misc/llppac

   1 #!/bin/sh
   2 set -e
   3
   4 font=$HOME/x/fnt/LastResort.ttf
   5 ! test -r $font || export LLPP_FALLBACK_FONT=$font
   6
   7 die() {
   8     echo "$1" >&2
   9     exit 1
  10 }
  11
  12 cachedir="${XDG_CACHE_HOME:-$HOME/.cache}/llpp"
  13 test -d "$cachedir" || die "cache directory '$cachedir' does not exist"
  14
  15 caspsuf=
  16 type=
  17
  18 executable_p() { command -v "$1" >/dev/null 2>&1; }
  19
  20 missing() {
  21     executable_p $1 || \
  22         eval "$1() { die \"$2 is needed for \$type conversion\"; }"
  23 }
  24
  25 text() {
  26     cat <<EOF
  27 <pre style="font-size: ${1}pt;">
  28 $1 <hr>
  29 EOF
  30     test -n "$2" && cat $2 || cat <<EOF
  31 S9 З3Ӡʒ I|l1 O0 0O o0 0o
  32 ~
  33 ЁЖЗ
  34
  35 abcdefghijklmnopqrstuvwxyz
  36 ABCDEFGHIJKLMNOPQRSTUVWXYZ
  37 абвгдеёжзийклмнопрстуфхцчшщэюя
  38 АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЭЮЯ
  39 инпипнишпнипшнипнпншпнпинпипнпипщнпипипин
  40 иннициатива шиорокошарикоподшипниковый взбзднуть гидроаэроионизация
  41 телегаммааппарат фельдъегерь четырёхсотпятидесятисемимиллиметровое
  42 превысокомногорассмотрительствующий вскользь забулдыжничество
  43 цецецница
  44 https://ru.wikipedia.org/wiki/Панграмма https://en.wikipedia.org/wiki/Pangram
  45 Экс-граф? Плюш изъят. Бьём чуждый цен хвощ!
  46 Waltz, bad nymph, for quick jigs vex
  47 ',.ფyfგcრლაოეუიდჰთნსყჯქxბმwვზ
  48 Σομε πσθδο γρεεκ
  49 Il1| Il|1 I1l| I1|l I|1l I|l1 lI1| lI|1 l1I| l1|I l|1I l|I1
  50 1lI| 1l|I 1Il| 1I|l 1|Il 1|lI |l1I |lI1 |1lI |1Il |I1l |Il1
  51 \`1234567890-=\\
  52 ~!@#\$$%^&*()_+|
  53 ₹₽£$€
  54 [];',.//|/\ &gt;- -&gt; ~~&lt; &lt;-- --&gt; == === ==&gt; &lt;==
  55 {}:"&lt;&gt;?"
  56 S in S9 59 S9 is like 5 also 8B B8 BB 88B8 88B8
  57 final fjord flair ft fff
  58 3ЗЭэзээз3ЗqgO0o ijklI1| bgqpykvwlliliiijil1 lLLl
  59 01234567890 3ДС ЗДС ӠДС ʒДС
  60 rnrmnnmrrrnmmr <i>Nürnberg</i> (rn -> m)
  61 Apay Арау
  62 örök für A̋ a̋ ő ű https://en.wikipedia.org/wiki/Diacritic
  63 молодо́й
  64 - - (~-) ~ ~ (-~) - ~ ~ - - ~ ~ ''"''""., 1.65Mb
  65 curly vs plain braces {} () {} ({ {( }) )} →
  66 Bjarne Stroustrup (/ˈbjɑːrnə ˈstraʊstrʊp/; Danish: [ˈbjɑːnə ˈsdʁʌwˀsdʁɔb]
  67 ○●🅼
  68 😁 😂 😃 😅 😆 😇 😈 😉 😊 😋 😌 😍 😎 😏 😐 😑
  69 😒 😓 😕 😗 😘 😙 😚 😛 😜 😞 😠 😡 😢 😥 😧 😨
  70 😩 😪 😫 😬 😭 😮 😯 😰 😱 😲 😳 😵 😶 😸 😹 😻
  71 😼 😽 😾 😿 🙀 🙁 🙂 🙃 🙄 🙆 🙇 🙉 🙊 🙌 🙍 🙎
  72 🙏 1F600 😀 1F621 😤 😦 😖 😄 😔 🙅 🙈 🙋 1F620
  73 1F627 😟 😴 😷 😺 1F609 1F60C 😝 😣 ✉ 📧
  74
  75 kerning
  76     echo \${@:5}
  77     life           [l i fe]
  78     lambdabot      [l a mb]
  79     Commutative    [Co mm utative]
  80     weird          [we i rd]
  81     Before         [Bef ore, Be for e]
  82     higher         [gh, h i gher, hi gher]
  83     multi          [mu lti]
  84     till           [ti ll]
  85     language       [an]
  86     plan           [la]
  87     when           [wh]
  88     kerning        [rn | ng]
  89     мне            [мн е]
  90     oiled          [oi led]
  91     mkfifo         [mk f i fo, mk f i fo]
  92     didn't         [di dn't]
  93     argumnet (sic) []
  94     more           [mo re]
  95     CmmCall        [CmmC a l l]
  96     -Wvla          [-Wv la]
  97     align          [a l i gn]
  98     Compiling      [Co mp iling]
  99     print          [pr int]
 100     fcmono         [f cmono]
 101     Mail           [M ail]
 102     xsrc           [x src]
 103     Pokorna
 104     buildglyphs.js [buil dg lyphs]
 105     Dollar         [Do ll ar]
 106     signed         [s i gned]
