help/html2wiki

   1 #!/bin/bash
   2 #
   3 # html to wiki script
   4 # Copyright (C) 2007-2010 Carlo Baldassi (the "Author") <carlobaldassi@gmail.com>.
   5 # All Rights Reserved.
   6 #
   7 # This program is free software; you can redistribute it and/or modify
   8 # it under the terms of the GNU General Public License as published by
   9 # the Free Software Foundation; either version 2 of the Licence, or
  10 # (at your option) any later version.
  11 #
  12 # This program is distributed in the hope that it will be useful,
  13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15 # GNU General Public License for more details.
  16 #
  17 # You should have received a copy of the GNU General Public License
  18 # along with this program; if not, see <http://www.gnu.org.licences/>.
  19 #
  20
  21 ### FUNCTIONS AND FILTERS ###
  22 #{{{
  23
  24 # REGEX RELATED FUNCTIONS
  25 #{{{
  26
  27 # anti-regex
  28 function anti
  29 { #{{{
  30         local id ind brid chbrid
  31         local STR S
  32         local OUTSTR
  33         local C T
  34         local BRANCHES NEWBRANCHES
  35         local BR OLDBR
  36         local CHBR CC
  37         local go c0 found chbrN
  38         local L="${#@}"
  39         #echo "L=$L"
  40         for id in $(seq 1 $L)
  41         do
  42                 S="$(eval "echo \"\$${id}\"")"
  43                 STR[$id]="$S"
  44                 N[$id]=$(echo -n "$S" | wc -c)
  45                 BR[$id]=1
  46                 OLDBR[$id]=1
  47         done
  48         BRANCHES=1
  49         T[1]=""
  50         OUTSTR=""
  51         ind=1
  52         go=1
  53         while [ $go -eq 1 ]
  54         do
  55                 go=0
  56                 for id in $(seq 1 $L)
  57                 do
  58                         [ $ind -le ${N[$id]} ] || continue;
  59                         go=1
  60                         C[$id]="$(echo "${STR[$id]}" | cut -c${ind})"
  61                         OLDBR[$id]=${BR[$id]}
  62                 done
  63                 [ $go -eq 0 ] && continue;
  64                 NEWBRANCHES=$BRANCHES
  65                 for brid in $(seq 1 $BRANCHES)
  66                 do
  67                         local CHBR=( )
  68                         local chbrN=0
  69                         for id in $(seq 1 $L)
  70                         do
  71                                 [ $ind -le ${N[$id]} ] || continue;
  72                                 [ ${OLDBR[$id]} -eq $brid ] || continue;
  73                                 c0="${C[$id]}"
  74                                 if [ $chbrN -eq 0 ]
  75                                 then
  76                                         CHBR[1]="$brid"
  77                                         CC[1]="$c0"
  78                                         let chbrN++
  79                                 else
  80                                         found=0
  81                                         for chbrid in $(seq 1 $chbrN )
  82                                         do
  83                                                 if [ "$c0" == "${CC[$chbrid]}" ]
  84                                                 then
  85                                                         found=1;
  86                                                         break;
  87                                                 fi
  88                                         done
