misc/llppac

   1 #!/bin/sh
   2 set -e
   3
   4 die() {
   5     echo "$1" >&2
   6     exit 1
   7 }
   8
   9 cachedir="${XDG_CACHE_HOME:-$HOME/.cache}/llpp"
  10 test -d "$cachedir" || die "cache directory '$cachedir' does not exist"
  11
  12 caspsuf=
  13 type=
  14
  15 executable_p() { command -v "$1" >/dev/null 2>&1; }
  16
  17 missing() {
  18     executable_p $1 || \
  19         eval "$1() { die \"$2 is needed for \$type conversion\"; }"
  20 }
  21
  22 text() {
  23     cat <<EOF
  24 <pre style="font-size: ${1}pt;">
  25 === $1 ===
  26 EOF
  27     test -n "$2" && cat $2 || cat <<EOF
  28 S9 З3Ӡʒ Il1
  29 ~
  30 ЁЖЗ
  31
  32 abcdefghijklmnopqrstwxyz
  33 ABCDEFGHIJKLMNOPQRSTWXYZ
  34 абвгдеёжзийклмнопрстуфхцчшщэюя
  35 АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЭЮЯ
  36 инпипнишпнипшнипнпншпнпинпипнпипщнпипипин
  37 иннициатива шиорокошарикоподшипниковый взбзднуть гидроаэроионизация
  38 телегаммааппарат фельдъегерь четырёхсотпятидесятисемимиллиметровое
  39 превысокомногорассмотрительствующий вскользь забулдыжничество
  40 цецецница
  41 Σομε πσθδο γρεεκ
  42 \`1234567890-=\\
  43 ~!@#\$$%^&*()_+|
  44 [];',./
  45 {}:"&lt;&gt;?"
  46 S in S9 59 S9 is like 5
  47 final fjord flair ft
  48 3ЗЭэзээз3ЗqgO0o ijklI1| bgqpykvwlliliiijil1 lLLl
  49 01234567890 3ДС ЗДС ӠДС ʒДС
  50 rnrmnnmrrrnmmr <i>Nürnberg</i> (rn -> m)
  51 Apay Арау
  52 örök für
  53 - - (~-) ~ ~ (-~)
  54 curly vs plain braces {} () {} →
  55 Bjarne Stroustrup (/ˈbjɑːrnə ˈstraʊstrʊp/; Danish: [ˈbjɑːnə ˈsdʁʌwˀsdʁɔb]
  56
  57 kerning
  58     echo \${@:5}
  59     life           [l i fe]
  60     lambdabot      [l a mb]
  61     Commutative    [Co mm utative]
  62     weird          [we i rd]
  63     Before         [Bef ore, Be for e]
  64     higher         [gh, h i gher, hi gher]
  65     multi          [mu lti]
  66     till           [ti ll]
  67     language       [an]
  68     plan           [la]
  69     when           [wh]
  70     kerning        [rn | ng]
  71     мне            [мн е]
  72     oiled          [oi led]
  73     mkfifo         [mk f i fo, mk f i fo]
  74     didn't         [di dn't]
  75     argumnet (sic) []
  76     more           [mo re]
  77     CmmCall        [CmmC a l l]
  78     -Wvla          [-Wv la]
  79     align          [a l i gn]
  80     Compiling      [Co mp iling]
  81     print          [pr int]
  82     fcmono         [f cmono]
  83     Mail           [M ail]
  84     xsrc           [x src]
  85     Pokorna
  86     buildglyphs.js [buil dg lyphs]
  87     Dollar         [Do ll ar]
  88     signed         [s i gned]
  89     Pine           [Pi ne]
  90     right          [r i ght]
  91     ugliness
  92     Nigeria        [Ni g eria]
  93     Nothing        [Nothi ng]
  94     modify         [mo dify]
  95     mplus          [mp lus]
  96     Apostolic      [A postolic]
  97     коде           [к о де]
  98     Прокофьевну    [Пра ко фьевну]
  99     Хэлп           [X элп]
 100     коснулся       [ко снулся]
 101     обьявляным     [об ья вленым]
 102     Queen          [Qu een]
 103     etymological   [et ymologoical]
 104     sigh           [s i gh]
 105     git            [gi t]
 106     некочевого     [не ко чевого]
 107     Розовое        [Ро зо вое]
 108     Trump          [Tru mp]
 109     покрытие       [по кр ытие]
