user-tools/tabularize

   1 #!/bin/bash
   2
   3 true <<EOF
   4 =pod
   5
   6 =head1 NAME
   7
   8 tabularize - Takes TAB-delimited lines of text and outputs formatted table.
   9
  10 =head1 SYNOPSIS
  11
  12 I<COMMAND> | tabularize [I<OPTIONS>]
  13
  14 =head1 OPTIONS
  15
  16 =over 4
  17
  18 =item -a, --ascii
  19
  20 7-bit ascii borders
  21
  22 =item -u, --unicode
  23
  24 borders with nice graphical chars
  25
  26 =item -H, --no-horizontal
  27
  28 no horizontal lines in the output
  29
  30 =item -M, --no-margins
  31
  32 no margins, ie. no right-most and left-most vertical borders
  33
  34 =item -v, --output-vertical-separator I<CHAR>
  35
  36 vertical separator character(s) in the output
  37
  38 =item -r, --align-right I<NUM>
  39
  40 align these columns (0-indexed) to the right,
  41 others are auto-detected and if deemed to contain numeric data,
  42 only then aligned to the right.
  43 this option is repeatable.
  44
  45 =item -l, --align-left I<NUM>
  46
  47 similar to --align-right option
  48
  49 =back
  50
  51 =head1 ENVIRONMENT
  52
  53 =over 4
  54
  55 =item PAGER
  56
  57 If B<$PAGER> is set and standard output is a terminal
  58 and the resulting table is wider than the terminal,
  59 then pipe the table through B<$PAGER>.
  60
  61 =back
  62
  63 =cut
  64
  65 EOF
  66
  67
  68 TAB=$'\t'
  69 IFS=''
  70
  71 set_ascii()
  72 {
  73         verticalBar='|'
  74         horizontalBar='-'
  75         topleftCorner='+'
  76         toprightCorner='+'
  77         bottomleftCorner='+'
  78         bottomrightCorner='+'
  79         leftCross='+'
  80         middleCross='+'
  81         rightCross='+'
  82         topCross='+'
  83         bottomCross='+'
  84 }
  85 set_unicode()
  86 {
  87         verticalBar='│'
  88         horizontalBar='─'
  89         topleftCorner='┌'
  90         toprightCorner='┐'
  91         bottomleftCorner='└'
  92         bottomrightCorner='┘'
  93         leftCross='├'
  94         middleCross='┼'
  95         rightCross='┤'
  96         topCross='┬'
  97         bottomCross='┴'
  98 }
  99
 100
 101 set_unicode
 102
 103 declare -a columnAlignment=()
 104 horizontalLines=yes
 105 margins=yes
 106
 107 while [ $# -gt 0 ]
 108 do
 109 case "$1" in
 110 -a|--ascii)
 111         set_ascii
 112         ;;
 113 -u|--unicode)
 114         set_unicode
 115         ;;
 116 -H|--no-horizontal)
 117         horizontalLines=''
 118         ;;
 119 -M|--no-margins)
 120         margins=''
 121         ;;
 122 -r|--align-right)
 123         shift
 124         columnAlignment[$1]=right
 125         ;;
 126 -l|--align-left)
 127         shift
 128         columnAlignment[$1]=left
 129         ;;
 130 -v|--output-vertical-separator)
 131         shift
 132         verticalBar=$1
 133         topleftCorner=$1
 134         toprightCorner=$1
 135         bottomleftCorner=$1
 136         bottomrightCorner=$1
 137         leftCross=$1
 138         middleCross=$1
 139         rightCross=$1
 140         topCross=$1
 141         bottomCross=$1
 142         ;;
 143 -h|--help)
 144         pod2text "$0"
 145         exit
 146         ;;
 147 -*)
 148         echo "$0: unknown option: $1" >&2
 149         exit -1
 150         ;;
 151 esac
 152 shift
 153 done
 154
 155
 156 explode()
 157 {
 158         local separator=$1
 159         local string=$2
 160         local new_string
 161         declare -g -a EXPLODE_ARRAY=()
 162
 163         while [ -n "$string" ]
 164         do
 165                 EXPLODE_ARRAY+=("${string%%$separator*}")
 166                 new_string=${string#*$separator}
 167                 if [ "$string" = "$new_string" ]
 168                 then
 169                         break
 170                 fi
 171                 string=$new_string
 172         done
 173
 174         # return $EXPLODE_ARRAY
 175 }
 176
 177 is_numeric_check_sep()
 178 {
 179         local thousands_sep=$2
 180         local fraction_sep=$3
 181         [[ $1 =~ ^[+-]?[0-9]+($thousands_sep[0-9]+)*($fraction_sep[0-9]+($thousands_sep[0-9]+)*)?$ ]]
 182 }
 183 is_numeric()
 184 {
 185         is_numeric_check_sep "$1" ',' . && return 0
 186         is_numeric_check_sep "$1" ' ' , && return 0
 187         is_numeric_check_sep "$1" '.' , && return 0
 188         return 1
 189 }
 190
 191
 192 # compute width of each column
 193
 194 column_width=()
 195 num_line=0
 196 text=''
 197 numericals_by_col=()
 198 non_numericals_by_col=()
 199 while read -r line
 200 do
 201         text="$text${text:+$'\n'}$line"
 202         explode $'\t' "$line"
 203
