|
| 1 | +#!/bin/bash |
| 2 | +# clean-html-js.sh |
| 3 | + |
| 4 | +VER="1.3" |
| 5 | +GREEN="\033[1;32m" |
| 6 | +RED="\033[0;31m" |
| 7 | +YELLOW="\033[1;33m" |
| 8 | +ENDCOLOR="\033[0m" |
| 9 | + |
| 10 | +tproc=`basename $0` |
| 11 | +echo -e $GREEN"$tproc version $VER"$ENDCOLOR |
| 12 | +echo "" |
| 13 | + |
| 14 | +usage() |
| 15 | +{ |
| 16 | + tproc=`basename $0` |
| 17 | + echo -e $YELLOW"usage:"$ENDCOLOR |
| 18 | + echo -e $GREEN" bash $tproc input.html"$ENDCOLOR |
| 19 | + echo "options:" |
| 20 | + echo " -c not clean comment (default = false);" |
| 21 | + echo " -n not verify type file (default = false);" |
| 22 | + echo " -o str output path (default = input.utf8.html);" |
| 23 | + echo " -r rewrite file (default = false);" |
| 24 | + echo " -h help." |
| 25 | + echo |
| 26 | + exit 1 |
| 27 | +} |
| 28 | + |
| 29 | +testcomponent() |
| 30 | +{ |
| 31 | + tnocomp="" |
| 32 | + tcomp="/usr/bin/iconv" |
| 33 | + tdeb="libc-bin_*.deb" |
| 34 | + if [ ! -f "$tcomp" ] |
| 35 | + then |
| 36 | + tnocomp="$tnocomp $tcomp($tdeb)" |
| 37 | + fi |
| 38 | + tcomp="/usr/bin/enca" |
| 39 | + tdeb="enca_*.deb" |
| 40 | + if [ ! -f "$tcomp" ] |
| 41 | + then |
| 42 | + tnocomp="$tnocomp $tcomp($tdeb)" |
| 43 | + fi |
| 44 | + tcomp="/bin/sed" |
| 45 | + tdeb="sed_*.deb" |
| 46 | + if [ ! -f "$tcomp" ] |
| 47 | + then |
| 48 | + tnocomp="$tnocomp $tcomp($tdeb)" |
| 49 | + fi |
| 50 | + tcomp="/bin/grep" |
| 51 | + tdeb="grep_*.deb" |
| 52 | + if [ ! -f "$tcomp" ] |
| 53 | + then |
| 54 | + tnocomp="$tnocomp $tcomp($tdeb)" |
| 55 | + fi |
| 56 | + if [ "+$tnocomp" != "+" ] |
| 57 | + then |
| 58 | + echo -e $RED"Not found $tnocomp !"$ENDCOLOR |
| 59 | + echo "" |
| 60 | + exit 0 |
| 61 | + fi |
| 62 | +} |
| 63 | + |
| 64 | +main() |
| 65 | +{ |
| 66 | +while getopts ":cno:rh" opt |
| 67 | +do |
| 68 | + case $opt in |
| 69 | + c) tcomment="1" |
| 70 | + ;; |
| 71 | + n) tnoverify="1" |
| 72 | + ;; |
| 73 | + o) tdest="$OPTARG" |
| 74 | + ;; |
| 75 | + r) trewrite="1" |
| 76 | + ;; |
| 77 | + h) usage |
| 78 | + ;; |
| 79 | + *) echo "Unknown option -$OPTARG" |
| 80 | + exit 1 |
| 81 | + ;; |
| 82 | + esac |
| 83 | +done |
| 84 | +shift "$(($OPTIND - 1))" |
| 85 | +SRCNAME="$1" |
| 86 | +if [ -z "$SRCNAME" ] |
| 87 | +then |
| 88 | + usage |
| 89 | +fi |
| 90 | +echo "$src" |
| 91 | +echo "" |
| 92 | +if [ -z "$tnoverify" ] |
| 93 | +then |
| 94 | + thtml=`file "$SRCNAME" | grep "HTML document"` |
| 95 | +else |
| 96 | + thtml="HTML document" |
| 97 | +fi |
| 98 | +if [ "+$thtml" != "+" ] |
| 99 | +then |
| 100 | + NEWNAME="${SRCNAME%.htm*}" |
