Bash Script to do quick counts of common source code types under the current or specified directory

dgSniffCode.sh

#!/bin/bash
# Bash Script to do quick counts of common source code types under the current or specified directory
# https://gist.github.com/3901863

# To quickly grab it, just type
# wget https://gist.github.com/raw/3901863/dgSniffCode.sh && chmod +x dgSniffCode.sh

# Features to add: 
#   multiple params ...A hack until I support * arguments:  ls | xargs -L 1 ~/dgSniffCode

# Colors for use with echo -e   -- see http://tldp.org/LDP/abs/html/colorizing.html
cBlueHi='\E[37;44m'"\033[1m"
cNormal="\033[0m"
cGreenLo='\E[40;32m'

disableColors()
{
  cBlueHi=""
  cNormal=""
  cGreenLo=""
}

summaryText=''
summaryOnly=0          # 1 if we only want the summary of non-test code

usage ()
{
  echo $0: 
  echo Bash Script to do quick counts of common source code 
  echo types under the current or specified directory
  echo " "
  echo Usage: 
  echo "  $0 -help             Show this text"
  echo "  $0 [-[-s]              Run the scan in the current directory"
  echo "  $0 [-s] [directory]  Run the scan in the specified directory"
}

addToSummary()
{
  # $1 = language, eg "Ruby"
  # $2 = count, eg 12121
  if [ ! $2 -eq 0 ]
  then
    printf -v summaryText2 "%s  %'d lines of %s\n" "$summaryText" "$2" "$1"
    summaryText=$summaryText2
  fi
}

printSummary()
{
  printf "${cBlueHi} Summary of non-test code lines: $cNormal\n"
  sortedSummary=`echo "$summaryText" | sort -n -r`
  printf "%s\n" "$sortedSummary"
}

countBadness ()
{
  # $1 = suffix being checked
  # $2 (option) = exclude files with this line
  flag=0
  for bad in TODO FIXME XXX '@todo' HACK hack 'sleep'
  do
    if [ -z $2 ]
    then
      badcount=`find . -name \*.$1 -exec cat '{}' \; | grep $bad | wc -l`
    else
      if [ $3 = "include" ]
      then 
        badcount=$(cat $(find . -name \*.$1 -print | grep $2) | grep $bad | wc -l)
      else
        badcount=$(cat $(find . -name \*.$1 -print | grep -v $2) | grep $bad | wc -l)
      fi
    fi
    if [ ! $badcount -eq 0 ]
    then 
      echo -en "    $cGreenLo $badcount $bad $cNormal"
      flag=1
    fi
  done
  if [ $flag -eq 1 ]
  then
    echo " "
  fi
}

countLines ()
{
  dgFiles=`find . -name \*.$2 -print | wc -l`
  if [ ! $dgFiles -eq "0" ]
  then
    if [ $summaryOnly -eq 0 ]
    then
      echo -e $cBlueHi $1: $cNormal
    fi
    # $3 is unit/test/spec indicator, eg /spec/
    if [ -z $3 ]
    then
      # "No spec"
      dgAcrossDirs=`find . -name \*.$2 -execdir pwd \; | uniq | wc -l`
      dgLines=`find . -name \*.$2 -exec cat '{}' \; | wc -l`
      addToSummary $1 $dgLines
      if [ $summaryOnly -eq 0 ]
      then
        printf "  %'d dgFiles $1 files across %'d directories\n" $dgFiles $dgAcrossDirs
        printf "  %'d total lines of $1" $dgLines
        printf " (Average lines per file = $(($dgLines/$dgFiles)))\n"
        countBadness $2
        printf "\n"
      fi
    else
      # "spec" provided
      # First the non-spec files...
      dgFiles2=`find . -name \*.$2 -print | grep -v $3 | wc -l`
      dgAcrossDirs=`find . -name \*.$2 -execdir pwd \; | grep -v $3 | uniq | wc -l`
      if [ $dgFiles2 -eq 0 ]
      then
        dgLines2=0
      else
        dgLines2=$(cat $(find . -name \*.$2 -print | grep -v $3) | wc -l)
      fi
      addToSummary $1 $dgLines2
      if [ $summaryOnly -eq 0 ]
      then
        printf "  %'d Non-test $1 files across %'d directories\n" $dgFiles2  $dgAcrossDirs
        printf "  %'d total lines of Non-Test $1" $dgLines2
        printf " (Average Lines per file = $(($dgLines2/$dgFiles2)))\n"
        countBadness $2 $3 exclude

        # Now the spec files
        dgAcrossDirs=`find . -name \*.$2 -execdir pwd \; | grep $3 | uniq | wc -l`
        dgFiles3=`find . -name \*.$2 -print | grep $3 | wc -l`
        printf "  + %'d TEST $1 files across $dgAcrossDirs directories\n" $dgFiles3
        if [ ! $dgFiles3 -eq "0" ]
        then
          dgLines3=$(cat $(find . -name \*.$2 -print | grep $3) | wc -l)
          printf "  + %'d total lines of TEST $1" $dgLines3
          printf " (Average Lines per file = $(($dgLines3/$dgFiles3)))\n"
          printf "  + $(((100*$dgLines3/($dgLines2+dgLines3)))) percent of $1 code is TEST $1\n"
          countBadness $2 $3 include
        fi
      printf "\n"
      fi
    fi
  fi
}

if [ "$1" = '-help' ]
then 
  usage
  exit
fi

if [ "$1" = '-m' ]
then 
  disableColors
  shift
fi


if [ "$1" = '-s' ]
then 
  summaryOnly=1
  shift
fi

if [ ! -z $1 ]
then
 pushd $1 > /dev/null
 dgNeedToPopd="yes"
fi

echo "=== Quick Code sniff of $(pwd) ==="

tstfilter='/spec/\|/test/\|/tests/\|/Test/\|/Tests/\|/app-test/\|/darwin-test\|-test-'

countLines Ruby rb $tstfilter
countLines Yml yml $tstfilter
countLines eRuby erb $tstfilter
countLines Javascript js $tstfilter 
countLines Flex mxml $tstfilter 
countLines ActionScript as $tstfilter 

countLines Java java $tstfilter
countLines JSP jsp $tstfilter
countLines HTML html $tstfilter
countLines CSS css $tstfilter
countLines Compass\(sCSS\) scss $tstfilter

countLines Scala scala $tstfilter
countLines Jade jade $tstfilter
countLines Shell sh $tstfilter
countLines C c $tstfilter
countLines Python py $tstfilter
countLines Go go _test.go
countLines Xml xml $tstfilter
countLines JSON json $tstfilter

if [ ! -z $dgNeedToPopd ]
then 
  popd > /dev/null
fi

printSummary
printf "\n"