 110     Julia          [J ulia | Jul i a]
 111
 112 almost kerning
 113     hope doesn't
 114     Illegal
 115     automatically effective
 116     Ilium is the Latin for Ilion.
 117
 118 (from UTF8-demo.txt)
 119 Box drawing alignment tests:                                          █
 120                                                                       ▉
 121   ╔══╦══╗  ┌──┬──┐  ╭──┬──╮  ╭──┬──╮  ┏━━┳━━┓  ┎┒┏┑   ╷  ╻ ┏┯┓ ┌┰┐    ▊ ╱╲╱╲╳╳╳
 122   ║┌─╨─┐║  │╔═╧═╗│  │╒═╪═╕│  │╓─╁─╖│  ┃┌─╂─┐┃  ┗╃╄┙  ╶┼╴╺╋╸┠┼┨ ┝╋┥    ▋ ╲╱╲╱╳╳╳
 123   ║│╲ ╱│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╿ │┃  ┍╅╆┓   ╵  ╹ ┗┷┛ └┸┘    ▌ ╱╲╱╲╳╳╳
 124   ╠╡ ╳ ╞╣  ├╢   ╟┤  ├┼─┼─┼┤  ├╫─╂─╫┤  ┣┿╾┼╼┿┫  ┕┛┖┚     ┌┄┄┐ ╎ ┏┅┅┓ ┋ ▍ ╲╱╲╱╳╳╳
 125   ║│╱ ╲│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╽ │┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▎
 126   ║└─╥─┘║  │╚═╤═╝│  │╘═╪═╛│  │╙─╀─╜│  ┃└─╂─┘┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▏
 127   ╚══╩══╝  └──┴──┘  ╰──┴──╯  ╰──┴──╯  ┗━━┻━━┛  ▗▄▖▛▀▜   └╌╌┘ ╎ ┗╍╍┛ ┋  ▁▂▃▄▅▆▇█
 128                                                ▝▀▘▙▄▟
 129
 130 https://mirrors.zju.edu.cn/CTAN/fonts/gofonts/doc/gofonts.pdf (page 2)
 131 """
 132 DIN Legibility Standard
 133
 134 The recent German DIN 1450 legibility standard
 135 recommends several features for font legibility, including
 136 differentiation of letter shapes to reduce confusion. The Go fonts
 137 conform to the 1450 standard by carefully differentiating zero from
 138 capital O; numeral 1 from capital I (eye) and lowercase l (ell);
 139 numeral 5 from capital S; and numeral 8 from capital B. The shapes of
 140 bowls of b d p q follow the natural asymmetries of legible Renaissance
 141 handwriting, aiding differentiation to reduce confusion. [5]
 142 """
 143
 144 Mistaking S for a 5 (at certain sizes) is more of a problem... but... Ze Germans
 145
 146 (https://en.wikipedia.org/wiki/Jabberwocky
 147  Idea from the aforementioned gofonts paper)
 148
 149     Бе сгладне и честлинните комбурси  Það leið að stekju, og slýgir greðlar
 150     търляха се и сврецваха във плите;  sig snældu og böluðu um slöffruna,
 151     съвсем окласни бяха тук щурпите    og angurvært sungu sópfiðrungar
 152     и отма равапсатваха прасурси.      við sífgelt týðmana svíræna.
 153
 154 (https://en.wikipedia.org/wiki/Typeface)
 155
 156 For the rational mind, type design can be a maddening game
 157 of drawing things differently in order to make them appear the
 158 same.
 159 </pre>
 160 EOF
 161 }
 162
 163 maketext() {
 164     # https://github.com/react-boilerplate/react-boilerplate/issues/1340
 165     test -d "$cachedir/fonts" || mkdir "$cachedir/fonts"
 166     filename="$(basename "$1")"
 167     textfilename="$2"
 168     cat >"$cachedir/fonts/text.html" <<EOF
 169 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
 170  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 171 <html>
 172   <head>
 173     <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 174     <title>$filename</title>
 175     <style type="text/css">
 176       @font-face { font-family: moo; src: url('$1'); }
 177       pre { font-family: moo; }
 178     </style>
 179   </head>
 180   $filename
 181   <body>
 182 EOF
 183     for size in 6 8 10 12 4 2; do
 184         text $size $textfilename >>"$cachedir/fonts/text.html"
 185     done
 186     cat >>"$cachedir/fonts/text.html" <<EOF
 187     <br/>
 188     $(fc-scan -f '%{fullname} %{style} %{slant} %{file} %{fontversion} %{capability} %{lang}' "$1" || file "$1")
 189   </body>
 190 </html>
 191 EOF
 192 }
 193
 194 trap 'test -n "$casp" && rm -f "$casp"' 0
 195
 196 while getopts s:c:m:t:fu opt; do
 197     case $opt in
 198         m) mime=$OPTARG;;
 199         t) type=$OPTARG;;
 200         f) force=true;;
 201         c) css="-s $OPTARG";;
 202         s) textfilename="$OPTARG";;
 203         ?) die "usage: $0 [-c css] [-m mime/type] [-t filter] [-f] [-s textfile] [path|url]";;
 204     esac
 205 done
 206 shift $(($OPTIND - 1))
 207 test -z "$1" && die "usage $0: path"
 208
 209 origin="$1"
 210 if ${force-test ! -e "$1"} && expr >/dev/null "$1" : "\(ftp\|https\?\)://"; then
 211     if executable_p wget; then
 212         dl() { wget -q $1 -O $2; }
 213     elif executable_p curl; then
 214         dl() { curl -L $1 -o $2; }
 215     else
 216         die "no program to fetch remote urls found"
 217     fi
 218
 219     hashof="$cachedir/$(basename "$1")"