  89                                         if [ $found -eq 0 ]
  90                                         then
  91                                                 let NEWBRANCHES++
  92                                                 let chbrN++
  93                                                 CHBR[$chbrN]="$NEWBRANCHES"
  94                                                 CC[$chbrN]="$c0"
  95                                                 BR[$id]=$NEWBRANCHES
  96                                                 T[$NEWBRANCHES]=${T[$brid]}
  97                                         fi
  98                                 fi
  99                         done
 100                         [ $chbrN -eq 0 ] && continue;
 101                         [ -n "$OUTSTR" ] && OUTSTR="${OUTSTR}|"
 102                         OUTSTR="${OUTSTR}${T[$brid]}[^"
 103                         for chbrid in $(seq 1 $chbrN )
 104                         do
 105                                 OUTSTR="${OUTSTR}${CC[$chbrid]}"
 106                                 T[${CHBR[$chbrid]}]="${T[${CHBR[$chbrid]}]}${CC[$chbrid]}"
 107                         done
 108                         OUTSTR="${OUTSTR}]"
 109                 done
 110
 111                 BRANCHES=$NEWBRANCHES
 112                 let ind++
 113         done
 114         echo "$OUTSTR"
 115 } #}}}
 116
 117 # anti-opentag regex
 118 function anti_open
 119 { #{{{
 120         anti "<$1>"
 121 } #}}}
 122
 123 # anti-closetag regex
 124 function anti_close
 125 { #{{{
 126         anti "</$1>"
 127 } #}}}
 128
 129 # anti-tag regex
 130 function anti_both
 131 { #{{{
 132         anti "<$1>" "</$1>"
 133 } #}}}
 134
 135 # skip whole tag-delimited parts
 136 function skipper
 137 { #{{{
 138         local id
 139         local L="${#@}"
 140         local S
 141         local OUTSTR=""
 142         for id in $(seq 1 $L)
 143         do
 144                 S="$(eval "echo \"\$${id}\"")"
 145                 OUTSTR="${OUTSTR}<$S>($(anti_both "$S"))*</$S>|"
 146         done
 147         local TL=""
 148         for id in $(seq 1 $L)
 149         do
 150                 S="$(eval "echo \"\$${id}\"")"
 151                 TL="${TL} <$S>"
 152         done
 153         OUTSTR="${OUTSTR}$(anti $TL)"
 154         echo $OUTSTR
 155 } #}}}
 156
 157 #}}}
 158
 159 # SUBSTITUTION FILTERS (FLAT)
 160 #{{{
 161
 162 # greedy remove
 163 function rmtags
 164 { #{{{
 165         sed -r "s@</?$1\>[^>]*>@@g";
 166 } #}}}
 167
 168 # substitute tags delimiters
 169 function subtags
 170 { #{{{
 171         local BODY_EL="($(skipper "$1"))"
 172         sed -r "s@<$1([[:space:]]+[^>]+)?>(${BODY_EL}*)</$1>@$2\2$3@g"
 173 } #}}}
 174
 175 # images
 176 function subimgs
 177 { #{{{
 178         sed -r "s/<img\>[^>]*src=\"([^>]*)\"[^>]*>/[[image \1]]/g"
 179 } #}}}
 180
 181 # links (external and internal)
 182 function sublinks
 183 { #{{{
 184         sed -r "s@<a href=\"(#[^\"]*)\">(($(anti_both "a"))*)</a>@[\1 \2]@g"    | \
 185         sed -r "s@<a name=\"([^\"]*)\">(($(anti_both "a"))*)</a>@[[# \1]]\2@g"
 186 } #}}}
 187
 188 #}}}
 189
 190 # SUBSTITUTION FILTERS (WITH NESTING)
 191 #{{{
 192
 193 # associate a depth to the given tags
 194 function indent
 195 { #{{{
 196         local LEV="$1"
 197         shift
 198         local BODY_EL="($(skipper "$@"))"
 199         local TAG="("
 200         local id
 201         local L="${#@}"
 202         for id in $(seq 1 $L)
 203         do
