populate.sh

   1 #!/bin/bash
   2 # set -x
   3
   4 #
   5 #   populate.sh
   6 #
   7 #   Copyright (C) 2007 nastasi@alternativeoutput.it
   8 #
   9 #  This program is free software; you can redistribute it and/or modify
  10 #  it under the terms of the GNU General Public License as published by
  11 #  the Free Software Foundation; either version 2 of the License, or
  12 #  (at your option) any later version.
  13 #
  14 #  This program is distributed in the hope that it will be useful, but
  15 #  WITHOUT ANY WARRANTY; without even the implied warranty of
  16 #  MERCHANTABLILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 #  General Public License for more details. You should have received a
  18 #  copy of the GNU General Public License along with this program; if
  19 #  not, write to the Free Software Foundation, Inc, 59 Temple Place -
  20 #  Suite 330, Boston, MA 02111-1307, USA.
  21 #
  22
  23 #
  24 # Description of tag behaviors
  25 # ----------------------------
  26 #
  27 #                         |                            |
  28 #       Template Tag      |         Assoc. Var         |  Description
  29 # ------------------------+----------------------------+-------------------
  30 #   #T_<tagname>#         | $T_<tagname>               |  simple tag
  31 #   ##T_<blkname>:start## | $B_<tagname>               |  block tag begin
  32 #        . . . . .        |                            |
  33 #      #T_<tagname>#      | $T_<blkname>_<tagname>     |  tag inside a block
  34 #      #T_<tagname>_X#    | $T_<blkname>_<tagname>_<#> |  indexed tag inside a block
  35 #                         |                            |  (the first in the alphab. order must
  36 #                         |                            |   exists to enable the entire block)
  37 #        . . . . .        |                            |
  38 #   ##T_<blkname>:stop##  |                            |  block tag finish
  39 #
  40
  41 # TODO
  42 #
  43 #  Aggiungere la filtratura per tipo di output
  44 #
  45 #  loop formati
  46 #+   loop lingue
  47 #+     loop for recursion (general)
  48 #+       get blocchi dal template
  49 #+       sost TAG nei blocchi con le VAR
  50 #+           loop for recursion (for each row of each block)
  51 #+             per ogni set di var per quel blocco (flat o con indice)
  52 #+             fino a quando la prima con indice non e' stringa vuota
  53 #+           fine loop for recursion (for each row of each block)
  54 #+       sostituire i TAG nel template con i blocchi
  55 #+       sostituire i TAG nel template con le VAR
  56 #
  57 #+       produrre il file in output
  58 #+     fine loop for recursion (general)
  59 #+   fine loop lingue
  60 #  fine loop formati
  61 #
  62 #
  63
  64 #  GLOBAL VARS
  65 # debug=1
  66 debug=""
  67 verbose=""
  68 NEWLI="#aCaPo#"
  69
  70 TMPFILE="/tmp/populate1_$$.tmp"
  71 TMP2FILE="/tmp/populate2_$$.tmp"
  72 TMP3FILE="/tmp/populate3_$$.tmp"
  73 TMP4FILE="/tmp/populate4_$$.tmp"
  74
  75
  76 #
  77 #  FUNCTIONS
  78 #
  79 function usage() {
  80     echo "Usage: $1 [-v] [-d] [-h] data-file"
  81     echo "  -v     Verbose enable."
  82     echo "  -d     Debug enable."
  83     echo "  -h     This help."
