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 A̋ a̋ ő ű https://en.wikipedia.org/wiki/Diacritic
  53 молодо́й
  54 - - (~-) ~ ~ (-~)
  55 curly vs plain braces {} () {} →
  56 Bjarne Stroustrup (/ˈbjɑːrnə ˈstraʊstrʊp/; Danish: [ˈbjɑːnə ˈsdʁʌwˀsdʁɔb]
  57
  58 kerning
  59     echo \${@:5}
  60     life           [l i fe]
  61     lambdabot      [l a mb]
  62     Commutative    [Co mm utative]
  63     weird          [we i rd]
  64     Before         [Bef ore, Be for e]
  65     higher         [gh, h i gher, hi gher]
  66     multi          [mu lti]
  67     till           [ti ll]
  68     language       [an]
  69     plan           [la]
  70     when           [wh]
  71     kerning        [rn | ng]
  72     мне            [мн е]
  73     oiled          [oi led]
  74     mkfifo         [mk f i fo, mk f i fo]
  75     didn't         [di dn't]
  76     argumnet (sic) []
  77     more           [mo re]
  78     CmmCall        [CmmC a l l]
  79     -Wvla          [-Wv la]
  80     align          [a l i gn]
  81     Compiling      [Co mp iling]
  82     print          [pr int]
  83     fcmono         [f cmono]
  84     Mail           [M ail]
  85     xsrc           [x src]
  86     Pokorna
  87     buildglyphs.js [buil dg lyphs]
  88     Dollar         [Do ll ar]
  89     signed         [s i gned]
  90     Pine           [Pi ne]
  91     right          [r i ght]
  92     ugliness
  93     Nigeria        [Ni g eria]
  94     Nothing        [Nothi ng]
  95     modify         [mo dify]
  96     mplus          [mp lus]
  97     Apostolic      [A postolic]
  98     коде           [к о де]
  99     Прокофьевну    [Про ко фьевну]
 100     Хэлп           [X элп]
 101     коснулся       [ко снулся]
 102     обьявляным     [об ья вленым]
 103     Queen          [Qu een]
 104     etymological   [et ymologoical]
 105     sigh           [s i gh]
 106     git            [gi t]
 107     некочевого     [не ко чевого]
 108     Розовое        [Ро зо вое]
 109     Trump          [Tru mp]
 110     покрытие       [по кр ытие]
 111     Julia          [J ulia | Jul i a]
 112     Singapore      [Si ng apore]
 113     Georgy         [Ge orgy]
 114     with           [wi th]
 115     Joe Armstrong  [J oe (rip :()]
 116     Yokozuna       [Y okozuna]
 117     подтверждение  [под тверждение]
 118
 119 almost kerning
 120     hope doesn't
 121     Illegal
 122     automatically effective
 123     Ilium is the Latin for Ilion.
 124
 125 (from UTF8-demo.txt)
 126 Box drawing alignment tests:                                          █
 127                                                                       ▉
 128   ╔══╦══╗  ┌──┬──┐  ╭──┬──╮  ╭──┬──╮  ┏━━┳━━┓  ┎┒┏┑   ╷  ╻ ┏┯┓ ┌┰┐    ▊ ╱╲╱╲╳╳╳
 129   ║┌─╨─┐║  │╔═╧═╗│  │╒═╪═╕│  │╓─╁─╖│  ┃┌─╂─┐┃  ┗╃╄┙  ╶┼╴╺╋╸┠┼┨ ┝╋┥    ▋ ╲╱╲╱╳╳╳
 130   ║│╲ ╱│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╿ │┃  ┍╅╆┓   ╵  ╹ ┗┷┛ └┸┘    ▌ ╱╲╱╲╳╳╳
 131   ╠╡ ╳ ╞╣  ├╢   ╟┤  ├┼─┼─┼┤  ├╫─╂─╫┤  ┣┿╾┼╼┿┫  ┕┛┖┚     ┌┄┄┐ ╎ ┏┅┅┓ ┋ ▍ ╲╱╲╱╳╳╳
 132   ║│╱ ╲│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╽ │┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▎
 133   ║└─╥─┘║  │╚═╤═╝│  │╘═╪═╛│  │╙─╀─╜│  ┃└─╂─┘┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▏
 134   ╚══╩══╝  └──┴──┘  ╰──┴──╯  ╰──┴──╯  ┗━━┻━━┛  ▗▄▖▛▀▜   └╌╌┘ ╎ ┗╍╍┛ ┋  ▁▂▃▄▅▆▇█
 135                                                ▝▀▘▙▄▟
 136
 137 https://mirrors.zju.edu.cn/CTAN/fonts/gofonts/doc/gofonts.pdf (page 2)
 138 """
 139 DIN Legibility Standard
 140
 141 The recent German DIN 1450 legibility standard
 142 recommends several features for font legibility, including
 143 differentiation of letter shapes to reduce confusion. The Go fonts
 144 conform to the 1450 standard by carefully differentiating zero from
 145 capital O; numeral 1 from capital I (eye) and lowercase l (ell);