 204                 S="$(eval "echo \"\$${id}\"")"
 205                 [ $id -gt 1 ] && TAG="${TAG}|"
 206                 TAG="${TAG}${S}"
 207         done
 208         TAG="${TAG})"
 209         sed -r "s@<${TAG}>(${BODY_EL}*)</\1>@<<I $LEV \1>>\2<</I $LEV \1>>@g"
 210 } #}}}
 211
 212 # invoke indent for all levels
 213 # up to the given one (in increasing order)
 214 function indent_alllevels
 215 { #{{{
 216         local LEV="$1"
 217         shift
 218         local LMAX="$LEV"
 219         [[ "$1" =~ ^@@[[:digit:]]+$ ]] && { LMAX=${1#@@}; shift; }
 220         if [ $LEV -gt 0 ]
 221         then
 222                 indent $((LMAX - LEV + 1)) "$@" |\
 223                 indent_alllevels $((LEV - 1)) "@@${LMAX}" "$@"
 224         else
 225                 cat
 226         fi
 227 } #}}}
 228
 229 # substitute tags delimiters
 230 # at a specified level
 231 function subtags_atlevel
 232 { #{{{
 233         local LEV="$1"
 234         shift
 235         local BODY_EL1="($(anti "<</I $LEV "))"
 236         #local BODY_EL1="(<<I [[:digit:]]+ [[:alpha:]]+>>($(anti "<<I " "<</I "))*<</I [[:digit:]]+ [[:alpha:]]+>>|$(anti "<</I $LEV "))"
 237         local BODY_EL2="($(skipper "$1"))"
 238         sed -r "s@<<I $LEV ([[:alpha:]]+)>>(${BODY_EL1}*)<</I $LEV \1>>@<<I $LEV \1>>\n<<MARKED>>\2\n<</I $LEV \1>>@g" | \
 239         sed -r "/^<<MARKED>>/s@<$1>(${BODY_EL2}*)</$1>@$2\1$3@g" | \
 240         sed -r "s@^<<MARKED>>@@" | \
 241         tr '\n' ' '
 242 } #}}}
 243
 244 # invoke subtags_atlevel for all levels
 245 # up to the given one (in decreasing order)
 246 # (delimiters are passed through array names)
 247 function subtags_multilevel
 248 { #{{{
 249         local L="$1"
 250         local TAG="$2"
 251         local ARRAY_B="$3"
 252         local ARRAY_E="$4"
 253
 254         if [ "$L" -gt 0 ]
 255         then
 256                 local BS="$(eval echo \"\${${ARRAY_B}[L - 1]}\")"
 257                 local ES="$(eval echo \"\${${ARRAY_E}[L - 1]}\")"
 258                 subtags_atlevel $L "$TAG" "$BS" "$ES"                           | \
 259                 subtags_multilevel $((L - 1)) "$TAG" "$ARRAY_B" "$ARRAY_E"
 260         else
 261                 cat
 262         fi
 263 } #}}}
 264
 265 # substitute indented tags
 266 function subindent
 267 { #{{{
 268         local LEV="$1"
 269         shift;
 270         local BODY_EL="($(anti "<</I $LEV $1>>"))"
 271         sed -r "s@<<I $LEV $1>>(${BODY_EL}*)<</I $LEV $1>>@$2\1$3@g"
 272 } #}}}
 273
 274 # invoke subindent for all levels
 275 # up to the given one (in increasing order)
 276 function subindent_alllevels
 277 { #{{{
 278         local LEV="$1"
 279         shift
 280         local LMAX="$LEV"
 281         [[ "$1" =~ ^@@[[:digit:]]+$ ]] && { LMAX=${1#@@}; shift; }
 282         if [ $LEV -gt 0 ]
 283         then
 284                 subindent $((LMAX - LEV + 1)) "$@" |\
 285                 subindent_alllevels $((LEV - 1)) "@@${LMAX}" "$@"
 286         else
 287                 cat
 288         fi
 289 } #}}}
 290
 291 #}}}
 292
 293 # FORMAT AND CLEANUP FILTERS
 294 #{{{
 295
 296 # remove unnecessary spaces
 297 function rmspace
 298 { #{{{
 299         sed -r "s/[[:space:]]+/ /g"
 300 } #}}}
 301
 302 # remove spaces at the beginning
 303 # of the lines
 304 function rmspace_start
 305 { #{{{
 306         sed -r "s/^[[:space:]]*//"
 307 } #}}}
 308
 309 # collapse newlines marked as weak
 310 # if next to non-weak newlines
