manlifter man page

manlifter — mass-conversion script and test harness for doclifter


manlifter [-d option] [-e] [-f listfile] [-h] [-I mandir] [-m] [-M] [-o outdir] [-p patch-directory] [-P] [-q] [-v] [-s section] [-X exclude] name...

manlifter [-S]


manlifter is a script that sequences doclifter(1) to convert an entire manual-page tree to XML-Docbook, optionally also generating HTML from the XML. Another use is as a torture-test tool for doclifter; it logs errors to standard output and collects timings.

Called without any file arguments, manlifter tries to convert all eligible man pages installed on the system, placing the resulting xml files under xmlman in the current directory. Each successfully translated page foo.N is copied to manN/foo.xml beneath the output directory, regardless of what source directory it came from.

A manual page is considered ineligible for batch conversion if it contains text indicating it has been generated from DocBook masters of from Doxygen.

For each source file examined, if the destination file exists and is newer than the source, the conversion is skipped; thus, incremental runs of manlifter do the least work needed to keep the target XML tree up to date. Likewise, in -h mode derived HTML files are only made when necessary.

Stub pages that are just .so redirections are translated to corresponding symlinks of XML files (and, with -h, HTML files).

manlifter may also be called with a single file argument, which is interpreted as the stem name of a potential manual page. manlifter then searches all selected manual sections for a matching page and attempts to convert it. In this case, a copy of the man page and the converted version are dropped immediately beheath the output directory, with the names foobar.man and foobar.man.xml, respectively. This mode is normally only of interest only to doclifter developers for debugging that program.

In either of the above cases, manlifter will uncompress the file if it has a .gz, .bz2 or .Z suffix on the name.

Options are as follows:


Pass the string argument to each doclifter call as options. Each space-separated token in the string becomes a separate argument in the call.


Run in log-filter mode (mainly of interest to doclifter developers). In this mode, manlifter reads a test log from standard input and filters it in a a way dependent on the -f and -q options. If neither of these is given, messages from successful runs are stripped out and only errors passed through to standard output.


Normally, run doclifter on the files named by each line in the argument file. In error-filter mode the argument is instead interpreted as a filtering regular expression.


Also generate HTML translations into the output directory. DocBook citerefentry markup is transformed to hyperlinks in the directory, and a contents listing is generated to index.html.


Specify the root of the manual-page tree. By default this is /usr/share/man.


Make a patch to correct the last page fetched. It is copied, an editor is called on the copy (using the environment variable $EDITOR), and then diff(1) is called to drop the patch in the prepatch directory. Fails with an error if such a patch is already present.


Lift the specified files, then do the equivalent of the -m option.


Set the output directory into which XML-DocBook translations will be dropped. By default this is xmlman under the current directory in batch mode, or the current directory otherwise.


Interpret the argument as the name of a patch directory (the default name is prepatch under the current directory). Each file named foo.N.patch is interpreted as a patch to be applied to the manual page foo(N) before doclifter translates it.


Enable profiling using the Python hotshot module; this is only useful for tuning doclifter so it runs faster. Raw data is written to manlifter.prof, and a digested report is appended to the log on standard output. Warning: the raw data files can become huge, and the postprocessing for report generation can take as long as the actual processing (or longer!).


Normally, pass the -q (quiet) option to each doclifter call. In error-filter mode, return a list of files on which translation failed.


Pass the -v (verbose) option to each doclifter call. This option can be repeated to increase the verbosity level.


Specify a section to scan. Use this with an argument; it should not be necessary when doing a conversion of the entire tree.


Compile error statistics from a manlifter logfile presented on standard input. This option will be of interest mainly to doclifter developers.


In batch mode exclude pages listed in the argument file. Meant to be used for pages that are known good and take an extremely long time to lift, in order to cut down the time for a test run. (Most pages lift in less than a half second, but a few can take 15 minutes or longer.)

manlifter emits a logfile to standard output. The file begins with a timestamp line and a blank line, and ends with a line giving run time and various interesting statistics. Between these are stanzas, separated by blank lines, one for each file on which doclifter was run.

The first line of each stanza beguns with "! ", followed by the pathname of the source manual pager, followed by "=" and the return status of doclifter run on that file. Following that is a space and doclifter's runtime in seconds.

This initial line may be followed by information messages and the error output of the doclifter run.

manlifter must find a copy of doclifter in either the current directory or one of the command directories in your PATH in order to run.


HTML generation is painfully slow. Unfortunately, there is little we can do to remedy this, because XSLT engines are painfully slow.

See Also

doclifter(1), xmlto(1)


Eric S. Raymond <esr@thyrsus.com>

There is a project web page at http://www.catb.org/~esr/doclifter/.


Explore man page connections for manlifter(1).