sloccount - count source lines of code (SLOC)
sloccount [--version] [--cached] [--append] [ --datadir directory ]
[--follow] [--duplicates] [--crossdups] [--autogen] [--multiproject]
[--filecount] [--wide] [--details] [ --effort F E ] [ --schedule F E ]
[ --personcost cost ] [ --overhead overhead ] [ --addlang language ] [
--addlangall ] [--] directories
sloccount counts the physical source lines of code (SLOC) contained in
descendants of the specified set of directories. It automatically
determines which files are source code files, and it automatically
determines the computer language used in each file. By default it
summarizes the SLOC results and presents various estimates (such as
effort and cost to develop), but its output can be controlled by
If you give sloccount a list of two or more directories, the counts
will be broken up according to that list. There is one important
limitation: the basenames of the directories given as parameters must
be different, because those names are used to group summary
information. Thus, you can’t run "sloccount /usr/src/hello
/usr/local/src/hello". This limitation only applies to parameters of
sloccount - subdirectories descended from the top directories can have
the same basename.
If you give sloccount only a single directory, sloccount tries to
automatically find a reasonable breakdown for purposes of reporting (so
it’ll produce a useful report). In this case, if the directory has at
least two subdirectories, then those subdirectories will be used as the
breakdown. If the single directory contains files as well as
directories (or if you give sloccount some files as parameters), those
files will be assigned to the directory "top_dir" so you can tell them
apart from other contents. Finally, if there’s a subdirectory named
"src", then that subdirectory is again broken down, with all the
further subdirectories prefixed with "src_". Thus, if directory "X"
has a subdirectory "src", which contains subdirectory "modules",
sloccount will report a separate count for "src_modules".
sloccount normally considers all descendants of these directories,
though unless told otherwise it ignores symbolic links.
sloccount is the usual front-end of the package of tools named
"SLOCCount". Note that the name of the entire package has capital
letters, while the name of this front-end program does not.
sloccount will normally report estimates of schedule time, effort, and
cost, and for single projects it also estimates the average number of
active developers. These are merely estimates, not etched in stone;
you can modify the parameters used to improve the estimates.
--version Report the version number of SLOCCount and immediately
exit. This option can’t be usefully combined with any
--cached Do not recalculate; instead, use cached results from a
previous execution. Without the --cached or --append
option, sloccount automatically removes the data directory
and recreates it.
--append Do not remove previous calculations from the data
directory; instead, add the analysis to the current
contents of the data directory.
Store or use cached data in the given data directory;
default value is "~/.slocdata".
--follow Follow symbolic links.
Count all duplicates. Normally, if files have equal
content (as determined using MD5 hash values), only one is
--crossdups Count duplicates if they occur in different portions of the
breakdown. Thus, if the top directory contains many
different projects, and you want the duplicates in
different projects to count in each project, choose this
--autogen Count source code files that appear to be automatically
generated. Normally these are excluded.
The different directories represent different projects;
otherwise, it’s assumed that all of the source code belongs
to a single project. This doesn’t change the total number
of files or SLOC values, but it does affect the effort and
schedule estimates. Given this option, effort is computed
separately for each project (and then summed), and the
schedule is the estimated schedule of the largest project.
--filecount Display counts of files instead of SLOC.
--wide Display in the "wide" (tab-separated) format.
--details Display details, that is, results for every source code
--effort F E
Change the factor and exponent for the effort model.
Effort (in person-months) is computed as F*(SLOC/1000)^E.
--schedule F E
Change the factor and exponent for the schedule model.
Schedule (in months) is computed as F*(effort)^E.
Change the average annual salary to cost.
Change the overhead value to overhead. Estimated cost is
computed as effort * personcost * overhead.
Add a language not considered by default to be a
‘‘language’’ to be reported. Currently the only legal
values for language are "makefile", "sql", and "html".
These files are not normally included in the SLOC counts,
although their SLOCs are internally calculated and they are
shown in the file counts. If you want to include more than
one such language, do it by passing --addlang more than
once, e.g., --addlang makefile --addlang sql.
Add all languages not normally included in final reports.
As with many other programs using Unix-like options, directories whose
names begin with a dash (‘‘-’’) can be misinterpreted as options. If
the directories to be analyzed might begin with a dash, use the double-
dash (‘‘- -’’) to indicate the end of the option list before listing
Filenames with embedded newlines (in the directories or their
descendants) won’t be handled correctly; they will be interpreted as
separate filenames where the newlines are inserted. An attacker could
prevent sloccount from working by creating filenames of the form
/normal/directory ... NEWLINE/dev/zero. Such filenames are exceedingly
rare in source code because they’re a pain to work with using other
tools, too. Future versions of sloccount may internally use NUL-
separated filenames (like GNU find’s -print0 command) to fix this.
There are many more languages not yet handled by SLOCCount.
SLOCCount only reports physical source lines of code. It would be very
useful if it could also report logical lines of code, and perhaps other
common metrics such as McCabe’s complexity measures and complexity
density (complexity/SLOC for each function or procedure).
See the SLOCCount website at http://www.dwheeler.com/sloccount. Note
that more detailed documentation is available both on the website and
with the SLOCCount package.
David A. Wheeler (firstname.lastname@example.org).