Linux utility to gather statistics about a codebase

Question

Suppose I have some software project I'm working on, and I want to compute some statistics about it:

• Number of lines (total & per file)
• Comment, spacing and other lines
• Numbers of functions, classes, methods
• Average length of a file, a function, a class, a method
• Cyclomatic complexity
• Dependency DAG depths for files, types etc.

I'll settle for only some of these, of course - the more the merrier. I do not guarantee the repository has any form of source control (although personally I'm working with mercurial right now); and it's not a clone of an online repository (so nothing like gitstats).

Requirements:

• Libre
• Gratis
• Works on Linux (AMD x86_64 CPUs)
• Supports at the very least C, C++98, C++11
• Support for multiple output formats
• Can produce output as console "ASCII-art" charts (a-la, say, hg diff --stat)

Desirable feature:

• Small standalone utility
• Not too many dependencies on other programs and libraries
• Fast

Answer 1

CLOC - Count Lines of Code - does some (not all) of those things, meeting all the requirements. It's also fast, small, and depends only on perl and a couple of perl libs.

Sample outout:

http://cloc.sourceforge.net v 1.60  T=0.46 s (354.6 files/s, 64728.0 lines/s)
--------------------------------------------------------------------------------
Language                      files          blank        comment           code
--------------------------------------------------------------------------------
C/C++ Header                    108           2761           2452          10226
C++                              47           1626            772           9703
make                              1            476            193            931
Bourne Again Shell                3             50             25            346
CMake                             2             48             68            225
XML                               1              0              0             11
YAML                              1              1              6              8
Bourne Shell                      1              1              2              5
--------------------------------------------------------------------------------
SUM:                            164           4963           3518          21455
--------------------------------------------------------------------------------

Answer 2

There is a cross platform python metrics library that gives you metrics either on a single file, or can recurse a directory tree. It is available from the original project or my fork adds recursive option on github.

Running on an old project that I have to hand with the recurse option:

Files                       Language        SLOC Comment McCabe
----- ------------------------------ ----------- ------- ------
 102                            XML       10601    2415      0
   7                            VGL        4837      14     49
 588                              C      441797  135164  11715   
 302                              D        7750       0      0    
  27                           XSLT         691     313      0    
  13                         Python         881     203    141
   4                        XML+PHP       21489       0      0
  38                       Makefile        1747    3948     18
   7                      Batchfile          45       8      7
   2                           HTML           9      36      0    
  17                      Text only           0       0      0
   1                         POVRay           8      19      0
----- ------------------------------ ----------- ------- ------
1108                          Total      489855  142120  11930

• Libre Yes
• Gratis Yes
• Works on Linux (AMD x86_64 CPUs) Yes it's python
• Supports at the very least C, C++98, C++11 Yes
• Support for multiple output formats Yes Can output as text, xml or csv
• Can produce output as console "ASCII-art" charts (a-la, say, hg diff --stat) Yes See above
• Small standalone utility Small, (tiny), python library that adds command line utility
• Not too many dependencies on other programs and libraries Depends on python, its standard libraries & pygments
• Fast Reasonably so