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 =head1 SEE ALSO
  64
  65 column(1), untabularize(1)
  66
  67 =cut
  68
  69 EOF
  70
  71
  72 TAB=$'\t'
  73 IFS=''
  74
  75 set_ascii()
  76 {
  77         verticalBar='|'
  78         horizontalBar='-'
  79         topleftCorner='+'
  80         toprightCorner='+'
  81         bottomleftCorner='+'
  82         bottomrightCorner='+'
  83         leftCross='+'
  84         middleCross='+'
  85         rightCross='+'
  86         topCross='+'
  87         bottomCross='+'
  88 }
  89 set_unicode()
  90 {
  91         verticalBar='│'
  92         horizontalBar='─'
  93         topleftCorner='┌'
  94         toprightCorner='┐'
  95         bottomleftCorner='└'
  96         bottomrightCorner='┘'
  97         leftCross='├'
  98         middleCross='┼'
  99         rightCross='┤'
 100         topCross='┬'
 101         bottomCross='┴'
 102 }
 103
 104
 105 set_unicode
 106
 107 declare -a columnAlignment=()
 108 horizontalLines=yes
 109 margins=yes
 110
 111 while [ $# -gt 0 ]
 112 do
 113 case "$1" in
 114 -a|--ascii)
 115         set_ascii
 116         ;;
 117 -u|--unicode)
 118         set_unicode
 119         ;;
 120 -H|--no-horizontal)
 121         horizontalLines=''
 122         ;;
 123 -M|--no-margins)
 124         margins=''
 125         ;;
 126 -r|--align-right)
 127         shift
 128         columnAlignment[$1]=right
 129         ;;
 130 -l|--align-left)
 131         shift
 132         columnAlignment[$1]=left
 133         ;;
 134 -v|--output-vertical-separator)
 135         shift
 136         verticalBar=$1
 137         topleftCorner=$1
 138         toprightCorner=$1
 139         bottomleftCorner=$1
 140         bottomrightCorner=$1
 141         leftCross=$1
 142         middleCross=$1
 143         rightCross=$1
 144         topCross=$1
 145         bottomCross=$1
 146         ;;
 147 -h|--help)
 148         pod2text "$0"
 149         exit
 150         ;;
 151 -*)
 152         echo "$0: unknown option: $1" >&2
 153         exit -1
 154         ;;
 155 esac
 156 shift
 157 done
 158
 159
 160 explode()
 161 {
 162         local separator=$1
 163         local string=$2
 164         local new_string
 165         declare -g -a EXPLODE_ARRAY=()
 166
 167         while [ -n "$string" ]
 168         do
 169                 EXPLODE_ARRAY+=("${string%%$separator*}")
 170                 new_string=${string#*$separator}
 171                 if [ "$string" = "$new_string" ]
 172                 then
 173                         break
 174                 fi
 175                 string=$new_string
 176         done
 177
 178         # return $EXPLODE_ARRAY
 179 }
 180
 181 is_numeric_check_sep()
 182 {
 183         local thousands_sep=$2
 184         local fraction_sep=$3
 185         [[ $1 =~ ^[+-]?[0-9]+($thousands_sep[0-9]+)*($fraction_sep[0-9]+($thousands_sep[0-9]+)*)?$ ]]
 186 }
 187 is_numeric()
 188 {
 189         is_numeric_check_sep "$1" ',' . && return 0
 190         is_numeric_check_sep "$1" ' ' , && return 0
 191         is_numeric_check_sep "$1" '.' , && return 0
 192         return 1
 193 }
 194
 195
 196 # compute width of each column
 197
 198 column_width=()
 199 num_line=0
 200 text=''
 201 numericals_by_col=()
 202 non_numericals_by_col=()
 203 while read -r line
 204 do
 205         text="$text${text:+$'\n'}$line"
 206         explode $'\t' "$line"
 207
 208         column=0
 209         for cell in "${EXPLODE_ARRAY[@]}"
 210         do
 211                 width=${#cell}
 212                 if [ -z "${column_width[$column]}" ] || [ $width -gt "${column_width[$column]}" ]
 213                 then
 214                         column_width[$column]=$width
 215                 fi
 216
 217                 # guess if this cell has numerical content (skip 1st line as it's likely a header)
 218                 if [ $num_line -gt 0 -a -n "$cell" -a -z "${columnAlignment[$column]:-}" ]
 219                 then
 220                         if is_numeric "$cell"