  84     exit 0
  85 }
  86
  87 function vars_get() {
  88     # echo "vars_get start"
  89     sed -n 's/\(#T_[A-Z_]\+#\)/\n\1\n/gp' | grep '^#T_' | sort | uniq | sed 's/#\(T_[A-Z_]\+\)#/\1/g'
  90     # echo "vars_get stop"
  91     return 0
  92 }
  93
  94 function unsedify() {
  95
  96     (
  97         if [ $# -eq 1 ]; then
  98             echo "$1"
  99         else
 100             cat
 101         fi
 102     ) | sed 's/\(["\\\/\$]\)/\\\1/g'
 103
 104     return 0
 105 }
 106
 107 function formatify() {
 108     (
 109         if [ $# -eq 2 ]; then
 110             echo "$2"
 111         else
 112             cat
 113         fi
 114     ) | (
 115         if [ "$1" = "tex" ]; then
 116             sed 's/à/\\`a/g;s/è/\\`e/g;s/ì/\\`i/g;s/ò/\\`o/g;s/ù/\\`u/g;s/áéíóú/\\'"'"'a/g;
 117 s/é/\\'"'"'e/g;s/í/\\'"'"'i/g;s/ó/\\'"'"'o/g;s/ú/\\'"'"'u/g;s/°/\$\^\\circ\$/g;
 118 s/<i>/\\textit{/g;s/<b>/\\textbf{/g;s/<\/[ib]>/}/g;s/<url>/URL: \\textit\{/g;s/<\/url>/\}/g;
 119 s/“/``/g;s/”/'"''"'/g;s/<lang:tex>//g;s/<\/lang:tex>//g;s/<lang:[^>]*>.*<\/lang:[^>]*>//g'
 120         elif [ "$1" = "htm" ]; then
 121             sed "s/${NEWLI}${NEWLI}/<br>${NEWLI}/g;"'s/<url>\([^<]*\)<\/url>/<a href="\1">\1<\/a>/g;
 122 s/<lang:htm>//g;s/<\/lang:htm>//g;s/<lang:[^>]*>.*<\/lang:[^>]*>//g'
 123         fi
 124     )
 125
 126
 127 }
 128
 129 function debecho() {
 130     if [ -z $debug ]; then
 131        return 0
 132     fi
 133     echo "debug "$*
 134     return 0
 135 }
 136
 137 function populator() {
 138     local tmplname lng frmt
 139
 140     tmplname="$1"
 141     contname="$2"
 142     lng="$3"
 143     frmt="$4"
 144
 145     tmplname="${tmplname}.${frmt}"
 146     if [ $lng ]; then
 147         lngsfx="_`echo "$lng" | tr '[:lower:]' '[:upper:]'`"
 148         filesfx="_$lng"
 149     else
 150         lngsfx=""
 151         filesfx=""
 152     fi
 153
 154     oldifs="$IFS"
 155     IFS="
 156 "
 157     anotherloop=0
 158
 159     #
 160     #  Recursion loop, resolve TAG created with a tag substitution, on the entire document.
 161     #
 162     while [ $anotherloop -eq 0 ]; do
 163         debecho "ANOTHERLOOP"
 164         #
 165         # GET BLOCKS
 166         #
 167
 168         # add normal tag before each block, output to TMPFILE
 169         cat $tmplname | sed 's/^\(##\(T_[A-Z_0-9]\+\):start##\)$/#\2#\n\1/g' > $TMPFILE
 170
 171         # get all blocks from the template, output to TMP2FILE
 172         cat $TMPFILE | sed -n '/##T_[A-Z_0-9]\+:start##/,/##T_[A-Z_0-9]\+:end##/p' > $TMP2FILE
 173
 174         # remove blocks from the template, output to TMP3FILE
 175         cat $TMPFILE | sed '/##T_[A-Z_0-9]\+:start##/,/##T_[A-Z_0-9]\+:end##/d' > $TMP3FILE
 176
 177         # get line number of the start of all blocks
 178         blist="`cat $TMP2FILE | grep -n '^##T_[A-Z_0-9]\+:start##$' | cut -d : -f 1`"
 179
 180         # for each line number ...
 181         for i in `echo "$blist"`; do
 182             # extracts the current block
 183             block_cur="`cat $TMP2FILE | sed -n $i',/##T_[A-Z_0-9]\+:end##/p'`"
 184
 185             # extracts the name of the block variable
 186             block_var="`echo "$block_cur" | sed -n "1,1p" | sed 's/^\(##T\(_[A-Z_0-9]\+\):start##\)$/B\2/g'`"
 187             # remove first and last line (block delimiters) and double '\' char, substitute \n with $NEWLI
 188             block_val="`echo "$block_cur" | sed '1d;$d;s/\\\\/\\\\\\\\/g' | sed "s/\$/$NEWLI\\\\\\/g"`"
 189
 190             # assign the value to the var
 191             read $block_var <<< "$block_val"
 192
 193             debecho "vv------------------------------"
 194             debecho "BLOCK_VAR: $block_var"
 195             debecho "BLOCK_VAL: ${!block_var}"
 196             debecho "^^------------------------------"
 197         done
 198
 199         #
 200         #  SUBSTITUTE TAG INSIDE EACH BLOCK
 201         #
 202         if [ $debug ]; then
 203             echo "debug blocks: ${!B_*}"
 204         fi
 205         for i in ${!B_*}; do
 206             blk_var="`echo "$i" | sed 's/^B_/T_/g'`"
 207             debecho "Tag_var: $tag_var"
 208             blk_val=""
 209             outindex=1
 210             for idx in `seq 1 100`; do
 211                 debecho "ISFIRST YES"
 212                 isfirst="YES"
 213
 214                 blk_src="${!i}"
 215                 #
 216                 #  Recursion loop, resolve TAG created with a tag substitution, for each row of each block.