 220     dl "$1" "$hashof" || test -e "$md5of"
 221     shift
 222     set -- "$hashof" "$@"
 223 else
 224     hashof="$1"
 225 fi
 226
 227 test -z "$type" && { ft=$(file -L --mime-type -b "$1") || die "$ft"; }
 228
 229 case $ft in
 230     application/x-gzip | application/x-compress) dc=zcat;;
 231     application/x-xz) dc=xzcat;;
 232     application/x-bzip2) dc=bzcat;;
 233     *) unset dc || true;;
 234 esac
 235
 236 filt='"${dc-cat}" "$1" |'
 237
 238 typeofmime() {
 239     case "$1" in
 240         application/postscript) type=ps;;
 241         application/pdf) type=pdf;;
 242         image/vnd.djvu) type=djvu;;
 243         text/html) type=html;;
 244         text/plain) type=text;;
 245         application/msword) type=word;;
 246         application/vnd.openxmlformats-officedocument.*  \
 247             | application/vnd.ms-powerpoint              \
 248             | application/vnd.ms-excel                   \
 249             | application/vnd.oasis.opendocument.*) type=uno;;
 250         image/svg+xml) type=svg;;
 251         image/png | image/jpeg) test -n "$dc" && type="image" || type="image2";;
 252         image/*) type=image;;
 253         application/x-dvi) type=dvi;;
 254         application/x-font-ttf                  \
 255             | application/vnd.ms-opentype       \
 256             | application/font-sfnt) type=font;;
 257         *) return 1;;
 258     esac
 259     return 0
 260 }
 261
 262 if test -z "$type"; then
 263     test -z "$mime" &&                                          \
 264         mime=$(file -L --mime-type -b "$1" || die "$mime") ||   \
 265             $(file -L --mime-type -bz "$1" || die "$mime")
 266     typeofmime "$mime" || die "unhandled file type: '$mime'"
 267 fi
 268
 269 caspsuf=".pdf"
 270 case $type in
 271     ps) conv='ps2pdf - "$casp"';;
 272     image2|pdf) test -z "$dc" && exec llpp "$@" || conv='cat >"$casp"';;
 273     texi) {
 274         missing texi2html "texi2html(http://www.nongnu.org/texi2html/)"
 275         missing prince "PrinceXML(http://www.princexml.com/)"
 276         conv='texi2html - -o - | prince $css - -o "$casp"'
 277     };;
 278     djvu) {
 279         missing ddjvu "ddjvu(http://djvu.sourceforge.net/doc/man/ddjvu.html)"
 280         conv='ddjvu -format=pdf - "$casp"'
 281     };;
 282     html) {
 283         missing prince "Prince(http://www.princexml.com/)"
 284         conv='prince $css - -o "$casp"'
 285     };;
 286     html2epub) {
 287         missing pandoc "pandoc(http://pandoc.org)"
 288         caspsuf=".epub"
 289         conv='pandoc -r html - -w epub2 -o "$casp"'
 290     };;
 291     word) {
 292         if executable_p unoconv && test -z "$dc"; then
 293             unset filt
 294             conv='unoconv -o "$casp" "$1"'
 295         else
 296             missing antiword "antiword or unoconv"
 297             conv='antiword -m 8859-1.txt -a a4 - >"$casp"'
 298         fi
 299     };;
 300     uno) {
 301         test -n "$dc" && die "cannot convert compressed '$mime'"
 302         unset filt
 303         missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 304         conv='unoconv -o "$casp" "$1"'
 305     };;
 306     svg) {
 307         if executable_p inkscape && test -z "$dc"; then
 308             unset filt
 309             conv='inkscape -z -A "$casp" "$1"'
 310         else
 311             missing rsvg-convert "rsvg-convert"
 312             conv='rsvg-convert -f pdf -o "$casp"'
 313         fi
 314     };;
 315     font) {
 316         maketext "$1" "$textfilename"
 317         exec llpp -origin $1 -layout-height 0 "$cachedir/fonts/text.html"
 318     };;
 319     image) {
 320         missing convert "convert(http://www.imagemagick.org/script/convert.php)"
 321         conv='convert - pdf:"$casp"'
 322     };;
 323     dvi) {
 324         test -n "$dc" && die "cannot convert compressed '$mime'"
 325         unset filt
 326         missing dvipdf "dvipdf(http://ghostscript.com/)"
 327         conv='dvipdf "$1" "$casp"'
 328     };;
 329     text) {
 330         missing pandoc "pandoc(http://pandoc.org/)"
 331         conv='pandoc -t epub2 - -o "$casp"'
 332         caspsuf=.epub
 333     };;
 334     *) die "unhandled filter type: '$type'";;
 335 esac
 336
 337 hash=$(cksum "$hashof") || die "$hash"
 338 casp=$cachedir/${hash%% *}$caspsuf
 339
 340 { test -n "$force" || test ! -e "$casp"; } && eval "$filt" "$conv"
 341 shift
 342
 343 exec llpp -origin $origin "$@" "$casp"