 221                         then
 222                                 let numericals_by_col[$column]++
 223                         else
 224                                 let non_numericals_by_col[$column]++
 225                         fi
 226                 fi
 227                 let column++
 228         done
 229
 230         let num_line++
 231 done
 232
 233 if [ $num_line = 0 ]
 234 then
 235         # no input, no output
 236         exit 0
 237 fi
 238
 239
 240 set -u
 241
 242 # compose the format string
 243
 244 table_width=0
 245 cellFmt=''
 246 topFmt=''
 247 innerFmt=''
 248 bottomFmt=''
 249 segments=()
 250 sequentColumn=''
 251 for column in "${!column_width[@]}"
 252 do
 253         if [ -z "${columnAlignment[$column]:-}" ]
 254         then
 255                 if [ ${numericals_by_col[$column]:-0} -ge ${non_numericals_by_col[$column]:-0} ]
 256                 then
 257                         columnAlignment[$column]=right
 258                 else
 259                         columnAlignment[$column]=left
 260                 fi
 261         fi
 262
 263         table_width=$[table_width + (column == 0 ? (0${margins:+1} ? ${#verticalBar} : 0) : ${#verticalBar}) + ${column_width[$column]}]
 264
 265         cellFmt="${cellFmt:-${margins:+$verticalBar}}${sequentColumn:+$verticalBar}%*s"
 266         topFmt="${topFmt:-${margins:+$topleftCorner}}${sequentColumn:+$topCross}%${column_width[$column]}s"
 267         innerFmt="${innerFmt:-${margins:+$leftCross}}${sequentColumn:+$middleCross}%${column_width[$column]}s"
 268         bottomFmt="${bottomFmt:-${margins:+$bottomleftCorner}}${sequentColumn:+$bottomCross}%${column_width[$column]}s"
 269         segments+=('')
 270
 271         sequentColumn=yes
 272 done
 273 table_width=$[table_width + (0${margins:+1} ? ${#verticalBar} : 0)]
 274 cellFmt="$cellFmt${margins:+$verticalBar}"
 275 topFmt="$topFmt${margins:+$toprightCorner}"
 276 innerFmt="$innerFmt${margins:+$rightCross}"
 277 bottomFmt="$bottomFmt${margins:+$bottomrightCorner}"
 278
 279
 280 topGrid=''
 281 innerGrid=''
 282 bottomGrid=''
 283 if [ $horizontalLines ]
 284 then
 285         topGrid=`printf -- "$topFmt" "${segments[@]}"`
 286         topGrid=${topGrid// /$horizontalBar}$'\n'
 287         innerGrid=`printf -- "$innerFmt" "${segments[@]}"`
 288         innerGrid=${innerGrid// /$horizontalBar}$'\n'
 289         bottomGrid=`printf -- "$bottomFmt" "${segments[@]}"`
 290         bottomGrid=${bottomGrid// /$horizontalBar}$'\n'
 291 fi
 292
 293
 294 print_table_g()
 295 {
 296         line_no=1
 297         while read -r line
 298         do
 299                 if [ $line_no = 1 ]
 300                 then
 301                         echo -n "$topGrid"
 302                 else
 303                         echo -n "$innerGrid"
 304                 fi
 305
 306                 explode $'\t' "$line"
 307                 printf_args=()
 308                 for column in "${!column_width[@]}"
 309                 do
 310                         [ $column -lt ${#EXPLODE_ARRAY[@]} ] && cell=${EXPLODE_ARRAY[$column]} || cell=''
 311                         chars=${#cell}
 312                         bytes=`LANG=C; echo "${#cell}"`
 313                         swidth=$[ ${column_width[$column]} + ( $bytes - $chars ) ]
 314                         [ ${columnAlignment[$column]} = left ] && swidth=-$swidth
 315                         printf_args+=($swidth "$cell")
 316                 done
 317                 printf -- "$cellFmt\n" "${printf_args[@]}"
 318
 319                 let line_no++
 320         done <<<"$text"
 321
 322         echo -n "$bottomGrid"
 323 }
 324
 325
 326 if [ -t 1 -a -n "$PAGER" ]
 327 then
 328         terminal_cols=`tput cols`
 329         if [ "${terminal_cols:-0}" -le $table_width ]
 330         then
 331                 print_table_g | exec "$PAGER"
 332                 exit ${PIPESTATUS[1]}
 333         fi
 334 fi
 335
 336 print_table_g