 107     Pine           [Pi ne]
 108     right          [r i ght]
 109     ugliness
 110     Nigeria        [Ni g eria]
 111     Nothing        [Nothi ng]
 112     modify         [mo dify]
 113     mplus          [mp lus]
 114     Apostolic      [A postolic]
 115     коде           [к о де]
 116     Прокофьевну    [Про ко фьевну]
 117     Хэлп           [X элп]
 118     коснулся       [ко снулся]
 119     обьявляным     [об ья вленым]
 120     Queen          [Qu een]
 121     etymological   [et ymological]
 122     sigh           [s i gh]
 123     git            [gi t]
 124     некочевого     [не ко чевого]
 125     Розовое        [Ро зо вое]
 126     Trump          [Tru mp]
 127     покрытие       [по кр ытие]
 128     Singapore      [Si ng apore]
 129     Georgy         [Ge orgy]
 130     with           [wi th]
 131     Yokozuna       [Y okozuna]
 132     подтверждение  [под тверждение]
 133     Xwayland       [X wayland]
 134     callable       [ca l l ab le]
 135     Noah           [N oa h]
 136     Los            [L os]
 137     refused        [ref used]
 138     Julia          [J ulia | Jul i a]
 139     Joe Armstrong  [J oe (rip :()]
 140     James          [J ames]
 141     CAPTCHA        [CA PTCHA]
 142     KIZER          [KI ZER]
 143     Kavanugh       [K avaunaugh]
 144
 145 almost kerning
 146     hope doesn't
 147     Illegal
 148     automatically effective
 149     Ilium is the Latin for Ilion.
 150     jtanx
 151
 152 (from UTF8-demo.txt)
 153 Box drawing alignment tests:                                          █
 154                                                                       ▉
 155   ╔══╦══╗  ┌──┬──┐  ╭──┬──╮  ╭──┬──╮  ┏━━┳━━┓  ┎┒┏┑   ╷  ╻ ┏┯┓ ┌┰┐    ▊ ╱╲╱╲╳╳╳
 156   ║┌─╨─┐║  │╔═╧═╗│  │╒═╪═╕│  │╓─╁─╖│  ┃┌─╂─┐┃  ┗╃╄┙  ╶┼╴╺╋╸┠┼┨ ┝╋┥    ▋ ╲╱╲╱╳╳╳
 157   ║│╲ ╱│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╿ │┃  ┍╅╆┓   ╵  ╹ ┗┷┛ └┸┘    ▌ ╱╲╱╲╳╳╳
 158   ╠╡ ╳ ╞╣  ├╢   ╟┤  ├┼─┼─┼┤  ├╫─╂─╫┤  ┣┿╾┼╼┿┫  ┕┛┖┚     ┌┄┄┐ ╎ ┏┅┅┓ ┋ ▍ ╲╱╲╱╳╳╳
 159   ║│╱ ╲│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╽ │┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▎
 160   ║└─╥─┘║  │╚═╤═╝│  │╘═╪═╛│  │╙─╀─╜│  ┃└─╂─┘┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▏
 161   ╚══╩══╝  └──┴──┘  ╰──┴──╯  ╰──┴──╯  ┗━━┻━━┛  ▗▄▖▛▀▜   └╌╌┘ ╎ ┗╍╍┛ ┋  ▁▂▃▄▅▆▇█
 162                                                ▝▀▘▙▄▟
 163
 164 https://mirrors.zju.edu.cn/CTAN/fonts/gofonts/doc/gofonts.pdf (page 2)
 165 """
 166 DIN Legibility Standard
 167
 168 The recent German DIN 1450 legibility standard
 169 recommends several features for font legibility, including
 170 differentiation of letter shapes to reduce confusion. The Go fonts
 171 conform to the 1450 standard by carefully differentiating zero from
 172 capital O; numeral 1 from capital I (eye) and lowercase l (ell);
 173 numeral 5 from capital S; and numeral 8 from capital B. The shapes of
 174 bowls of b d p q follow the natural asymmetries of legible Renaissance
 175 handwriting, aiding differentiation to reduce confusion. [5]
 176 """
 177
 178 (https://en.wikipedia.org/wiki/Jabberwocky
 179  Idea from the aforementioned gofonts paper)
 180
 181     Бе сгладне и честлинните комбурси  Það leið að stekju, og slýgir greðlar
 182     търляха се и сврецваха във плите;  sig snældu og böluðu um slöffruna,
 183     съвсем окласни бяха тук щурпите    og angurvært sungu sópfiðrungar
 184     и отма равапсатваха прасурси.      við sífgelt týðmana svíræna.