 311 function rmweaknewlines
 312 { #{{{
 313         sed -r "s/(<<WNL>>[[:space:]]*)+<<NL>>/<<NL>>/g" | \
 314         sed -r "s/<<WNL>>/<<NL>>/g"
 315 } #}}}
 316
 317 # apply space formattings
 318 function subspace
 319 { #{{{
 320         sed -r "s/[[:space:]]*<<NL>>[[:space:]]*/\n/g"          | \
 321         sed -r "s/<<SP9>>/         /g"                          | \
 322         sed -r "s/<<SP8>>/        /g"                           | \
 323         sed -r "s/<<SP7>>/       /g"                            | \
 324         sed -r "s/<<SP6>>/      /g"                             | \
 325         sed -r "s/<<SP5>>/     /g"                              | \
 326         sed -r "s/<<SP4>>/    /g"                               | \
 327         sed -r "s/<<SP3>>/   /g"                                | \
 328         sed -r "s/<<SP2>>/  /g"                                 | \
 329         sed -r "s/<<SP>>/ /g"
 330 } #}}}
 331
 332 #}}}
 333
 334 # CUSTOM FILTERS
 335 #{{{
 336
 337 function custom_filters
 338 { #{{{
 339         sed -r "s@(\[\[image[[:space:]]+)([^/]*/)?([^]]*\]\])@\1\3@g"
 340 } #}}}
 341
 342 #}}}
 343
 344 #}}}
 345
 346 ### CONSTANTS AND SUBSTITUTION TABLES
 347 #{{{
 348
 349 MAXLEV=4
 350
 351 DARRAY_DT_B=( '* //' '<<SP>>* //' '<<SP2>>* //' '<<SP3>>* //' '<<SP4>>* //' '<<SP5>>* //' '<<SP6>>* //' '<<SP7>>* //' '<<SP8>>* //' '<<SP9>>* //' )
 352 DARRAY_DT_E=( './/' './/' './/' './/' './/' './/' './/' './/' './/' )
 353
 354 DARRAY_DD_B=( '' '' '' '' '' '' '' '' '' )
 355 DARRAY_DD_E=( '<<NL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' '<<WNL>>' )
 356
 357 DARRAY_LI_B=( '* ' '<<SP>>* ' '<<SP2>>* ' '<<SP3>>* ' '<<SP4>>* ' '<<SP5>>* ' '<<SP6>>* ' '<<SP7>>* ' '<<SP8>>* ' '<<SP9>>* ' )
 358 DARRAY_LI_E=( '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' '<<NL>>' )
 359
 360 DARRAY_DIV_B=( '' '' '' '' '' '' '' '' '' )
 361 DARRAY_DIV_E=( ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' )
 362
 363 DARRAY_P_B=( '' '' '' '' '' '' '' '' '' )
 364 DARRAY_P_E=( ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' )
 365
 366 #}}}
 367
 368 ### MAIN FILTER ###
 369 #{{{
 370
 371 cat | \
 372         tr "\n" " "                                                             | \
 373         tr "\t" " "                                                             | \
 374         rmspace                                                                 | \
 375         rmtags "!doctype"                                                       | \
 376         rmtags "html"                                                           | \
 377         rmtags "META"                                                           | \
 378         rmtags "head"                                                           | \
 379         rmtags "link"                                                           | \
 380         rmtags "body"                                                           | \
 381         rmtags "table"                                                          | \
 382         rmtags "tr"                                                             | \
 383         rmtags "td"                                                             | \
 384         rmtags "center"                                                         | \
 385         rmtags "font"                                                           | \
 386         subimgs                                                                 | \