 146 numeral 5 from capital S; and numeral 8 from capital B. The shapes of
 147 bowls of b d p q follow the natural asymmetries of legible Renaissance
 148 handwriting, aiding differentiation to reduce confusion. [5]
 149 """
 150
 151 Mistaking S for a 5 (at certain sizes) is more of a problem... but... Ze Germans
 152
 153 (https://en.wikipedia.org/wiki/Jabberwocky
 154  Idea from the aforementioned gofonts paper)
 155
 156     Бе сгладне и честлинните комбурси  Það leið að stekju, og slýgir greðlar
 157     търляха се и сврецваха във плите;  sig snældu og böluðu um slöffruna,
 158     съвсем окласни бяха тук щурпите    og angurvært sungu sópfiðrungar
 159     и отма равапсатваха прасурси.      við sífgelt týðmana svíræna.
 160
 161 (https://en.wikipedia.org/wiki/Typeface)
 162
 163 For the rational mind, type design can be a maddening game
 164 of drawing things differently in order to make them appear the
 165 same.
 166 </pre>
 167 EOF
 168 }
 169
 170 maketext() {
 171     # https://github.com/react-boilerplate/react-boilerplate/issues/1340
 172     test -d "$cachedir/fonts" || mkdir "$cachedir/fonts"
 173     filename="$(basename "$1")"
 174     textfilename="$2"
 175     cat >"$cachedir/fonts/text.html" <<EOF
 176 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
 177  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 178 <html>
 179   <head>
 180     <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 181     <title>$filename</title>
 182     <style type="text/css">
 183       @font-face { font-family: moo; src: url('$1'); }
 184       pre { font-family: moo; }
 185     </style>
 186   </head>
 187   $filename
 188   <body>
 189 EOF
 190     for size in 6 8 10 12 4 2; do
 191         text $size $textfilename >>"$cachedir/fonts/text.html"
 192     done
 193     cat >>"$cachedir/fonts/text.html" <<EOF
 194     <br/>
 195     $(fc-scan -f '%{fullname} %{style} %{slant} %{file} %{fontversion} %{capability} %{lang}' "$1" || file "$1")
 196   </body>
 197 </html>
 198 EOF
 199 }
 200
 201 trap 'test -n "$casp" && rm -f "$casp"' 0
 202
 203 while getopts s:c:m:t:fu opt; do
 204     case $opt in
 205         m) mime=$OPTARG;;
 206         t) type=$OPTARG;;
 207         f) force=true;;
 208         c) css="-s $OPTARG";;
 209         s) textfilename="$OPTARG";;
 210         ?) die "usage: $0 [-c css] [-m mime/type] [-t filter] [-f] [-s textfile] [path|url]";;
 211     esac
 212 done
 213 shift $(($OPTIND - 1))
 214 test -z "$1" && die "usage $0: path"
 215
 216 origin="$1"
 217 if ${force-test ! -e "$1"} && expr >/dev/null "$1" : "\(ftp\|https\?\)://"; then
 218     if executable_p wget; then
 219         dl() { wget -q $1 -O $2; }
 220     elif executable_p curl; then
 221         dl() { curl -L $1 -o $2; }
 222     else
 223         die "no program to fetch remote urls found"
 224     fi
 225
 226     hashof="$cachedir/$(basename "$1")"
 227     dl "$1" "$hashof" || test -e "$md5of"
 228     shift
 229     set -- "$hashof" "$@"
 230 else
 231     hashof="$1"
 232 fi
 233
 234 test -z "$type" && { ft=$(file -L --mime-type -b "$1") || die "$ft"; }
 235
 236 case $ft in
 237     application/x-gzip | application/x-compress) dc=zcat;;
 238     application/x-xz) dc=xzcat;;
 239     application/x-bzip2) dc=bzcat;;
 240     *) unset dc || true;;
 241 esac
 242
 243 filt='"${dc-cat}" "$1" |'
 244
 245 typeofmime() {
 246     case "$1" in
 247         application/postscript) type=ps;;
 248         application/pdf) type=pdf;;
 249         image/vnd.djvu) type=djvu;;
 250         text/html) type=html;;
 251         text/plain) type=text;;
 252         application/msword) type=word;;
 253         application/vnd.openxmlformats-officedocument.*  \
 254             | application/vnd.ms-powerpoint              \
 255             | application/vnd.ms-excel                   \