| 101 | + NEWNAME="$NEWNAME.utf8.html" |
| 102 | + echo " $SRCNAME -> $NEWNAME" |
| 103 | + |
| 104 | + CODENAME=$(/usr/bin/enca -i "$SRCNAME") |
| 105 | + if [ "+$CODENAME" = "+" -o "+$CODENAME" = "+???" ] |
| 106 | + then |
| 107 | + CODENAME="UTF-8" |
| 108 | + elif [ "+$CODENAME" != "+UTF-8" ] |
| 109 | + then |
| 110 | + echo " Convert: $CODENAME -> UTF8" |
| 111 | + fi |
| 112 | + cat "$SRCNAME" | iconv -c -f ${CODENAME} -t UTF8 > "$NEWNAME" |
| 113 | + |
| 114 | +echo -e -n "[4]: 0.." |
| 115 | + |
| 116 | +echo -e -n "1.." |
| 117 | + |
| 118 | + sed -i -e ' |
| 119 | + s/\x0D$// |
| 120 | + s/<body/\n\L&/ig |
| 121 | + s/<iframe/\n\L&/ig |
| 122 | + s/<object/\n\L&/ig |
| 123 | + s/<d/\n\L&/ig |
| 124 | + s/<h/\n\L&/ig |
| 125 | + s/<l/\n\L&/ig |
| 126 | + s/<m/\n\L&/ig |
| 127 | + s/<p/\n\L&/ig |
| 128 | + s/<t/\n\L&/ig |
| 129 | + s/<\/body/\n\L&/ig |
| 130 | + s/<\/h/\n\L&/ig |
| 131 | + s/<script/\n\L&/ig |
| 132 | + s/<\/script>/\n\L&\n/ig |
| 133 | + s/<noscript/\n\L&/ig |
| 134 | + s/<\/noscript>/\n\L&\n/ig |
| 135 | + s/<iframe/\n\L&/ig |
| 136 | + s/<\/iframe>/\n\L&\n/ig |
| 137 | + s/<object/\n\L&/ig |
| 138 | + s/<\/object>/\n\L&\n/ig |
| 139 | + s/<ins/\n\L&/ig |
| 140 | + s/<\/ins>/\n\L&\n/ig |
| 141 | + s/<\!--/\n\L&/ig |
| 142 | + s/-->/\n\L&\n/ig |
| 143 | + ' "$NEWNAME" |
| 144 | + |
| 145 | +echo -e -n "2.." |
| 146 | + |
| 147 | + sed -i -e ' |
| 148 | + /^<script/,/^<\/script>/d |
| 149 | + /^<noscript/,/^<\/noscript>/d |
| 150 | + /^<iframe/,/^<\/iframe>/d |
| 151 | + /^<object/,/^<\/object>/d |
| 152 | + /^<ins/,/^<\/ins>/d |
| 153 | + ' "$NEWNAME" |
| 154 | + |
| 155 | +echo -e -n "3.." |
| 156 | + |
| 157 | + if [ -z "$tcomment" ] |
| 158 | + then |
| 159 | + sed -i -e ' |
| 160 | + /^<\!--/,/^-->/d |
| 161 | + ' "$NEWNAME" |
| 162 | + fi |
| 163 | + |
| 164 | +echo -e -n "4.." |
| 165 | + |
| 166 | + sed -i -e ' |
| 167 | + s/content="text\/html; charset=.*">/content="text\/html; charset=utf-8">/ig |
| 168 | + s/[ \t]*$// |
| 169 | + /^$/d |
| 170 | + ' "$NEWNAME" |
| 171 | + |
| 172 | +flgpre=`grep -i "<pre" "$NEWNAME"` |
| 173 | +if [ "+$flgpre" = "+" ] |
| 174 | +then |
| 175 | + echo -e -n "(5).." |
| 176 | + |
| 177 | + sed -i -e ' |
| 178 | + s/^[ \t]*// |
| 179 | + /^$/d |
| 180 | + ' "$NEWNAME" |
| 181 | +fi |
| 182 | + |
| 183 | +echo -e "END" |
| 184 | +if [ ! -z "$trewrite" ] |
| 185 | +then |
| 186 | + mv -fv "$NEWNAME" "$SRCNAME" |
| 187 | +fi |
| 188 | + exit 0 |
| 189 | +else |
| 190 | + file "$1" |
| 191 | + exit 1 |
| 192 | +fi |
| 193 | +} |
| 194 | + |
| 195 | +testcomponent |
| 196 | +main "$@" |
0 commit comments