 387         sublinks                                                                | \
 388         subtags "title" '<<NL>>+ ' '<<NL>>'                                     | \
 389         subtags "h1" '<<NL>>+ ' '<<NL>>'                                        | \
 390         subtags "h2" '<<NL>>++ ' '<<NL>>'                                       | \
 391         subtags "h3" '<<NL>>+++ ' '<<NL>>'                                      | \
 392         subtags "h4" '<<NL>>++++ ' '<<NL>>'                                     | \
 393         subtags "b" '**' '**'                                                   | \
 394         subtags "i" '//' '//'                                                   | \
 395         subtags "u" '__' '__'                                                   | \
 396         subtags "code" '<<NL>>[[code]]<<NL>>' '<<NL>>[[/code]]<<NL>>'           | \
 397         indent_alllevels $MAXLEV "dl" "ul" "ol"                                 | \
 398         subtags_multilevel $MAXLEV "dt" DARRAY_DT_B DARRAY_DT_E                 | \
 399         subtags_multilevel $MAXLEV "dd" DARRAY_DD_B DARRAY_DD_E                 | \
 400         subtags_multilevel $MAXLEV "li" DARRAY_LI_B DARRAY_LI_E                 | \
 401         subtags_multilevel $MAXLEV "div" DARRAY_DIV_B DARRAY_DIV_E              | \
 402         subtags_multilevel $MAXLEV "p" DARRAY_P_B DARRAY_P_E                    | \
 403         subtags "div" '<<NL>>' '<<NL>>'                                         | \
 404         subtags "div" '<<NL>>' '<<NL>>'                                         | \
 405         subtags "p" '<<NL>>' '<<NL>>'                                           | \
 406         subindent_alllevels $MAXLEV "dl" '<<NL>>' ''                            | \
 407         subindent_alllevels $MAXLEV "ul" '<<NL>>' '<<NL>>'                      | \
 408         subindent_alllevels $MAXLEV "ol" '<<NL>>' '<<NL>>'                      | \
 409         rmspace_start                                                           | \
 410         rmspace                                                                 | \
 411         rmweaknewlines                                                          | \
 412         custom_filters                                                          | \
 413         subspace                                                                | \
 414         cat -s
 415 echo
 416
 417 #}}}
 418
 419 ### END ###
 420
 421 # OBSOLETE / NONWORKING STUFF
 422 #{{{
 423
 424 #function anti_open
 425 #{#{{{
 426 #       local ind
 427 #       local STR="${1}>"
 428 #       local N=$(echo -n "$STR" | wc -c)
 429 #       local OUTSTR="[^<]"
 430 #       local T="<"
 431 #       local C
 432 #       for ind in $(seq 1 $N)
 433 #       do
 434 #               C="$(echo "$STR" | cut -c${ind})"
 435 #               OUTSTR="${OUTSTR}|${T}[^$C]"
 436 #               T="${T}${C}"
 437 #       done
 438 #
 439 #       echo "$OUTSTR"
 440 #}#}}}
 441
 442 #function anti_close
 443 #{#{{{
 444 #       echo "$(anti_open "/$1")"
 445 #}#}}}
 446
 447 #function anti_both
 448 #{#{{{
 449 #       local ind
 450 #       local STR1="${1}>"
 451 #       local STR2="/${1}>"
 452 #       local N=$(echo -n "$STR2" | wc -c)
 453 #       local C1 C2
 454 #       C1="$(echo "$STR1" | cut -c1)"
 455 #       local OUTSTR="[^<]|<[^${C1}/]"
 456 #       local T1="<${C1}"
 457 #       local T2="</"