 217                 #
 218                 blk_anotherloop=0
 219                 while [ $blk_anotherloop -eq 0 ]; do
 220                     debecho "BLK_ANOTHERLOOP $i $idx"
 221                     sedargs=""
 222                     tlist="`echo "$blk_src" | vars_get`"
 223                     debecho "TILIST $tlist"
 224
 225                     for e in $tlist; do
 226                         debecho "LOO1"
 227
 228                         tag_var="`echo "$i $e" | sed 's/^B_\([A-Z_]\+\) T_\([A-Z_]\+\)/T_\1_\2/g'`"
 229                         echo "$tag_var" | grep -q '_X$'
 230                         if [ $? -eq 0 ]; then
 231                             # fault back to original language if not defined
 232                             for ll in $lngsfx ""; do
 233                                 tag_ele="`echo "$tag_var" | sed 's/X$//g'`${idx}${ll}"
 234                                 debecho "tag_ele: $tag_ele"
 235                                 debecho "  BRA3"
 236                                 declare -p $tag_ele >/dev/null 2>&1
 237                                 if [ $? -eq 0 ]; then
 238                                     break
 239                                 else
 240                                     if [ "$isfirst" = "YES" -a -z "$ll" ]; then
 241                                         outindex=0
 242                                         break 4
 243                                     fi
 244                                     if [ -z "$ll" ]; then
 245                                         echo "WARN: undefined [$tag_ele]"
 246                                     elif [ "$verbose" ]; then
 247                                         tag_fback="`echo "$tag_var" | sed 's/X$//g'`${idx}"
 248                                         echo "WARN: undefined [$tag_ele]; fault back to main [${!tag_fback}]"
 249                                     fi
 250                                 fi
 251                                 if [ -z "$ll" ]; then
 252                                     debecho "  ISFIRST NO"
 253                                     isfirst="NO"
 254                                 fi
 255                             done
 256
 257                         else
 258                             debecho "  BRA4"
 259                             # faultback to original language if not defined
 260                             for ll in $lngsfx ""; do
 261                                 tag_ele="${tag_var}${ll}"
 262                                 debecho "  tag_ele: $tag_ele"
 263
 264                                 declare -p $tag_ele >/dev/null 2>&1
 265                                 if [ $? -eq 0 ]; then
 266                                     break
 267                                 else
 268                                     if [ -z "$ll" ]; then
 269                                         echo "WARN: undefined [$tag_ele]"
 270                                     elif [ "$verbose" ]; then
 271                                         echo "WARN: undefined [$tag_ele]; fault back to main [${!tag_var}]"
 272                                     fi
 273                                 fi
 274                             done
 275                         fi
 276                         tag_val="`echo "${!tag_ele}" | sed "s/\$/$NEWLI/g" | tr -d '\n' | sed "s/$NEWLI\$//g" \
 277                                 | formatify $frmt`"
 278                         debecho "  tag_val: $tag_val"
 279
 280                         # TAG replacement
 281                         tag_val="`unsedify "$tag_val"`"
 282                         sedargs="${sedargs}s/#${e}#/$tag_val/g;"
 283                     done # for e ...
 284                     debecho "----------------------------------"
 285                     debecho "$sedargs"
 286                     blk_val="${blk_val}`echo "$blk_src" | sed "$sedargs"`"
 287                     if [ $? -ne 0 ]; then
 288                         echo "$sedargs" > sedargs.txt
 289                         exit 111
 290                     fi
 291
 292                     echo "$blk_val" | grep -q '#\{1,2\}T_[A-Za-z_:]\+#\{1,2\}'
 293                     if [ $? -eq 0 ]; then
 294                         # echo "$blk_val" | grep '#\{1,2\}T_[A-Za-z_:]\+#\{1,2\}'
 295                         # echo "POST BLK GREP"
 296                         # read
 297                         blk_src="$blk_val"
 298                         blk_val=""
 299                         blk_anotherloop=0
 300                     else
 301                         blk_anotherloop=1
 302                     fi
 303                 done # while [ $blk_another_loop ....