 185
 186 (https://en.wikipedia.org/wiki/Typeface)
 187
 188 For the rational mind, type design can be a maddening game
 189 of drawing things differently in order to make them appear the
 190 same.
 191 </pre>
 192 EOF
 193 }
 194
 195 maketext() {
 196     # https://github.com/react-boilerplate/react-boilerplate/issues/1340
 197     test -d "$cachedir/fonts" || mkdir "$cachedir/fonts"
 198     filename="$(basename "$1")"
 199     textfilename="$2"
 200     cat >"$cachedir/fonts/text.html" <<EOF
 201 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
 202  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 203 <html>
 204   <head>
 205     <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 206     <title>$filename</title>
 207     <style type="text/css">
 208       @font-face { font-family: moo; src: url('$1'); }
 209       pre { font-family: moo; }
 210     </style>
 211   </head>
 212   $filename
 213   <body>
 214 EOF
 215     for size in 6 8 10 12 4 2; do
 216         text $size $textfilename >>"$cachedir/fonts/text.html"
 217     done
 218     cat >>"$cachedir/fonts/text.html" <<EOF
 219     <br/>
 220     $(fc-scan -f '%{fullname} %{style} %{slant} %{file} %{fontversion} %{capability} %{lang}' "$1" || file "$1")
 221   </body>
 222 </html>
 223 EOF
 224 }
 225
 226 trap 'test -n "$casp" && rm -f "$casp"' 0
 227
 228 while getopts s:c:m:t:fu opt; do
 229     case $opt in
 230         m) mime=$OPTARG;;
 231         t) type=$OPTARG;;
 232         f) force=true;;
 233         c) css="-s $OPTARG";;
 234         s) textfilename="$OPTARG";;
 235         ?) die "usage: $0 [-c css] [-m mime/type] [-t filter] [-f] [-s textfile] [path|url]";;
 236     esac
 237 done
 238 shift $(($OPTIND - 1))
 239 test -z "$1" && die "usage $0: path"
 240
 241 origin="$1"
 242 if ${force-test ! -e "$1"} && expr >/dev/null "$1" : "\(ftp\|https\?\)://"; then
 243     if executable_p wget; then
 244         dl() { wget -q $1 -O $2; }
 245     elif executable_p curl; then
 246         dl() { curl -L $1 -o $2; }
 247     else
 248         die "no program to fetch remote urls found"
 249     fi
 250
 251     hashof="$cachedir/$(basename "$1")"
 252     dl "$1" "$hashof" || test -e "$md5of"
 253     shift
 254     set -- "$hashof" "$@"
 255 else
 256     hashof="$1"
 257 fi
 258
 259 test -z "$type" && { ft=$(file -L --mime-type -b "$1") || die "$ft"; }
 260
 261 case $ft in
 262     application/x-gzip | application/x-compress) dc=zcat;;
 263     application/x-xz) dc=xzcat;;
 264     application/x-bzip2) dc=bzcat;;
 265     *) unset dc || true;;
 266 esac
 267
 268 filt='"${dc-cat}" "$1" |'
 269
 270 typeofmime() {
 271     case "$1" in
 272         application/postscript) type=ps;;
 273         application/pdf) type=pdf;;
 274         image/vnd.djvu) type=djvu;;
 275         text/html) type=html;;
 276         text/plain) type=text;;
 277         application/msword) type=word;;
 278         application/vnd.openxmlformats-officedocument.*  \
 279             | application/vnd.ms-powerpoint              \
 280             | application/vnd.ms-excel                   \
 281             | application/vnd.oasis.opendocument.*) type=uno;;
 282         image/svg+xml) type=svg;;
 283         image/png | image/jpeg) test -n "$dc" && type="image" || type="image2";;
 284         image/*) type=image;;
 285         application/x-dvi) type=dvi;;
 286         application/x-font-ttf                  \
 287             | application/vnd.ms-opentype       \