 256             | application/vnd.oasis.opendocument.*) type=uno;;
 257         image/svg+xml) type=svg;;
 258         image/png | image/jpeg) test -n "$dc" && type="image" || type="image2";;
 259         image/*) type=image;;
 260         application/x-dvi) type=dvi;;
 261         application/x-font-ttf                  \
 262             | application/vnd.ms-opentype       \
 263             | application/font-sfnt) type=font;;
 264         *) return 1;;
 265     esac
 266     return 0
 267 }
 268
 269 if test -z "$type"; then
 270     test -z "$mime" &&                                          \
 271         mime=$(file -L --mime-type -b "$1" || die "$mime") ||   \
 272             $(file -L --mime-type -bz "$1" || die "$mime")
 273     typeofmime "$mime" || die "unhandled file type: '$mime'"
 274 fi
 275
 276 caspsuf=".pdf"
 277 case $type in
 278     ps) conv='ps2pdf - "$casp"';;
 279     image2|pdf) test -z "$dc" && exec llpp "$@" || conv='cat >"$casp"';;
 280     texi) {
 281         missing texi2html "texi2html(http://www.nongnu.org/texi2html/)"
 282         missing prince "PrinceXML(http://www.princexml.com/)"
 283         conv='texi2html - -o - | prince $css - -o "$casp"'
 284     };;
 285     djvu) {
 286         missing ddjvu "ddjvu(http://djvu.sourceforge.net/doc/man/ddjvu.html)"
 287         conv='ddjvu -format=pdf - "$casp"'
 288     };;
 289     html) {
 290         missing prince "Prince(http://www.princexml.com/)"
 291         conv='prince $css - -o "$casp"'
 292     };;
 293     html2epub) {
 294         missing pandoc "pandoc(http://pandoc.org)"
 295         caspsuf=".epub"
 296         conv='pandoc -r html - -w epub2 -o "$casp"'
 297     };;
 298     word) {
 299         if executable_p unoconv && test -z "$dc"; then
 300             unset filt
 301             conv='unoconv -o "$casp" "$1"'
 302         else
 303             missing antiword "antiword or unoconv"
 304             conv='antiword -m 8859-1.txt -a a4 - >"$casp"'
 305         fi
 306     };;
 307     uno) {
 308         test -n "$dc" && die "cannot convert compressed '$mime'"
 309         unset filt
 310         missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 311         conv='unoconv -o "$casp" "$1"'
 312     };;
 313     svg) {
 314         if executable_p inkscape && test -z "$dc"; then
 315             unset filt
 316             conv='inkscape -z -A "$casp" "$1"'
 317         else
 318             if executable_p rsvg-convert; then
 319                 conv='rsvg-convert -f pdf -o "$casp"'
 320             else
 321                 test -n "$dc" && die "cannot convert compressed '$mime'"
 322                 unset filt
 323                 missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 324                 conv='unoconv -o "$casp" "$1"'
 325             fi
 326         fi
 327     };;
 328     font) {
 329         maketext "$1" "$textfilename"
 330         exec llpp -origin $1 -layout-height 0 "$cachedir/fonts/text.html"
 331     };;
 332     image) {
 333         missing convert "convert(http://www.imagemagick.org/script/convert.php)"
 334         conv='convert - pdf:"$casp"'
 335     };;
 336     dvi) {
 337         test -n "$dc" && die "cannot convert compressed '$mime'"
 338         unset filt
 339         missing dvipdf "dvipdf(http://ghostscript.com/)"
 340         conv='dvipdf "$1" "$casp"'
 341     };;
 342     text) {
 343         missing pandoc "pandoc(http://pandoc.org/)"
 344         conv='pandoc -t epub2 - -o "$casp"'
 345         caspsuf=.epub
 346     };;
 347     *) die "unhandled filter type: '$type'";;
 348 esac
 349
 350 hash=$(cksum "$hashof") || die "$hash"
 351 casp=$cachedir/${hash%% *}$caspsuf
 352
 353 { test -n "$force" || test ! -e "$casp"; } && eval "$filt" "$conv"
 354 shift
 355
 356 exec llpp -origin $origin "$@" "$casp"