 304                 if [ $outindex -eq 0 ]; then
 305                     break
 306                 fi
 307             done
 308
 309             debecho "++++++++++++++++++++++++++++++++++"
 310             debecho "BLK_VAL: $blk_val"
 311             debecho "**********************************"
 312             # read $blk_var <<< "$blk_val"
 313
 314             # TAG replacement
 315             blk_val="`unsedify "$blk_val" | sed "s/$NEWLI\$//g"`"
 316             debecho "BLK_VAR: $blk_var: $blk_val"
 317             debecho "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
 318             cat $TMP3FILE | sed "s/#${blk_var}#/$blk_val/g"  >$TMP4FILE
 319             mv $TMP4FILE $TMP3FILE
 320         done
 321
 322         # cat, sed to avoid multitag on single line, grep of tags, sort, uniq, sed to remove '#' chars
 323         t_vars="`cat $TMP3FILE | vars_get`"
 324
 325         for tag_var in $t_vars; do
 326             # echo "TAG_VAR: $tag_var"
 327             for ll in $lngsfx ""; do
 328                 tag_ele="${tag_var}${ll}"
 329                 declare -p $tag_ele >/dev/null 2>&1
 330                 if [ $? -eq 0 ]; then
 331                     break
 332                 else
 333                     if [ -z "$ll" ]; then
 334                         echo "WARN: undefined [$tag_ele]"
 335                     elif [ "$verbose" ]; then
 336                         echo "WARN: undefined [$tag_ele]; fault back to main [${!tag_var}]"
 337                     fi
 338                 fi
 339             done
 340             # TAG replacement
 341             tag_val="`echo "${!tag_ele}" | sed "s/\$/$NEWLI/g" | tr -d '\n' | sed "s/$NEWLI\$//g" \
 342                   | formatify $frmt | unsedify`"
 343
 344
 345             cat $TMP3FILE | sed "s/#${tag_var}#/$tag_val/g" >$TMP4FILE
 346             mv $TMP4FILE $TMP3FILE
 347         done
 348
 349         cat $TMP3FILE | grep -q '#\{1,2\}T_[A-Za-z_:]\+#\{1,2\}'
 350         if [ $? -eq 0 ]; then
 351             # cat $TMP3FILE | grep -q '#\{1,2\}T_[A-Za-z_:]\+#\{1,2\}'
 352             # echo POSTGREP
 353             # read
 354             anotherloop=0
 355             cat $TMP3FILE | sed "s/$NEWLI/\n/g" > $TMP4FILE
 356             tmplname="$TMP4FILE"
 357         else
 358             anotherloop=1
 359         fi
 360     done #  while [ $anotherloop -eq 0 ]; do
 361     cat $TMP3FILE | sed "s/$NEWLI/\n/g" > ${contname}${filesfx}.$frmt
 362
 363     rm -f $TMPFILE $TMP2FILE $TMP3FILE $TMP4FILE
 364
 365     IFS="$oldifs"
 366
 367     return 0
 368 }
 369
 370 #
 371 #  MAIN
 372 #
 373
 374 # maybe parametrizable
 375 # tmplname="temptest"
 376 tmplname="cv_template"
 377 while [ $# ]; do
 378     case $1 in
 379         -v) verbose="Y"
 380             ;;
 381         -d) debug="Y"
 382             ;;
 383         -h) usage $0
 384             ;;
 385         *)  break
 386     esac
 387     shift
 388 done
 389 contname="$1"
 390
 391 rm -f $TMPFILE $TMP2FILE $TMP3FILE $TMP4FILE
 392
 393 if [ ! -f ${contname}.data ]; then
 394     echo "File ${contname}.data not exists"
 395     exit 1
 396 fi
 397
 398 # source the data file from template scope
 399 . ${tmplname}.data
 400
 401 # source the data file from user scope
 402 . ${contname}.data
 403
 404 for frmt in tex htm; do
 405     for lng in "" eng; do
 406         echo "Format: $frmt  Lang: $lng"
 407         populator "$tmplname" "$contname" "$lng" "$frmt"
 408     done
 409 done
 410
 411 exit 0