 288             | application/font-sfnt) type=font;;
 289         *) return 1;;
 290     esac
 291     return 0
 292 }
 293
 294 if test -z "$type"; then
 295     test -z "$mime" &&                                          \
 296         mime=$(file -L --mime-type -b "$1" || die "$mime") ||   \
 297             $(file -L --mime-type -bz "$1" || die "$mime")
 298     typeofmime "$mime" || die "unhandled file type: '$mime'"
 299 fi
 300
 301 caspsuf=".pdf"
 302 case $type in
 303     ps) conv='ps2pdf - "$casp"';;
 304     image2|pdf) test -z "$dc" && exec llpp "$@" || conv='cat >"$casp"';;
 305     texi) {
 306         missing texi2html "texi2html(http://www.nongnu.org/texi2html/)"
 307         missing prince "PrinceXML(http://www.princexml.com/)"
 308         conv='texi2html - -o - | prince $css - -o "$casp"'
 309     };;
 310     djvu) {
 311         missing ddjvu "ddjvu(http://djvu.sourceforge.net/doc/man/ddjvu.html)"
 312         conv='ddjvu -format=pdf - "$casp"'
 313     };;
 314     html) {
 315         missing prince "Prince(http://www.princexml.com/)"
 316         conv='prince $css - -o "$casp"'
 317     };;
 318     html2epub) {
 319         missing pandoc "pandoc(http://pandoc.org)"
 320         caspsuf=".epub"
 321         conv='pandoc -r html - -w epub2 -o "$casp"'
 322     };;
 323     word) {
 324         if executable_p unoconv && test -z "$dc"; then
 325             unset filt
 326             conv='unoconv -o "$casp" "$1"'
 327         else
 328             missing antiword "antiword or unoconv"
 329             conv='antiword -m 8859-1.txt -a a4 - >"$casp"'
 330         fi
 331     };;
 332     uno) {
 333         test -n "$dc" && die "cannot convert compressed '$mime'"
 334         unset filt
 335         missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 336         conv='unoconv -o "$casp" "$1"'
 337     };;
 338     svg) {
 339         if executable_p inkscape && test -z "$dc"; then
 340             unset filt
 341             conv='inkscape -z -A "$casp" "$1"'
 342         else
 343             if executable_p rsvg-convert; then
 344                 conv='rsvg-convert -f pdf -o "$casp"'
 345             else
 346                 test -n "$dc" && die "cannot convert compressed '$mime'"
 347                 unset filt
 348                 missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 349                 conv='unoconv -o "$casp" "$1"'
 350             fi
 351         fi
 352     };;
 353     font) {
 354         maketext "$1" "$textfilename"
 355         exec llpp -origin $1 -layout-height 0 "$cachedir/fonts/text.html"
 356     };;
 357     image) {
 358         missing convert "convert(http://www.imagemagick.org/script/convert.php)"
 359         conv='convert - pdf:"$casp"'
 360     };;
 361     dvi) {
 362         test -n "$dc" && die "cannot convert compressed '$mime'"
 363         unset filt
 364         missing dvipdf "dvipdf(http://ghostscript.com/)"
 365         conv='dvipdf "$1" "$casp"'
 366     };;
 367     text) {
 368         missing pandoc "pandoc(http://pandoc.org/)"
 369         conv='pandoc -t epub2 - -o "$casp"'
 370         caspsuf=.epub
 371     };;
 372     *) die "unhandled filter type: '$type'";;
 373 esac
 374
 375 hash=$(cksum "$hashof") || die "$hash"
 376 casp=$cachedir/${hash%% *}$caspsuf
 377
 378 { test -n "$force" || test ! -e "$casp"; } && eval "$filt" "$conv"
 379 shift
 380
 381 exec llpp -origin $origin "$@" "$casp"