stripping page headers and footers), section and subsection headers only, TkMan, [tn]roff, RTF, SGML (soon--I finally found a DTD), HTML, MIME, LaTeX, LaTeX 2e, Perl 5's pod. Previously <I>PolyglotMan</I> required pages to be formatted by nroff prior to its processing; with version 3.0, it prefers [tn]roff source and usually can produce results that are better yet. PolyglotMan improves upon other man page filters in several ways: (1) its analysis recognizes the structural pieces of man pages, enabling high quality output, (2) its modular structure permits easy augmentation of output formats, (3) it accepts man pages formatted with the variant macros of many different flavors of UNIX, and (4) it doesn't require modification of or cooperation with any other program. PolyglotMan is a rewrite of TkMan's man page filter, called bs2tk. (If you haven't heard about TkMan, a hypertext man page browser, you should grab it via anonymous ftp from ftp.cs.berkeley.edu: /ucb/people/phelps/tkman.tar.Z.) Whereas bs2tk generated output only for TkMan, PolyglotMan generalizes the process so that the analysis can be leveraged to new output formats. A single analysis engine recognizes section heads, subsection heads, body text, lists, references to other man pages, boldface, italics, bold italics, special characters (like bullets), tables (to a degree) and strips out page headers and footers. The engine sends signals to the selected output functions so that an enhancement in the engine improves the quality of output of all of them. Output format functions are easy to add, and thus far average about about 75 lines of C code each. A note for HTML consumers: This filter does real (heuristic) parsing-- no <PRE>! Man page references are turned into hypertext links. The files <URL:ftp://ftp.cs.berkeley.edu/ucb/people/phelps/tcltk/sgi-ls.1.html> and <URL:ftp://ftp.cs.berkeley.edu/ucb/people/phelps/tcltk/ksh.1.html> are examples of the quality of output produced entirely automatically (no retouching) by PolyglotMan. These translations were produced by PolyglotMan starting with the [tn]roff source (again no retouching). Several people have extended World Wide Web servers to format man pages on the fly. Check the README file in the contrib directory for a list. CHANGES in 3.0 * [tn]roff source preferred for superior results, when roff macros are sufficiently recognized. Autodetection of source or formatted input. * New software license that makes it free for any use CHANGES in 2.5 * SGML output format that adheres to Davenport DocBook v2.3 DTD (NOT READY IN CURRENT VERSION!) * MIME output format, for e-mail and Emacs 19.29's enriched mode (Neal Becker) * port to Macintosh by Matthias Neeracher * list of valid volume names can be given as a parameter (Dag Nygren) * updated to LaTeX2e (H. Palme) * debugging scaffolding erected (at the end of software's development cycle!) CHANGES in 2.2 * when in SEE ALSO, hyphens would confuse man page-reference finder, so re-linebreak if necessary to eliminate them (!) (Greg Earle & Uri Guttman) CHANGES in 2.1 * gets() replaced by custom code. gets() deprecated since it reads until \0, introducing security problems. (Robert Withrow) * TkMan module revised for Tk 4.0