 204         column=0
 205         for cell in "${EXPLODE_ARRAY[@]}"
 206         do
 207                 width=${#cell}
 208                 if [ -z "${column_width[$column]}" ] || [ $width -gt "${column_width[$column]}" ]
 209                 then
 210                         column_width[$column]=$width
 211                 fi
 212
 213                 # guess if this cell has numerical content (skip 1st line as it's likely a header)
 214                 if [ $num_line -gt 0 -a -n "$cell" -a -z "${columnAlignment[$column]:-}" ]
 215                 then
 216                         if is_numeric "$cell"
 217                         then
 218                                 let numericals_by_col[$column]++
 219                         else
 220                                 let non_numericals_by_col[$column]++
 221                         fi
 222                 fi
 223                 let column++
 224         done
 225
 226         let num_line++
 227 done
 228
 229 if [ $num_line = 0 ]
 230 then
 231         # no input, no output
 232         exit 0
 233 fi
 234
 235
 236 set -u
 237
 238 # compose the format string
 239
 240 table_width=0
 241 cellFmt=''
 242 topFmt=''
 243 innerFmt=''
 244 bottomFmt=''
 245 segments=()
 246 sequentColumn=''
 247 for column in "${!column_width[@]}"
 248 do
 249         if [ -z "${columnAlignment[$column]:-}" ]
 250         then
 251                 if [ ${numericals_by_col[$column]:-0} -ge ${non_numericals_by_col[$column]:-0} ]
 252                 then
 253                         columnAlignment[$column]=right
 254                 else
 255                         columnAlignment[$column]=left
 256                 fi
 257         fi
 258
 259         table_width=$[table_width + (column == 0 ? (0${margins:+1} ? ${#verticalBar} : 0) : ${#verticalBar}) + ${column_width[$column]}]
 260
 261         cellFmt="${cellFmt:-${margins:+$verticalBar}}${sequentColumn:+$verticalBar}%*s"
 262         topFmt="${topFmt:-${margins:+$topleftCorner}}${sequentColumn:+$topCross}%${column_width[$column]}s"
 263         innerFmt="${innerFmt:-${margins:+$leftCross}}${sequentColumn:+$middleCross}%${column_width[$column]}s"
 264         bottomFmt="${bottomFmt:-${margins:+$bottomleftCorner}}${sequentColumn:+$bottomCross}%${column_width[$column]}s"
 265         segments+=('')
 266
 267         sequentColumn=yes
 268 done
 269 table_width=$[table_width + (0${margins:+1} ? ${#verticalBar} : 0)]
 270 cellFmt="$cellFmt${margins:+$verticalBar}"
 271 topFmt="$topFmt${margins:+$toprightCorner}"
 272 innerFmt="$innerFmt${margins:+$rightCross}"
 273 bottomFmt="$bottomFmt${margins:+$bottomrightCorner}"
 274
 275
 276 topGrid=''
 277 innerGrid=''
 278 bottomGrid=''
 279 if [ $horizontalLines ]
 280 then
 281         topGrid=`printf -- "$topFmt" "${segments[@]}"`
 282         topGrid=${topGrid// /$horizontalBar}$'\n'
 283         innerGrid=`printf -- "$innerFmt" "${segments[@]}"`
 284         innerGrid=${innerGrid// /$horizontalBar}$'\n'
 285         bottomGrid=`printf -- "$bottomFmt" "${segments[@]}"`
 286         bottomGrid=${bottomGrid// /$horizontalBar}$'\n'
 287 fi
 288
 289
 290 print_table_g()
 291 {
 292         line_no=1
 293         while read -r line
 294         do
 295                 if [ $line_no = 1 ]
 296                 then
 297                         echo -n "$topGrid"
 298                 else
 299                         echo -n "$innerGrid"
 300                 fi
 301
 302                 explode $'\t' "$line"
 303                 printf_args=()
 304                 for column in "${!column_width[@]}"
 305                 do
 306                         [ $column -lt ${#EXPLODE_ARRAY[@]} ] && cell=${EXPLODE_ARRAY[$column]} || cell=''
 307                         chars=${#cell}
 308                         bytes=`LANG=C; echo "${#cell}"`
 309                         swidth=$[ ${column_width[$column]} + ( $bytes - $chars ) ]
 310                         [ ${columnAlignment[$column]} = left ] && swidth=-$swidth
 311                         printf_args+=($swidth "$cell")
 312                 done
 313                 printf -- "$cellFmt\n" "${printf_args[@]}"
 314
 315                 let line_no++
 316         done <<<"$text"
 317
 318         echo -n "$bottomGrid"
 319 }
 320
 321
 322 if [ -t 1 -a -n "$PAGER" ]
 323 then
 324         terminal_cols=`tput cols`
 325         if [ "${terminal_cols:-0}" -le $table_width ]
 326         then
 327                 print_table_g | exec "$PAGER"
 328                 exit ${PIPESTATUS[1]}
 329         fi
 330 fi
 331
 332 print_table_g