 458 #       for ind in $(seq 2 $((N - 1)) )
 459 #       do
 460 #               C1="$(echo "$STR1" | cut -c${ind})"
 461 #               C2="$(echo "$STR2" | cut -c${ind})"
 462 #               OUTSTR="${OUTSTR}|${T1}[^${C1}]|${T2}[^${C2}]"
 463 #               T1="${T1}${C1}"
 464 #               T2="${T2}${C2}"
 465 #       done
 466 #       C2="$(echo "$STR2" | cut -c${N})"
 467 #       OUTSTR="${OUTSTR}|${T2}[^${C2}]"
 468 #
 469 #       echo "$OUTSTR"
 470 #}#}}}
 471
 472 #function skipper
 473 #{#{{{
 474 #       echo "<$1>($(anti_both "$1"))*</$1>|$(anti_open "$1")"
 475 #}#}}}
 476
 477 #function rmtagsbody
 478 #{#{{{
 479 #       sed -r "s@<$1\>[^>]*>[^<]*</$1>@@g";
 480 #}#}}}
 481
 482 #function subdesclist
 483 #{#{{{
 484 #       subtags "dt" '* ' '.'           | \
 485 #       subtags "dd" '' '<<NL>>'        | \
 486 #       subtags "dl" '<<NL>>' ''
 487 #}#}}}
 488
 489 #function worknest
 490 #{#{{{
 491 #       local BODY_EL="($(anti_both "$1"))"
 492 #       sed -r "s@<$1>(${BODY_EL}*)</$1>@<$1{{$2}}>\1</$1{{$2}}>@g"
 493 #}#}}}
 494
 495 #function subtags_upperlev
 496 #{#{{{
 497 #       local BODY_EL="($(skipper "$1"))"
 498 #       sed -r "s@<$1>(${BODY_EL}*)</$1>@$2\1$3>@g"
 499 #}#}}}
 500
 501 #function subtags_skip
 502 #{#{{{
 503 #       local BODY_EL="((<$1>($(anti_both "$1"))*</$1>)|($(anti "<$1>" "<$2>")))"
 504 #       #local BODY_EL="($(skipper "$1")|$(skipper "$2"))"
 505 #       sed -r "s@<$2>(${BODY_EL}*)</$2>@$3\1$4@g"
 506 #}#}}}
 507
 508 #function subtags_without
 509 #{#{{{
 510 #       local BODY_EL="($(anti "<$1>" "</$1>" "<$2>" "</$2>"))"
 511 #       sed -r "s@<$1>(${BODY_EL}*)</$1>@$3\1$4@g"
 512 #}#}}}
 513
 514 #function subtags_within_
 515 #{#{{{
 516 #       #local BODY_EL1="($(skipper "$1"))"
 517 #       #local BODY_EL2="($(skipper "$1"))"
 518 #       local BODY_EL1="(<$1>($(anti_both "$1"))*</$1>|$(anti "<$1>" "<$2>"))"
 519 #       local BODY_EL2="($(skipper "$1" "$2"))"
 520 #       #local BODY_EL2="($(anti "$2"))"
 521 #       local BODY_EL3="($(skipper "$1"))"
 522 #       sed -r "s@<$1>(${BODY_EL1}*)<$2>(${BODY_EL2}*)</$2>(${BODY_EL3}*)</$1>@<$1>\1$3\4$4\8</$1>@g"
 523 #}#}}}
 524
 525 #function subtags_within
 526 #{#{{{
 527 #       local N="$1"
 528 #       shift;
 529 #       if [ $N -gt 0 ]
 530 #       then
 531 #               subtags_within_ "$@" | subtags_within $(( N - 1 )) "$@"
 532 #       else
 533 #               cat
 534 #       fi
 535 #}#}}}
 536
 537 #function subtags_atlevel__
 538 #{#{{{
 539 #       local LEV="$1"
 540 #       shift
 541 #       local NONINDTAG="(<<I [[:digit:]]+ [[:alpha:]]+>>($(anti "<<I " "<</I "))*<</I [[:digit:]]+ [[:alpha:]]+>>|$(anti "<<I " "<</I " "<$1>"))"
 542 #       local BODY_EL="($(skipper "$1"))"
 543 #       local NONIND="(<<I [[:digit:]]+ [[:alpha:]]+>>($(anti "<<I " "<</I "))*<</I [[:digit:]]+ [[:alpha:]]+>>|$(anti "<</I $LEV "))"
 544 #       sed -r "s@<<I $LEV ([[:alpha:]]+)>>(${NONINDTAG}*)<$1>(${BODY_EL}*)</$1>(${NONIND}*)<</I $LEV \1>>@<<I $LEV \1>>\2$2\5$3\8<</I $LEV \1>>@g"
 545 #}#}}}
 546
 547 #function subtags_atlevel_
 548 #{#{{{
 549 #       local N="$1"
 550 #       shift;
 551 #       if [ $N -gt 0 ]
 552 #       then
 553 #               subtags_atlevel__ "$@" | subtags_atlevel_ $(( N - 1 )) "$@"
 554 #       else
 555 #               cat
 556 #       fi
 557 #}#}}}
 558
 559 #function subtags_atlevel
 560 #{#{{{
 561 #       subtags_atlevel_ 10 "$@"
 562 #}#}}}
 563
 564 #}}}
 565